Physics of organic electronic devices

SOLID STATE PHYSICS, VOL. 55

Physics of Organic Electronic Devices I. H. CAMPBELL AND D. L. SMITH Los Alamos National Laboratory, Los Alamos, New Mexico

I.

II.

III.

IV.

V.

VI.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Organic Electronic Materials . . . . . . . . . . . . . . . . . . . . . . 2. Organic Electronic Devices . . . . . . . . . . . . . . . . . . . . . . 3. Scope of This Article . . . . . . . . . . . . . . . . . . . . . . . . . . Electronic Properties of re-Conjugated Organic Materials . . . . . . . . . . 4. E l e m e n t a r y Picture of the Electronic Structure . . . . . . . . . . . . . 5. E l e c t r o n - I o n and E l e c t r o n - E l e c t r o n Interactions . . . . . . . . . . . . 6. Solid State Properties . . . . . . . . . . . . . . . . . . . . . . . . . M e t a l / O r g a n i c Interface Electronic Structure . . . . . . . . . . . . . . . . 7. Built-in Potentials in M e t a l / O r g a n i c / M e t a l Structures ......... 8. Built-in P o t e n t i a l and Schottky Energy Barrier M e a s u r e m e n t . . . . . . Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9. Built-in P o t e n t i a l and Schottky Energy Barrier M e a s u r e m e n t . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10. Solid State and M o l e c u l a r Properties . . . . . . . . . . . . . . . . . . 11. M a n i p u l a t i n g Schottky Energy Barriers Using Dipole Layers . . . . . . Electrical T r a n s p o r t Properties . . . . . . . . . . . . . . . . . . . . . . . 12. Time-of-Flight Mobility Measurements . . . . . . . . . . . . . . . . . 13. M o b i l i t y from Single-Carrier S C L D i o d e I-V Characteristics . . . . . . 14. M o b i l i t y Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . Organic Diodes and Field-Effect Transistors . . . . . . . . . . . . . . . . . 15. Organic Diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16. Field-Effect Transistors . . . . . . . . . . . . . . . . . . . . . . . . S u m m a r y and F u t u r e Directions . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 4 8 11 12 13 16 21 24 26 29 32 45 49 54 57 63 66 77 79 108 113 117

I. Introduction

This article discusses the basic device physics that governs the operation of organic electronic devices. Organic electronic devices are a new class of solid state electrical devices that have been the subject of intense research in the last decade. The two most widely studied devices are light-emitting diodes (LEDs) and field-effect transistors (FETs). These organic devices are attractISBN 0-12-607755-X ISSN 0081-1947/01 $35.00

Copyright 9 2001 by Academic Press All rights of reproduction in any form reserved.

2

I.H. CAMPBELL AND D. L. SMITH

ing considerable interest because they have processing and performance advantages for low-cost and/or large-area applications. An organic lightemitting diode consists of a thin-film of a luminescent organic material contacted by metal electrodes on the top and bottom of the film. One electrode serves as an electron injecting contact and the other as a hole injecting contact. When a sufficient voltage bias is applied to the metal contacts, electrons and holes are injected into the organic material. The injected electrons and holes recombine in the organic material, emitting light. Organic field-effect transistors are lateral devices consisting of a conducting gate contact, a gate insulator, source and drain electrodes electrically isolated from the gate contact by the gate insulator, and an organic film in contact with the source and drain electrodes. When a bias is applied between the gate and source-drain electrodes, charge is injected into the organic film from the source and drain contacts. With an additional bias applied between the source and drain contacts a current flows laterally between these two contacts. The source to drain current is modulated by the gate voltage producing the field-effect transistor action. Organic electronic devices use undoped, insulating organic materials as the light-emitting and charge-transporting layers. The charge carriers in the devices are injected from the contacts. Electronic devices based on doped organic materials have not been developed in a manner analogous to doped inorganic semiconductor devices. To date, it has proven difficult to obtain robust n- and p-type doping of organic materials suitable for electronic device development. Doped, conducting organic materials, such as aciddoped polymers, have electrical properties similar to low conductivity metals. They have been used in organic electronic devices as a metallic contact. This article considers organic electronic devices that use thin films of undoped, conjugated organic materials for the active layer. These devices are the focus of both current scientific research and commercial development. References [1-4] present recent reviews of the electronic properties and Refs. [5-9] 1 A. J. Heeger, S. Kivelson, J. R. Schrieffer, and W. P. Su, Reviews of Modern Physics 60, 781 (1988). 2 N. C. Greenham and R. H. Friend, in Solid State Physics (H. Ehrenreich and F. Spaepen, eds.), Vol. 49, Academic Press, New York (1995). 3 C. E. Swenbergand M. Pope, Electronic Processes of Organic Crystals and Polymers, Oxford University Press, Oxford (1999). 4 p. M. Borsenberger and D. S. Weiss, Organic Photoreceptorsfor Xerography, Marcel Dekker, New York (1998). 5 L. J. Rothberg and A. J. Lovinger,J. Materials Research 11, 3174 (1996). 6 p. E. Burrows, G. Gu, V. Bulovic, Z. Shen, S. R. Forrest, and M. E. Thompson,IEEE Trans. on Electron Devices 44, 1188 (1997). 7 C. H. Chen, J. Shi, and C. W. Tang, Macromolecular Symposia 125, 1 (1998). 8 Semiconducting Polymers: Chemistry, Physics and Engineering, (G. Hadziioannou and P. Van Hutten, eds.), John Wiley & Sons, New York (2000).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

3

present recent reviews of device applications of organic electronic materials. Organic electronic devices have improved dramatically since their invention, and a wide range of products are now available or in development.lo-13 Organic LEDs have been reported with luminous efficiencies of 20-30 lm/W and with external quantum efficiencies of 7 - 8 % . 1 4 - i 7 These etiiciencies compare favorably with other emissive display technologies which typically have luminous efficiencies of about 5 lm/W 18. An external quantum efficiency of 7% corresponds to an internal quantum efficiency of about 35%. In organic LEDs the injected electrons and holes form excitons, which then recombine. Both singlet and triplet excitons are formed but, usually, only the singlet excitons recombine radiatively. The branching ratio for singlet/triplet formation is not known, but if governed by spin statistics alone then three triplets would be formed for every singlet. Therefore, statistics alone would argue that the maximum internal quantum efficiency would be 25%. The reported 35% internal quantum efficiency implies that spin statistics alone do not limit the maximum internal quantum efficiency. Recent experiments using phosphorescent organic molecules, in which both the singlet and triplet excitons can recombine radiatively, suggest that it may be possible to produce organic LEDs with close to 100% internal quantum efficiency. 17'19-22 The development of organic field-effect transistors has progressed more slowly than that of organic LEDs. However, organic FETs are now beginning to achieve performance levels suitable for circuit appli9 G. Horowitz, Advanced Materials 10, 365 (1998). lo j. R. Sheats, H. Antoniadis, M. Hueschen, W. Leonard, J. Miller, R. Moon, D. Roitman, and A. Stocking, Science 273, 884 (1996). 11 D. B. Roitman, H. Antoniadis, J. Sheats, and F. Pourmirzaie, Laser Focus Worm 34, 163 (1998). 12 p. E. Burrows, G. Gu, V. Bulovic, Z. Shen, S. R. Forrest, and M. E. Thompson, IEEE Trans. Elec. Device 44, 1188 (1997). 13 R. H. Friend, R. W. Gymer, A. B. Holmes, J. H. Burroughes, R. N. Marks, C. Taliani, D. D. C. Bradley, D. A. DosSantos, J. L. Bredas, and M. Logdlund, Nature 397, 121 (1999). 14 Results from many corporations. See, for example, Cambridge Display Technology

(www.cdtltd.co.uk) or Uniax Corporation (www.uniax.com). 15 j. Kido and Y. Iizumi, Appl. Phys. Lett. 73, 2721 (1998). 16 S. E. Shaheen, G. E. Jabbour, B. Kippelen, N. Peyghambarian, J. D. Anderson, S. R. Marder, N. R. Armstrong, E. Bellmann, and R. H. Grubbs, Appl. Phys. Lett. 74, 3212 (1999). 17 M. A. Baldo, S. Lamansky, P. E. Burrows, M. E. Thompson, and S. R. Forrest, Appl. Phys. Lett. 75, 4 (1999). 18 S. Matsumoto, in Electronic Display Devices (S. Matsumoto, ed.), Chapter 1, John Wiley & Sons, New York (1990). 19 M. A. Baldo, D. F. OBrien, Y. You, A. Shoustikov, S. Sibley, M. E. Thompson, and S. R. Forrest, Nature 395, 6698 (1998). 20 D. F. OBrien, M. A. Baldo, M. E. Thompson, and S. R. Forrest, Appl. Phys. Lett. 74, 442 (1999). 21 R. C. Kwong, S. Sibley, T. Dubovoy, M. Baldo, S. R. Forrest, and M. E. Thompson, Chemistry of Materials 11, 3709 (1999).

4

I.H. CAMPBELL AND D. L. SMITH

cations. 23-32 Transistors with field-effect mobilities of a b o u t 1 cm2/Vs, c o m p a r a b l e to a m o r p h o u s silicon, have been reported. 3~ H i g h mobility n-channel organic F E T s a n d c o m p l e m e n t a r y n- a n d p-channel organic circuits have been developed. 3x The use of high dielectric c o n s t a n t gate insulators has p r o d u c e d transistors with operating voltages below 5 V that are compatible with conventional inorganic circuitry. 32

1. ORGANIC ELECTRONIC MATERIALS The organic materials used in this new class of electronic devices are n-conjugated materials, either small molecules or polymers. T h e y have energy gaps ranging from a b o u t 1.5 eV to 3.5 eV, 2'3 are u n d o p e d , a n d therefore have essentially no free carriers at r o o m temperature. F i g u r e 1.1 shows the chemical structure of three representative materials: the small molecule t r i s - ( 8 - h y d r o x y q u i n o l a t e ) - a l u m i n u m [Alq], the p o l y m e r poly(pphenyelene vinylene) [PPV-I, a n d the small molecule pentacene. Tris-(8h y d r o x y q u i n o l a t e ) - a l u m i n u m was used in the first organic light-emitting diodes. 33'a4 Poly (p-phenylene vinylene) was the active material used in the first p o l y m e r light-emitting diodes. 35 It is an insoluble polymer, i.e. it does not dissolve in organic solvents, a n d is p r e p a r e d by t h e r m a l c o n v e r s i o n of

22 V. Cleave, G. Yahioglu, P. LeBarny, R. H. Friend, and N. Tessler, Advanced Materials 11, 285 (1999). 23 H. Sirringhaus, N. Tessler, and R. H. Friend, Science 280, 1741 (1998). 24 H. Sirringhaus, P. J. Brown, R. H. Friend, M. M. Nielsen, K. Bechgaard, B. M. W. LangeveldVoss, A. J. H. Spiering, R. A. J. Jannssen, and E. W. Meijer, Nature 401, 685 (1999). 25 Z. Bao, A. Dodabalapur, and A. J. Lovinger, Appl. Phys. Lett. 69, 4108 (1996). 26 Z. Bao, A. J. Lovinger, and A. Dodabalapur, Appl. Phys. Lett. 69, 3066 (1996). 27 A. Dodabalapur, Z. Bao, A. Makhija, J. G. Laquindanum, V. R. Raju, Y. Feng, H. E. Katz, and J. Rogers, Appl. Phys. Lett. 73, 142 (1998). 2s R. Hajlaoui, G. Horowitz, F. Garnier, A. ArceBrouchet, L. Laigre, A. E1Kassmi, F. Demanze, and F. Kouki, Advanced Materials 9, 389 (1997). 29 C. J. Drury, C. M. J. Mutsaers, C. M. Hart, M. Matters, and D. M. deLeeuw, Appl. Phys. Lett. 73, 108 (1998). 30 S. F. Nelson, Y.-Y. Lin, D. J. Glundlach, and T. N. Jackson, Appl. Phys. Lett. 72, 1854 (1998). 31 y..y. Lin, A. Dodabalapur, R. Sarpeshkar, Z. Bao, W. Li, K. Baldwin, V. R. Raju, and H. E. Katz, Appl. Phys. Lett. 74, 2714 (1999). 32 C. D. Dimitrakopoulos, S. Purushothaman, J. Kymissis, A. Callegari, and J. M. Shaw, Science 283, 822 (1999). 33 C. W. Tang and S. A. Van Slyke, Appl. Phys. Lett. 51, 913 (1987). 34 C. W. Tang, S. A. Van Slyke, and C. H. Chen, J. Appl. Phys. 65, 3610 (1989). 35 j. H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. Mackay, R. H. Friend, P. L. Burns, and A. B. Holmes, Nature 347, 539 (1990).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

5

---%

Fro. 1.1. Chemical structures of Alq (left), PPV (center) and pentacene (fight).

a precursor polymer. Soluble derivatives of PPV, such as poly [2-methoxy, 5-(2'-ethyl-hexyloxy)-l,4-phenylene vinylene] ( M E H - P P V ) , 36 which can be spin cast from organic solutions, are now more widely used. Pentacene is widely used in organic thin-film transistors. 3~ These materials all have large conjugated units, i.e. regions with resonant single and double bonds, which determine their conduction and valence energy levels and thus their energy gaps. The main differences between the small molecule and polymer materials are their processing methods and mechanical properties. Thin films of small molecules, such as Alq and pentacene, are usually prepared by vacuum evaporation, whereas thin polymer films are formed by solution processing methods such as spin casting. In both cases, the resulting films are amorphous or small grain polycrystalline (5-20 nm crystallites) and are highly disordered, z'33'37 Figure 1.2 shows the crystal structure of P P V determined from x-ray diffraction measurements. 38 These organic films have densities of about 1 g/cm 3, thermal conductivities of about 1 x 10 .3 W/cm-K, and specific heats of about 1 J/g-K. 39'4~ Polymer films have better mechanical integrity than films of small molecules. The polymer chain is held together by strong covalent bonds and, in a polymer film, the chains are typically entangled, which further increases the mechanical strength of the film. By 36 D. Braun and A. J. Heeger, Appl. Phys. Lett. 58, 1982 (1991). 37 D. J. Gundlach, Y.-Y. Lin, T. N. Jackson, S. F. Nelson, and D. G. Schlom, IEEE Electron Device Letters 18, 87 (1997). 3s D. Chen, M. J. Winokur, M. A. Masse, and F. E. Karasz, Phys. Rev. B41, 6759 (1990). 39 y. Agari, M. Shimada, and A. Ueda, Polymer 38, 2649 (1997). 40 C. H. M. Maree, R. A. Weller, L. C. Feldman, K. Pakbaz, and H. W. H. Lee, J. Appl. Phys. 84, 4013 (1998).

6

I.H. CAMPBELL AND D. L. SMITH

Fia. 1.2. Crystal structure parameters for PPV (courtesy of M. J. Winokur).

contrast, there is only a weak Van der Waals attraction between molecules in a molecular film. The electrical and optical properties of thin films of small molecules and polymers are generally similar. From an electronic structure point of view, the molecular films can be considered as a collection of distinct molecular sites. For the polymer films, the extended polymer chain is broken into independent sites by a combination of structural and chemical defects. In these organic films, electronic conduction occurs by hopping from one localized site to another. Hopping conduction is an essential aspect of the electrical transport in these materials. The optical absorption and electroluminescence spectra of a 50-nm M E H - P P V film are shown in Fig. 1.3. The absorption coefficient at the peak of the spectrum is about 2 x 10 5 cm-1.41 The absorption band is about 0.5 eV wide and the luminescence band is about 0.2 eV wide. The peak of the emission spectrum is red shifted from the peak of the absorption spectrum. Phonon replicas are apparent in both the absorption and emission spectra. The spectral width and the energy shift between the two spectra are due to a combination of structural relaxation and site energy disorder effects in the solid film. Because the absorption and emission spectra do not overlap appreciably, re-absorption of the emitted light is not significant in organic light-emitting diodes. The electroluminescence and photoluminescence spectra (not shown) are essentially identical, indicating that the emitting excited states formed by electrical and optical excitation are the same. The 41 M. G. Harrison, J. Gruner, and G. C. W. Spencer, Phys. Rev. B55, 7831 (1997).

7

PHYSICS OF ORGANIC ELECTRONIC DEVICES

1.0

0.4

r~

0.8 x~

0.3 9 l

l

0.6 0.2~ r~

t~

0.4

0

.,..~

o

r~ ~,,,l o

0.1

/

~ 02

~

",

9

0.0

1.5

2.0

2.5

3.0

0.0

Photon Energy (eV) FIr. 1.3. Electroluminescence (solid line) and optical absorption spectrum (dashed line) of MEH-PPV. The chemical structure of MEH-PPV is shown in the inset.

general features in the optical spectra, shown in Fig. 1.3, such as the strength and width of the absorption spectrum and the shift between the peak of the absorption and luminescence, are similar for most conjugated organic materials. The energy gap of these materials is typically between 1.5 eV and 3.5 eV and therefore the emission spectrum of organic light-emitting diodes can span the visible spectral region. 2 The intrinsic radiative lifetime of the luminescent, singlet excitons in these materials is between 1 ns and 20 ns. 2'42 The solid state, photoluminescence quantum efficiency varies from about 10% to n e a r 1 0 0 % . 2 ' 4 3 - 4 7 The optical index of refraction and static

42 R. Priestley, A. D. Walser and R. Dorsinville, Optics Communications 158, 93 (1998). 43 D. Z. Garbuzov, V. Bulovic, P. E. Burrows, and S. R. Forrest, Chem. Phys. Lett. 249, 433 (1996). 44 M. Remmers, D. Neher, J. Gruner, R. H. Friend, G. H. Gelinck, J. M. Warman, C. Quattrocchi, D. A. dos Santos, and J. L. Bredas, Macromolecules 29, 7432 (1996). 45 N. C. Greenham, I. D. W. Samuel, G. R. Hayes, R. T. Phillips, Y. Kessener, S. C. Moratti, A. B. Holmes, and R. H. Friend, Chem. Phys. Lett. 241, 89 (1995). 46 G. KoppingGrem, G. Leising, M. Schimetta, F. Stelzer, and A. Huber, Synthetic Metals 76, 53 (1996). 47 R. G. Sun, Y. Z. Wang, D. K. Wang, Q. B. Zheng, E. M. Kyllo, T. L. Gustafson, and A. J. Epstein, Appl. Phys. Lett. 76, 634 (2000).

8

I.H. CAMPBELL AND D. L. SMITH

FIG. 1.4. Organic light-emitting diode device structure.

dielectric constant of these materials are typically about 1.7 and 3, respectively.48 - 51

2. ORGANIC ELECTRONIC DEVICES The two principal organic electronic devices are light-emitting diodes and field-effect transistors. Organic light-emitting diodes are vertical structures that utilize electrical transport across a thin organic film. In contrast, organic field-effect transistors are lateral structures that utilize transport in the plane of an organic thin film. This distinction is significant because the distance through which current is driven in the two types of structures differs by orders of magnitude and the electrical transport properties of the organic thin films are often anisotropic. This is particularly true for polymer films, where the polymer chains lie predominantly in the plane of the film. 52 Figure 1.4 shows the structure of a typical organic light-emitting diode. The diode consists of a transparent metal/organic film/metal structure. The transparent metal contact is usually indium tin oxide (ITO) or a thin metal (e.g., 10 nm of Au). The organic film, either small molecule or polymer, is typically about 100 nm thick. The top metal contact is a low work function metal or metal alloy such as Ca or Mg:Ag. The diode structure is typically 48 V. Bulovic, V. B. Khalfin, G. Gu, P. E. Burrows, D. Z. Garbuzov, and S. R. Forrest, Phys. Rev. B58, 3730 (1998). 49 A. J. Campbell, D. D. C. Bradley, J. Laubender, and M. Sokolowski, J. Appl. Phys. 86, 5004 (1999). 50 S. J. Martin, D. D. C. Bradley, P. A. Lane, H. Mellor, and P. L. Burn, Phys. Rev. B59, 15133 (1999). 51 W. Holzer, M. Pichlmaier, E. Drotleff, A. Penzkofer, D. D. C. Bradley, and W. J. Blau, Optics Communications 163, 24 (1999). 52 D. McBranch, I. H. Campbell, D. L. Smith, and J. P. Ferraris, Appl. Phys. Lett. 66, 1175 (1995).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

9

fabricated as follows: A glass substrate is sputter coated with semi-transparent ITO, the organic film is then either evaporated or spin cast onto the semi-transparent contact, and finally the top contact is evaporated onto the organic film. The bottom and top contacts are patterned so that their spatial overlap defines the area of the LED. The bottom contact is generally patterned using optical lithography and the top contact is usually patterned using a shadow mask. The top contact is not patterned using photolithography because of the sensitivity of the organic thin-film to the processing procedures. Many variations of the basic organic light-emitting diode device structure have been investigated. Organic LEDs with two transparent contacts have been used to construct transparent displays 53-55 and microcavity designs have been used to adjust organic LED spectral properties. 56'57 Doped polymer anodes 58'59 and thin (1 nm) inorganic layers at the metal/organic interface 6~ have been used to improve device lifetime. The basic organic LED structure can also be used to produce organic photodetectors and photovoltaic cells. 2 ' 6 2 - 6 4 To date, these devices have not been completely successful because it is difficult to separate the optically generated charge carriers and produce a photocurrent. 2'64-67 Organic transistors are in a comparatively early stage of development and there is considerable variety in the details of their structure and fabrication. Figure 1.5 shows the structure of a prototype organic thin-film transistor fabricated on a crystalline Si substrate. The transistor consists of an organic thin-film on top of a gate insulator contacted by metal source and drain 53 G. Gu, V. Bulovic, P. E. Burrows, S. R. Forrest, and M. E. Thompson, Appl. Phys. Lett. 68, 2606 (1996). 54 G. Parthasarathy, P. E. Burrows, V. Khalfin, V. G. Kozlov, and S. R. Forrest, Appl. Phys. Lett. 72, 2138 (1998). 55 G. Gu, G. Parthasarathy, and S. R. Forrest, Appl. Phys. Lett. 74, 305 (1999). 56 p. E. Burrows, V. Khalfin, G. Gu, and S. R. Forrest, Appl. Phys. Lett. 73, 435 (1998). 57 R. H. Jordan, L. J. Rothberg, A. Dodabalapur, and R. E. Slusher, Appl. Phys. Lett. 69, 1997 (1996). 5a S. A. Carter, M. Angelopoulos, S. Karg, P. J. Brock, and J. C. Scott, Appl. Phys. Lett. 70, 2067 (1997). 59 j. Gao, A. J. Heeger, J. Y. Lee, and C. Y. Kim, Synthetic Metals 82, 221 (1996). 60 L. S. Hung, C. W. Tang, and M. G. Mason, Appl. Phys. Lett. 70, 152 (1997). 61 F. Li, H. Tang, J. Anderegg, and J. Shinar, Appl. Phys. Lett. 70, 1233 (1997). 62 G. Yu and A. J. Heeger, J. Appl. Phys. 78, 4510 (1995). 63 j. j. M. Halls, K. Pichler, R. H. Friend, S. C. Moratti, and A. B. Holmes, Appl. Phys. Lett. 68, 3120 (1996). 64 M. Granstrom, K. Petritsch, A. C. Arias, and R. H. Friend, Synthetic Metals 102, 957 (1999). 65 S. Barth, H. Bassler, H. Rost, and H. H. Horhold, Phys. Rev. B56, 3844 (1997). 66 S. Barth and H. Bassler, Phys. Rev. Lett. 79, 4445 (1997). 67 A. Kohler, D. A. dosSantos, D. Beljonne, Z. Shuai, J. L. Bredas, A. B. Holmes, A. Kraus, K. Mullen, and R. H. Friend, Nature 392, 903 (1998).

10

I.H. CAMPBELL AND D. L. SMITH Metal Source

Organic Thin Film /

V//'////////'///////////A

p+ Gate Contact

/

Metal Drain

Gate Insulator Si Substrate

FIG. 1.5. Organic field-effect transistor device structure.

electrodes. The gate insulator is often silicon dioxide about 100 nm thick grown on a doped silicon wafer that serves as the gate contact. The organic film is usually made from small molecules that typically have higher carrier mobility than polymers due to increased molecular order. 9'26'28'3~ 72 In operation, current is carried in a thin, approximately 5 nm, region of the organic thin-film next to the gate insulator. 73 The properties of the first few layers of the organic film are thus critical in determining the transistor performance. Most organic FETs are p-type devices that use Au for the source and drain contacts. The channel length, i.e. the spacing between the source and drain contacts, has been varied from about 0.1 #m to over 100~m. 9'72 A typical prototype transistor structure is fabricated as follows: A heavily doped silicon wafer is oxidized to produce the gate insulator, the source and drain contacts are vacuum evaporated and defined using optical lithography, and finally the organic film is thermally evaporated covering the gate insulator, and metal source and drain contacts. Organic transistors have also been fabricated on plastic substrates using both sputter deposited oxides and spin cast polymers as the gate insulator. In some cases, the fabrication sequence is changed so that the organic thin-film is deposited prior to the source and drain contacts. The source and drain contacts are then evaporated onto the organic film using shadow masks to define their structure.

68 L. Torsi, A. Dodabalapur, L. J. Rothberg, A. W. P. Fung, and H. E. Katz, Phys. Rev. B57, 2271 (1998). 69 G. Horowitz, F. Garnier, A. Yassar, R. Hajlaoui, and F. Kouki, Advanced Materials 8, 52 (1996). 70 R. Hajlaoui, D. Fichou, G. Horowitz, B. Nessakh, M. Constant, and F. Garnier, Advanced Materials 9, 557 (1997). 71 F. Garnier, G. Horowitz, D. Fichou, and A. Yassar, Supramolecular Science 4, 155 (1997). 72 j. A. Rogers, A. Dodabalapur, Z. N. Bao, and H. E. Katz, Appl. Phys. Lett. 75, 1010 (1999). 73 M. A. Alam, A. Dodabalapur, and M. R. Pinto, IEEE Transactions on Electron Devices 44, 1332 (1997).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

11

3. SCOPE OF THIS ARTICLE

This article discusses the basic device physics that governs the operation of organic light-emitting diodes and field-effect transistors. The field of organic electronic device physics is relatively new and is rapidly developing. Not all of the issues in the field are settled and some points remain controversial, but the essential device physics is becoming relatively clear. The operating principles of organic light-emitting diodes are fundamentally distinct from conventional inorganic semiconductor-based LEDs. The rectification and light-emitting properties of inorganic LEDs are due to the electrical junction between oppositely doped, p- and n-type regions of the inorganic semiconductor. TM In contrast, organic light-emitting diodes are formed using an undoped, insulating organic material. The rectification and light-emitting properties of the polymer LED are caused by the use of asymmetric metal contacts. One metal contact is only able to inject electrons efficiently and the other contact only injects holes efficiently. The injected electrons and holes recombine in the undoped organic film, emitting light. High work function metals inject holes more efficiently than electrons and, similarly, low work function metals inject electrons more efficiently than holes. Therefore, the high work function metal is the anode and the low work function metal is the cathode. This defines the forward and reverse bias directions of the diode and leads to current rectification, bipolar charge injection, and light emission in organic LEDs. The essential physical processes that need to be understood to describe organic LEDs are charge injection into the organic material from the metallic contacts, charge transport in the organic layer, and electron-hole recombination processes in the organic film. The operation of organic field-effect transistors is similar to that of inorganic thin-film transistors. In both devices, an undoped thin-film is contacted by metallic source and drain electrodes and separated from a gate electrode by a thin gate insulator. Applying a bias to the gate contact induces charge in the thin-film from the source and drain contacts. The conductivity of the thin-film is determined by the amount of induced charge, which is controlled by the gate bias. The main differences between the organic and inorganic structure are the type of metallic contacts used and the transport properties of the undoped thin film. Inorganic structures typically use doped semiconductor contacts made from the same material used for the insulating thin film. TM Using the same material for the contacts and the insulating film ensures good electrical contact between the electrodes and the thin film. In contrast, organic FETs use metallic contacts 74 S~ M. Sze, Physics of Semiconductor Devices, John Wiley & Sons, New York (1981).

12

I.H. CAMPBELL AND D. L. SMITH

instead of doped organic materials. This can result in significant contact resistance effects in the transistor current-voltage characteristics. The transport properties of thin organic films are dominated by carrier hopping from site to site in the disordered organic film. This leads to an effective carrier mobility that depends on the density of charge induced by the gate in the organic film. The essential physical processes that need to be understood to describe organic FETs are the behavior of the metallic contacts and the charge transport properties of the organic film. This article describes the physics of organic electronic devices by presenting the essential electronic structure and charge transport properties of organic materials separately and then combining them to describe the operating characteristics of devices. The article is organized as follows: Section II reviews the electronic structure of re-conjugated organic materials; Section III presents the essential features of metal/organic interface electronic structure; Section IV describes the electrical transport properties of thin organic films; Section V presents a device model of organic lightemitting diodes and discusses field-effect transistor operation; Section VI summarizes the conclusions.

II. Electronic Properties of n-Conjugated Organic Materials For the materials used to fabricate the organic electronic devices discussed in this article, the electronic states that play the essential roles in device operation are derived primarily from the p orbitals of carbon in an sp2-p hybrid configuration. The hybridized sp 2 orbitals form a bonds essential to the molecular structure, but electronic excitations or charged states involving these orbitals are very high in energy. The low-energy neutral excitations and charged states, of interest for electronic devices, are formed from the p orbitals involved in ~ bonding. 2-5 The low-energy electronic excitations involving these p orbitals are typically in the 1.5-3.5 eV range for the materials used in devices. Both conjugated polymers and smaller conjugated molecules are used to make organic electronic devices. Although there are significant fabrication differences for these two classes of materials, their basic electrical behavior is similar. Devices fabricated from a polymer and a corresponding oligomer (an oligomer is a molecule consisting of a small number of the repeat units of a polymer, essentially a very short chain polymer) have similar electrical properties. 75 This result is not surprising because structural and chemical 75 M. D. Joswick,I. H. Campbell, N. N. Barashkov, and J. P. Ferraris, J. Appl. Phys. 80, 2883 (1996).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

13

defects limit the length over which conjugation is maintained in conjugated polymers. From an electronic structure point of view the polymer is like a group of oligomers bonded together, with weak interactions between the electronic states in the n-bonded manifold of the different conjugated segments. The electronic states of interest for devices are localized on the individual molecules or in the case of polymers on the separate segments over which conjugation is maintained. Organic electronic devices are formed from dense thin films of the conjugated molecules or polymers. The electronic properties of the dense films are not always the same as those of the isolated molecules or single polymer chains. The interactions between the molecules in the films can have important effects on the electronic structure. Current quantum chemical techniques can accurately describe many aspects of the electronic structure of isolated molecules. A theme of current research in this area is aimed at determining which features of the electronic structure of the isolated molecules are maintained and which are modified in the condensed phase. The section is organized as follows: First, an elementary picture of the electronic structure of n-conjugated hydrocarbons is presented; electron-ion and electron-electron interactions not included in the simple picture are then discussed; and, modifications of the molecular electronic structure in the solid state are described.

4. ELEMENTARY PICTURE OF THE ELECTRONIC STRUCTURE

An elementary molecular orbital (MO) picture of the electronic structure of n-conjugated materials is first discussed. Because many of the conjugated materials used in organic electronic devices contain phenylene (benzene) rings, benzene and biphenyl, a molecule consisting of two benzene rings bonded together, are used as examples. Figure II.1 is a schematic of the molecular structure of benzene. The solid lines represent the a bonds formed

2

H \

/

H

3

1H - ~ O ~ H / H 2'

\

213 4

Iv AI

H 3'

111 "

"

-~ -213

FIG. II.1. Chemical structure of benzene and its n-electron molecular orbital energies.

14

I.H. CAMPBELL AND D. L. SMITH

primarily from the sp 2 hybridized orbitals of carbon and the s orbitals of hydrogen. Neutral excitations and charged states involving these orbitals lie outside of the energy range of interest. The Pz orbitals of carbon, which are orthogonal to the plane of the page, form the rt bonds indicated by the circle. The low-energy excitations and charged states of device interest are formed from these orbitals. Six spatial states (twelve including spin) can be formed from the Pz orbitals. In a simple molecular orbital picture with nearest neighbor interactions, the 6 MOs (not normalized) are (1,1,1,1,1,1), (0,1,1,0,- 1 , - 1), (2,1,- 1 , - 2 , - 1,1), (-2,1,1,-2,1,1), (0,1-1,0,1,- 1) and ( 1 , - 1,1,- 1,1,- 1) with corresponding MO energies -2fl, -fl, -fl, fl, fl, and 2ft. Here the site ordering for the M Os is (1,2,3,4,3',2'), where the site numbers are indicated in the figure and fl is the magnitude (a positive number) of the hopping integral. The MO energy levels are schematically illustrated in Fig. II.1. MOs 2 and 3, and 4 and 5 are degenerate. In the electronic ground state of the molecule, MOs 1, 2, and 3 are each occupied with two electrons. MOs number 2 and 5 have no amplitude on sites 1 and 4. The energy gap between occupied and empty MOs is 2fl and the sum of the occupied M O energies is -8ft. Figure 11.2 shows a schematic of the molecular structure of biphenyl, which consists of two benzene molecules bound together. A simple pertur-

2f5 + C/6

213- C/6 13+ C/3

13- C/3

_•

-~ + c/3

_~

-~-c/3

_~

-213+ C/6

_~

-213- C/6

FIG. 11.2. Chemical structure of biphenyl and its n-electron molecular orbital energies.

PHYSICS OF ORGANIC ELECTRONIC DEVICES

15

bation theory description of the MO energies of biphenyl is shown next to the molecular structure schematic. The lowest energy biphenyl MOs are made up from the first benzene MO (L1 for the first MO on the left benzene and R1 for the first MO on the right benzene): (L1 + R1) and ( L 1 - R1) with energies - 2 f l - C/6 and - 2 / / + C/6, respectively. Here C is the magnitude of the hopping integral between carbon-4 on the left benzene and carbon-1 on the right benzene. In the same way symmetric and antisymmetric biphenyl MOs can also be made from benzene MOs 3, 4 and 6: (LN + RN) and (LN - RN), where N = 3, 4, or 6 with energies - f l + C/3, - f l - C/3 for N = 3; fl - C/3,/3 + C/3 for N = 4; and 2fl + C/6, 2 / / - C/6 for N = 6. Because benzene MOs 2 and 5 have no amplitude on carbons 1 and 4, they do not mix when forming biphenyl in the simple nearest neighbor interaction model. The inter-benzene hopping matrix element C depends on the angle between the planes of the two benzene molecules, C = ColcOs 01, where 0 is the angle between planes of the two molecules and Co is the magnitude of the matrix element when the benzene molecules are coplanar. The 6 lowest energy MOs are each occupied by 2 electrons in the neutral ground state. The energy gap between occupied and empty MOs is 2 ( f l - C/3) and the sum of the occupied MO energies is - 1 6 f l The energy gap of biphenyl is lowered compared with benzene by - 2 C / 3 . This lowering of the energy gap is a consequence of the more spatially extended nature of the states in biphenyl. The extent of this gap lowering depends on the angle between the benzene molecules, but the sum of the MO energies does not depend on the angle at this level of approximation. The lowest energy state with an extra electron added to the biphenyl molecule has the extra electron in the (L4 + R4) MO with energy f l - C/3. The sum of the occupied MO energies for this negatively charged state is - 15/3 - C/3. The lowest energy state with an electron removed from the biphenyl molecule has a hole in the (L3 + R3) MO with energy - f l + C/3. The sum of the occupied MO energies for this positively charged state is also - 1 5 / / - C/3. The sum of the occupied MO energies depends on the angle between the benzene molecules for both the positively and negatively charged states, but not for the neutral state. This leads to a coupling between the angular orientation and electronic charges. This molecular orbital picture is highly idealized but its qualitative results are preserved in more complete calculations. Figure II.3 shows the result 76 of a quantum chemical calculation (AM1 level 77) of the angular dependence of the energy of the biphenyl molecule (arbitrary energy zero) in the ground 76 Z. G. Yu, D. L. Smith, A. Saxena, R. L. Martin, and A. R. Bishop, Phys. Rev. Lett. 84, 721 (2000). 77 H. J. S. Dewar, E. G. Zoebisch, and E. F. Healy, J. Am. Chem. Soc. 107, (1985).

16

I . H . C A M P B E L L AND D. L. SMITH 0.6

0.3

~

0.0

w -0.3

anion . . . . neutml cation ,,

-0.6

o

.

.

.

9'0

.

.

.

.

.

2- o

380

Angle (degree) FIG. 11.3. AM1 results for the total energy as a function of torsion angle for biphenyl. Solid, dashed, and dot-dashed lines are for anion, neutral, and cation states, respectively (from Ref. 76).

electronic state for the neutral, positive, and negative ions. For the neutral molecule, there is a weak dependence of the energy on angle (in the simple description above the energy is independent of this angle). The lowest energy angle is approximately 40 degrees. The angular dependences of the energy for the two ions are very similar, they are much stronger than for the neutral molecule and have a minimum at 0 and a maximum at 90 degrees. The shape of the angular dependence curves is similar to Icos 01 expected from the simple MO model. The minimum energy angle of the neutral molecule does not coincide with that for the ions so that a rotation will occur when an electron or hole is transferred to a neutral molecule to form an ion.

5. ELECTRON--ION AND ELECTRON--ELECTRON INTERACTIONS

The lowest energy geometry of a molecule depends on its electronic configuration. The minimum energy bond lengths and angles are different for the excited and ground electronic states of a neutral molecule and for the ground states of a neutral and a charged molecule. Structural relaxation of the molecule occurs when the electronic state of the molecule is changed. This relaxation affects, for example, the optical absorption and emission spectra near the fundamental absorption edge. As seen in Fig. 1.3, vibrational replicas are observed in both the absorption and emission spectra of MEH-PPV. These replicas occur because vibrational excitations can be induced in the final state of the transition. The theory describing optical

PHYSICS OF ORGANIC ELECTRONIC DEVICES

17

((= (-> (--> FIG. 11.4.Benzenoid (top), quinoid (middle), and electron polaron (bottom) bond configurations in PPV.

transitions in which the initial and final electronic states have different lowest energy geometries is well established. 1-4 Because of the strong coupling between electronic states and molecular geometry that occurs in the =-conjugated materials used for electronic devices and also because of the molecular disorder usually present in dense films of these materials, the optical absorption and emission spectra are typically broad. Structural relaxation, compared to the neutral ground state, also occurs for charged states. ~-4 The double bonds that occur in re-conjugated materials are delocalized and flexible in that it is possible to shift between different configurations comparatively easily. Figure 11.4 schematically shows two double bonding configurations, the benzenoid (top) and quinoid

18

I.H. CAMPBELL AND D. L. SMITH

(middle) configurations, for PPV. The benzenoid configuration has the lower energy for the neutral state. When an electron is added or removed from the molecule, the double bonding is disrupted, leading to an admixture of the two configurations for the charged states. The situation is schematically illustrated in the bottom of Fig. 11.4. An extra electron is put into a p, orbital to form a lone pair, which disrupts the double bonding scheme. The p, orbital that had been paired with the one containing the extra electron in the benzenoid configuration now pairs with an orbital in a benzene ring to form a quinoid configuration. Because the quinoid configuration is higher energy than the benzenoid configuration, the system must eventually return to the benzenoid configuration. A singly occupied, unpaired Pz orbital results when the configuration changes. (The figure is very schematic; the electrons aren't actually localized in a single orbital and the configuration doesn't abruptly change from one form to another.) The structurally relaxed negatively charged state is called an electron polaron. If an electron were removed from the neutral molecule, a hole polaron would be formed. The schematic for the hole polaron is similar to that shown for the electron polaron except that the doubly occupied Pz orbital is replaced with an empty one. The relaxation energy depends on the length of the conjugated segment, decreasing as the length increases. Theoretical estimates for PPV give relaxation energies varying from about 0.15 eV for a three-ring oligomer to about 0.05 eV for a long chain. 78 Structural relaxation occurs very rapidly when an electron is added to or removed from a n-conjugated organic molecule and from a device point of view the charged states that result can essentially always be considered as structurally relaxed. The singly charged states of device interest are electron and hole polarons. For simplicity these states are usually just called electrons and holes. The structural relaxation that occurs around charged states can lead to an effective attractive interaction between two polarons. If this attractive interaction is stronger than the repulsive Coulomb interaction for charges with the same sign, a doubly charged bound state, which has been called a bipolaron, can result. Bipolarons in conjugated polymers have been extensively discussed. ~'79-85 A schematic for an electron bipolaron is similar to 78 Z. Shuai, J. L. Bredas, and W. P. Su, Chem Phys. Lett. 228, 301 (1994). 79 S. A. Brazovskii and N. Kirova, Synthetic Metals 57, 4385 (1993). ao S. Brazovskii, N. Kirova, Z. G. Yu, A. R. Bishop, and A. Saxena, Optical Materials 9, 502 (1998). 81 A. Saxena, S. Brazovskii, N. Kirova, Z. G. Yu, and A. R. Bishop, Synthetic Metals 101, 325 (1999). a2 Z. G. Soos, S. Ramasesha, D. S. Galvao, and S. Etemad, Phys. Rev. B47, 1742 (1993). 83 y. Shimoi and S. Abe, Synthetic Metals 78, 219 (1996). a4 y. Shimoi and S. Abe, Phys. Rev. B50, 14781 (1994). a5 y. Shimoi and S. Abe, Phy. Rev. B49, 14113 (1994).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

19

that for the electron polaron in the bottom of Fig. II.4 except that both of the p, orbitals explicitly shown are doubly occupied. For the bipolaron state there is only one region of the higher energy quinoid configuration, rather than two as would occur for two isolated polarons. The reduction of the high-energy quinoid configuration is the physical origin of the attractive interaction. A hole bipolaron schematic is similar to that for the electron bipolaron except that both of the p, orbitals shown are empty. If charged bipolarons were strongly bound, they could have a significant influence on the behavior of organic electronic devices. 79'86'87 Section III discusses possible device implications of these kinds of states and approaches to detecting them. However, for the materials typically used for organic electronic devices, bipolarons are found not to be strongly bound and not to significantly influence device behavior, as There is an attractive interaction between an electron polaron and a hole polaron due both to the Coulomb interaction and also to the structural relaxation around the charged states. A bound state results that is usually called an exciton. (This state is sometimes called a neutral bipolaron if the interaction due to structural relaxation is emphasized.) There is an extensive literature on excitons in re-conjugated organic materials. 78'82'83'85'89-1~

86 N. Kirova and S. Brazovskii, Synthetic Metals 76, 229 (1996). 87 p. S. Davis, A. Saxena and D. L. Smith, J. Appl. Phys. 78, 4244 (1995). 88 I. H. Campbell, T. W. Hagler, D. L. Smith, and J. P. Ferraris, Phys. Rev. Lett. 76, 1900 (1996). 89 M. Chandross, S. Mazumdar, S. Jeglinski, X. Wei, Z. V. Vardeny, E. W. Kwock, and T. M. Miller, Phys. Rev. BS0, 14702 (1994). 90 M. Chandross, S. Mazumdar, M. Liess, P. A. Lane, Z. V. Vardeny, M. Hamaguchi, and K. Yoshino, Phys. Rev. B55, 1486 (1997). 91 M. Chandross and S. Mazumdar, Phys. Rev. B55, 1497 (1997). 92 p. G. DaCosta and E. M. Conwell, Phys. Rev. B48, 1993 (1993). 9 3 y. N. Gartstein, M. J. Rice, and E. M. Conwell, Phys. Rev. B52, 1683 (1995). 94 E. M. Conwell, Synthetic Metals, 83, 101 (1996). 95 E. Moore, B. Gherman, and D. Yaron, J. Chem Phys. 106, 4216 (1997). 96 D. Yaron, E. E. Moore, Z. Shuai, and J. L. Bredas, J. Chem Phys. 108, 7451 (1998). 97 E. E. Moore and D. Yaron, J. Chem Phys. 109, 6147 (1998). 98 E. Moore and D. Yaron, Synthetic Metals 85, 1023 (1997). 99 Z. Shuai, S. K. Pati, W. P. Su, J. L. Bredas, and S. Ramasesha, Phys. Rev. B55, 15368 (1997). loo M. J. Rice and Y. N. Gartstein, Phys. Rev. Lett. 73, 2504 (1994). lOl y. N. Gartstein, M. J. Rice, and E. M. Conwell, Phys. Rev. B52, 1683 (1995). lo2 S. Brazovskii, N. Kirova, A. R. Bishop, V. Klimov, D. McBranch, N. N. Barashkov, and J. P. Ferraris, Optical Materials 9, 472 (1998). lo3 j. L. Bredas, J. Cornil, and A. J. Heeger, Advanced Materials 8, 447 (1996). lO4 H. S. Woo, O. Lhost, S. C. Graham, D. D. C. Bradley, R. H. Friend, C. Quattrocchi, J. L. Bredas, R. Schenk, and K. Mullin, Synthetic Materials 59, 13 (1996). lO5 D. Beljonne, Z. Shuai, R. H. Friend, and J. L. Bredas, J. Chem. Phys. 102, 2041 (1995).

20

I.H. CAMPBELL AND D. L. SMITH

The spin-orbit interaction is very small in conjugated organic materials because light elements are involved and spin is a good quantum number. The spins of the electron and hole involved in an exciton can be either singlet or triplet paired. There is often an allowed optical transition between the electronic ground state and the lowest energy singlet exciton, but not between the electronic ground state and a triplet exciton. The triplet exciton is more strongly bound than the singlet exciton. The magnitude of the energy splitting between the singlet and triplet excitons can be an important parameter in organic LEDs because it helps determine the stability of the triplet exciton. The molecular orbital picture is based on a mean field treatment of electron-electron interactions and assumes that electron correlation effects do not dominate the electronic structure. Electron correlation can be significant in n-conjugated organic materials. If electron correlation does not change the qualitative nature of the ground state, low-energy neutral excited states, and low-energy charged states, its effects can usually be incorporated into a mean field description parametrically. However, there are cases in which electron correlation does qualitatively change the nature of these states. For example, in some polyenes (chains consisting of (CH2) n units) electron correlation causes an optical dipole forbidden state to be the lowest energy excited singlet. Using symmetry labels appropriate for polyenes (C2h point group), the ground state has Ag symmetry and the lowest dipole allowed singlet excited state has B u symmetry. This Bu state is expected to be the lowest energy singlet excited state on the basis of mean field electronic structure theory. When electron correlation is taken into account another singlet excited state, with Ag symmetry, (not well described in mean field theory) can be pulled to lower energy than the B ~ state. 9~176 The Ag excited state does not have an allowed dipole transition to the electronic ground state; there is an allowed two-photon transition between these states. The B~ excited state is stabilized relative to the Ag excited state by the phenylene rings that occur in many of the materials used for electronic devices. 1~ For LEDs, it is important that the lowest singlet excited state have an allowed dipole transition to the ground 106 B. Lawrence, W. E. Torruellas, M. Cha, M. L. Sundheimer, G. I. Stegeman, J. Meth, S. Etemad, and G. Baker, Phys. Rev. Lett. 73, 597 (1994). lo7 Z. G. Soos, S. Ramasesha, and D. S. Galvao, Phys. Rev. Lett. 71, 1609 (1993). lo8 Z. G. Soos, D. S. Galvao, and S. Etemad, Advanced Materials 6, 280 (1994). lo9 Z. G. Soos, S. Ramasesha, D. S. Galvao, R. G. Kepler, and S. Etemad, Synthetic Metals 54, 35 (1993). 11o M. Chandross, Y. Shimoi, and S. Mazumdar, Chem. Phys. Lett. 280, 85 (1997). 111 A. Chakrabarti and S. Mazumdar, Phys. Rev. B59, 4839 (1999). 112 D. Yaron and R. Silbey, Phys. Rev. B45, 11655 (1992).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

21

state so that materials in which the mg state is the lowest energy singlet exciton cannot be used for organic LEDs.

6. SOLID STATE PROPERTIES The electronic structure of individual molecules can be significantly modified in the dense solid state films used for devices. Quantum chemical techniques can accurately describe many aspects of the electronic structure of individual molecules. Infinite-chain polymers can present a problem for these methods, but the conjugation breaks that occur for real polymers limit the size of the systems that are of practical interest. Understanding which features of the electronic structure of the isolated molecules are maintained and which are significantly modified in the condensed phase and deducing the electronic properties of the condensed phase materials from those of the individual molecules or single-chain polymers is a current theoretical challenge. In many cases, the properties of localized neutral excited states, such as excitons, are not strongly modified going from the individual molecule to the condensed phase. 95'97'98'113 Single molecule calculations have been successful in describing near edge optical absorption spectra. By contrast the energies of charged states in dense films are strongly modified by screening. As a result, single molecule calculations of the ionization potential or electron affinity do not accurately describe the corresponding observations in dense films. Because the energies of charged states are strongly modified in the condensed phase, it is not straightforward to extract the energy gap of a dense film from single molecule calculations. Because exciton binding energies can be significant, extracting the energy gap from optical absorption data is also not straightforward. The energy gap is an important material parameter for device design. Polarons and perhaps bipolarons are the important charged states in organic materials. If bipolarons were strongly bound, so that a bipolaron had a significantly lower energy than two isolated polarons, they would have an important influence on the behavior of electrical devices. For example, they would play an important role in Schottky barrier formation. 79'86'87 Because of the possibility of strongly bound bipolarons, it is convenient to define two energy gaps, one corresponding to polaron formation and a second corresponding to bipolaron formation. The singleparticle energy gap, corresponding to polaron formation, is analogous to the energy gap in a conventional semiconductor. It is the energy difference between the electronic ground state and a state consisting of spatially 113 R. L. Martin, J. D. Kress, I. H. Campbell, and D. L. Smith, Phys. Rev. (accepted).

22

I.H. CAMPBELL AND D. L. SMITH

separated electron and hole polarons. The charge transfer energy gap, corresponding to bipolaron formation, is smaller than the single-particle energy gap by the bipolaron binding energy. As discussed in Section III, it is the maximum equilibrium built-in potential that can be supported by the organic material. The materials used for organic electronic devices are, for the most part, highly disordered. The disorder is the result of different local environments of the molecules, different molecular geometries imposed by steric interactions with nearest neighbors, and chemical defects that break up conjugation. As a result of this disorder there is an ensemble of energies for both neutral excitations and charged states. Strong structural relaxation of the molecule occurs when a charge is added or removed and the electron transfer integrals between neighboring molecules are not large. As a result the low-energy charged electronic states of device interest are localized on individual molecules or on the individual conjugated segments of polymers. Band structure calculations have been performed for some of the conjugated materials of device interest. 114-116 These calculations do not include disorder or structural relaxation for charged states. They indicate narrow bands corresponding to intermolecular transport. Such calculations are of interest for describing these materials in an ordered limit. There are examples of particularly well-ordered organic materials in which delocalized bandlike states are important. 117 But the low-energy states of device interest in the materials typically used for organic electronic devices are localized and electrical transport results from hopping between these localized states. In crystalline semiconductors, covalent chemical bonds are necessarily broken at nonepitaxial interfaces. The lattice matching conditions necessary for epitaxial interfaces are very restrictive. The electronic structure of nonepitaxial interfaces involving crystalline semiconductors are complex and often defect dominated. For example, interface or defect states in the energy gap of the semiconductor usually dominate the electronic properties of metal/semiconductor interfaces. In contrast, covalent chemical bonds are not necessarily broken at interfaces involving organic electronic materials and restrictive lattice matching conditions analogous to those for crystalline semiconductors do not arise. These less restrictive properties are critical to the success of organic electronic devices. They affect the type of substrate that can be used and the properties of both organic/organic and metal/ organic interfaces. Because lattice matching is not an issue, thin films of 114 p. Vogl and D. K. Campbell, Phys. Rev. B41, 12797 (1990). 115 p. G. DaCosta, R. G. Dandrea, and E. M. Conwell, Phys. Rev. B47, 1800 (1993). 116 M. Rohlfing and S. G. Louie, Phys. Rev. Lett. 82, 1959 (1999). 117 j. H. Schon, S. Berg, Ch. Kloc, and B. Batlogg, Science 287, 1022 (2000).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

23

organic electronic materials may be prepared on almost any substrate. Plastic, low-cost, transparent, and/or flexible substrates have been investigated. Organic/organic interfaces are analogous to inorganic semiconductor heterojunction interfaces except that restrictive lattice matching conditions do not apply and, in many cases, the properties of the organic molecules at the interface are not strongly perturbed. It is thus straightforward to produce near ideal organic/organic heterojunctions simply by depositing (by spin casting or thermal evaporation) one organic material onto another. Analogous to semiconductor heterostructures, these junctions can be critical to efficient device design. At the organic/organic heterojunction, there can be discontinuities in material properties such as the conduction and valence energy levels and the carrier mobility. These discontinuities are used, for example, to confine charge carriers to light-emitting layers to enhance LED efficiency. Metal/organic interfaces are very important for organic devices because they determine the type and amount of charge injected into the undoped, insulating organic layer. Metal/organic interfaces are designed to produce good electrical contacts. It is essential that the metal/organic Schottky energy barriers be made small enough to make good electrical contacts. If the Schottky energy barriers were pinned, as usually happens in inorganic semiconductors, it would be impossible to produce small energy barriers for efficient injection of both carrier types. In some cases, there is effectively a weak interaction between the metal and the organic at the interface. For these interfaces, the electronic structure of the interface components is not strongly modified and it is possible to produce good electrical contacts for electrons by using low work function metals and for holes by using high work function metals. There is strong evidence of chemical reactions at some metal/organic interfaces. 118-121 However, when an interfacial chemical reaction occurs, it does not necessarily introduce states in the energy gap, formed from the rc orbitals, that would pin the Schottky barrier. States associated with tr bonding are usually in a different energy region than the rc orbitals and are unlikely to influence the energy gap region. A disruption of the rc bonding could produce states in the energy gap, but more often leads to an alternate rc bonding scheme without making a state in the energy gap. The electronic structure of organic interfaces is thus completely different from that of interfaces involving inorganic 11s T. Kugler, A. Johansson, I. Dalsegg, U. Gelius, and W. R. Salanceck, Synthetic Metals 91, 143 (1997). 119 C. Fredriksson and J. L. Bredas, J. Phys. Chem. 98, 4253 (1993). 12o p. Dannetun, M. Fahlman, C. Fauquet, K. Kaerijama, Y. Sonoda, R. Lazzaroni, J. L. Bredas, and W. R. Salaneck, Synthetic Metals 67, 133 (1994). 121 R. Lazzaroni, J. L. Bredas, P. Dannetun, C. Fredriksson, S. Stafstrom, and W. R. Salaneck, Electrochimica Acta 39, 235 (1994).

24

I.H. CAMPBELL AND D. L. SMITH

semiconductors, and qualitatively different interface electrical properties are observed.

III. Metal/Organic Interface Electronic Structure Organic electronic devices consist of undoped, insulating thin films of conjugated organic materials into which charged carriers are injected from metallic electrodes. The operation of the devices is determined by the charge injection properties of the metal/organic interface and the electrical properties of electron and hole polarons in the organic film. One of the most basic questions concerning the electronic structure of a metal/organic interface is the energy required to inject electrons and holes from the metal contact into the organic material; that is, the difference between the Fermi energy of the metal contact and the energies of the electron and hole polaron states of the organic material. These energy differences are called the electron and hole Schottky energy barriers in analogy with the corresponding injection barriers at metal/semiconductor contacts. In inorganic semiconductors, such as Si and GaAs, Schottky energy barriers are weakly dependent on the type of metal contact for a given semiconductor; i.e., the Schottky energy barriers for various metals on Si are similar. 74 This is not the case at metal/organic interfaces; 88'122-124 there is a qualitative difference in the observed behavior of Schottky energy barriers in conjugated organic materials compared with inorganic semiconductors. Indeed, the essential operating principle of organic diodes is based on the asymmetry in the Schottky energy barriers of the two contacts making up the structure. Organic diode structures are fully depleted metal/organic/metal structures. At zero bias there is a built-in potential in these structures equal to the difference between the electron (or hole) Schottky energy barriers of the two metal contacts. 88'122-124 There is an electric field in the organic layer that is equal to this built-in potential divided by the thickness of the organic film. Measurements of built-in potentials, in combination with Schottky energy barrier measurements for a few specific metals, provide a method to determine the Schottky energy barriers for a wide range of metal contacts. Schottky energy barriers and built-in potentials play an important role in determining light-emitting diode and field-effect transistor characteristics, as discussed in Section V, and it is important to measure them directly to understand device performance. 122 X. Wei, S. A. Jeglinski, and Z. V. Vardeny, Synthetic Metals 85, 1215 (1997). 123 G. G. Malliaras, J. R. Salem, P. J. Brock, and J. C. Scott, J. Appl. Phys. 84, 1583 (1998). 124 I. D. Parker, J. AppL Phys. 75, 1656 (1994).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

25

The charged states of organic thin films, e.g. electron and hole polarons, have not been as widely studied as neutral excited states, and the energies of charged states in organic thin films are just becoming established. Optical measurements directly probe charge-neutral excited states, such as excitons, but charged states are only indirectly probed by optical spectroscopy and other experimental approaches must be used to determine their properties. For example, it is not straightforward to determine the single-particle energy gap of a conjugated organic material by optical absorption spectroscopy because the exciton binding energy can be significant and the excitons do not have a hydrogenic spectrum in which the threshold for absorption into the continuum states can be easily recognized. 2'5~ However, Schottky energy barrier and built-in potential measurements in device structures can be used to determine the electronic energy structure of organic films. The single-particle energy gap, analogous to the energy bandgap in inorganic semiconductors, is the energy difference between the electronic ground state and a state consisting of an electron polaron and a hole polaron that are spatially separated. It can also be viewed as the sum of the energies required to (1) remove an electron from the material, leaving a hole polaron and putting the electron in a specific reference state, and (2) take an electron from the same reference state and put it into the organic material to form an electron polaron. If the reference state is the Fermi energy of a metal contact, these two energies are the electron and hole Schottky energy barriers and thus, the single-particle energy gap is the sum of the electron and hole Schottky barriers of a metal contact. The charge transfer energy gap, which can be different from the single-particle energy gap if bipolarons are strongly bound, can be determined from built-in potential measurements. In the absence of interface states that pin the Schottky energy barriers, the maximum built-in potential in an organic structure is a direct measure of the charge transfer energy gap. This section discusses built-in potentials and Schottky energy barriers in metal/organic/metal structures for three representative organic materials: M E H - P P V , pentacene, and Alq. Built-in potential and Schottky energy barrier measurements are presented and the use of these results to determine the electronic energy structure of the organic film is described. The electronic energy structure is then interpreted using molecular electronic structure calculations. Finally, based on these results, an approach to controllably manipulating Schottky barriers at metal/organic interfaces is demonstrated. 125 j. L. Bredas, J. Cornil, and A. J. Heeger, Advanced Materials 8, 447 (1996). 126 j. Cornil, A. J. Heeger, and J. L. Bredas, Chem. Phys. Lett. 272, 463 (1997). 127 E. A. Silinsh and V. Capek, Organic Molecular Crystals, AlP Press, New York (1994).

26

I.H. CAMPBELL AND D. L. SMITH

7. BUILT-IN POTENTIALS IN METAL/ORGANIC/METAL STRUCTURES As demonstrated by capacitance-voltage (C-V) measurements, undoped metal/organic/metal structures can be fabricated in which the organic layer is fully depleted at reverse, zero, and small forward bias. 49'128'129 This means that the electric charge in the material is small enough that it does not significantly perturb the electric field in the structure and, therefore, the electric field is spatially uniform across the device. At zero bias, there is a built-in potential (Vbi) in these structures equal to the difference between the electron (or hole) Schottky energy barriers of the two metal contacts. There is an electric field in the organic layer that is equal to this buit-in potential divided by the thickness of the organic film. These fully depleted structures can be used to make measurements that determine the built-in potentials and Schottky energy barriers for different metals. Figure III.1 (top) shows an energy level diagram of a metal/organic/metal structure with asymmetric metal contacts in thermal equilibrium. ~3~ The vertical axis is energy and the horizontal axis is position. The slanting solid lines (slanting dashed lines) on the top and bottom represent the formation energies for negatively and positively charged polarons (bipolarons), respectively. The solid horizontal line represents the spatially constant electrochemical potential. The ovals on each side of the diagram represent the energy levels of possible, localized interface states. The formation energies are functions of position because of the electrostatic potential change across the structure. There is a built-in electrostatic potential (Vbi) in the device equal to the difference between the electron (tke) (or hole [tkh]) Schottky energy barriers of the two metal contacts, i.e. Vb~ = tk~,l- 4~e,r or, equivalently, ~bh,r -- ~bh,l. The single-particle energy gap is the sum of the electron and hole Schottky energy barriers of a single contact, e.g. ~be,l + tkh,1. In thermal equilibrium, the electrochemical potential is constant across the structure. The electrochemical potential can be divided into the sum of two pieces, the electrostatic potential and the chemical potential. By measuring the built-in electrostatic potential change across the structure at equilibrium (Vbi), one can determine the change in chemical potential. The charge transfer energy gap between the bipolaron levels is the maximum change in chemical potential that can be supported by the organic material. If the chemical potential goes above the electron or below the hole bipolaron formation energy per particle, these intrinsic states will charge to 128 I. H. Campbell, D. L. Smith, and J. P. Ferraris, Appl. Phys. Lett. 66, 3030 (1995). 129 y. F. Li, J. Gao, G. Yu, Y. Cao, and A. J. Heeger, Chem. Phys. Lett. 287, 83 (1998). 13o I. H. Campbell, P. S. Davids, J. P. Ferraris, T. W. Hagler, C. M. Heller, A. Saxena, and D.L. Smith, Synthetic Metals 80, 105 (1996).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

27

FIG. III.1. Energy level diagrams of an asymmetric metal/organic-film/metal structure at equilibrium. Energy is on the vertical and position is on the horizontal axis. The solid lines represent the formation energy of charged polarons and the dashed lines represent the formation energies per particle of charged bipolarons. The energy levels are shown including the built-in electrostatic potential (top) and with the built-in electrostatic potential subtracted (bottom). The Schottky energy barriers for electrons (tke) and holes (~bh) metal contacts are indicated (top). The empty (filled) ovals represent the energies of electron (hole) trap interface states. (Reprinted from Ref. 130, copyright 1996, with permission from Elsevier Science.)

a high density. Figure III.1 (bottom) shows the same energy level diagram when the effect of the electrostatic potential has been subtracted. The slanting solid line is the spatially varying chemical potential. The net change in chemical potential across the structure is equal to the built-in electrostatic potential. The chemical potential cannot be pushed above the formation energy per particle of the lowest energy intrinsic negatively charged excitation or below the formation energy per particle of the highest energy intrinsic positively charged excitation. Neither can the chemical potential be pushed above a high density of interface states that can charge negatively or below a high density of interface states that can charge positively.

28

I . H . C A M P B E L L AND D. L. SMITH I

!

> .m

t-

-

jJ

.

S

O

n

9 ~176. . . . . . .

S

9 Bm

t-

o aB

"3" .i.=, .m

f

j e~

J

oB oo ~

S

0

1

2

Work Function Difference (eV)

FIG. 111.2. Calculated built-in potential as a function of metal work function difference for an organic film with a single-particle energy gap of 2.4 eV; weakly bound bipolarons and no traps (solid line), bipolarons with a 0.5 eV binding energy and no traps (dashed line), and weakly bound bipolarons with 1020cm-3 electron traps in the middle of the single-particle gap (dotted line) (from Ref. 130).

If there is no state charging, the built-in potential across a metal/organic/ metal structure will be the difference between the work functions of the two metals. A smaller built-in potential implies that charging has occurred. A smaller built-in potential could be due to charging of either intrinsic states of the organic material or interface states. If small built-in potentials not related to intrinsic charged excitation energies are observed, it is a clear indication that interface state charging is the limiting process. For the organic materials used for electronic devices, the built-in potentials are often found to scale directly with the metal work functions over a wide range and built-in potentials nearly as large as the single-particle energy gap are observed. A model describing built-in potentials in metal/organic/metal structures, limited by charged intrinsic excitations and not by interface states, is presented in Ref. [87]. Polarons and bipolarons are the important charged excitations. The model includes possible extrinsic charged states such as those introduced into M E H - P P V by doping with C6o. Fig. III.2 shows the calculated built-in potential across a 50-nm organic film sandwiched between 2 metal contacts as a function of the difference in work function between the metals forming the contacts. 87 The single-particle energy gap is 2.4 eV. The work function of one of the contacts coincides with the negative polaron formation energy and the work function of the other contact is varied. The calculations are for room temperature. The solid line is for negligible bipolaron binding energy and no traps, the dashed line is for a

PHYSICS OF ORGANIC ELECTRONIC DEVICES

29

0.5-eV bipolaron binding energy and no traps, and the dotted line is for negligible bipolaron binding and 1.10 20 cm -3 electron accepting traps at an energy midway between the negatively and positively charged polaron levels. Because of thermal occupation of the polaron levels, essentially no built-in potential occurs until the work function drops to about 0.2 eV below the negative polaron formation energy. The built-in potential then increases approximately linearly with work function until it saturates at about 2.1 V. The saturation value of the built-in potential is smaller than the single-particle energy gap because of thermal occupancy of the polaron levels. When bipolarons with a significant binding energy are included, the maximum built-in potential decreases by slightly less than the bipolaron binding energy. The inclusion of a high density of trapping sites, such as occurs by C6o doping of MEH-PPV, also decreases the maximum built-in potential. These theoretical results provide a reference point for the experimental results presented following. 8. BUILT-IN POTENTIAL AND SCHOTTKY ENERGY BARRIER MEASUREMENT TECHNIQUES Capacitance-voltage measurements show that the charge density in the organic films can be low enough that the electric field across the bulk of the film is uniform. The size of the uniform electric field, and thus the built-in potential, can be measured by electroabsorption. The electroabsorption response of the organic film at a given photon energy is proportional to the imaginary part of the nonlinear susceptibility, Imz~3)(hv), and the square of the electric field A~(hv) ~:

-AT T (hv) oc Imzt3)(hv)E 2

(3.1)

where ~ is the absorption coefficient, hv is the photon energy, T is the transmission, and E is the electric field. TM In the experiment, the electric field consists of a DC component and an applied AC component E = Edc + Eac cos(~t)

(3.2)

and the electroabsorption response is -AT ---T--(hv) oc Imzt3)(hv)(E2c(1 + cos(2f~t))/2 + 2EacEdc cos(f~t) + E2c)

(3.3)

where Edc is the DC electric field, Eae is the amplitude, and f~ is the angular 131 D. E. Aspnes and J. Rowe, Phys. Rev. BS, 4022 (1972).

30

I.H. CAMPBELL AND D. L. SMITH

frequency of the applied AC electric field. In the presence of a DC electric field, the electroabsorption response is modulated at both the fundamental and the second harmonic frequency of the applied AC bias. The response at the fundamental frequency is -AT

(hv; ~) ~ Imz(3)(hv)(2E acEdc) COS(~t)

(3.4)

and the response at the second harmonic of the AC frequency is -AT (hv; 2f~) ~ Imz~3)(hv)(E2c/2)cos (2f~t) T

(3.5)

If there is no DC electric field then the response is modulated only at the second harmonic frequency of the applied AC bias. The DC field consists of the electric field from the equilibrium built-in potential and an applied DC bias. The size of the built-in potential can be determined by measuring the ratio of the electroabsorption response at the fundamental and at the second harmonic frequency for a known AC bias and no applied DC bias. The function describing the material optical properties Imz~3)(hv) is the same at both the fundamental and the second harmonic frequencies and divides out when the ratio is taken. The size of the built-in potential can also be found by applying an external DC bias and monitoring the electroabsorption signal at the fundamental frequency of the AC bias to determine the applied DC bias required to cancel the built-in potential. In the second method, it is necessary that significant charge injection not occur at the bias voltage necessary to cancel the built-in potential. This is usually the case. The first approach is more versatile than the second and can be used to measure, for example, electric fields in multilayer devices under high carrier injection conditions. The second method is easier to apply and more accurate when it is appropriate. The two approaches give consistent results when they can both be used. To use electroabsorption to measure the built-in potential, the material must have a sufficiently large nonlinear response; i.e., Imx~3)(hv) must be large enough to yield significant absorption changes. For some organic molecules used in devices, such as Alq, this condition is not satisfied. For these materials, a second approach, measuring the photocurrent as a function of bias, can be used to determine the built-in potential. In this approach the photocurrent resulting from above gap optical absorption is measured as a function of bias on the structure. The photocurrent signal changes sign when the applied bias reverses the sign of the electric field in the material and vanishes when the bias cancels the equilibrium built-in field. Because there are significant injection currents under bias, modulation

PHYSICS OF ORGANIC ELECTRONIC DEVICES

31

techniques must be used to distinguish the photocurrent. When both electroabsorption and photocurrent vs bias techniques can be used, they give consistent results. Internal photoemission directly measures individual Schottky energy barriers 132 rather than just the difference between Schottky barriers as determined from built-in potential measurements. However, internal photoemission in organic materials works for a limited range of Schottky barrier values. In internal photoemission, optically generated hot electrons (or holes) in a metal contact yield a photocurrent as they traverse a metal/ insulator interface. The photocurrent yield is Yield oz (hv

-

t~s) 2

(3.6)

where 4~s is the Schottky barrier and hv is the photon energy. 132 The Schottky barrier is determined by extrapolating the photocurrent yield to zero as a function of photon energy. In fully depleted organic materials internal photoemission can be used to determine both electron and hole Schottky energy barriers in the same structure by changing the bias direction. This is in contrast to inorganic semiconductors, where an n-type sample must be used to find the electron Schottky barrier and a p-type sample must be used to find the hole Schottky barrier. In one bias direction, the photocurrent is due to electrons excited over the electron Schottky barrier whereas in the other bias direction, the photocurrent is due to holes excited over the hole Schottky barrier. The single-particle energy gap is the sum of the electron and hole Schottky barriers. Because these two barriers can be measured in the same device, the single-particle energy gap can be determined from measurements on a single device structure. This minimizes problems with irreproducible device fabrication. There is a limit to the range of the Schottky barriers that can be measured using internal photoemission. If the Schottky barrier is too large, it is necessary to use such high-energy photons that photocurrent from absorption in the low-energy tail of the organic material absorption spectrum becomes a problem. If the Schottky barrier is too small, injection currents become a problem. Thus only Schottky barriers near the center of the single-particle energy gap can be determined using internal photoemission. The photocurrent thresholds determined in the presence of an electric field are smaller than the zero electric field Schottky barriers because of the image charge potential created when an electron or hole leaves the metal. The electric field lowers photocurrent thresholds by A~bs = e(ee/e) 1/2

(3.7)

132 R. Williams, in Injection Phenomena (R. K. Willardson and A. C. Beer, eds.), Academic Press, New York, (1970), Chapter 2.

32

I.H. CAMPBELL AND D. L. SMITH

where E is the electric field in the organic material and e is its static dielectric constant. 132 It is therefore important to know the electric field in the device structure, which is directly determined by built-in potential measurements. The electric field dependence of the photoemission threshold can be investigated by making measurements as a function of bias voltage. These measured results are found to agree with Eq. 111.7 after accounting for the built-in electric field and using a value for the dielectric constant independently determined from capacitance measurements. 9. BUILT-IN POTENTIAL AND SCHOTTKY ENERGY BARRIER MEASUREMENT RESULTS This section presents the results of internal photoemission and built-in potential measurements of metal/organic Schottky energy barriers in three representative organic electronic materials: MEH-PPV, pentacene, and Alq. In addition to determining the Schottky energy barriers to charge injection, which are critical for device applications, these measurements are also used to determine the energies of the fundamental charged states in these organic materials. The MEH-PPV results are presented in detail to serve as a model for the analysis of the other materials. The results for pentacene and Alq molecules are then summarized in light of the discussion of the MEH-PPV results. Electroabsorption measurements of C6o doped MEH-PPV are presented to illustrate the effects of extrinsic electronic states within the energy gap of an organic material. a. M E H - P P V

Figure 111.3 shows the electroabsorption spectrum near the absorption edge of an A1/MEH-PPV/A1 structure at the fundamental (upper panel) and second harmonic (lower panel) of the applied AC bias. 13~ The absorption spectrum is shown for comparison. The magnitude of the electroabsorption signal at the fundamental frequency depends on the size of the DC bias, as shown in the upper panel of Fig. 111.3, whereas the magnitude of the electroabsorption signal at the second harmonic frequency does not depend on the size of the DC bias. The spectral shape is the same at both the fundamental and second harmonic frequencies. Because there is no built-in potential in this symmetric structure, the electroabsorption signal at the fundamental frequency vanishes when no DC bias is applied.

133 I. H. Campbell, J. P. Ferraris, T. W. Hagler, M. D. Joswick,I. D. Parker, and D.L. Smith, Polymers for Advanced Technologies 8, 417 (1997).

PHYSICS OF ORGANIC ELECTRONIC DEVICES 1.0

0.4

0.5

0.3

0.0

0.2

-0.5 "3

0.1 ;~

-1.0 ~, 1.0 [-

0.0 ~. 0.4

i

33

o"

t~

0.5

0.3

0.0

0.2

-0.5 0.1 -1.0 1.8

2.0

2.2

2.4

2.6

0.0 2.8

Energy (eV) FIG. 111.3. Electroabsorption signal as a function of photon energy for an A1/MEH-PPV/A1 structure measured at the fundamental (upper panel) and at the second harmonic frequency (lower panel) of the applied AC bias. The three curves in the upper panel are for different values of an applied DC bias. The electroabsorption signal at the second harmonic frequency does not depend on DC bias. The absorption spectrum is shown in both panels for comparison (from Ref. 130).

Figure III.4 (upper panel) is a plot of the magnitude of the electroabsorption signal as a function of DC bias for a series of metal/MEH-PPV/A1 structures. 8a'la~ The electroabsorption signal is nulled at a voltage corresponding to the bias necessary to cancel the internal electric field produced by the different metal contacts. The signal is nulled when the applied bias equals the built-in potential, V b i . The DC bias is referenced to the A1 contact, i.e. the Au/A1 structure is nulled when the A1 is biased - 1 V relative to the Au contact. The bias required to null the electroabsorption signal corresponds closely to the difference between the metal work functions except for the saturation that occurs in the Sm-A1 structure. (The metal work functions are listed in Fig. III.7.) Figure III.4 (lower panel) is an analogous plot of the magnitude of the electroabsorption signal as a function of DC bias for a series of metal/MEH-PPV/Ca structures. 88'13~ The bias is referenced to the metal contact, i.e. the A1-Ca structure is nulled when the A1 is biased 1.3 V relative to the Ca contact. The bias required to null the electroabsorption signal corresponds closely to the difference

34

I.H. CAMPBELL AND D. L. SMITH

FIG. 111.4. Magnitude of the electroabsorption response at a photon energy of 2.1 eV as a function of bias for metal/MEH-PPV/A1 structures (upper panel) and metal/MEH-PPV/Ca structures (lower panel) (from Ref. 88).

between the metal work functions except for the saturation that occurs in the Pt-Ca structure. The Pt-Ca structure requires 2.1 V to cancel the built-in electrostatic potential. Thus, MEH-PPV can support a change in chemical potential at least as large as 2.1 eV at room temperature so that the separation between the formation energy per particle of the lowest energy intrinsic negatively charged excitation and of the highest energy intrinsic positively charged excitation must be larger than 2.1 eV. For an organic material with a single-particle energy gap of 2.4 eV (typical of MEH-PPV as discussed following) and no charged excitations with lower energy than the polarons, one expects saturation of the built-in potential at about 2.1 eV (see Fig. III.2). These results demonstrate that bipolarons are not strongly bound in MEH-PPV. 88 Figure III.5 shows the result of internal photoemission measurements in which the square root of the photocurrent yield is plotted as a function of photon energy for an A1/MEH-PPV/Ca structure in reverse bias. 88 In

PHYSICS OF ORGANIC ELECTRONIC DEVICES

35

FIG. III.5. Internal photoemission response as a function of photon energy for an A1/MEHPPV/Ca structure biased to collect electrons. The inset shows the electric field dependence of the photoresponse threshold (from Ref. 88).

reverse bias, the photocurrent is due to electrons excited over the electron Schottky energy barrier depicted in the inset in Fig. III.5. The solid line is a least squares fit to the photocurrent that extrapolates to 1.0 eV. Figure III.6 is a plot of the square root of the photocurrent as a function of photon energy for the same A1/MEH-PPV/Ca structure as in Fig. III.5, and a C u / M E H - P P V / C a structure, both in forward bias. 88 In forward bias, the photocurrent is due to holes excited over the hole Schottky energy barrier as depicted in the inset in Fig. III.6. The solid lines are least squares fits to the photocurrents that extrapolate to 1.1 eV and 0.80 eV for A1 and Cu, respectively. The electric field dependence, taking account of the built-in potential in the structures, of the photoemission thresholds shows the expected behavior from Eq. III.7, as shown in the inset of Fig. III.5. The

FIG. 111.6.Internal photoemission response as a function of photon energy for an A1/MEHPPV/Ca structure and a Cu/MEH-PPV/Ca structure biased to collect holes (from Ref. 88).

36

I.H. CAMPBELL AND D. L. SMITH Work Function Sm 2.7 Ca 2 . 9 A1 Ag Cu Au Pt

4.3 4.3 ~ 4.65.1 5.6 - -

MEH-PPV 2.90

5.30

Fermi Energy Sm, Ca 3.0 AI --Ag Cu --Au Pt

4.3 4.3 4.6 5.1 5.2

Fio. III.7. Energy level diagram for MEH-PPV and a series of metal contacts deduced from the electroabsorption and internal photoemission measurements. The line at 2.9 eV (5.3 eV) corresponds to electron (hole) polarons in MEH-PPV. The measured Fermi energies of the metals in contact with MEH-PPV are shown on the right. The work functions of the metals are shown on the left (from Ref. 88).

zero field Schottky barriers are A1 electrons 1.2 eV, A1 holes 1.2 eV, and Cu holes 0.9 eV. The single-particle energy gap of M E H - P P V , determined by the measured electron and hole Schottky barriers on the A1 structure, is 2.4 eV. 88 The single-particle energy gap is about 0.2 eV larger than the absorption threshold of 2.2 eV. The energy difference of 0.2 eV between the single-particle energy gap and the absorption threshold is the exciton binding energy. 88'134 The electroabsorption and photoemission results provide a consistent picture of the electronic structure of M E H - P P V . Figure 111.7 is an electronic energy diagram of M E H - P P V derived from these measurements. The uncertainty in energy values is a b o u t 4-0.1 eV. The metal contacts to M E H - P P V are accurately described by the ideal Schottky picture, in which the electron (hole) Schottky barrier is determined by the energy difference between the work function of the metal and the electron (hole) p o l a r o n level of the material. The metal work functions listed in Fig. 111.7 were measured in situ using Kelvin probe techniques. The Kelvin probe measurements were relative to (111) Au that was taken to be 5.3 eV; 135 the Au films used for contacts were polycrystalline and had a slightly smaller work function. The Kelvin probe results were close to the standard literature values. 135 The electron (hole) polaron level is at 2.9 eV (5.3 eV). The charged bipolaron binding energies are less than 0.1 eV.

134 S. F. Alvarado, P. F. Seidler, D. G. Lidzey, and D. D. C. Bradley, Phys. Rev. Lett. 81, 1082 (1998). 135 H. B. Michaelson, in CRC Handbook of Chemistry and Physics (R. C. Weast and M. J. Astle, eds.) CRC Press, Boca Raton, Florida (1982), p. E-79.

PHYSICS OF ORGANIC ELECTRONIC DEVICES

>~" 9

1

o,..~

o

'

'

'

'

C'a

37

'

~-1 -1 0 1 Work Function Difference (eV)

FIG. 111.8. Calculated (solid line) and measured (points) potential difference across metal/ MEH-PPV/A1 structures as a function of the work function difference of the contacts (from Ref. 88).

Figure III.8 shows the calculated and the measured built-in potential as a function of the work function difference with respect to A1. The calculations used the energy levels of Fig. III.7 and the model of Ref. [87]. Figure III.9 is a similar plot with the work function difference referenced to Ca. The calculated results are not sensitive to bipolarons for bipolaron binding energies less than 0.1 eV. There is good agreement between the measured and calculated results, indicating that the energy level scheme provides a quantitative description of both the internal photoemission and electroabsorption results. C6o-doped MEH-PPV structures were investigated to show the effect of an extrinsic trapping level. C6o introduces an electron trap below the electron polaron level in MEH-PPV, but does not add charged carriers.

~ 0 :

'

,

~ -1 -

.

,

.

-'1

'

AI

,,-.4

"==-2 -~

'

-

2

'

0

Work Function Difference (eV) FIG. III.9. Calculated (solid line) and measured (points) potential difference across metal/ MEH-PPV/Ca structures as a function of the work function difference of the contacts (from Ref. 88).

38

I.H. CAMPBELL AND D. L. SMITH

FIG. III.10. Schematic representation of the built-in potentials in undoped (a) and C6o-doped (b) MEH-PPV. The upper panels show the relative alignment of the electron and hole polaron energy levels of MEH-PPV, the electron acceptor level of C6o, and the work functions of Pt and Ca metals before the metals and the polymer are in contact. The lower panels show the built-in potential for the structures after contact (from Ref. 136).

Figure 111.10 is a series of schematic energy level diagrams indicating the effect of C60 doping on the built-in potential in MEH-PPV. 63'136-138 The upper panel in Fig. III.10 shows the relative alignment of the electron and hole polaron energy levels of MEH-PPV and the work functions of Pt and Ca before the metals and the polymer are in contact. The Pt/MEH-PPV/Ca structure after contact is shown in the lower panel of Fig. III.10. After contact, there is a built-in potential in the polymer film, Vbi, only slightly smaller than the energy gap of MEH-PPV. Figure III.10 is analogous to Fig. III.10 except for C60-doped MEH-PPV. Before contact, the relative alignment of the energy levels is the same with the addition of the C60 electron acceptor energy level that lies within the energy gap of MEH-PPV. The energy separation between the hole polaron level and the C60 electron accepting level is labeled AC60. After contact, there is a built-in potential in the structure slightly smaller than AC60. In this case, the chemical potential at the Ca contact is pinned near the electron acceptor level of C60 and the Pt contact is pinned near the hole polaron level of MEH-PPV as before. 136 C. M. Heller, I. H. Campbell, D. L. Smith, N. N. Barashkov, and J. P. Ferraris, J. Appl.

Phys. BS1, 3227 (1997). 137 N. S. Sariciftci and A. J. Heeger, Synthetic Metals 70, 1349 (1995). 138 E. Maniloff, D. Vacar, D. McBranch, H. L. Wang, B. Mattes, and A. J. Heeger, Synthetic Metals 84, 547 (1997).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

39

FI~. III.11. Magnitude of the electroabsorption signal at a photon energy of 2.1 eV as a function of external DC bias for metal/MEH-PPV/Ca structures (upper panel) and for metal/C6o-doped MEH-PPV/Ca structures (lower panel) (from Ref. 136).

By measuring the built-in potential in structures employing metal contacts that are not pinned at the hole polaron level of M E H - P P V it is possible to determine the energy difference between the C6o electron acceptor level and the work function of the metals. Figure III.11 shows the magnitude of the electroabsorption signal as a function of DC bias for both metal/MEH-PPV/Ca and metal/C6o-doped M E H - P P V / C a structures. 136 For all of the C6o-doped structures, the built-in potential is reduced by about 0.6 eV compared to the results for undoped MEH-PPV. It is clear that the chemical potential is pinned at the low work function Ca contact because changing the higher work function metal changes the built-in potential demonstrating that the high work function contact is not pinned. Figure III.12 shows the calculated built-in potential across both an undoped and a C6o-doped M E H - P P V film sandwiched between two metal contacts as a function of the difference in work function between the high work function metal and the fixed Ca contact. 136 The experimental built-in

40

I.H. CAMPBELL AND D. L. SMITH I

I

I

/.

2.0-

~

I .~..~

I

pIt

1.5M/MEH-PPV/Ca

..."" Au

~

_

oo Z.......( "O.0

0.5

""~M/MEHPPV+C60/Ca

1.0

1.5

2.0

2.5

WorkFunctionDifference (WM-Wc~)(eV) FIG. 111.12.Calculated and measured built-in potential as a function of metal work function difference for undoped and C6o-doped MEH-PPV. The calculated built-in potential for undoped (upper solid line) and C6o-doped (lower dashed line) MEH-PPV are in good agreement with the measured built-in fields for undoped (squares) and for C6o-doped (diamonds) metal/polymer/Ca structures (from Ref. 136).

potentials are also shown in Fig. III.12. The electron and hole polaron energy levels of M E H - P P V and the work functions of the metals used in the calculation are those shown in Fig. III.7. The C6o molecular density was 4.1019 cm -3. The model describes the data well for a C6o acceptor energy level 1.7 eV above the M E H - P P V hole polaron level. F o r M E H - P P V , there are no intrinsic charged excitations in an energy range almost as large as the single-particle energy gap; that is, between the formation energy of negatively and positively charged polarons. F o r the metals investigated, the size of the built-in potential in M E H - P P V films is not limited by charged interface states. That is, interface electron traps, which could charge negatively, are not introduced significantly below the metal work function and interface hole traps, which could charge positively, are not introduced significantly above the metal work function when the metal work function is in the single-particle gap. This observation does not imply that there is no chemical interaction between the M E H - P P V and the metals, and there is evidence that chemical reactions do occur, li8'139'14~ but 139 T. Kugler, W. R. Salaneck, H. Rost, and A. B. Holmes, Chem. Phys. Lett. 310, 391 (1999).

PHYSICS OF ORGANIC ELECTRONIC DEVICES .

1.0

.

.

41

.

\\

0.5

08'0

| 0.5

| 1.0 1.5 Bias (V)

Pt/ea 2.0

2.5

FIG. 111.13. Magnitude of the electroabsorption signal at a photon energy of 1.8 eV as a function of external DC bias for metal/pentacene/Ca structures.

rather that interface states which limit the built-in potential are not generated by the reactions. C6o-doping introduces an electron accepting state in the single-particle energy gap of MEH-PPV. This state limits the built-in potential in C 6 o - d o p e d MEH-PPV films and therefore it can be identified by built-in potential measurements. The work function of some metals, such as Pt, lie outside of the single-particle energy gap of MEHPPV. The built-in potential of a device structure consisting of an MEH-PPV film with one of the electrodes being Pt will therefore saturate and be less than the difference in work functions of the two contacts. b. Pentacene

Figure III.13 shows the electroabsorption signal as a function of DC bias voltage for three metal/pentacene/metal structures with increasing built-in potentials. The built-in potential is about 1.3 V for A1/Ca and about 1.8 V for both Au/Ca and Pt/Ca structures. Figure III.14 shows a plot of the square root of the measured photocurrent yield for the electron Schottky barrier of Ca (Ca top contact) and the hole barrier of A1 (A1 top contact) for pentacene. The measurements were performed with a bias applied to the test structures producing an electric field of about 2 x 10 5 V/cm in the pentacene layer. The zero electric field Schottky barriers are about 0.6 eV for electrons from Ca and about 0.6 eV for holes from A1. The energy level diagram for metal/pentacene contacts deduced from the internal photoemission and electroabsorption measurements is shown in Fig. III.15. The work functions of the metals are shown on the left and the 14o I. G. Hill, A. Rajagopal, A. Kahn, and Y. Hu, Appl. Phys. Lett. 73, 662 (1998).

42

I.H. CAMPBELL AND D. L. SMITH

1.0

Ca/Ca

/ /

0.5 r~

o I t"

0.0 1.0

I

"+

A1/AI

,,/*

O O

0.5 ~r 0.0

d

"+

,

0.5

I

1.0

1.5

Photon Energy (eV) FIG. 111.14. Internal photoemission response as a function of photon energy for a Ca/ pentacene/Ca structure biased to collect electrons (upper panel) and an A1/pentacene/A1 structure biased to collect holes (lower panel).

Work Function

Ca

2.9

A1

4.3

Au Pt

5.1 5.6

Pentacene

2.4

Fermi Energy

~

Ca3.0 .... A14.3 Au, Pt 4.8

4.9

FIG. 111.15. Energy level diagram for pentacene and a series of metal contacts deduced from the electroabsorption and internal photoemission measurements. The line at 2.4 eV (4.9 eV) corresponds to electron (hole) polarons in pentacene. The measured Fermi energies of the metals in contact with pentacene are shown on the right. The work functions of the metals are shown on the left.

PHYSICS OF ORGANIC ELECTRONIC DEVICES

43

1.0 Sm/Ca t "~ 0.5

Ca 0.0 -0.5

0.0

0.5

1.0 1.5 Bias (V)

2.0

2.5

FIG. 111.16. Magnitude of the photocurrent signal as a function of bias voltage at a photon energy of 2.7 eV for metal/Alq/metal structures. The photocurrent signal goes to zero when the applied bias cancels the built-in potential (from Ref. 141).

measured energy positions of the metals in contact with pentacene are shown on the right. The ionization potential (hole polaron level) of pentacene is about 4.9 eV, as determined from UPS measurements. 127 From the internal photoemission and photocurrent vs bias measurements, the electron polaron level is about 2.4 eV and the hole polaron level is about 4.9 eV; the energy gap is about 2.5 eV. The absorption threshold is 1.7 eV and therefore the exciton binding energy is about 0.8 eV. No evidence of bipolarons is observed, as expected for small molecules. c. Alq

The electroabsorption signal in Alq is small and photocurrent as a function of bias voltage was used to determine the built-in potential for a series of metal/Alq/metal structures. Figure III.16 shows the AC photocurrent signal as a function of DC bias voltage for five structures with increasing built-in potentials. 141 The built-in potential in the structure is the bias at which the photocurrent is a minimum. The built-in potential is near zero for Ca/Ca and Sm/Ca structures, about 0.4 V for A1/Ca, about 1.6 V for Au/Mg, and about 2 V in Pt/Ca structures. The built-in potential of 2 V in Pt/Ca structures shows that the difference between the electron Schottky barriers of Pt and Ca is 2 eV. Figure 111.17 is a plot of the square root of the measured photocurrent yield for electron Schottky barriers of Ca, Mg, and A1 (using Ca top contacts) and a similar plot for the hole Schottky barriers of Cu and Au 141 I. H. Campbell and D. L. Smith, Appl. Phys Lett. 74, 561 (1999).

44

I.H. CAMPBELL AND D. L. SMITH ~" 1.0

!

!

w

M g / f e a

Elect~On

! Ban'ier

~0.5

0.0 1.0 O

~

0.5

r~ 0.0

i

0.5

I

I

,f

I

1.0 1.5 Photon Energy (eV)

!

2.0

FI6. III.17. Internal photoemission response as a function of photon energy for metal/Alq/ Ca structures biased to collect electrons (upper panel) and metal/Alq/A1 structure biased to collect holes (lower panel) (from Ref. 141).

(using A1 top contacts). 141 The measurements were performed with a bias applied to the test structures producing an electric field of about 2 x 10 5 V/cm in the Alq layer. The solid line is a least-squares fit to the data that extrapolates to electron injection thresholds of about 0.5 eV for Ca and Mg, and 0.9 eV for A1, and hole injection thresholds of 0.7 eV and 1.3 eV for Au and Cu, respectively. The image charge potential lowers the extrapolated threshold in the films by about 0.1 eV for each case. The energy level diagram for metal/Alq contacts deduced from the internal photoemission and photocurrent vs bias measurements is shown in Fig. 111.18. The work functions of the metals are shown on the left and the measured energy positions of the metals in contact with Alq are shown on the right. The ionization potential (hole polaron level) of Alq is about 6.0 eV as determined from ultraviolet photoemission measurements. 142'143 F r o m the internal photoemission and photocurrent vs bias measurements, the electron polaron level is about 3.0 eV and the hole polaron level is about 6.0 eV; the energy gap is about 3.0 eV. The optical absorption threshold of Alq is about 2.7 eV 43 and therefore the exciton binding energy is roughly 142 A. Schmidt, M. L. Anderson, and N. R. Armstrong, J. Appl. Phys. 78, 5619 (1995). 143 A. Rajagopal, C. I. Wu, and A. Kahn, J. Appl. Phys. 83, 2649 (1998).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

Work Function Sm 2.7 e V ~ Ca 2.9 eV Mg 3.6 e V ~ A14.3 eV Cu 4.6 e V ~

Fermi Energy Ec 3.0 eV Ca, Sm, Mg 3.6 eV A14.0 eV Alq3

Au 5.1 e V ~ Pt 5.6 eV ~

45

Cu 4.6 eV Au 5.2 eV

Ev 6.0 eV

Pt 5.6 eV

Fro. 111.18. Energy level diagram for Alq and a series of metal contacts deduced from the photocurrent and internal photoemission measurements. The line at 3.0 eV (6.0 eV) corresponds to electron (hole) polarons in Alq. The measured Fermi energies of the metals in contact with Alq are shown on the right. The work functions of the metals are shown on the left (from Ref. 141).

0.3 eV. The three low work function metals investigated, Sm, Ca, and Mg, all had an electron Schottky barrier of about 0.6 eV. In contrast to MEH-PPV, the electron Schottky barrier in Alq is pinned for these low work function metals. For metals with work functions larger than Mg the ideal Schottky model provides a generally accurate description of the energy barrier. The smallest barrier to electron injection was about 0.6 eV for Ca, Mg, and Sm, and the smallest barrier to hole injection was about 0.4 eV for Pt. No evidence of bipolarons is observed, as expected for small molecules.

10. SOLID STATE AND MOLECULAR PROPERTIES

The built-in potential and internal photoemission measurements give important information about the energies of the excited and charged states in the organic materials. Specifically the exciton binding energies of MEH-PPV, pentacene, and Alq were found to be 0.2, 0.8, and 0.3 eV, respectively; and charged bipolarons in MEH-PPV are weakly bound, if at all. The exciton binding energy of pentacene was previously known from optical measurements of charge transfer states, 3 but this approach could not be applied to MEH-PPV or Alq because the corresponding charge transfer states do not appear in their optical spectra. The device results agree with the earlier optical result for pentacene. Previous estimates of the exciton binding energy of MEH-PPV varied over a wide range, from essentially zero to well over 1 eV. The disagreement was partially the result of different definitions. In some

46

I.H. CAMPBELL AND D. L. SMITH

theoretical discussions the exciton binding energy was taken as a measure of the importance of correlation effects in the molecule. To estimate exciton binding energies theoretically, it is necessary to correctly describe both the exciton, a neutral excited state, and a state consisting of electron and hole polarons spatially separated, essentially two charged states. Single molecule calculations can describe the exciton state reasonably accurately without large corrections from solid state effects. However, there are large polarization effects to the energies of charged states that must be added to molecular calculations to describe the charged states. When these polarization effects are taken into account, the measured charged state energies can be described and the measured exciton binding energy explained. This point has been illustrated in a series of calculations on Alq. In Ref. [-113], a hybrid density-functional-theory approach was used to calculate the ground state electronic properties and a time-dependent density-functional-theory approach was used to investigate the excited state electronic properties of the Alq molecule. The calculated molecular results were compared with measurements on dense solid state films of Alq. The molecular calculations describe the optical absorption spectrum near the fundamental absorption threshold without significant corrections from solid state effects, but large dielectric corrections must be included for the molecular calculations to describe the measured ionization potential and single-particle energy gap. When these dielectric corrections are made, using the calculated molecular polarizability, both the measured ionization potential and single-particle energy gap are well described. Vertical excitation energies and oscillator strengths were calculated for the first ten excited singlet states of Alq. The optical absorption spectrum of a thin solid film of Alq is shown in Fig. III.19.113 For comparison, graphical representations of the calculated singlet excitation energies and oscillator strengths are also shown in Fig. III.19. Overall, the agreement between calculation and experiment is very good for the low-energy transitions. The calculated onset of absorption is at 2.77 eV, in close agreement with the observed onset. The most intense transitions are underestimated by a few tenths of an eV. Above about 3.5 eV the agreement with experiment is not as satisfactory, probably due to a basis set artifact. Overall, the fundamental absorption edge in the dense solid state film is well described by the molecular calculations. The calculated molecular ionization potential (IPm) is 6.60 eV and the calculated molecular electron affinity (EArn) is 0.83 eV for vertical transitions in which the molecular geometry of the ground state is also used for the ions. The calculated structural relaxation energies are 0.09 eV for the positive ion and 0.11 eV for the negative ion. The static polarizability was found to be basically isotropic with ~ - 327 a.u. The experimental solid state ionization potential of Alq is between 5.6 and

PHYSICS OF ORGANIC ELECTRONIC DEVICES

47

1.0

0.5 0

< 0.0 2.0

!

2.5

3.0

3.5

4.0

Photon Energy (eV) FIG. 111.19. The near gap optical absorption spectrum of Alq. The vertical lines represent the energies and oscillator strengths calculated using time-dependent DFT (from Ref. 113).

6.0 eV, 142-145 nearly 1 eV smaller than the calculated molecular ionization potential. The measured solid state energy gap, 3.0 eV, is more than 2 eV smaller than the computed difference between the molecular ionization potential and electron affinity, I P m - E A m = 5.8 eV. These differences are due to the additional stabilization associated with the charged states in the solid state film resulting from polarization of the neighboring molecules and from structural relaxation. Polarization stabilizes both the positive ion, decreasing the ionization potential of the solid, and the negative ion, increasing the electron affinity. Such polarization corrections are much less important for the neutral excited states that appear in optical property calculations. The solid state ionization potential and single-particle energy gap of the Alq film can be calculated using the theoretical electron affinity and ionization potential of the molecule and correcting for solid state polarization effects and structural relaxation. The permanent dipole moments of the various Alq molecules are oriented so that their interaction energy with the localized charge cancels and on average makes no net contribution to the stabilization energy. This interaction of the charge with the permanent dipole moments leads to an inhomogeneous broadening, but not a net shift, of the energy distribution. The induced dipoles are oriented by the localized charge and are all directed toward (or away from) the charge and have the 144 y. Hamada, T. Sano, M. Fujita, T. Fujii, Y. Nishio, and K. Shibata, Jpn. J. Appl. Phys. 32, L514 (1993). 145 M. Probst and R. Haight, Appl. Phys. Lett. 71, 202 (1997).

48

I.H. CAMPBELL AND D. L. SMITH

same sign energy contribution. The stabilization energy due to interaction with the induced dipoles is 1

e~ i

Epolar : ~ ~ di "r~.2

(3.8)

Z

where d~ is the induced dipole moment at molecule i and r~ is the intermolecular distance. To estimate this induced polarization correction we enclose the charged Alq molecule of interest in a spherical cavity also containing the ten nearest Alq molecules in the x-ray crystal structure. The radius of this cavity is r o = 1.1 nm. The polarization energy from the ten nearest neighbor molecules is calculated directly. The additional stabilization associated with the remaining molecules is treated using a continuum approximation and the Clausius-Mossotti relation. The stabilization energy is then ~e2

e2 ( e - l )

(3.9)

where the sum is over the ten neighbor molecules and the second term is the contribution from the more distant molecules. The solid state ionization potential and single-particle energy gap are then IP s = IP m - -

Epola

Eg = ( I P m - E A m )

r --

Estruct(h

)

- 2Epola r - Estruct(e ) - Estruet(h )

(3.10) (3.11)

where Estruet(e,h ) is the structural relaxation energy for the electron (hole) polaron. The sum above, to the ten nearest molecules, gives a contribution to the stabilization energy of 0.70 eV, and the continuum contribution is 0.44 eV. The total polarization correction for each ion is 1.14 eV. The calculated structural relaxation for the hole polaron is 0.09 eV. The solid state ionization potential is then 6.60 eV - 1.14 eV - 0.09 eV = 5.37 eV, in reasonable agreement with experimental values ranging from 5.6 to 6.0 eV. The sum of the calculated structural relaxation energies is 0.20 eV. The energy gap is then 6.60 e V - 0.83 e V - (2 • 1.14 e V ) - 0 . 2 eV = 3.29 eV. This is in reasonable agreement with the experimental result of 3.0 eV + 0.2 eV. These results show that single molecule calculations can describe localized neutral excitations in solid state films reasonably well, but that large polarization corrections are necessary to describe charged states in solid state films using single molecule calculations. When these corrections are included, the measured ionization potential, energy gap, and exciton binding energy are reasonably described. m

PHYSICS OF ORGANIC ELECTRONIC DEVICES

49

11. MANIPULATING SCHOTTKY ENERGY BARRIERS USING DIPOLE LAYERS The Schottky energy barriers between a metal and a conjugated organic material are important parameters for device operation. Small Schottky barriers are required for efficient electrical injection. For some materials, such as MEH-PPV, small barriers can be reached using common metals for electrodes, although reactive low work function metals such as Ca are needed to achieve small electron barriers. For other materials such as Alq, small energy barriers cannot be reached using common metals for the electrodes. It would be useful to controllably manipulate Schottky barriers so that small barriers can be reached on all materials and the use of reactive metals can be avoided. 146-149 Self-assembled monolayer (SAM) techniques can be used to attach a monolayer of polar molecules to the surface of a metal. Because of the ordering inherent in SAM structures, the molecular dipoles are oriented relative to the metal surface. The metal work function can therefore be controlled using the oriented SAM dipole layer. Because the Schottky model holds for many organic materials, Schottky barriers can also be controlled using the SAM dipole layer. The schematic energy level diagrams shown in Fig. 111.20 illustrate the basic idea. 146'147 Figure III.20a represents the untreated metal/organic interface (i.e., there is no SAM dipole layer on the metal). Figures. III.20b and III.20c show the effect of inserting an oriented dipole layer between the metal and the organic film. In Fig. III.20a, the dipole layer is oriented so that the electron Schottky energy barrier is decreased, and in Fig. III.20b the dipole layer is oriented so that the electron Schottky energy barrier is increased. Fig. III.20d is a magnified view of the interface showing a SAM with an electric field across it representing the effect of the dipole layer. Schottky barrier control was demonstrated using three alkane-thiol adsorbates to form the self-assembled monolayers: CH3(CH2)9SH [CH 3 SAM], NHz(CH2)loSH [ N H 2 SAM], and CF3(CF2)7(CHz)zSH ICE 3 SAM]. The chemical structure of the CH 3 SAM is shown at the top of Fig. III.21; the chemical structure of the other molecules is similar except the end groups are changed to give different dipole moments. 146 When the thiol adsorbate forms a monolayer film on the surface of a column Ib metal, the hydrogen attached to the sulfur in the molecule is removed and the sulfur bonds to the metal. These adsorbates were chosen because their self146 I. H. Campbell, S. Rubin, T. A. Zawodzinski, J. D. Kress, R. L. Martin, D. L. Smith, N. N. Barashkov, and J. P. Ferraris, Phys. Rev. B54, R14321 (1996). 147 I. H. Campbell, J. D. Kress, R. L. Martin, D. L. Smith, N. N. Barashkov, and J. P. Ferraris, Appl. Phys. Lett. 71, 3528 (1997). 148 F. Nuesch, Y. Li, and L. J. Rothberg, Appl. Phys. Lett. 75, 1799 (1999). 149 F. Nuesch, F. Rotzinger, L. SiAhmed and L. Zuppiroli, Chem. Phys. Lett. 288, 861 (1998).

50

I.H. CAMPBELL AND D. L. SMITH

FIG. 111.20.Schematic energy level diagrams of metal/organic interfaces: panel a, untreated interface; panel b (panel c), dipole layer that decreases (increases) the electron Schottky energy barrier; and panel d, magnified view of the interface (from Ref. 146).

assembly properties have been extensively studied and they form dense, well-ordered monolayers, and because the set includes molecules with dipole moments of both signs. Silver was chosen as an electrode because its work function is near the center of the M E H - P P V energy gap so the electron Schottky energy barrier can be either increased or decreased. The change in work function of the chemically treated electrodes was measured with respect to pristine Ag using a Kelvin probe. In the Kelvin probe technique, the substrate metal surface and a metallic, vibrating probe tip constitute the two plates of a capacitor. The vibration of the probe tip induces an AC current. This current is nulled when the voltage applied to the tip is equal to the difference in surface potential between the tip and the substrate. Figure 111.21 shows the Kelvin probe current as a function of relative substrate bias (~rAp p -- VAg) for a pristine Ag electrode and for Ag electrodes modified by the three SAMs. ~46 The difference Vgpp -- FAg repre-

PHYSICS OF ORGANIC ELECTRONIC DEVICES

51

FIG. 111.21. Kelvin probe current as a function of substrate bias for a pristine Ag electrode and for Ag electrodes modified by the three SAMs. The applied substrate bias has been shifted by the difference in the surface potentials of the Kelvin probe tip and the pristine Ag substrate. The molecular structure of the CH3 SAM and its calculated dipole moment is shown above the panel (from Ref. 146).

sents the change in surface potential with respect to the pristine Ag electrode. The CH 3, NH2, and CF 3 SAMs change the surface potential with respect to pristine Ag by - 0 . 7 0 V, -0.45 V and 0.85 V, respectively. The CH 3 and NH 2 SAMs decrease the effective work function and the CF3 SAM increases the effective work function of the Ag electrode. The expected surface potential shift due to a molecular dipole layer has the form

A0 -- N (fl~~ + #Ag+S- )

(3.12)

where N is the areal density of molecules, #mol is the dipole moment of the molecule, #Ag+S- is the screened dipole moment of the Ag+S- bond, and e is a static dielectric constant. For these SAMs, N takes on values between about 3 and 5.1014 cm -2 and e is between 2 and 3. The value of #Ag+S- is difficult to determine precisely, but is expected to be nearly the same for the three SAMs. The dipole moments of the SAMs were calculated using quantum chemistry techniques. The calculated dipole moments were 2.24D, 1.77D, and - 1.69D for the CH3, NH 2, and CF 3 SAMs, respectively. These calculated dipole moments give the observed trends and approximate magnitudes of the surface potential shifts.

52

I.H. CAMPBELL AND D. L. SMITH 1.0 r~

_

/"

"~ 0.5 /

__< 0.0

-1.0 -0.5 0.0 0.5 1.0 Diode Bias (VApp- VAg)

FIG. 111.22. Electroabsorption signal as a function of diode bias for Ag/MEH-PPV/Ca structures with a pristine Ag electrode and for Ag electrodes modified by the three SAMs. The applied substrate bias has been shifted by the built-in potential in the pristine Ag/polymer/Ca structure (from Ref. 146).

To determine the effect of the SAM layers on Schottky barriers, the chemically treated electrodes were incorporated in diode structures with a MEH-PPV layer roughly 50 nm thick and a top Ca contact. Figure III.22 shows the measured electroabsorption signal as a function of diode bias (VApp- VAg) for a pristine Ag electrode and for Ag electrodes modified by the SAMs. ~46 Because the calcium/polymer Schottky energy barrier is constant, the difference, VApp- VAg, represents the change in electron Schottky energy barrier with respect to the pristine Ag electrode. The CH 3, NH 2, and CF 3 SAMs change the electron Schottky energy barrier by - 0 . 6 0 V, -0.45 V, and 0.50 V, respectively. These results demonstrate tuning of the Schottky energy barrier of Ag on MEH-PPV in an organic diode structure over a range of more than 1 eV. Figures III.21 and III.22 show that changing the effective work function of the Ag substrate produces a corresponding change in the Ag/polymer electron Schottky energy barrier. The shift of the surface potentials seen in Fig. III.21 is essentially the same as that of the Schottky energy barriers seen in Fig. III.22, except for the CF 3 SAM. The effective work function of the Ag/CF a SAM film is greater than the ionization potential of MEH-PPV so the electron Schottky energy barrier saturates at a shift of 0.50 V. These alkane-thiol molecules have large energy gaps that block charge injection. As a result, the diodes with alkane-thiol modified contacts did not have favorable injection properties even though, in some cases, they had small Schottky barriers. Conjugated-thiol SAMs, with modest energy gaps, can improve charge injection. The conjugated-thiol-based SAMs both modify the metal/organic Schottky energy barrier and are sufficiently transparent to electrons to allow improved charge injection. This point is

PHYSICS OF ORGANIC ELECTRONIC DEVICES

,,s

53

F

"~19o ~, 0.06 ~" ~o.~

;01_00

9~

I

[

eo.op 0"000

"

1.o

15

20

DiodeBias (V) /

5

/

/

/

/

10

15

Diode Bias (V) FIG. 111.23.Current-voltage characteristics of Cu/MEH-PPV/Ca structures for a pristine Cu electrode and for Cu electrodes modified by the two conjugated SAMs. The fluorinated SAM improves charge injection and the non-fluorinated SAM degrades charge injection. The inset shows electroabsorption measurements of the device built-in potentials; the F SAM increases the built-in potential and the H SAM decreases the built-in potential consistent with the Kelvin probe measurements. The molecular structure of the fluorinated SAM is shown above the panel (from Ref. 147).

emphasized in Fig. III.20d, which shows the electron and hole energy levels of two different SAMs represented by the pairs of solid and dashed lines. The large energy gap alkane-thiol blocks charge injection, while the smaller energy gap conjugated-thiol does not significantly impede electron transfer. Two thiol-adsorbates were used to form conjugated self-assembled monolayers: H S ( C 6 H 4 C 2 ) 2 C 6 H 4 - F [F SAM] and H S ( C 6 H 4 C 2 ) 2 C 6 H 4 - H [H SAM]. The chemical structure of the F SAM adsorbate is shown at the top of Fig. III.23; the H SAM chemical structure is equivalent except the fluorine atom is replaced by hydrogen. ~47 These adsorbates were chosen because they are known to form dense, well-ordered monolayers and the two molecules shift the Schottky energy barrier in opposite directions. The Cu electrode has a work function within the energy gap of MEH-PPV, which allows the hole Schottky energy barrier to be either increased or decreased, to either improve or degrade charge injection. The inset in Fig. III.23 shows the measured electroabsorption signal as a function of diode bias for a pristine Cu electrode and for Cu electrodes modified by the two S A M s . 147 The built-in potential of the pristine Cu/polymer/Ca structure is about 1.5 V; i.e., the electroabsorption signal is zero at 1.5 V. The built-in potential increases to about 1.7 V for the F SAM structures and decreases to about 1.2 V for the H SAM structures. The difference in the built-in

54

I.H. CAMPBELL AND D. L. SMITH

potentials represents the change in hole Schottky energy barrier. Figure 111.23 shows the current density as a function of diode bias for a pristine Cu/polymer/Ca diode and for analogous diodes employing Cu electrodes modified by the two S A M s . 147 The structure with the F SAM has substantially higher current than the pristine Cu structure for a given voltage and, similarly, the current is considerably reduced in the H SAM structure. Because the calcium electrode and the polymer thickness is the same for each diode, these current-voltage characteristics are representative of the change in hole injection from the Cu electrodes. They demonstrate significant improvement in hole injection for the fluorinated SAM and, similarly, a substantial decrease in hole injection for the non-fluorinated SAM. These results demonstrate control and improvement of charge injection in organic electronic devices by utilizing self-assembled monolayers to manipulate the Schottky energy barrier between a metal electrode and the organic electronic material.

IV. Electrical Transport Properties The charge injection and carrier transport properties of organic materials govern the electrical characteristics of organic electronic devices. Without charge injection, conjugated organic materials have negligible intrinsic carrier concentrations and very high room temperature resistivity. In these insulating, disordered organic films, the carrier mobility is dominated by hopping transport between localized molecular sites. 3'4 Although organic materials can be electronically doped, the dopant ions significantly modify the intrinsic properties of the organic material, making carrier mobilities determined in doped materials largely irrelevant to the undoped films used in devices. 38 Therefore, measurements of the mobility must be performed on undoped, insulating films. Conventional Hall effect measurements have not been useful for determining carrier mobilities in these insulating organic materials and photo-Hall measurements are complicated by short carrier lifetimes and relatively large exciton binding energies. Two general approaches are used to measure the carrier mobility: (1) measuring the transit time of optically injected carriers across thin films 4'~5~ ~52 xso I. H. Campbell, D. L. Smith, C. J. Neef, and J. P. Ferraris, Appl. Phys. Lett, 74, 2809 (1999). 151 M. Redecker, D. D. C. Bradley, M. Inbasekaran, and E. P. Woo, Appl. Phys. Lett. 73, 1565 (1998). 151 M. Redecker, D. D. C. Bradley, M. Inbasekaran, and E. P. Woo, Appl. Phys. Lett. 74, 1400 (1999).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

55

and (2) fitting the current-voltage characteristics of devices. 9'24'68'15 3 The electronic structure of conjugated organic thin films consists of a distribution of localized electronic states with different energies. The site energy distributions are believed to be approximately Gaussian with standard deviations typically between 0.1 eV and 0.2 eV. 4 The localized sites are either individual molecules or isolated conjugation segments of a polymer chain. Electrical transport results from carrier hopping between neighboring sites. At room temperature, equilibration between neighboring sites of different energy is generally fast enough that carrier transport can be described using a mobility picture. 4 However, at lower temperatures or for systems with very large disorder, the mobility description may not be valid. 4 Hopping transport in a disordered system leads to a mobility that can depend strongly on both electric field and carrier density. 4 The mobility increases with increasing electric field and with higher carrier densities. As the electric field increases, more states become available for energetically favorable hopping transitions, thus increasing the mobility. At high carrier densities, the mobility is increased because charge transport occurs predominantly in a region with a higher density of states and therefore increased number of energetically favorable hopping sites. Organic diodes and organic field-effect transistors depend on transport normal to and parallel to the plane of the film, respectively. Because the molecular packing in organic thin films is often asymmetrical, particularly for polycrystalline and polymer films, they are likely to have different mobilities normal to and parallel to the plane of the film. 162 Organic diodes and FETs also operate in distinct electric field and charge density regimes. Organic diodes typically operate at electric fields of several times 10 5 V/cm and at carrier densities of up to a few 1017 cm-3.163 In contrast, field-effect transistors operate at lateral electric fields of a few 104 V/cm and at carrier -

161

153 G. G. MaUiaras, J. R. Salem, P. J. Brock, and C. Scott, Phys. Rev. B58, R13411 (1998). 154 p. E. Burrows, Z. Shen, V. Bulovic, D. M. McCarthy, S. R. Forrest, J. A. Cronin, and M. E. Thompson, J. Appl. Phys. 79, 7991 (1996). 155 A. J. Campbell, D. D. C. Bradley, and D. G. Lidzey, J. Appl. Phys. 82, 6326 (1997). 156 A. J. Campbell, M. S. Weaver, D. G. Lidzey, and D. D. C. Bradley, J. Appl. Phys. 84, 6737 (1998). 157 p. W. M. Blom, M. J. M. DeJong, and J. M. Vleggaar, Appl. Phys. Lett. 68, 3308 (1996). 158 p. W. M. Blom, M. J. M. DeJong, and M. G. Van Munster, Phys. Rev. B55, R656 (1997). 159 H. C. F. Martens, H. B. Brom, and P. W. M. Blom, Phys. Rev. B60, R8489 (1999). 16o L. Bozano, S. A. Carter, J. C. Scott, G. G. Malliaras, and P. J. Brock, Appl. Phys. Lett. 74, 1132 (1999). 161 M. A. Lampert and P. Mark, Current Injection in Solids, Academic, New York (1970). 162 C. Y. Yang, F. Hide, M. A. DiazGarcia, A. J. Heeger, and Y. Cao, Polymer 39, 2299 (1998). 163 B. K. Crone, I. H. Campbell, P. S. Davids, D. L. Smith, C. J. Neef, and J. P. Ferraris, J. Appl. Phys. 86, 5767 (1999).

56

I.H. CAMPBELLAND D. L. SMITH

densities above 10 ~8 c m - 3 . 74 For carrier densities below about 10 ~7 c m -3, typical of diode operation, the mobility does not depend strongly on the carrier density. 76 In this low density limit, interactions between carriers and state filling effects are not significant. In contrast, for carrier densities above about 10 ~8 cm -3, typical of FET operation, occupation of the lower energy hopping sites begins to enhance the carrier mobility. 76 Because of the anisotropy in molecular structure and differences in the operating regimes, the mobilities determined from LED and FET measurements are not necessarily equivalent. Three techniques have been used to determine carrier mobilities in organic electronic materials: time-of-flight (TOF) current transient measurements, 4 fitting of single-carrier space charge limited (SCL) diode currentvoltage (I-V) characteristics, 9'24'68'~53-161 and analysis of field-effect transistor current-voltage characteristics. 9 These three mobility measurement techniques sample different electric field and charge density regimes. The time-of-flight technique can measure the carrier mobility for electric fields from about 5 x 104 V/cm to 1 • 106 V/cm but is restricted to very low volume averaged carrier densities, about 10 ~3 cm -3 or less. Because the TOF technique is restricted to very low carrier densities, it is difficult to measure the carrier mobility in the presence of extrinsic trap states. Fitting single-carrier space charge limited diode current-voltage characteristics can probe the mobility for electric fields from about 10 5 to 106 V/cm and for carrier densities from about 1016 to 10 x8 cm -3. These measurements are much less sensitive to trapping effects than TOF measurements because the comparatively large density of injected carriers can fill moderate densities of traps without significantly perturbing the mobility measurement. In SCL diode measurements, the electric field and carrier density are functions of position within the device structure. Therefore, fitting the measured I-V characteristics requires assumptions about the carrier density and electric field dependences of the mobility. If both TOF, which requires low trap densities, and SCL diode mobility measurements can be performed, they usually yield consistent results. In FET mobility measurements, the lateral electric fields that are responsible for current flow are typically in the 104 V/cm range and the charge is confined to a thin region of the organic film adjacent to the gate insulator. This leads to very high charge carrier densities of about 1019 cm -a. These carrier densities are high enough to significantly modify the mobility due to changes in the occupation of hopping sites. In addition, the results are sensitive to the local molecular structure near the interface that may differ significantly from that typical of bulk films. 164-168 The FET mobilities are often significantly larger than 164

D. D. C. Bradley,M. Grell, A. Grice, A. R. Tajbakhsh, D. F. OBrien,and A. Bleyer, Optical

Materials 9, 1 (1998).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

57

TOF and SCL diode results on the same material. The higher mobility inferred from FET measurements may be due to the different carrier density regimes in which the measurements are made or to variations in the local molecular structure in the two kinds of devices. FET mobility results will be discussed in the transistor part of Section V. The carrier mobilities of different organic materials, as measured by TOF and SCL diode techniques, vary over a large range, from 10 -8 cm2/~s to 10 -2 cm2/Vs. 4 They are orders of magnitude smaller than those in typical inorganic semiconductors. TM The low mobility of the organic materials plays a critical role in determining the characteristics of organic devices. Measured carrier mobilities for films made from a given organic material can differ both because of different processing conditions for the films and because of different synthesis conditions used to make the organic material. 163 It is difficult to compare detailed mobility results between different research groups who may use different processing conditions and materials made under different synthetic conditions. It is possible to standardize processing conditions so that consistent mobility results can be achieved for films fabricated from material made in a given synthetic run. However, especially for conjugated polymers, films fabricated from materials made in different synthetic runs, even when using the same equipment and nominally the same approach for the synthesis, often give somewhat different mobility results. The electrical properties of the polymer film are sensitive to the film morphology, which can be affected by small changes in the polymer molecular weight distribution that varies slightly from batch to batch. This section focuses on mobility measurements of three representative conjugated organic materials: MEH-PPV, Alq, and pentacene. Carrier mobilities determined by TOF and SCL techniques are compared. The electric field dependence of the mobilities is discussed and theoretical models used to interpret the measurements are described.

2. TIME-OF-FLIGHT MOBILITY MEASUREMENTS

Time-of-flight is an established technique to measure carrier mobilities in insulating materials. In this technique, a semitransparent blocking contact/ insulating film/blocking contact structure is used. An optical pulse incident 165 S. Guha, W. Graupner, R. Resel, M. Chandrasekhar, H. R. Chandrasekhar, R. Glaser, and G. Leising, Phys. Rev. Lett. 82, 3625 (1999). 166 L. Athouel, R. Resel, N. Koch, F. Meghdadi, G. Froyer, and G. Leising, Synthetic Metals 101, 627 (1999). 167 p. A. Lane, M. Liess, X. Weis, J. Partee, J. Shinar, A. J. Frank, and Z. V. Vardeny, Chemical Physics 227, 57 (1998). 168 H. J. Schon, C. Kloc, R. A. Laudise, and B. Batlogg, Appl. Phys. Lett. 73, 3574 (1998).

58

I.H. CAMPBELL AND D. L. SMITH

on the material through the semitransparent contact creates a thin sheet of electron-hole pairs next to that contact and, depending on the sign of the applied bias, electrons or holes are driven across the sample. The absorption depth of the optical excitation must be small compared to the film thickness and the optical pulse duration must be short compared to the transit time of the charged carriers across the sample. Low-intensity optical pulses are used so that the photogenerated charge carrier density does not significantly perturb the spatially uniform electric field in the structure. The carrier mobility, #, is determined from the measured carrier transient time, z, by d2

# =

zV

(4.1)

where d is the film thickness and V is the applied voltage. The structures used for TOF measurements consisted of a thin, semitransparent A1 contact on a glass substrate, an organic film a few #m thick, and a top, thick A1 contact. A nitrogen laser pumped dye laser producing 500 ps pulses was tuned to the peak of the absorption coefficient for the organic material so that the absorption depth of the optical pulse was much smaller than the film thickness. The bandwidth of the current preamplifier was two orders of magnitude greater than the reciprocal of the transit time. The product of the structure capacitance and amplifier input impedance was at least two orders of magnitude smaller than the transit time. The total charge injected into the film was about 0.01 CV, where C is the capacitance of the structure and V the applied voltage. Figure IV.1 is a log-log plot of the TOF hole current density in MEH-PPV as a function of time after optical excitation for applied biases of 10 V, 40 V, and 100 V at room temperature. 15~ The transit time was determined by the intersection of the asymptotes to the plateau and declining slope of the current transient. The MEH-PPV film was 1.8 #m thick. Figure IV.2 shows the hole mobility as a function of electric field determined from the TOF measurements (markers) and a least-squares fit (solid line) to the Poole-Frenkel form # = #o exp

(4.2)

where E is the electric field and #o and E o are parameters describing the mobility. The Poole-Frenkel form for the electric field dependent mobility is frequently observed in organic molecular solids and polymers 4. Figure IV.2 shows that the Poole-Frenkel form describes the measured TOF results reasonably well. The fit to the TOF data yielded the parameters #o = 2.1 x 10-7 cm2/Vs and E o = 8.7 x 104 V/cm. The TOF current transi-

PHYSICS OF ORGANIC ELECTRONIC

' ' ' ' I

'

'

'

'

' ' ' ' I

'

'

'

DEVICES

'

' ' ' ' I

59

'

IOV

10.8

10-9

.... , 10 4

........

i lj

i

i

!

, 10 -3

i1 i

........

i i i i

i

I

I

, 10 -a i

i

,

i i ij

i

40v .,..~ c~

10-7

r.)

I||l

l

I

I

I

i

fill

104 I

l

i

i

i

i

,Ill

10 .4

'

'

'

'

'

'''I

i

i

i

i

i

llll

l

10 .3

'

'

'

'

'

~''I

l

l

i

,

i

,l,l

'

'

'

10 -5

10"6

104

10"4

Time (s) FI6. IV.1. Time-of-flight hole current transients for M E H - P P V at three applied voltages. The structure was semitransparent A1 (10 n m ) / M E H - P P V (1.8 (m)/A1 (100 nm) (from Ref. 150).

ents are close to the dispersive regime at the largest electric fields 4. In linear-linear current transient plots it becomes difficult to distinguish the current plateau at electric fields above about 4 x 105 V/cm. The error bars on the mobility shown in Fig. IV.2 are estimated from T O F measurements on several different devices. It was not possible to measure hole T O F transients at significantly lower temperatures. At 250K the T O F plateau and falling edge could no longer be clearly distinguished over a significant range of electric fields. Neither was it possible to measure electron T O F transients. Space charge limited diode I-V measurements, discussed following, show that the electron mobility of M E H - P P V is much smaller than the hole mobility, and trapping effects may be significant for electrons.

60

I . H . C A M P B E L L A N D D. L. SMITH

3.0 2.5 ~~ 2.0 ,?

o

1.5

;~ 1.0 o

0.5 ,

0.O

,,

I

I

1

2

~

I

I

4

5

6

Electric Field (105 Wcm) FIG. IV.2. The hole mobility of M E H - P P V as a function of electric field derived from T O F measurements. The markers are T O F results and the solid line is a least-squares fit of Eq. 4.2 to the T O F results (from Ref. 150).

Figure IV.3 is a log-log plot of the T O F electron current in Alq as a function of time after optical excitation for applied biases of 20 V, 50 V, and 100 V at room temperature. 113 The transit time was determined by the intersection of the asymptotes to the plateau and declining slope of the current transient. Figure IV.4 shows the electron mobility as a function of electric field determined from the T O F measurements (markers) and a least-squares fit (solid line) to the mobility assuming the Poole-Frenkel form. Figure IV.4 shows that the Poole-Frenkel form describes the measured T O F results reasonably well. The fit to the T O F data yielded /t o = 1.5 • 10-8cm2/Vs and E o = 1.5 • 104 V/cm. It was not possible to measure the hole mobility in Alq using the T O F technique. Significant hole trapping that obscures the current transients is widely observed in Alq. 169 Figure IV.5 shows the measured T O F hole mobility of pentacene at room temperature as a function of electric field (markers). The mobility is essentially independent of field over the range of electric fields considered. The mean value of the T O F data yielded # = 1.2 x 10 -3 cm2/Vs. It was not possible to measure the electron mobility in pentacene using the T O F technique in these films. 169 R. G. Kepler, P. M. Beeson, S. J. Jacobs, R. A. Anderson, M. B. Sinclair, V. S. Valencia, and P. A. Cahill, Appl. Phys. Lett. 66, 3618 (1995).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

.

.

.

.

.

.

.

.

|

.

.

.

.

.

.

.

.

|

61

*

0.1

10 .4

10 -2

10 .3

iov

. . . . . . . . . . . . . . . . . . .

r,r

0.1

0.01 104

2

4

68

2

4

68

10-4

2

4

2

4

10-3

0.1

2

10 -6

4

68

2

4

10-5

68

10 -4

Time (s) Fro. IV.3. Time-of-flight electron current transients for Alq at three applied voltages. The structure was semitransparent A1 (10 nm)/Alq (2 (m)/A1 (100 nm) (from Ref. 113).

These TOF mobility results are typical of organic electronic materials with a low density of charge carriers in the film. Strong field dependence and low mobility, typical of hopping conductivity in a broad density of states, is observed for both holes in MEH-PPV and electrons in Alq. For more ordered systems, the mobility is higher and less strongly field dependent. For example, the pentacene hole mobility is significantly larger and has much weaker field dependence than the MEH-PPV hole and the Alq electron mobility.

62

I.H. CAMPBELL AND D. L. SMITH

5 O

4

~,O

3

2 1

0 0

1

2

!

!

3

4

5

Electric Field (105V/cm) FIG. IV.4. The electron mobility of Alq as a function of electric field derived from TOF measurements. The markers are TOF results and the solid line is a least-squares fit of Eq. 4.2 to the TOF results (from Ref. 113).

2.0

O

1.5

O

1.0

Z~ O

0.5

~

!

!

2

3

Electric Field (105V/cm) FIG. IV.5. The hole mobility of pentacene as a function of electric field derived from TOF measurements.

PHYSICS OF ORGANIC ELECTRONIC DEVICES

63

13. MOBILITY FROM SINGLE-CARRIER S C L DIODE I-V CHARACTERISTICS

The sensitivity of the time-of-flight technique to small densities of extrinsic traps frequently prevents its use to measure carrier mobilities. This problem can be overcome by using single-carrier diode current-voltage characteristics in the space charge limited current flow regime. To demonstrate the validity of this technique, we compare measured and calculated currentvoltage characteristics of MEH-PPV and pentacene hole only devices using the TOF hole mobility measurements presented preceding. The independently measured hole mobilities were used, without adjustable parameters, to calculate the current-voltage characteristics of device structures with space charge limited hole current. The I-V characteristics were described using the device model of Ref. [170]. For the SCL contacts, the model reduces to a numerical evaluation of space charge limited current with a field dependent mobility that includes carrier drift and diffusion (the diffusion component is small). Figure IV.6 shows measured (solid) and calculated (dashed) currentvoltage characteristics for a Pt/MEH-PPV/A1 structure. 15~ The TOF mobility was used and there were no adjustable parameters in the calculation. 17o p. S. Davids, I. H. Campbell, and D. L. Smith, J. Appl. Phys. 82, 6319 (1997).

10 -3

PffA1 10 -4

~

10 4

~ 10_6 10 -7

10-8

0

20

40

10

! 80

1 100

Bias (V) FIG. IV.6. Measured (solid) and calculated (dashed) current-voltage characteristics for a Pt (10 nm)/MEH-PPV (2 #m)/A1 (100 nm) structure. Positive bias corresponds to space charge limited hole injection from Pt. The calculation used the fit to the TOF mobility shown in Fig. IV.2 without adjustable parameters (from Ref. 150).

64

I.H. CAMPBELL AND D. L. SMITH 0.25 0.20

r

~

,.

0.15

~176/ ___13.13fl/ 0

,

1

-

,

,

,

2

3

4

5

Bias (V) FIG. IV.7. Measured (solid) and calculated (dashed) current-voltage characteristics for two Pt/pentacene/Ca structures. Positive bias corresponds to space charge limited hole injection from Pt. The calculations used the TOF mobility shown in Fig. IV.5 without adjustable parameters.

Positive bias corresponds to space charge limited hole injection from the Pt contact and negligible electron injection from the A1 contact that has a large electron Schottky barrier. There is good agreement between the measured and calculated I-V characteristics over 5 orders of magnitude in current. The current-voltage measurements were made using devices with 2-#m thick M E H - P P V films prepared identically to those used for the T O F measurements to ensure that the microstructures of the organic films were equivalent. 162 The average charge density in the T O F measurement is about 2 orders of magnitude smaller than that in the space charge limited I-V characteristic. At an applied bias of 50 V the average calculated hole density for the diode was about 1015 cm -3, whereas the average hole density in the T O F measurement was about 1013 cm-3. Because the mobility derived from the T O F measurements at low density accurately describes the I-V characteristics at much higher density, trapping effects are not important for holes in this material. The carrier density in these 2-#m thick devices under space charge limited current flow is about 2 orders of magnitude smaller than in typical polymer LEDs that have organic layers about 100 nm thick. Figure IV.7 shows measured (solid) and calculated (dashed) currentvoltage characteristics for 150-nm and 300-nm thick Pt/pentacene/Ca structures. The T O F mobility was used and there were no adjustable parameters in the calculation. Positive bias corresponds to space charge limited hole injection from the Pt contact and negligible electron injection from the Ca

PHYSICS OF ORGANIC ELECTRONIC DEVICES

65

contact that has a large electron Schottky barrier. There is good agreement between the measured and calculated I-V characteristics. The highest charge density in the TOF measurement is about 4 orders of magnitude smaller than that in the space charge limited I-V characteristic. Because the mobility derived from the TOF measurements at low density accurately describes the I-V characteristics at much higher density, trapping effects are not important for holes in this material. The results for hole dominated diodes made from MEH-PPV and pentacene show that if a mobility is known from TOF measurements it can be used to accurately describe the current-voltage characteristics of singlecarrier SCL diodes. Single-carrier SCL diode current-voltage characteristics are next used to determine the mobility for cases in which trapping interfered with the TOF measurements. To determine the electron mobility of MEH-PPV, for which it was not possible to measure the electron mobility using the time-of-flight technique, a series of electron only structures with space charge limited current flow was measured and fit using an electric field dependent and charge density independent electron mobility. Current-voltage characteristics were measured for a series of Ca/MEHPPV/Ca electron only devices in which the polymer thickness was varied. The current is space charge limited because the energy barrier to injection of electrons from Ca into MEH-PPV is small. Figure IV.8 shows current density versus bias voltage for a thickness series of Ca/Ca electron only devices. 163'171 The experimental results are shown as solid lines and the model results as dashed lines. The same electron mobility parameters, #o = 5 x 10 -12 cm2/Vs, and E o = 1.0x 104 V/cm, were used for all the structures. The model describes current-voltage characteristics for Ca/Ca devices over a range of thicknesses, and over several orders of magnitude of current density. The thickness scaling is n o t V2/L 3 as expected from the analytic result for space charge limited current that does not include the field dependence of the mobility. 161 Figure IV.9 shows the calculated electron density and electric field profiles for the 100 nm Ca/Ca device at a bias such that the current density is 5 x 10 -2 A/cm 2. Electrons are injected from the left contact and collected at the right contact. These profiles, for a single bias point, demonstrate the range over which the electric field and charge density vary in SCL conditions. For single-carrier SCL diodes, the carrier mobility is the main physical quantity that determines the current-voltage characteristics. Of course, the device geometry and the dielectric constant must also be known, but they can be independently determined. Therefore, the carrier mobility can be 171 B. K. Crone, P. S. Davids, I. H. Campbell, and D. L. Smith, Appl. Phys. Lett. 84, 833 (1998).

66

I.H. CAMPBELL AND D. L. SMITH 0.1

I

Ca/Ca

2

m

0.05

o,,J

10 Bias (V)

0

20

10-1

~

0"2

10.3 r,.) 104 10

1

Bias (V) FIG. IV.8. Measured (solid line) and calculated (dashed line) current-voltage characteristics for 25-, 60-, and 100-nm thick Ca/MEH-PPV/Ca electron only devices on linear (upper panel) and log-log (lower panel) scales (from Ref. 163).

determined from the I-V curves of these devices. If SCL contacts cannot be made, as is the case for both electrons and holes in Alq, it is also necessary to know the injection properties of the contact. In the diode part of Section V, it is shown that if the Schottky barrier is known, single-carrier diode I-V curves can be used to find the carrier mobility for non-space charge limited cases.

14. MOBILITY MODELS The Poole-Frenkel form for the field dependence of mobility was first observed in TOF measurements on molecularly doped polymers. 3'4'172,173 172 D. M. Pai, J. Chem. Phys. 52, 2285 (1970). 173 W. D. Gill, J. App. Phys. 43, 5033 (1972).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

67

1020 , '?'~ 1019 ~ ,,,,i ~

~10

Ca/Ca 5x 102A]cm 2

18

~1017 1016 2 Ca/Ca 5xl

~

I

l

1

00

|

!

50

100

Position (nm)

FIG. IV.9. Calculated electron density (upper panel) and electric field (lower panel) profiles for a ll0-nm Ca/MEH-PPV/Ca device biased to provide 5 x 10 -2 A/cm 2 device current density. The electron injecting contact is on the left (from Ref. 163).

These materials consist of isolated molecular dopants in an inert polymer host. The dopant molecules do not introduce charged carriers, i.e they do not dope the material in the usual semiconductor sense of the term, but they provide low-energy sites that the injected carriers occupy. In the molecularly doped polymers, transport results from carrier hopping between the dopant molecules, the host polymer provides a matrix for the active dopant sites. Because similar field dependence is observed in molecularly doped polymers and the solid state films used for electronic devices, it is plausible that the physical mechanisms controlling the electrical transport in these two classes of materials are similar. B~issler and coworkers extensively studied mobility in these types of materials using Monte Carlo simulations of the Gaussian disorder model (GDM).174,175 In the GDM, electrical transport results from carrier hopping between localized sites with the site energies randomly distributed according to a Gaussian distribution. The GDM describes some features of 174 H. Bassler, Phys. Stat. Sol. (b) 175, 15 (1993). 175 D. Hertel, H. Bassler, U. Scherf, and H. H. Horhold, J. Chem. Phys. 110, 9214 (1999).

68

I.H. CAMPBELL AND D. L. SMITH

the observed mobility. However, the magnitude of the electric fields over which strong field dependence was found in the simulations did not correspond well with experiment. The magnitude of the fields at which the G D M showed strong field dependence, qualitatively similar to the experimental observations, was significantly larger than the field regimes at which that behavior was observed experimentally. Gartstein and Conwel1176 showed that a spatially correlated site energy distribution for the carriers can explain the observed field dependence. Physically, strong field dependence occurs when the potential drop ( e E l ) across a relevant length scale (f) for the system becomes comparable to kT. With uncorrelated site energies, the only relevant length scale is the distance between hopping sites. This distance is relatively small and strong field dependence only occurs at high fields. Spatial correlations introduce a new longer length scale, the length over which the site energies are correlated. Thus spatial correlations can lead to strong field dependence of the mobility at lower fields than would occur for uncorrelated site energies. Many of the dopant molecules used in molecularly doped polymers have large permanent electric dipole moments. Dunlap and coworkers 177-181 proposed a model for the mobility of molecularly doped polymers based on the long-range nature of the interaction between charged carriers and the dipole moments of the molecular dopants. In this model the energetic site disorder is the result of different electrostatic potentials at the various sites due to the random distribution in orientation of the dipole moments of the nearby dopant molecules. Because the charge-dipole interaction is long range, sites that are spatially close also have nearly the same energy so that there is a correlation between site position and energy. This model has been successful in describing many aspects of the mobility of molecularly doped polymers. Some of the materials used for organic electronic devices, such as Alq, also have large permanent dipole moments and the charge-dipole interaction model probably applies to them as well as to the molecularly doped polymers for which it was originally designed. 1~3 However, other materials used for organic devices, such as PPV, do not have permanent dipole moments and therefore this model does not appear to apply to them in detail.

y. N. Gartstein and E. M. Conwell, Chem. Phys. Lett. 217, 41 (1994). 177 D. n. Dunlap, P. E. Parris, and V. M. Kenkre, Phys. Rev. Lett. 77, 542 (1996). 178 S. V. Novikov, D. H. Dunlap, V. M. Kenkre, P. E. Parris, and A. V. Vannikov, Phys. Rev. Lett. 81, 4472 (1998). 179 p. E. Parris, M. Kus, D. H. Dunlap, and V. M. Kenkre, Phys. Rev. E56, 5295 (1997). 18o p. E. Parris, D. H. Dunlap, and V. M. Kenkre, J. Polymer Science B35, 2803 (1997). 181 V. M. Kenkre, M. Kus, D. H. Dunlap, and P. E. Parris, Phys. Rev. E58, 99 (1998). 176

PHYSICS OF ORGANIC ELECTRONIC DEVICES

69

A second physical mechanism, fluctuations in molecular geometry such as the phenylene ring-torsion in PPV, also leads to spatially correlated site energies and applies to conjugated materials without permanent dipoles. The spatial energy correlation is the result of strong intermolecular restoring forces for ring-torsion fluctuations in dense films of closely packed molecules. By contrast, the intramolecular restoring force for a ring-torsion is small as seen, for example, in the AM1 calculations for biphenyl shown in Fig. 11.3. For neutral biphenyl, the energy is almost independent of the torsion angle. Because the restoring force is primarily intermolecular, ring-torsions on neighboring molecules tend to move together. If an extra electron or hole is added to the system, the energy of the charged state depends strongly on the torsion angle, as also seen in the AM1 calculations of biphenyl in Fig. 11.3. Thus, there is a strong coupling between the energy of a carrier at a site and the ring orientation at that site. Different sites will have different ring orientations and therefore this coupling leads to disorder in the site energies. Because the rings on near neighbors move together, there is a spatial correlation in the site energies. Details of the site energy density of states and spatial correlation functions for this model are worked out in Ref. [76]. The results are found to be similar to those for the charge-dipole interaction model. Both models give a site energy density of states that is nearly Gaussian and a two-site energy correlation function that falls off as the reciprocal of distance between sites. As a result the field dependence of mobility predicted by the two models is very similar, although the physical origin of the energy disorder is different. There is a difference in expected temperature dependence in the two models because in the charge-dipole interaction model the disorder is independent of temperature whereas in the molecular geometry fluctuation model, the disorder increases with increasing temperature. A 1D Master equation with nearest neighbor hopping that can be used to describe field-dependent mobility has been exactly solved by Derrida. 182 In the continuum limit, the mobility can be calculated using this solution to the Master equation. 177 The result is # =

#o

(4.3)

Y fo dye-~r (e-~(~ where 7 = fleE, E is the electric field, e is the magnitude of the electron charge, e(y) is the site energy at position y, #o is the mobility if all site energies were the same, and fl = 1/kT. Using a Gaussian approximation to 182 B. Derrida, J. Statistical Phys. 31, 433 (1982).

70

I.H. CAMPBELL AND D. L. SMITH

calculate the correlation function and the (l/y) dependence for the correlation function (e(y)e(O)) given by both the charge-dipole interaction and the molecular geometry fluctuation models gives -- fl20-2

#~

(4.4)

# = (2fla~/fleEa)K l(2fla~/fleEa) where cr is essentially the standard deviation of the density of site energies, (l/a) is a momentum cutoff, and Kl(z ) is the first-order modified Bessel function of the third kind. (Slightly different prescriptions were used for cutting off momentum integrals when going to the continuum limit in Refs. [177] and [76], which leads to a small difference in numerical factors for the forms stated in those papers.) In the charge-dipole disorder model the standard deviation of the site energy distribution is c r - ~/eZPZno/12rceZa where P is the dipole moment and n o is the density of dipoles. In the molecular geometry fluctuation model a = v/v2kT/4rtKa where v is the linear coupling constant between site energies and the ring-torsion and K is the angular restoring force constant for the ring-torsion. In the chargedipole interaction model a is temperature independent, whereas in the molecular geometry fluctuation model ~r is proportional to ~ . Using an asymptotic expansion for K~(z), the high field mobility becomes ~77

(4.5)

]2 '~' ~,-112tr2e(2~crx/~eEa)

In this 1D solution both models give the same result for the field-dependence for the mobility, In # ~ x/~; that is, the Poole-Frenkel form for the field dependence. Because a has different temperature dependence in the two models, the mobilities have different temperature dependencies. The 1D results are not generally valid for dense three-dimensional (3D) systems. In 3D, the carriers can take optimal paths to avoid high-energy barriers, whereas in the 1D model there is only one path the carriers can take. The steady state Master equation describing the carrier transport for this system is 0 = ~ Iog~jPj(1 - Pi) - r176

- Pj)I

(4.6)

J Here Pi is the probability for the polaron to be on site i and coo is the polaron hopping rate from site j to site i. Double occupation at a site is excluded. After finding the solution for P~, the average carrier velocity is found from ~'ij ('~ -- Pi)Rij -~ = (4.7) ZjPj where/~j is the position difference between sites j and i, and the mobility is

PHYSICS OF ORGANIC ELECTRONIC DEVICES

10-'

E

71

m

10-5

io f fl

O v

::L ~s

10 -6

B

~B

,

200

I

400El~

~B

i

I

1 600

|

I

800

i

1000

(V ~cm-1~)

FIG. IV.10. Calculated logarithm of mobility as a function of E 1/2 with different polaron-torsion couplings. Solid, dashed, and dot-dashed lines correspond to v = 0.1, 0.2, and 0.3 eV, respectively (from Ref. 76).

found from ~ = #E. The 3D Master equation is in general too complex for analytic solution and two numerical approaches are commonly used, Monte Carlo simulation and direct solution of the Master equation using sparse matrix techniques. Compared with Monte Carlo simulations, the sparse matrix approach has some advantages: it guarantees the steady state solution; it is more convenient for considering density-dependent effects; and it is often numerically more efficient. For cases in which both approaches can be used they give the same result. The field-dependent mobility in the dilute limit can be found by linearizing the Master equation. Figure IV.10 shows a calculation of mobility as a function of electric field for the molecular geometry fluctuation model with three values for the coupling parameter, v, between the ring-torsion and the site energy. The ring-torsion restoring force, K, was chosen to describe the measured hole mobility of MEH-PPV using v = 0.3 eV. The values for these parameters are consistent with quantum chemical estimates using model systems. 76 The curves are reasonably close to linear, showing that the model gives approximately the Poole-Frenkel form. Figure IV.11 shows the density of states for the site energies with the same values of the coupling parameters as in Fig. IV.10. For a system with a stronger coupling and therefore a broader density of site energies, the mobility is low and has strong field dependence, whereas for a system with weak coupling and therefore a narrower density of site energies, the mobility is higher and has weaker field dependence.

72

I . H . C A M P B E L L A N D D. L. S M I T H 16.0

..... ....

12.0

v--O.1 V=0.2 V--0.3

"E to >

%

8.0

qr-" v

v

4.0

0.0 -0.4

-0.2

0

0.2

0.4

r~(ev) FIG. IV.ll. Calculated site density of states as a function of energy for different polaron-torsion couplings. Solid, dashed, and dot-dashed lines correspond to v = 0.1, 0.2, and 0.3 eV, respectively (from Ref. 76).

The results of Fig. IV.10 suggest that there is correlation between the magnitude of the mobility and the strength of the field dependence because stronger energetic disorder leads to both lower magnitude of the mobility and stronger field dependence. Molecular geometry fluctuations, such as ring-torsions, are a source of energetic disorder. If the molecular structure is constrained in such a way that the restoring force for ring-torsions is increased or their coupling to the site energies is reduced, energetic disorder is reduced. As a result, the magnitude of the mobility is enhanced and its field dependence is weakened. Comparing TOF hole mobility measurements in MEH-PPV and poly(9,9-dioctylfluorene) (PFO) shows this correlation. The phenylene rings in an isolated neutral MEH-PPV chain can rotate easily. In PFO, two rings are fixed together by bridging bonds so they can only rotate together. As a result, both the intermolecular restoring force to ring-torsion is increased, because two rings rather than one collide with a neighboring molecule, and the coupling of the ring-torsion to the site energy is reduced, because the charge can more easily delocalize on two rings than on a single ring. Figure IV.12 compares mobility measurements of MEHPPV 15~ and PFO 15~ with calculations based on the molecular geometry fluctuation model. The observed hole mobility in PFO is about two orders higher than that for MEH-PPV and the field dependence is much weaker. This is the expected qualitative behavior. In the calculations, the hole mobility data of MEH-PPV was fit by adjusting the parameter K describing the ring-torsion intermolecular restoring force and the parameter v descri-

PHYSICS OF ORGANIC ELECTRONIC DEVICES

73

bing the strength of the coupling between the ring-torsion and the site energy around the values estimated from AM l-level quantum chemistry calculations of model systems. The value of K was then increased to fit the P F O data. In principle, both an increase in K and a decrease in v should occur when going from MEH-PPV to PFO. But such changes in either of the two parameters give very similar results. Good fits to the mobilities of both materials are achieved for physically reasonable values of the parameters. The mobility can depend on carrier density, because when some carriers fill deep potential sites from which hopping is difficult, the other carriers become more mobile. The density dependence of mobility can be studied by solving the nonlinear Master equation. Figure IV.13 illustrates the carrier density effects on the mobility in the molecular geometry fluctuation model with parameters appropriate for holes in MEH-PPV. The mobility is enhanced by almost one order of magnitude with increase of the carrier density to n = 6.9 x 10 ~a cm-3 at E -~ 4 x 104 V/cm. In the low-field regime, where the field-assisted carrier hopping is less efficient than in the high-field regime, the carrier density effect on mobility is more pronounced. Diode measurements show that at electric fields of a few times 10 6 V/cm there is not a strong carrier density dependence of the mobility in MEH-PPV for densities up to about 10 ~8 cm -3. By contrast, field effect transistor measurements have suggested that the mobility increases strongly with increasing

1 0 -s

,

, ....

1 0 -3

u~ E v

o

lo-'-

1 0 -6

J |

200

|

|

400 600 800 E 1~ (vl/2cm -1/2)

. . . .

200

|

400

,

6;0

800

E 1~ (Vl~cm-l~)

FIG. IV.12. Logarithm of hole mobility as a function of E 1/2. The left panel shows experimental (dots) and calculated (solid line) results for MEH-PPV. The fight panel shows experimental (dots) and calculated (solid line) results for PFO (from Ref. 76).

74

I.H. CAMPBELL AND D. L. SMITH

i

i

9

i ~"

~"

I

.'"

.,"

~

sS

~

/

" t j

jf

9

sS

7f

s s SS

4~ 6

I

~00

*

400

I

600

E~

(V~/~em-le)

,

I

~00

1000

FIG. IV.13. Calculated logarithm of mobility as a function of E 1/2 with different carrier densities. Dotted, short-dashed, long-dashed, and dot-dashed lines correspond to carrier densities n = 0.08, 0.5, 2, and 6.9 x 1018 cm -3, respectively. The solid line shows the results of solving the linearized Master equation (from Ref. 76).

carrier density at low fields for carrier densities above about 10 is c m -3. The calculated results in Fig. IV.13 are consistent with these device measurements and explain why this behavior is expected when the different field/ density regimes are sampled. Figure IV.14 shows the effect of deep traps on the mobility. Traps are randomly distributed with a concentration 2 x 10 ~7 cm-3 and a trap energy level 0.5 eV below the center of the Gaussian site energy density of states. The molecular geometry fluctuation model with parameters appropriate for holes in MEH-PPV is used. Because of the traps, the mobility is small in the low-field regime for small carrier densities. When the carrier density is sufficiently large to saturate the traps, the mobility is enhanced dramatically. These results show how a small density of traps can have a very large effect on TOF measurements, in which the carrier densities are very small, but do not significantly affect single-carrier SCL diode measurements, in which the carrier densities are much larger. In 3D systems, a carrier can optimize its path to avoid high-energy barriers and achieve a higher mobility. Figure IV.15 illustrates current patterns in the low-field and the high-field regions. The molecular geometry fluctuation model with parameters appropriate for holes in M E H - P P V is used. The figure shows a projection of the 3D lattice onto the x - y plane (the field is in the x direction) by summing over the currents in the z direction.

75

PHYSICS OF ORGANIC ELECTRONIC DEVICES

10 -5

r

.

,

9

i

9

,

9

I

,

]

0_6

E ~J"

sf

10 -7 ~

2c~0

,

0

I

40 1~2(V~40m0-~)

,

,000

FIo. IV.14. Calculated logarithm of mobility as a function of E 1/2 with different carrier densities for a system with randomly distributed traps. The trap concentration is 2 x 1017 cm-3 and the trap level is - 0 . 5 eV. Short-dashed, long-dashed, dot-dashed, and dotted lines correspond to carrier densities n = 4.7, 2.4, 1.2, 0.3 x 1017 cm -3, respectively. The solid line shows the results of solving the linearized Master equation without traps (from Ref. 76).

The width of each line in the figure is proportional to the current across the bond. Darker lines indicate that the current is opposite to the standard directions (from left to right and from down to up). In the low-field regime, the carriers take complex paths involving many chains. When such irregular paths occur, a 1D model, where the path is always along the field, is not appropriate. In the high-field regime, where the field is strong enough to overcome the energy barriers, the carrier paths are essentially one-dimensional. Because the energetic disorder is electrostatic in the charge-dipole interaction model, it should be the same, except for a sign reversal, for electrons and holes. Thus the charge-dipole interaction model predicts that the mobility for electrons and holes should have similar electric field dependence. This model should be applicable if the energetic disorder is dominated by the random orientation of the molecular dipoles. For Alq, the molecular dipole moment is large, 5.3 Debye, 113 and the intermolecular spacing is about 1 nm, which leads to disorder with a standard deviation of about 0.1 eV, comparable to the total energetic disorder. Thus, it is likely that dipolar disorder is the major source of energetic disorder in Alq films. Although it is not possible to measure both carrier mobilities of Alq using TOF, the device measurements presented in Section V give similar electron

76

I . H . CAMPBELL AND D. L. SMITH

FIG. IV.15. Current patterns for different applied fields. The width of a line is proportional to the current across the bond. Upper and lower panels are for E = 0.5 x 105 and 2 x 106 V/cm, respectively (from Ref. 76).

and hole mobilities with similar electric field dependence as expected if dipolar disorder is dominant. In MEH-PPV, holes have a higher mobility and a weaker electric field dependence than electrons, suggesting that the energetic disorder is greater for electrons than for holes. Dipolar disorder is not expected to be a major source of the energetic disorder in MEH-PPV films. MEH-PPV has a flexible molecular structure and molecular geometry fluctuations such as ring-torsion should be a significant source of energetic disorder. For the

PHYSICS OF ORGANIC ELECTRONIC DEVICES

77

simple example of the biphenyl molecule discussed in Section II, there is an electron-hole symmetry and the coupling of electron and hole states to ring-torsion is the same. This electron-hole symmetry is broken in more complex molecules such as MEH-PPV and it is not expected that electrons and holes have the same field dependence. The hole mobility in pentacene is comparatively high and essentially independent of electric field. (The electron mobility in pentacene is not well studied.) This is the behavior expected from a material that does not have spatially correlated energetic disorder. Pentacene does not have a permanent dipole moment and it is structurally rigid. As a result, neither charge-dipole interactions nor molecular geometry fluctuations are expected to make major contributions to energetic disorder. Therefore the energetic disorder in pentacene is expected to be comparatively weak and not spatially correlated.

V. Organic Diodes and Field-Effect Transistors A number of organic electronic devices have been proposed, but organic diodes, in particular LEDs, and FETs, have been the most extensively explored. Both of these classes of devices use thin films of undoped insulating materials. Organic diodes are vertical transport devices consisting of an organic film sandwiched between metal contacts. The carriers are transported vertically from one contact to the other across the thickness of the film. Organic FETs are lateral devices consisting of a conducting gate contact, a gate insulator, source and drain electrodes electrically isolated from the gate contact by the gate insulator, and an organic film in contact with the source and drain electrodes. Carriers are transported laterally along the organic film across the gap between source and drain electrodes. For both classes of devices, carriers in the organic film originate from the metallic contacts. Organic diodes are quite different from conventional semiconductor p-n junction diodes. Organic FETs are more nearly similar to inorganic thin-film FETs. Detailed device models have been demonstrated for organic diodes 153-155'157'158'170'171'183'184 but device models specifically for organic FETs are less extensively d e v e l o p e d . 9'73'185-187 p. W. M. Blom, M. J. M. DeJong, and S. Breedijk, Appl. Phys. Lett. 71, 930 (1997). B. K. Crone, P. S. Davids, I. H. Campbell, and D. L. Smith, J. Appl. Phys. 87, 1974 (2000). L. Torsi, A. Dodabalapur, and H. E. Katz, J. Appl. Phys. 78, 1088 (1995). G. Horowitz, R. Hajlaoui, H. Bouchriha, R. Bourguiga, and M. Hajlaoui, Advanced Materials 10, 923 (1998). 187 R. Tecklenburg. G. Pasch, and S. Scheinert, Advanced Materials for Optics and Electronics 8, 285 (1998). 183 184 185 186

78

I.H. CAMPBELL AND D. L. SMITH

The materials used for organic electronic devices are disordered. As a result there is an ensemble of energies for the charged states. The low-energy charged states of device interest, polarons, are localized on individual sites and electrical transport results from polaron hopping between these localized states. At room temperature, equilibration between neighboring sites of different energy is generally fast enough that carrier transport can be described using a mobility picture. Hopping transport in a disordered system leads to a mobility that can depend strongly on both electric field and carrier density. Organic diodes and organic FETs operate in distinct electric field and charge density regimes. Organic diodes operate in a high-field, low-carrier density region whereas the FETs operate in a lowfield, high-carrier density region. For the operating conditions relevant to organic diodes, the carrier density dependence of the mobility can be neglected, but the field dependence must be included. The carrier density dependence of the mobility must be included for the operating conditions relevant to organic FETs. The field dependence of the mobility at the low carrier densities relevant to organic diodes has been determined, for some materials, by a combination of TOF and single-carrier SCL diode currentvoltage measurements. In organic devices, carriers are injected into an undoped film from metal contacts. The Schottky energy barriers between the contact metal Fermi energy and the polaron levels in the organic material have been determined for a variety of cases. To determine interfacial current densities, it is in general also necessary to know transition rates for electrons and holes to cross the metal/organic interface. Reliably determining such transition rates, either by direct measurement or by microscopic theory, is a challenging task.188-193 Fortunately, for most cases of interest the current densities at which the devices are operated are not large enough to drive carrier densities at the metal/organic interface far out of local equilibrium. As a result, carrier injection enters device models as a boundary condition for interface carrier densities with these interface densities determined by local thermodynamic equilibrium. 170 An understanding of the microscopic details of carrier transport at the interface is thus not required for the device model. This is a major simplification that has allowed successful device modeling. In organic LEDs, electrons are injected from one side of the organic film and holes are injected from the other. The electrons and holes form excitons 188 189 19o 191 192 193

V. I. Arkhipov, E. V Emelianova, Y. H. Tak, and H. Bassler, J. Appl. Phys. 84, 848 (1998). U. Wolf, V. I. Arkhipov, and H. Bassler, Phys. Rev. B59, 7507 (1999). V. I. Arkhipov, U. Wolf, and H. Bassler, Phys. Rev. B59, 7514 (1999). U. Wolf, S. Barth, and H. Bassler, Appl. Phys. Lett. 75, 2035 (1999). E. M. Conwell and M. W. Wu, FPhys. Rev. Lett. 70, 1867 (1997). y. N. Gartstein and E. M. Conwell, Chem. Phys. Lett. 255, 93 (1996).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

79

that recombine, perhaps radiatively. For an organic LED device model, it is necessary to include a description of carrier recombination. The recombination process is bimolecular. A Langevin form for the bimolecular recombination coefficient is commonly used. 194 The section first describes a device model for organic diodes. The model is then compared with experimental results on single-layer, single-carrier structures; with single-layer bipolar structures; and with multilayer structures. Experimental results on organic FETs are then discussed. 15. ORGANIC DIODES

a. Device Model The device model 17~ for organic diodes includes charge injection, transport, recombination, and space charge effects in the organic material. It can describe contact limited current, space charge limited current, and cases in between. Because of the geometry of organic diodes, a one-dimensional device model is appropriate. The transport of electrons and holes in the organic film are described by time dependent continuity equations coupled to Poisson's equation On c3t

1 t3J. = G- R e t?x

Op

1 c3Jp

c3t ~ e c3x = G - g t3E 4he t3----x-.........e (p - n)

(5.1)

(5.2) (5.3)

with drift-diffusion forms used for the current densities,

J. = e#,, nE + ~

(5.4)

and

Jp= e#p (pE

kT Op) e ~xx

(5.5)

Here, n (19) is the electron (hole) density, J. (Jp) is the electron (hole) current density, G (R) is the carrier generation (recombination) rate, #. (#p) is the electron (hole) mobility, e is the magnitude of the electron charge, E is the 194 V. N. Abakumov, V. I. Perel, and I. N. Yassievich, in Nonradiative Recombination in

Semiconductors,North-Holland, Amsterdam (1991), p. 108.

80

I.H. CAMPBELL AND D. L. SMITH

electric field, e is the static dielectric constant, k is Boltzmann's constant, T is temperature, x is the position coordinate along the film growth axis, and the diffusivities have been written in terms of the mobilities using the Einstein relation. The carrier mobilities are field dependent in the organic materials used to form organic LEDs and are reasonably well described by the Poole-Frenkel form # = #o exp

(5.6)

The Einstein relation between diffusivity and mobility does not necessarily apply in regions where the mobility has strong field dependence. This question was studied for the specific case of the charge-dipole interaction model for the mobility, discussed in Section IV, in Ref. [180]. The Einstein relation was found to apply at high fields, in slightly modified form, for this model. The same arguments and conclusions apply to the molecular geometry fluctuation model. In practice, the diffusion terms make a very small contribution to the calculated device current. 195 Electron-hole recombination is bimolecular, R = 7 (np). A Langevin form is used for the recombination coefficient 4roe# m

7= - -

(5.7)

where #m is the larger of #, or #p. The generation rate is determined from the recombination rate using detailed balance, G = V(nePe), where (nePe) is the product of equilibrium electron and hole densities. For the materials used in organic diodes, typically with energy gaps near 2 eV, carrier generation is very small. The equations are spatially discretized using the Scharfetter-Gummel approach. The resulting first-order differential equations are integrated forward in time using, for example, Gear's method. To find the steady state solution at a given applied voltage bias, a time-dependent potential ramp that stops at the desired voltage is applied to the right contact and the equations are integrated forward in time starting from thermal equilibrium until steady state is reached. The position independence of the total particle current J = J, + Jp is used to verify that steady state has been reached. Poisson's equation is integrated, with the calculated steady state values for electron and hole densities, to verify that the correct value of the potential drop across the device for the applied voltage bias is achieved. To calculate 195 j. M. Lupton and I. D. W. Samuel, J. of Phys. D. - Appl. Phys. 32, 2973 (1999).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

81

a current-voltage (I-V) curve a series of voltage ramps is applied. In performing the time integrations it is useful to have an explicit expression for the time dependence of the electric field

t?E(x)c3t

L1 c~r

(

4nee J(x) - -s

J(x) dx

)

(5.8)

where L is the length of the device and r is the electrostatic potential whose value at the right contact, r is set by the voltage ramp and whose value at the left contact, r is fixed at zero. Equation 5.8 follows from the time derivative of Poisson's equation. It is used to find E(0) to start the spatial integration of Poisson's equation at each new time step. The time-dependent formulation is used to avoid the difficulties that arise with two-point boundary conditions in a steady state formulation. For the equilibrium description, the material is viewed as made up of a series of localized electronic sites. Nondegenerate cases are considered so that the occupation probability of a site by electrons or holes is significantly less than unity. Let no be the density of localized sites times the number of ways that a site can be occupied by an electron or hole (i.e., its degeneracy). If E~ (E,) is the energy of the electrons (holes) (so that, ( E ~ - E~) is the energy gap of the polymer) the equilibrium densities are

ne = no e-(E~-

edp(x) -

O)/kr

(5.9)

and

Pe -" no e(E~ -

e4)(x) -

~)/kT

(5.10)

where ~ is the Fermi energy. To find the equilibrium solution, Poisson's equation is integrated using these expressions for the carrier densities. In organic materials, polaron states are disordered with an approximately Gaussian density of states. For simplicity, a single polaron energy is used in the device model. The effect of including a Gaussian density of states in the device model was investigated in Ref. [163]. It was found that the effects of the Gaussian state density could be incorporated by a small, temperaturedependent, shift of the effective barrier height. The size of the shift was smaller than the experimental uncertainties in the measured barrier heights. The contacts enter the model as spatial boundary conditions specifying the particle currents at the boundaries x - 0 and x = L. There are three particle current components at each metal/organic material interface: a thermionic emission current component from the metal into the organic material, a backflowing interface recombination current component that is the time-reversed process of thermionic emission, and possibly an athermal current component such as a tunneling current. Specifically consider the

82

I.H. CAMPBELL AND D. L. SMITH

electron current at x = O, Jn(O) -- S t h -

Jir ~- Stu

(5.11)

where Jth is the thermionic emission current density, J~ is the interface recombination current density, and Jt, is the tunneling current density. The thermionic emission current density has the general form Jth -- - e ~ i,:

O'i,:

P,f~ - ~ (1 - P:),

(5.12t

where i labels electronic states in the metal incident on the interface, Pi is the probability that this state is occupied, fi is the electron flux carried by the state, : labels electron polaron states in the organic material, tri,: is the capture cross section for the state i incident on the interface to be captured into the polaron state :, A is the interface area, and Pe is the probability the polaron state g is occupied. The backflowing interface recombination current density has the general form

1(1)

ffir --" -- e ~ P: ~ i,:

(1 - P~),

(5.13)

where (1/'C:,i) is the transition rate for a polaron in state : to go into the state i in the metal. (1/~:,~) and ai,: are related by detailed balance

where i* is the time-reversed state of i; that is, if an electron in state i is going toward the interface, an electron in state i* is going away from the interface. Thus (Ti,:

Jth -- ffir = - - e Z L Z i,:

(Pi-

P:)

(5.15)

The electron distribution in the metal is not driven out of equilibrium by the current flowing in the device, so that the first term in Eq. 5.15 balances a quasi-equilibrium distribution in the organic material near the interface Jth-

O'i,:

Jir "- - e ~ fi ~

(pQEq _ p:)

(5.16)

i,:

where pqEq is the quasi-equilibrium distribution in the organic material. If intersite hopping is fast enough to keep the carrier distribution in the organic material in local quasi-equilibrium (a necessary condition for the mobility description in the first place), Eq. 5.16 takes the form Jth -- Jir = -- g(g(0) QEq - g(0))

(5.17)

PHYSICS OF ORGANIC ELECTRONIC DEVICES

83

where the kinetic coefficient, K, has the form fie/ pQEq

e 2 f,-y K =

i,e

N(O) QEq

(5.18)

N(O) QEq is the electron polaron density at the interface determined by

quasi-equilibrium with the metal and N(O) is the actual electron polaron density at the interface. The boundary condition for electron current at x =Ois

Jn(O) --

- K ( N ( 0 ) QEq - N(0)) + Jtu

(5.19)

For most cases of interest the kinetic coefficient, K, and the interface densities are very large and the boundary condition reduces to the requirement of quasi-equilibrium at the interface N(0) QEq ~ N(0)

(5.20)

In the device model calculations, specific forms for the three current components are used. 17o However, in almost all cases of interest the results reduce to the quasi-equilibrium result. This is a major simplification because the results do not then depend on the detailed microscopic forms for the current components, which are somewhat idealized. The other three interfacial particle currents, which constitute the remaining three boundary conditions, have forms analogous to the x = 0 electron boundary condition. Because of the image force, the interfacial energy barrier depends on the electric field at the interface, /elE_(0)[

(5.21)

where q) is the Schottky energy barrier at zero field and e is the static dielectric constant of the organic material. (The image force barrier lowering term is only included when the electric field has the correct sign for barrier lowering.) As seen in Fig. III.5, this barrier lowering shows up directly in the internal photoemission measurements of barrier height. It can have significant consequences on the calculated device behavior and it is important that it be included in the device model. At steady state, the continuity equations can be integrated spatially to obtain a recombination current J,

J, = jo'~ eR. dx = Jn(L) - J,(O) = Jp(O) - Jp(L)

(5.22)

Electrons (holes) injected at x = 0 (x = L) to give J,(0) (Jp(L)) either recombine in the device and contribute to the recombination current, J,, or

84

I.H. CAMPBELL AND D. L. SMITH

completely traverse the device and contribute to J,(L) (Jp(O)). Both electron current at the hole injecting contact J,,(L) and hole current at the electron injecting contact Jp(O) result in a parasitic loss which lowers the quantum efficiency. In the ideal case of unity quantum efficiency J r - IJI, and J,,(L) = Jp(O) = O. Two important figures of merit for organic LEDs are the recombination efficiency r/, (the ratio of the recombination current J, to the total device current ]J]) and the recombination power efficiency r/,e (the ratio of power output from recombination to electric power input). These have the form J' Jr "Egap firP = ij[ . Vapplied

(5.23) (5.24)

The total recombination current Jr will be higher than the radiative recombination current because some excitons recombine nonradiatively. Spatial dependence of the radiative and nonradiative recombination rates is possible. For example, dipolar quenching of radiative recombination can occur near the metallic interfaces and a spatial variation in the density of nonradiative recombination centers can occur during d e p o s i t i o n . 196'197 Carrier density dependence of the recombination processes is also possible, for example by saturation of recombination centers. However, the spatial and density dependence of radiative and nonradiative rates are not known so the ratio of the radiative recombination to the total recombination (Q) is taken as constant. The quantities r/r and r/re are related to the quantum efficiency and power efficiency, respectively, by multiplying qr and r/w by the ratio of radiative to total recombination, r/q = Qqr and qp = Qr/re, where r/q is the quantum efficiency and qp is the power efficiency. In a simple case, where all the singlet excitons decay radiatively, all the triplet excitons decay nonradiatively, and 1/4 of the excitons formed are singlets and 3/4 are triplets, the ratio of radiative to total recombination will be Q = 1/4.

b. Single-Carrier Devices Device model results for representative single-carrier organic diodes, using material parameters similar to those of MEH-PPV, are first presented to illustrate the importance of the Schottky energy barrier on device behavior. Then detailed comparisons between measured and calculated current-volt196

V. Choong, Y. Park, Y. Gao, T. Wehrmeister, K. Mullen, B. R. Hsieh, and C. W. Tang,

Appl. Phys. Lett. 69, 1492 (1996). 197 H. Becker, S. E. Burns, and R. H. Friend, Phys. Rev. B56, 1893 (1997).

PHYSICS OF ORGANIC ELECTRONIC DEVICES 0.10 p..

%

'

~'/'

0.08

'

~,~, 0.06

.;

, 0.04

!/' ' ;

0.02

i

9;

~ #9

/"

,

i

14,

g -~

/; ;

10-13

1/

j/

10-4 0

/

;

/ ,'" .,o9

."

.../ , /"

aB

eaw

/

z

..." 9

eoe

/

5

i

/.o

..

/

l

/ / e""

f

i9

-

"

,

#"

r

!

/

. i#

#"

/

/

o;

f,-" 10-2

I

/

, j..-

%

,

:

/

J

#

/ /.

I

9

j

'

t

85

:"

..." i

i

10

15

20

Bias (V) FIG. V.1. Linear-linear (upper panel) and log-linear (lower panel) plots of calculated current density as a function of bias voltage for 120-nm MEH-PPV devices with a 1.4 eV barrier to electron injection and 0.1, 0.2, 0.3, 0.4, 0.5, and 0.6 eV barriers to hole injection (from Ref. 170).

age characteristics are presented. Calculated results for a 120-nm thick device with a barrier for electron injection fixed at 1.4 eV approximately corresponding to an A1 contact (at this large value for the barrier few electrons are injected) and the barrier for hole injection varied from 0.1 to 0.6 eV in 0.1 eV steps is shown in Fig. V.1.17~ The upper panel shows a linear-linear plot and the lower panel a log-linear plot of the calculated I-V curves. The calculated results for 0.1, 0.2, and 0.3 eV barriers are nearly the same. For these cases, the current flow is essentially space charge limited. As the energy barrier is further increased the current is decreased, indicating that the current flow becomes injection limited. These results show that for typical organic diode device parameters, it is important that injection barriers be kept less than about 0.4 eV. Figure V.2 shows the calculated hole densities for four bias voltages as a function of position for a device with a 0.6 eV barrier to hole injection and a 1.4 eV barrier to electron injection, approximately corresponding to a Cu/ MEH-PPV/A1 structure (upper panel), and a device with a 0.1 eV barrier to hole injection and a 2.3 eV barrier to electron injection, approximately corresponding to a Au/MEH-PPV/Au structure (lower panel). 17~ The bias voltages in the upper panel are 20 V (solid line), 15 V (dotted line), 10 V

86

I.H. CAMPBELL AND D. L. SMITH

Cu/MEH-PPV/AI

1016

i

i

i

i

i

03 |

E o 1015 e--

a

1014

0

-1-

1013

0

200 400 600 800 1000 1200 Position (Angstroms) Au/MEH-PPV/Au

1019 03 !

E ~ 1018 121 ~

1017

0

I

10160

200 400 600 800 1000 Position (Angstroms)

FIG. V.2. Calculated hole density as a function of position for a Cu/MEH-PPV/A1 device (upper panel) and the Au/MEH-PPV/Au device (lower panel). In the upper panel the bias voltages are 20 V (solid line), 15 V (dotted line), 10 V (dashed line), and 5 V (dot-dash line). In the lower panel the bias voltages are 8 V (solid line), 6 V (dotted line), 4 V (dashed line) and 2 V (dot-dash line). The hole injecting contact is at the right (from Ref. 170).

(dashed line), and 5 V (dot-dash line). In the lower panel the bias voltages are 8 V (solid line), 6 V (dotted line) 4 V (dashed line), and 2 V (dot-dash line). In both panels, the hole injecting contact is at the right. The hole density for the device in the upper panel is essentially constant spatially. The hole density increases rapidly with increasing bias but for all the bias values shown is rather small and does not significantly influence the electric field in the device. For the bias range shown the hole density in the device is the quasi-equilibrium value at the hole injecting contact, including barrier height lowering by the image force. The increase in hole density with bias is due to the image force barrier lowering. The behavior shown in the upper panel is characteristic of contacted limited diodes. The hole density for the

PHYSICS OF ORGANIC ELECTRONIC DEVICES

87

Cu/MEH-PPV/AI

2

|

,

|

|

|

E CO O ,s,.v

-o

I

I.I.

._o L-

9

o

(I) l.IJ I

00

200

I

I

400

600

I

I

800 1000 1200

Position (Angstroms)

10

.

Au/MEH-PPV/Au . . . .

E

~, 6! ........ "o ._~

4r. ........

" ........ ,

',

ILl m

... o 0

200 400 600 800 Position (Angstroms)

1000

FIG. V.3. Calculated electric field as a function of position for a Cu/MEH-PPV/A1 device (upper panel) and the Au/MEH-PPV/Au device (lower panel). The values of voltage bias are the same as in Fig. V.2. The hole injecting contact is at the fight (from Ref. 170).

device in the lower panel varies strongly with position. The hole density at the hole injecting contact is equal to the true equilibrium value. (Because of the sign of the electric field in this case, there is no image force lowering of the injection barrier and no tunneling injection). The hole density changes significantly with bias and it is large enough to strongly influence the electric field in the device. The behavior shown in the lower panel is characteristic of space charge limited diodes. Figure V.3 shows the calculated electric fields as a function of position for the Cu/MEH-PPV/A1 structure in the upper panel and the Au/MEH-PPV/ Au structure in the lower panel at the same bias voltages as in Fig. V.2.170 For the device in the upper panel, the electric field is an essentially constant function of position, whereas for the device in the lower panel the electric field is a strongly varying function of position. For the device in the upper panel, the electric field at the hole injecting contact has the correct sign to lead to image force lowering of the injection barrier. Because of the high hole density near the hole injecting contact in the device in the lower panel, the sign of the

88

I.H. CAMPBELL AND D. L. SMITH

/ / 00i ! 0

10

20

30

40

Bias (V) FIG. V.4. Measured (solid lines) and calculated (dashed lines) current density as a function of voltage bias for MEH-PPV devices about 110 nm thick with Au as the electron injecting contact and Pt, Au, Cu, and A1 as the hole injecting contact (from Ref. 198).

electric field is reversed and it does not have the correct sign to lead to image force lowering of the injection barrier. These results are characteristic of contact limited and space charge limited diodes, respectively. The dependence of the device current on organic film thickness can be used to distinguish contact limited from space charge limited current flow. The device current in the contact limited regime scales as (Vapplied- Vbi)/L because the electric field is constant across these devices, and the mechanisms that determine device current--carrier mobility, injection barrier lowering, and carrier density--all scale with electric field. In the space charge limited regime, the electric field is not constant across the device, and the current scaling is complex and depends more strongly on device thickness. It does not follow that ( V a p p l i e d - - Vbi)2/L3, as predicted by constant mobility space charge limited current calculations, because of the strong electric field dependence of the carrier mobilities. 161 It is therefore straightforward to experimentally distinguish contact limited from space charge limited current flow by the length scaling of device currents. Now, detailed comparisons between measured and calculated I-V characteristics for single-carrier diodes are presented. First the effect of changing barrier heights by using different contact metals is considered. Figure V.4 compares measured and calculated I-V curves for M E H - P P V devices about 110 nm thick in which the cathode contact is fixed to be Au and the anode contact is varied, including Pt, Au, Cu, and A1.198 The mobility parameters #o and E o were determined by fitting to these data. The same mobility parameters were used for the calculation of all four devices. The Schottky 198 I. H. Campbell, P. S. Davids, D. L. Smith, N. N. Barashkov, and J. P. Ferraris, Appl. Phys.

Lett. 72, 1863 (1998).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

89

lo6

':' L 1

0

2

-

1

3 6

~ 8 Bias (V)

4

10

12

102 i'M

E rj v~

1 10 -2 10-4 5

10

15

20

Bias (V)

,

L

,

Interface Recombination

4[ AVAu 9

i

..-"

10"6P''T,~ ..... 9 • ~'~ BB~

~

25

30

_1

Thermionic 1

.

r~'" Tu~r~l~ng Dev~c'~d~Current, 20

.~

35

1 40

Bias (V) FIG. V.5. Calculated values of the injection current components and the total device current as a function of bias for the Pt (upper panel), Cu (central panel), and A1 (lower panel) hole injecting contact devices (from Ref. 198).

energy barriers for hole injection into MEH-PPV were essentially zero for Pt, 0.2 eV for Au, 0.6 eV for Cu, and 1.1 eV for A1. The I-V curves for the devices with hole injection from Pt and from Au are essentially the same. This is expected because the energy barriers for hole injection are small enough in both cases that the current is space charge limited. When the Schottky barrier is increased in the device with a Cu contact, the current at a given bias voltage is much reduced, showing that the contact limits the current flow in this case. The current is further reduced when an A1 contact with a still larger barrier is used. The magnitude of the three components of the calculated injection current, that is the thermionic, interface recombination, and tunneling current components, together with the net device current for the devices with Pt, Cu and A1 contacts, is shown in Fig. V.5.198 The upper panel shows

90

I.H. CAMPBELL AND D. L. SMITH

the case of a Pt injecting contact, where the energy barrier to injection is very small (0.05 eV was used in the calculation) and the current flow is space charge limited. In this case the magnitude of the thermionic emission and interface recombination currents are both very large and they almost exactly cancel each other. The net device current, which is the difference in magnitude between the thermionic emission and interface recombination currents, is much smaller than either of these two components separately. There is no tunneling current because the electric field at the contact has the wrong sign for tunneling. The thermionic emission current and interface recombination currents are independent of bias because the applied electric field is completely screened by a large carrier density at the interface. The calculated behavior of the injection current components for the case of the Au injecting contact (not shown), which has a 0.2 eV Schottky barrier, is qualitatively similar to that for the Pt contact. This behavior of the injection current components is characteristic of space charge limited diodes. The middle panel of Fig. V.5 is for the device with a Cu contact where the energy barrier is 0.6 eV and the current flow is contact limited. The magnitude of the thermionic emission and interface recombination components are still much larger than the net device current. The electric field near the contact is no longer screened by a high density of carriers near the interface as for the device with a Pt contact, and there is image force lowering of the barrier. Therefore the thermionic emission current is bias dependent and in addition tunneling is possible. Over most of the bias range shown, thermionic emission is much larger than tunneling, but tunneling is a stronger function of bias than is thermionic emission, and at the highest bias shown these two current components become comparable. In the range where thermionic emission dominates tunneling, a combination of thermionic emission and interface recombination (the time-reversed process of thermionic emission) keeps the interface hole density at its quasi-equilibrium value. In this regime the microscopic details of the current components do not influence the device model results. The behavior of the injection current components for the Cu/MEH-PPV/Au device is characteristic of contact limited diodes with moderate injection barriers. The lower panel of Fig. V.5 is for a device with an A1 contact where the hole barrier is 1.1 eV. All current components are strongly reduced by the large Schottky barrier and it is necessary to go to much larger bias to get a reasonably large current. Because tunneling is a more rapidly increasing function of bias than thermionic emission, tunneling is larger than thermionic emission at the bias required for significant current flow in this device with a large energy barrier. Interface recombination nearly cancels the tunneling injection current. The net device current is approximately proportional to the tunneling component but is much smaller in absolute

91

PHYSICS OF ORGANIC ELECTRONIC DEVICES

0.01

i

!

I ,

10 -2

I I o

I 10 -3 m !

I

I r~ 0

e

10 -4

!

2

! #

/ o e

/ 0

.,e"

0

i

10

5

15

Bias (V) FI6. V.6. Measured (solid line) and calculated (dashed line) current density versus bias voltage for a Ca/Alq/Ca electron only device with a 100-nm thick Alq layer. The inset shows the same results on a log-log plot. Also shown is a calculated I-V characteristic assuming a SCL contact with a 0.1 eV energy barrier (dot-dash) (from Ref. 113).

magnitude. In this regime, the hole density at the injecting contact is not determined by quasi-equilibrium and the microscopic details of the current components are important for the device model results. The behavior of the injection current components for the A1/MEH-PPV/Au device is characteristic of contact limited diodes with large injection barriers. For practical diodes, it is very desirable to use small injection barriers so the large injection energy barrier regime is not of great practical interest. For Alq, neither electron nor hole contacts are SCL for common metal electrodes. The current-voltage characteristics of electron only Alq diodes were measured and compared to model calculations using the independently determined energy barrier (from internal photoemission, Section III) and mobility (from TOF, Section IV). Figure V.6 shows measured (solid) and calculated (dashed) current-voltage characteristics for a Ca/Alq/Ca structure with a 100-nm thick Alq layer. 113 There is good agreement between the measured and calculated I-V characteristics. For the Ca contact, an energy barrier of 0.62 eV was used to describe the measured I-V characteristic. This energy barrier is in good agreement with the energy barrier of 0.6 eV ___0.1 eV determined from internal photoemission and built-in potential measurements. Also shown in Fig. V.6 is the calculated I-V characteristic assuming a SCL contact with a 0.1 eV energy barrier (dot-dash). The space charge

92

I.H. CAMPBELL AND D. L. SMITH ,~ 0.10

I

|

005 tJ

00:100t

t

'

(V-Vbi)/d

0.00

|

' -10

0' 10 Bias (V)

2'0

FIG. V.7. Measured (solid) and calculated (dashed) current density versus bias voltage for a Pt/Alq/Ca hole only device with 50-nm, 100-nm, and 150-nm thick Alq layers. The inset shows that the current scales with electric field including the built-in potential (from Ref. 141).

limited contact would give a substantially higher current density for a given voltage. As discussed in Section IV, the charge-dipole disorder model implies that the electric field dependence of the electron and hole mobilities of Alq should be comparable. It was not possible to measure the hole mobility of Alq using time-of-flight techniques and there are no space charge limited hole contacts to Alq. To estimate the hole mobility, the I-V characteristics of a thickness series of Pt/Alq/Ca devices were measured and fit using the hole mobility as the only adjustable parameter (the other parameters were determined in Sections III and IV). Figure V.7 shows the measured and calculated I-V characteristics for 50-nm, 100-nm and 150-nm thick Alq layers. T M The inset shows that the current scales with electric field including the built-in potential. The fit to the hole mobility accurately describes the measured characteristics as a function of applied bias and film thickness. The hole mobility determined from this procedure is similar to the electron mobility but the exact mobility parameters are sensitive to the precise value of the hole Schottky barrier. The measured hole Schottky barrier is 0.4eV ___0.1 eV and this uncertainty in the barrier leads to relatively large uncertainties in the fit mobility. Nevertheless, within the experimental error, the electric field dependence of the electron and hole mobilities are similar as expected by the charge-dipole interaction model. The single-carrier single-layer device model results have a simple form when the current is contact limited. As an example, for an electron only device the electron current density is independent of position and equal to the net device current density, Jd. When the current is contact limited both the electric field and the electron density are nearly constant across the device and the device current density is well approximated by the drift

PHYSICS OF ORGANIC ELECTRONIC DEVICES

93

component of the electron current density, Jd -" e#N(O)E, where # depends on the electric field. From Eq. 5.19 the electron density is Jtu

S ( 0 ) = S ( 0 ) Q~~ + ~

-

K

Jd

(5.25)

The device current density is negligible compared to the sum of the thermionic and tunneling current densities and can be dropped in the preceding equation. There are two limiting regimes: one in which the thermionic emission current density is large compared to the tunneling current density and one in which the tunneling current density is large compared to the thermionic emission current density. In the first limit, the carrier density at the contact is given by its quasi-equilibrium value. This is a particularly simple case because all of the complexities of the injection process have canceled out and the interface carrier density is determined by statistics alone. In this case both N(0), because of the image force barrier lowering, and # have e,/~ field dependence. The second limit is more complex because the interface carrier density depends on the details of the injection processes. Summarizing, for typical organic LED device parameters, current is space charge limited if the Schottky energy barrier to injection is less than about 0.4 eV and contact limited if it is greater than that. Injection currents have a component due to thermionic emission and a component due to tunneling. Thermionic emission is more important for smaller injection barrier structures. If thermionic emission dominates injection, a combination of thermionic emission and interface recombination establishes a quasi-equilibrium carrier density at the interface. This is the case of interest for most practical devices. If tunneling dominates, a combination of tunneling and interface recombination establishes the carrier density at the interface. The net device current is small compared to the largest injection component. Interface recombination almost exactly cancels the injection current; that is, most carriers that are injected into the polymer fall back into the metal contact. As a result of the low mobilities of conjugated polymers only a very small fraction of the injected carriers are extracted from the contact region.

c. Bipolar Devices SCL electron only Ca/MEH-PPV/Ca diodes were discussed in Section IV, and were used to extract the electron mobility parameters. Having determined both electron and hole mobilities from single-carrier diodes, these mobilities are used in the device model to describe bipolar, light-emitting structures. Figure V.8 shows current density versus bias voltage for a series of Pt/Ca bipolar devices with MEH-PPV layer thickness from 40 nm to 110

94

I.H. CAMPBELL AND D. L. SMITH 0.1 Pt/Ca

i

l

40 nm !

I

' 50 nm !

0.05

_// t

100

/:/:,,/,I gl;

~,~ l 0

I

B

/

J

.I /

ll0nm

4

0

Bias (V) 10-1

Pt/Ca

10.2 o~.~

10-3

10-4 10 Bias (V) FIG. V.8. Measured (solid line) and calculated (dashed line) current density versus bias voltage for 40-, 50-, 100-, and ll0-nm thick Pt/MEH-PPV/Ca bipolar devices on linear (upper panel) and log-log (lower panel) scales (from Ref. 163).

nm. 163 The Pt and Ca contacts provide low-energy barriers for hole and electron injection, respectively. The data is described using the device model with the carrier mobility parameters determined from the single-carrier devices with no additional fitting parameters. The current is dominated by holes in these devices because the hole mobility is much larger than the electron mobility. The model describes the data over a range of device thicknesses and over several orders of magnitude of device current. Figure V.9 shows current density versus bias voltage for Pt/Ca and Cu/Ca bipolar devices, as well as Pt/A1 and Cu/A1 hole only devices. 163 The Pt/Ca and Cu/A1 devices are 100 nm thick, the Pt/A1 is 90 nm thick, and the Cu/Ca is 80 nm thick. The Pt/Ca device has space charge limited contacts for both electrons and holes, whereas the Cu/Ca device has space charge limited contacts for electrons, but holes are contact limited due to the larger energy

PHYSICS OF ORGANIC ELECTRONIC DEVICES 0.1

I

95

I

!

Pt/A1

' P t/Ca Cu/

9 v,,.4

0.05

7

r..)

o

,,

~

I

10

o

15

20

Bias (V) ft..,,

10 l

l

!

u

i

!

!

!

C .~

10"2 -

10_3 r,.)

J

-

/ 4

4 ,7 /

/ .7 ....

104 1

10 Bias (V)

FIG. V.9. Measured (solid line) and calculated (dashed line) current density versus bias voltage for MEH-PPV devices with various contacts: Pt/A1 and Cu/A1 hole only devices and Pt/Ca and Cu/Ca bipolar devices on linear (upper panel) and log-log (lower panel) scales. Holes are injected from the Pt, Au, and Cu electrodes. Electrons are injected from the Ca electrode. Devices are about 100 nm thick (from Ref. 163).

barrier to injection of holes from Cu into MEH-PPV. The model describes the data well over several orders of current density using the mobility parameters determined from single-carrier devices. The Pt/Ca and Pt/A1 devices have similar currents, the Pt/A1 current is somewhat higher because it is thinner and has a smaller built-in potential. The Cu/Ca device has a substantially larger current than the Cu/A1 devices. This is due in part to the thickness difference, but primarily because the Cu/Ca device current has contributions from both electrons and holes, whereas the Cu/A1 device current is hole only. Figure V.10 shows calculated carrier density and electric field profiles for Pt/Ca and Cu/Ca bipolar devices, for biases that give a current density of 6 x 10 -2 A/cm2.163 In both cases electrons are injected from the left at x = 0 and holes from the right at x = L. The electron and hole densities are given by the quasi-thermal equilibrium values at the injecting contacts. For the

96

I.H. CAMPBELL AND D. L. SMITH ~ , 10 20

,

|

e~

Pt/Ca 6xl02AJcm 2

---

p n

i Cu/Ca ~ 6xl02A/cm 2

- - -

---p

n

~.,lO

~

~ 1016 ~-

I i~ I I I

rJ 1014 . 1!

I

I t

~

2

~

1

~

o o

!

50

Position (nm)

100 0

!

40

80

Position (rim)

FIG. V.10. Calculated hole (solid line) and electron (dashed line) carrier density (upper panel) and electric field (lower panel) profiles for Pt/MEH-PPV/Ca and Cu/MEH-PPV/Ca devices. The electron injecting contact is at the left and the hole injecting electrode is at the right (from Ref. 163).

Pt/Ca device the electrons and holes are space charge limited and have high carrier densities at the injecting contacts that suppress the electric field. The electron density drops rapidly across the device. The slope of this drop is determined by a combination of the electron mobility and carrier recombination. The holes dominate the current density across virtually the entire device. For the Cu/Ca device the electron contact is space charge limited, whereas the holes are contact limited. The electric field and carrier densities near the hole injecting contact at x = L are relatively constant. The electron density is high near the electron injecting contact at x = 0, and screens the electric field. At this current density, the electron density is about 3 orders of magnitude larger than the hole density, however the electron mobility is about a factor of 300 lower than the hole mobility. The electrons and holes both contribute significantly to the device current. Figure V.11 shows calculated current density profiles for electrons and holes for the Cu/Ca device at applied biases of 6V, 9.5V, and 13V, with current densities of 7 • 10 -5 A/cm 2, 3 x 10 -3 A/cm 2, and 6 x 10 -2 A/cm 2, respectively. 163 The bipolar devices emit light and the measured luminance can be

PHYSICS OF ORGANIC ELECTRONIC DEVICES 10-1

I

10"2

"

13V,

10-3 f,.._____

. . . .

- .

,

...

-- hole ~~

1 0 -4

rj

10 5

97

---

-

"~ ....

electron 6 V, 7xl0-5A/cm 2 ,~,

I.

0

~

2

9.5 V, 3xl0-3A/cm 2

_

10-6

-

.

40

.

.

I -i

.

80

Position (nm) FIG. V.11. Calculated electron (dashed line) and hole (solid line) current density profiles for a Cu/MEH-PPV/Ca device. The current density profiles are shown for bias voltages of 6, 9, and 13V, corresponding to current densities of 7 • 10-s, 3 x 10-3, and 6 x 10-2A/cm2 respectively (from Ref. 163).

compared with the expectations of the device model. Figure V.12 shows the measured and calculated external luminance as a function of device current for 40-nm and l l 0 - n m thick P t / C a devices. 163 The symbols are the measured device luminance using a silicon photodiode placed flush against the L E D substrate. The lines are calculated external luminance obtained by multiplying the calculated recombination current, Jr, by the optical energy gap and by a factor ( that is the fraction of recombination events leading to externally measurable light emission. ( includes nonradiative recombination, total internal reflection inside the organic layer, and absorption as the emitted light goes through the semitransparent contact. It is taken to be = 1/110 to fit the measured luminance for the l l 0 - n m device. It is difficult to determine ( quantitatively, but this value is a reasonable estimate. The model reproduces the linear behavior of luminance as a function of current density and the decrease in luminance with decreasing device thickness. However, the model underestimates the magnitude of the drop in luminescence with decreasing thickness. This difficulty with the model may be due to the assumption that radiative recombination efficiency is uniform across a device and the same for different device thickness.

98

I.H. CAMPBELL AND D. L. SMITH 0.1

Pt/Ca ~

~

~

m

..... 9

ll0nm 40 nm

0.05

1 o

o

0.025 Current Density (A/crn2)

0.05

FIG. V.12. Measured (symbols) and calculated (lines) external luminance versus device current density for 40-nm and ll0-nm thick Pt/MEH-PPV/Ca devices (from Ref. 163).

d. Multilayer Devices The calculated carrier profiles shown in Fig. V.10 illustrate a potential problem for organic LEDs. Because the electron mobility is much smaller than the hole mobility for M E H - P P V , most of the device current is carried by holes and the recombination is strongly peaked near the electron injecting contact. These properties are undesirable for two reasons: M a n y of the injected holes traverse the device without recombining (i.e., Jp(L) is comparable to Jp(0) in Eq. V.22); and recombination takes place near a contact where dipole quenching and nonradiative losses can reduce luminescence efficiency. Similar problems occur if it is only possible to make space charge limited contacts for one of the carrier types. In general, asymmetric injection or transport properties between electrons and holes causes one of the carriers to dominate the current flow. A dominant carrier type usually results in parasitic currents that do not produce recombination. Multilayer structures can be used to overcome these difficulties. 33'34'x84'199-2~ F o r example, a bilayer structure can be used to present an energy barrier to the dominant carrier and prevent it from traversing the device without recombining. Multilayer structures can also prevent the region of high recombination from occurring near an electrode; they are typically used to confine the carrier recombination to a thin region (10 nm) near an internal organic/ 199 L. S. Hung and C. W. Tang, Appl. Phys. Lett. 74, 3209 (1999). 200 M. Strukelj, F. Papadimitrakopoulos, T. M. Miller, and L. J. Rothberg, Science 267, 1969 (1995). 2ol C. Giebeler, H. Antoniadis, D. D. C. Bradley, and Y. Shirota, J. Appl. Phys. 85, 129 (1999). 202 E. Bellmann, S. E. Shaheen, R. H. Grubbs, S. R. Marder, B. Kippelen, and N. Peyghambarian, Chemistry of Materials 11, 399 (1999).

PHYSICS OF ORGANIC ELECTRONIC DEVICES 3.0eV

99

I ...... 3.0eV Organic A Organic B I I

5. leV

I

- - 5.3eV

I~ 0-0.6eV

!

5.4-6.0eV

5.4eV

0.1 :

I~

I

!

:~

,

"

!

"~ 0.05

,

"

~[ 0

0

"'"

I

,

]/ I #

I

,

f

"

-

.....

,' 10

/ 0.6eV

I

+' 20

Bias (V) Fro. V.13. Energy level diagram for single-carrier bilayer devices with a variable energy barrier to hole transport (above the panel). The holes are injected into organic B from a metal with an energy level at 5.3 eV that provides space charge limited current. Calculated current-voltage characteristics for the hole only bilayer devices as a function of the hole energy level discontinuity at the organic heterojunction (from Ref. 184).

organic interface. The thin region where the recombination occurs is often doped with a small luminescent organic molecule that serves as the dominant radiative recombination path. This is a convenient method of varying the optical emission energy in organic LEDs. To illustrate functions that can be achieved with multilayer structures, consider a bilayer hole only device with the energy level diagram shown in Fig. V.13. The conduction energy level is 3.0 eV for both organic materials. The valence energy level is 5.4 eV for material B, and is varied from 5.4 to 6.0 eV in 0.2 eV increments for material A. The two organic layers are 50 nm thick and have the same hole mobility parameters. The left (right) metal contact has an energy level of 5.1 eV (5.3 eV). The I-V characteristics are

100

I.H. CAMPBELL AND D. L. SMITH 1020

T

Organic A

m,• 101

[

Organic B

a

8

9~1016 .......................................i 0.2eV

-

I

No 10 .14 B

-] 0.4eV

1012

J 0.6eV

I

4

I

0.6eV

~

3

N

2-

~

1 ................................... i 0.2eV 0eV

-

_1 ~ 0

I

~ 0.4eV

50 Position (nm)

-

-

100

FIG. V.14. Calculated hole density (upper panel) and electric field profiles (lower panel) for hole only bilayer devices at a current density of 0.1 A/cm2 as a function of the hole energy level discontinuity at the interface. The holes are injected from the right at x = 100 nm and the organic heterojunction is at x = 50 nm (from Ref. 184).

calculated for forward bias where the holes are injected on the right and there is an energy barrier for holes to traverse the structure going from material B into material A. (Details of how current flow at the heterostructure interface is included in the device model are discussed in Ref. [184].) Figure V.13 shows the calculated current-voltage characteristics for the hole only two-layer devices as a function of the energy barrier between the two organic materials. 184 Successively higher voltages are needed to achieve a given current density as the energy barrier is increased. Even the modest 0.2 eV heterojunction barrier causes a significant increase in the voltage required to obtain a given current. Figure V.14 shows the calculated hole density and electric field profiles at a current density of 0.1 A/cm 2 for the hole only bilayer devices as a function of the energy barrier between the two organic materials. 184 The holes are injected into material B at the right contact. The hole density profile in material B remains unchanged as the energy barrier is increased, except near the heterojunction interface where the hole density increases as the barrier increases. The hole density in material A is relatively uniform across the

101

PHYSICS OF ORGANIC ELECTRONIC DEVICES

20 E

,

,

,

.. .,

2

4

..

,

. ,

j

15-

to o

-o ._~

10

._o

5

IJ. o

UJ

0 ~.~0

Luminescent Layer

.,

6 Bias

,

8

'

10

(V)

FIG. V.15. Calculated electric field as a function of bias in the center of the blocking layer and in the center of the luminescent layer of a 0.5 eV single-carrier barrier structure (from Ref.

204).

layer and its magnitude decreases as the barrier increases. For a 0.4 eV barrier the hole density is over 2 orders of magnitude lower than for the case with no barrier. The field across material B does not change significantly as the energy barrier is increased. However, the field in organic material A increases and is nearly constant spatially across the layer. The increased voltage required to maintain a given current density is dropped across material A. A large field is needed to maintain a constant current density when the carrier density in material A decreases because of the energy barrier. For devices with a hole barrier there is a large accumulation of holes at the interface. The spike in the hole density at the interface causes a rapid change in the electric field at the interface. The field in the hole barrier layer (material A) is significantly larger than in the hole injection layer (material B). The electric field in the middle of both the blocking layer and the hole injection layer is plotted as a function of bias for a 0.5 eV barrier device in Fig. V.15. As the bias is increased, the electric field in the blocking layer increases rapidly. In contrast, the electric field in the hole injection layer remains relatively small and increases slowly with bias. The addition of a blocking layer causes an accumulation of charge at the blocking interface and, because of this charge accumulation, changes the electric field distribution in the structure. The effect of charge accumulation at a blocking layer in a multilayer organic device structure has been directly observed using electroabsorption

102

I.H. CAMPBELL AND D. L. SMITH I

I

6

E

5

4 3 2 J

~D 0

f

f Luminescent Layer

-

0

I

I

2

4

6

Bias (V) FIG. V.16. Measured electric field as a function of bias in the blocking layer and in the luminescent layer of a 200-nm single-carrier barrier structure (from Ref. 204).

techniques. 2~ To illustrate the approach, a structure was investigated that consisted of a large energy gap electron blocking layer, poly(p-phenylene diamine), and a smaller energy gap luminescent layer, MEH-PPV, each 100-nm thick sandwiched between Ca contacts. The internal electron energy barrier is about 1 eV. The Ca/luminescent layer contact has a negligible Schottky energy barrier and the Ca/blocking layer contact has large electron and hole Schottky energy barriers. Figure V.16 shows the measured electric field in each of the layers as a function of forward bias, i.e. electrons are injected into the luminescent layer. This is an analogous situation to the calculations shown in Fig. V.15 except that, in this case, electrons are blocked rather than holes. These experimental results demonstrate the effects of blocking layers by directly measuring the redistribution of electric field in the structure due to the accumulation of charge at the blocking layer/luminescent layer interface. Inserting an organic layer near a contact can also enhance current flow by serving as a transport layer. The transport layer can have increased carrier mobility, a reduced Schottky barrier, or both. The upper panel of Fig. V.17 shows the effect of increasing electron mobility in a layer near the electron contact of a two-layer electron only device. 2~ (The energy barrier to electron injection at the metallic contact is 0.5 eV and there is no energy barrier at the heterojunction interface.) The solid line is the calculated I-V characteristic when the electron mobilities of the two layers are the same. 203 I. H. Campbell, M. D. Joswick, and I. D. Parker, Appl. Phys. Lett. 67, 3171 (1995). 2o4 I. H. Campbell and D. L. Smith, in Semiconducting Polymers." Chemistry, Physics and Engineering (G. Hadziioannou and P. Van Hutten, eds.), John Wiley & Sons, New York (2000).

PHYSICS OF ORGANIC ELECTRONIC 0.10

,

,

|

,

DEVICES

103

| ;

0.08 Od

E

--..../

Enhanced mobility

0.06

i

0.04

.."

0.02

A

0 4

0

6

8

10

12

Bias (V)

r

E

2.0

0

%

12 V

1.5

v

i"',

!""..........,....,

Enhanced m ~

"~ E a

1.0

E

0.5

o ~

1| i

, ......................

i

m

m

0

!

0

20

!

4'0

60

8'o

~00

Position (nm) FIG. V.17. Calculated current density as a function of bias (upper panel) and electron density as a function of position at 12 V bias (lower panel) for a two-layer electron only 0.5 eV barrier structure. The mobility in the left layer is increased by a factor of 10 in the enhanced mobility structure (dotted line) (from Ref. 204).

The dotted line is the calculated I-V characteristic when the electron mobility in the layer near the electron injecting contact is increased by a factor of ten. The lower panel of Fig. V.17 shows the calculated electron density as a function of position at a 12 V bias. 2~ The electron current is constant across the structure. Because the electron mobility changes abruptly at the interface between layers, the electron density must also change abruptly at this interface. The electron density is larger in the lower mobility material at the right of the device. In a two-carrier structure, the holes will also be concentrated in the right layer. Thus the enhanced electron mobility helps to concentrate the electrons in the region where the hole density is high. To illustrate the effects of multilayer structures on device efficiency, consider a two-carrier device made from a two-layer structure using an electron contact on the left with a 0.5 eV injection barrier and a hole contact on the right with a 0.1 eV injection barrier. If there is no heterostructure barrier for holes, the hole current is space charge limited because of the

104

I . H . C A M P B E L L A N D D. L. SMITH 0.10

' Total Cu;rent

~

'

0.08

E

0.06

v

0.04

Single

0.02 0

0

Laye~

I BlockingL a ~ ~ ~ ~ ~1~

2

4

6

8

lO

Bias (V) FIG. V.18. Calculated current (solid line) and recombination current (dashed line) density as a function of voltage bias for a single-layer structure, a two-layer structure with a hole blocking layer, and a two-layer structure in which the hole blocking layer also serves as an electron transport layer (from Ref. 204).

small energy barrier to hole injection. The electron current, however, is contact limited because of the comparatively large energy barrier to electron injection. If the electron and hole mobilities are the same, the device current will be dominated by holes. Such a structure will not be an efficient LED because most of the injected holes will traverse the device without recombining with the comparatively small number of injected electrons. If a heterojunction energy barrier for holes is included, the holes are confined in the layer on the right by this barrier and the number of holes traversing the device without recombining is reduced. If in addition the layer on the left has increased electron mobility, electron current will increase and the voltage necessary to reach a given current will decrease. Figure V.18 shows the calculated current density (solid lines) and recombination current density (dashed lines) as a function of bias for a single layer structure, a structure with a 0.3 eV hole barrier, and a structure with a 0.3 eV hole barrier and also with a factor of 10 enhanced electron mobility in the left material layer. 2~ For the single-layer device the recombination current density is only about a fifth of the total device current; the current is dominated by holes and only about a fifth of them recombine. The other four-fifths of the holes traverse the device without recombining and are lost at the electron injecting contact. When the 0.3 eV hole blocking layer is included, the holes cannot easily cross the energy barrier. The recombination current density is essentially equal to the total current density, meaning that essentially all the carriers injected into the device recombine. In this case the device quantum efficiency is limited by the efficiency of radiative recombination. However, the bias necessary to reach a given current density has been increased by the hole blocking layer. When the electron mobility

PHYSICS OF ORGANIC ELECTRONIC DEVICES

105

of the organic layer on the left is also increased, so that it serves as both a hole blocking layer and also as an electron transport layer, both the total current and the recombination current are increased. The enhanced electron mobility has increased the electron current. In this structure the current in the left layer is carried only by electrons; the holes are effectively blocked from this layer. The electrons recombine when they enter the layer on the right, which has a large density of holes. It is not necessary to use an electron blocking layer in this structure because the hole injection is high so that there are a large number of holes in the structure with which the electrons can recombine. In the model calculations previously presented, the energy offset between the electron and hole energy levels at organic heterojunctions was taken as an adjustable parameter. In organic materials, the heterojunction energy offsets are generally given by the difference between the energy levels of the constituent materials with respect to vacuum. 143'2~ Although there can be small energy changes due to polarization effects at the interface, 143'2~ this energy alignment is expected for the weak van der Waals interactions between the organic molecules. However, organic heterojunction interfaces can have excited state intermolecular interactions. For example, excited states called exciplexes can be formed consisting of an electron in one organic layer and a hole in the other material. 2~176 The exciplexes can have optical properties very different from the constituent materials and they can be the dominant recombination state in multilayer devices where most of the electrons and holes recombine at a heterojunction interface.

e. Transient Response and High Current Density Operation In m a n y applications, such as video displays, the frequency response of the diode is important. The response time of an organic L E D is determined by the carrier transit and recombination times. 2~ 21o The carrier transit time is approximately (z = L2/#V, which is about 10 -6 s for typical organic device parameters. Although the carrier mobility in disordered organic films is very low, the length that the carriers traverse is short and they can therefore be modulated at relatively high frequencies. The turnoff time for 2o5 I. G. Hill and A. Kahn, J. Appl. Phys. 84, 5583 (1998). 206 D. D. Gebler, Y. Z. Wang, J. W. Blatchford, S. W. Jessen, D. K. Fu, T. M. Swager, A. G. MacDiarmid, and A. J. Epstein, Appl. Phys. Lett. 70, 1644 (1997). 207 D. D. Gebler, Y. Z. Wang, D. K. Fu, M. Swager, and A. J. Epstein, J. Chem. Phys. 108, 7842 (1998). 208 D. J. Pinner, R. H. Friend, and N. Tessler, J. Appl. Phys. 86, 5116 (1999). 209 V. R. Nikienko, V. I. Arkhipov, Y. H. Tak, J. Pommerehne, H. Bassler, and H. H. Horhold, J. Appl. Phys. 81, 7514 (1997). 21o V. I. Nikitenko, Y. H. Tak, and H. Bassler, J. Appl. Phys. 84, 2334 (1998).

106

I.H. CAMPBELL AND D. L. SMITH

light emission is considerably faster because the carrier recombination times are between 1 ns and 20 ns. The devices previously described were operated at steady state with current densities of the order 0.1 A/cm 2 or below. Steady state operation is of interest for many display applications. However, there are also applications where high-intensity pulsed operation is desired. In this mode of operation considerably higher electric fields and carrier densities occur in the diodes. The extent to which material parameters, such as mobility, and device behavior extracted from measurements performed at much lower current densities can be extended to these high bias conditions is addressed using pulsed high-current measurements. The M E H - P P V structures used for the pulse bias measurements were designed to minimize series resistance. The devices consist of thin M E H - P P V films ( < 1 0 0 nm) sandwiched between thick Pt and Ca electrodes. Test structures made with thinner Ca films had similar I-V characteristics at low current density, but were dominated by series resistance effects at high current density. The electron mobility of M E H - P P V is much smaller than the hole mobility and because of this large mobility difference, holes dominate the current flow in these structures. The electrical properties of the devices are essentially space charge limited hole only devices. The injected electrons recombine close to the cathode and do not significantly alter space charge in the device. The measured current-voltage characteristics were acquired using both steady state and pulsed techniques at room temperature. The pulse duration and the rise time of the electrical excitation were 250 ns and 15 ns, respectively, and the repetition rate was 10 Hz. The transient current was measured using an inductively coupled current probe. The devices were designed so that Joule heating was not significant. Figure V.19 is a log-linear plot of the measured (solid and crosses) and calculated (dashed) current density as a function of applied bias for structures with 40-nm and 60-nm thick M E H - P P V layers. 211 The steady state I-V measurements are shown as solid lines and the pulsed measurements as crosses. There is good agreement between the measured and calculated I-V characteristics over six orders of magnitude in current. The fit to the I-V measurements yielded mobility parameters #o = 7.8 x 10-7 cm2/Vs and E o = 4 x 104 V/cm. Figure V.20 shows the calculated hole density and electric field as a function of position for the 40-nm device at biases of 3 V (0.1 A/cm 2) and 15 V (1 x 103 A/cm2). 211 The hole density is shown on the left vertical axis and the electric field on the right. The hole injecting contact (Pt) is at the origin and the electron injecting contact (Ca) is at the other end of the device (40 nm). In both cases, the electric field and hole density are nonuniform due to the 211 I. H. Campbell, D. L. Smith, C. J. Neef, and J. P. Ferraris, Appl. Phys Lett. 75, 841 (1999).

PHYSICS OF ORGANIC ELECTRONIC DEVICES

107

103 Pt/Ca ~

40 nm +,f+

102 +,,,

~

101

J

~ -I-It + I+

,4.. -v

60 n m

f

=~ 10~ 10 "1 L) 10-2 10 -3 _ 0

t,

, 10

5

. . . . . 15

20

Bias (V) FIG. V.19. Measured (solid, CW, and crosses, pulsed) and calculated (dashed) currentvoltage characteristics for Pt/MEH-PPV/Ca structures with polymer thickness of 40 nm and 60 nm. Positive bias corresponds to hole injection from Pt (from Ref. 211).

space c h a r g e of the injected holes a n d the hole density is b e l o w 1018 c m - 3 in the b u l k of the material. T h e electron density ( n o t s h o w n ) is only significant n e a r the e l e c t r o n injecting c o n t a c t d u e to the low m o b i l i t y of electrons in this material. These calculated carrier density profiles do n o t d e p e n d s t r o n g l y on the a s s u m e d c h a r g e carrier mobility. F o r a given device , 1019

, ~"

~

101 8

~

017 I 1 "-

,.

--

]4 -3

~,

-~ 2

1016 0

~ ~- -"/ -' 10

"--'---""~ ' 20

' 30

1 0

~

-1 40

Position (nm) FIG. V.20. Calculated carrier density (solid) and electric field (dashed) profiles for the 40-nm thick device of Fig. V.19 at applied biases of 3V (0.1 A/cm 2) and 15V (103 A/cm2), respectively. The Pt contact is at left and the Ca contact is at the right (from Ref. 211).

108

I.H. CAMPBELL AND D. L. SMITH

geometry and applied voltage the magnitude of the current depends on the value of the mobility but the injected charge density does not. The maximum electric field and carrier density is about 4 x 10 6 g / c m and 1 x 1018 cm-3, respectively. These results demonstrate that an electric field dependent mobility, without carrier density dependence, provides an accurate description of hole transport in this polymer over this range of field and carrier density. 16. FIELD-EFFECTTRANSISTORS Organic field-effect transistors have been investigated since the late 1980s. Recently, they have achieved performance comparable to inorganic thin-film transistors. The structure of an organic FET is shown in Fig. 1.5. The organic FET operates in a manner analogous to inorganic thin-film transistors employing undoped semiconductor layers and doped semiconductor contacts. Although a detailed, quantitative model for organic FETs has not yet been demonstrated, much of the essential device physics is clear. Organic FETs operate in a charge injection mode where the charge is injected into the organic material from the metallic source and drain contacts. The charge injection mode is distinct from both charge accumulation and charge inversion modes of operation. Conventional doped, inorganic FETs operate in a charge inversion regime. For example, a p-channel transistor uses p-type semiconductor contacts and an n-type doped semiconductor layer in which the channel is formed. 74 When sufficient gate bias is applied to the structure, an inversion layer is formed consisting of a thin p-type region in the n-type doped semiconductor adjacent to the gate insulator (the channel). The p-type doped semiconductor contacts make good electrical contact only to this inversion layer. In the on state, the current flows through the inversion layer. In the off state, the leakage current is very low because the structure consists of a reverse biased diode in series with a forward biased diode. This high off state resistance is one of the principle advantages of doped inorganic transistors. In contrast, organic FETs operate in a charge injection regime. Because the organic material in which the channel is formed is undoped, all of the charge in the channel is injected from the contacts. Space charge limited contacts, as described in Section III, are used. At zero gate bias, the organic material contains no free charges except for a thin region (a few nm) near the contacts that contains charge thermally excited from the metals into the organic material. When a gate bias is applied to the structure, charges are injected into the organic material from both source and drain contacts, forming a thin sheet of charge adjacent to the gate insulator (the channel). In the on state, the current flows through this thin sheet of charge. In the

PHYSICS O F O R G A N I C E L E C T R O N I C DEVICES

109

off state, the leakage current is low because of the high intrinsic resistivity of the undoped organic film. The charge injection regime is not equivalent to a charge accumulation regime. Transistors operating in an accumulation regime utilize doped organic or semiconductor layers but, instead of forming an inversion layer, the channel is formed by biasing the gate to accumulate charge of the same type as the layer doping. For example, a p-type transistor has the gate biased so that additional holes accumulate adjacent to the gate insulator. Devices operating in the accumulation regime have large leakage currents and their total current is sensitive to the organic layer thickness. The poor leakage current and sensitivity to film thickness occur because the contacts are not rectifying, as they are in the inversion regime, so current can flow through the doped regions of the organic material independent of the gate bias. It is therefore not desirable to operate in an accumulation regime. A device model of an organic FET needs to include the charge transport properties of the organic material and the electrical properties of the metallic contacts. The most important properties that must be included are the charge density and electric field dependence of the carrier mobility and the contact interface electronic structure. Because the charge density and electric field in the organic material vary by many orders of magnitude, including an accurate description of the mobility is important for calculating the charge density and potential profiles within the device. At present, the charge transport properties of organic materials, particularly at the high carrier densities typical of FETs, are not very well understood. A detailed numerical device model is likely to help explain discrepancies between conventional FET models and organic device measurements. To date, organic FETs have been analyzed using inorganic thin-film transistor models that use electric field and charge density independent mobilities and ignore the details of the contacts. 9 The drain-source current in such a model is W IDS =-~--

IzCi[V6s- VoJVDs Linear Region

W IDS = ~-~-,#Ci[Vcs- Vo] 2 z/_.

Saturation Region

(5.26) (5.27)

where W is the device width, L is the channel length, Ci is the capacitance per unit area of the insulating layer, Vcs is the gate-source voltage, Vos is the drain-source voltage and Vo is the offset voltage. The linear region applies for Vos << (Vas - Vo) and the saturation regime for Vos > Vas where los is constant, independent of Vos. The offset voltage accounts for several factors, including the effect of work function differences between the gate- and

110

I.H. CAMPBELL AND D. L. SMITH -100

Vcs = -5 V

-80 -4 V

-60 . ~ -40

-3 V " -20

-2 V

I

0

-1

I

I

-2 -3 VDS (V)

O,+IV -4 -5 I

FIG. V.21. Transistor current-voltage curves for a pentacene FET fabricated on a Si substrate that serves as a gate electrode, with a YSZ gate insulator and Pt source and drain electrodes.

drain-source contacts and trapped charges in the gate insulator. It is typically determined by fitting to measured FET I-V characteristics. The charge carrier mobility is usually determined in the saturation region from the slope of I~/s2 vs Vcs and the known structure parameters. Using this approach, the best organic FET mobilities at room temperature are about 1 cm2/Vs, comparable to amorphous silicon. To show an example of organic F E T properties, devices were fabricated on p-type Si substrates using yttria stabilized zirconia (dielectric constant about 25) as the gate insulator and pentacene as the organic material. The structures used Pt source and drain contacts, a 150-nm thick gate insulator, and a gate length of 10 #m with W/L = 350. The drain-source current as a function of gate bias is shown in Fig. V.21. The offset voltage is close to zero as expected, because the ionization potential of pentacene and the valence band of silicon are both about 5 eV and the insulator is relatively free of trapped charge. The mobility determined from these I-V characteristics using Equations (5.26) and (5.27) is about 10-1 cm2/Vs in both the linear and saturation regimes of the transistor. This mobility is almost two orders of magnitude higher than that determined from the time-of-flight and space charge limited current-voltage measurements shown in Section IV. The origin of this higher mobility may be due either to an increase in the mobility with carrier density or to an anisotropic mobility characteristic of

PHYSICS OF ORGANIC ELECTRONIC DEVICES

111

-5

-4

-3 -3 ~-2

-2 V

-1

-1 V

0 0

-1

-2

-3

-4

vo FIG. V.22. Transistor current-voltage curves for a pentacene FET fabricated on a transparent polycarbonate substrate. (Reprinted with permission from Ref. 32. Copyright 1999 American Association for the Advancement of Science. Readers may view, browse, and/or download this material for temporary copying purposes only, provided these uses are for noncommercial personal purposes. Except as provided by law, this material may not be further reproduced, distributed, transmitted, modified, adapted, performed, displayed, published, or sold in whole or in part, without prior written permission from AAAS.)

the film structure. Reliably assessing the film structure, e.g. using x-ray diffraction, is difficult because the conduction occurs in a layer only a few nm thick adjacent to the insulator. Organic FETs can be made flexible, low cost, lightweight, and transparent. High-performance pentacene FETs have been fabricated on transparent, plastic polycarbonate substrates. The current-voltage characteristics of these transistors are shown in Fig. V.22. 32 The mobility of these FETs was 0.2 cm2/Vs as determined in the saturation regime. Mechanically flexible, organic transistors with good performance have also been fabricated. 29 These transistors used polymers for the substrate, the gate, drain, and source contacts, the gate insulator, and the active material in which the channel is formed. The most important properties of field-effect transistors are their frequency response and transconductance. The response time of organic FETs is determined by the carrier transit time. The transit time is ~ = LZ/#VD, which is about 10 -6 s for typical F E T parameters. The response time of the F E T is therefore similar to that of LEDs. The higher mobility observed in

112

I.H. CAMPBELLAND D. L. SMITH

FETs compensates for the greater distance between the electrodes. The transconductance, i.e. C~ID/t?Vcat constant drain voltage, of the organic FETs (in the linear region), again using a constant mobility model, is g - W#CiVD/L. For the high dielectric constant FETs operating at low voltage previously discussed the transconductance is about 10-6 mhos. Organic transistors are being explored for chemical sensor applications. Because the current in an organic transistor is carried in a thin organic layer about 3 nm thick, it is straightforward to produce FETs with very thin active organic layers that can be directly exposed to the ambient. The current-voltage characteristics of the transistor are sensitive to vapor molecules. The vapor molecules can change both the mobility and density of the charge carriers in the organic film. Each molecule produces distinct changes in the transistor response that can be used to identify the vapor molecule. The chemical specificity and strength of the sensing interaction can be tuned by adding chemical functional groups to the active organic molecules or by incorporating other functional organic materials into or on top of the active organic layer. An example of organic transistor response is shown in Fig. V.23. The I-V characteristics are shown for the following exposure conditions: ambient, and 1 part-per-thousand each of methanol, ethanol, p-xylene, o-xylene, and tetrahydrofuran (THF). In all cases, exposure to the vapor decreases the drain-source current. The changes in the drain-source current are a function of the drain-source voltage, i.e. the shapes of the I-V curves are different. The change in shape of the I-V characteristic shows that it is possible to distinguish molecules from their distinct responses as a function of voltage. To illustrate this, consider the differential response, R, of the transistor. The differential response is the derivative of the drain-source current with respect to the drain-source voltage of the reference ambient response divided by the derivative of the drain-source current with respect to the drain-source voltage of the sensor gas response at constant gate voltage or

( e

~

dlR~

( dls~

(5.28)

where IR(s)[VRts)] is the drain-source current [drain-source voltage] of the reference ambient (sensor gas) response. The lower panel in Fig. V.23 shows the differential response of the transistor upon exposure to different vapors. The xylene, THF, and alcohol molecules have substantially different differential response as a function of voltage. The 2 xylene molecules have peaks at about 2-3 V and the 2 alcohol molecules have peaks at about 5-6

113

PHYSICS OF ORGANIC ELECTRONIC DEVICES

-6

Reference Ambient

-5

/ _/

Ethanol Methanol

-4

o-Xylene ~p-Xylene

-2 -1

_•... ,,

0 0

~ o

.,.

,..,.

.

.

.

.

.

.

.

.

.

I

.

.

.

.

.

.

THF .

.

.

.

.

,-..

-

i

,.

-2

-'.

-8

,

lO

VDS (V) ,

__ -

-

~. X .lene ~.- , ,

~I /

r i= Q

r~

o-Xylene~

|

,

,,,

,

THF ..-.'-~.. ~ . . M e t h a n o l .\ ~ / / ~..-~ Ethanol

"Y / ",/

/

"~ " - Y

/

X'X-\

It t

~3

ol--t

1 0

-2

-4

-6

-8

-10

VDs (V) FIG. V.23. Effect on the current-voltage curves of exposing a pentacene FET to various organic solvents (upper panel), differential response (Eq. 5.27) of the transistor current-voltage curves (lower panel).

V. The THF molecule has a sharper peak at about 3 V. The large amount of information available from a single transistor in a chemical sensor array will add significantly to its ability to discriminate and identify molecules. The response time of these organic thin-film transistors is a few seconds at concentrations of about 1 part-per-thousand.

Vl. Summary and Future Directions This article discussed the essential device physics governing the operation of organic electronic devices focusing on aspects relevant to light-emitting

114

I . H . C A M P B E L L AND D. L. SMITH

diodes and field-effect transistors. In about a decade of research, most of the basic physical mechanisms governing organic LED performance have been identified and many important problems have been resolved. Organic electronic devices use undoped, insulating organic materials as light-emitting and charge-transporting layers. Carriers are injected into the insulating organic materials from metallic contacts. The organic materials are conjugated small molecules or polymers that are either vacuum evaporated or cast from solution to produce disordered thin films. There are no lattice matching issues for substrates or heterostructures such as occur for inorganic semiconductors. The ease and flexibility of fabrication are major advantages of organic electrical devices. The basic device physics of organic electronic devices is distinct from doped inorganic crystalline semiconductor based devices. The rectification of inorganic diodes is due to the electrical junction between oppositely doped, p- and n-type regions of the inorganic semiconductor. In contrast, the rectification of organic diodes is caused by the use of asymmetric metal contacts. One metal contact is only able to inject electrons efficiently and the other contact only injects holes efficiently. Therefore, the properties of metal/organic contacts, rather than doping profiles, determine the behavior of the device. Transistor action in conventional, inorganic semiconductor FETs is due to the formation of an inversion layer in the doped semiconductor under the gate insulator. In undoped organic FETs, transistor action is due to injection of charge into the insulating organic film. The injected carriers form a thin sheet of charge adjacent to the gate insulator that is the conducting channel. Charge transport in the disordered organic thin films occurs by hopping from site to site rather than by band transport. This hopping mechanism leads to mobilities in organic films that are orders of magnitude lower than inorganic semiconductor mobilities and that depend strongly on electric field and carrier density. For the conditions typical of organic LEDs, the carrier mobilities vary from about 10 -8 cmZ/Vs to 10-2 cmZ/Vs. These low mobilities require LED device designs that utilize transport normal to thin films (about 100 nm thick) rather than parallel to the film. Organic LEDs have been demonstrated with display level brightness at bias voltages below 3 V, internal quantum efficiencies above 25%, and continuous operating lifetimes in excess of 9 years. They are now used in a number of display products and their development for commercial applications is accelerating. Organic field-effect transistors have not yet been commercially developed. To be technologically useful, FETs require carrier mobilities of about 1 cmZ/Vs or higher. To achieve these relatively high mobilities in organic thin films it is necessary to use either high electric fields, well-ordered thin organic films, or high carrier densities. Organic FETs have recently been

PHYSICS OF ORGANIC ELECTRONIC DEVICES

115

produced with carrier mobilities over 1 c m 2 / V s and operating voltages below 5 V, comparable to widely used amorphous Si thin-film transistors. These devices are becoming promising for large-area, low-cost electronics applications. The scientific and technological interest in these materials and devices is accelerating and new products and applications continue to emerge. Examples include optically pumped solid state organic lasers and perhaps electrically pumped organic diode lasers, 212-221 solid state light-emitting electrochemical cells, 222-226 and broad spectrum white light-emitting diodes. 227'228 The basic device physics of organic LEDs is becoming relatively clear and the focus of much device research is shifting to understanding the unique aspects of organic FETs. 229 An organic transistor device model incorporating the relevant charge injection and transport properties of organic materials is needed. The success of organic electronic devices has led to wide interest in designing new organic materials with properties suited for specific applications. The objective is to be able to design an organic molecule or polymer 212 F. Hide, M. A. DiazGarcia, B. J. Schwartz, M. R. Andersson, Q. B. Pei, and A. J. Heeger,

Science 273, 1833 (1996). 213 F. Hide, B. J. Schwartz, M. A. DiazGarcia, and A. J. Heeger, Chem. Phys. Lett. 256, 424 (1996). 214 M. D. McGehee, M. A. DiazGarcia, F. Hide, R. Gupta, E. K. Miller, D. Moses, and A. J. Heeger, Appl. Phys. Lett. 72, 1536 (1998). 215 N. Tessler, G. J. Denton, and R. H. Friend, Nature 382, 695 (1996). 216 N. Tessler, N. T. Harrison, and R. H. Friend, Advanced Materials 10, 64 (1998). 217 N. Tessler, D. J. Pinner, V. Cleave, D. S. Thomas, G. Yahioglu, P. LeBarny, and R. H. Friend, Appl. Phys. Lett. 74, 2764 (1999). 218 V. G. Kozlov, P. E. Burrows, G. Parthasarathy, and S. R. Forrest, Appl. Phys. Lett. 74, 1057 (1999). 219 C. Zenz, W. Graupner, S. Tasch, G. Leising, K. Mullen, and U. Scherf, Appl. Phys. Lett. 71, 2566 (1997). 220 S. V. Frolov, M. Shkunov, Z. V. Vardeny, and K. Yoshino, Phys. Rev. B56, R4363 (1997). 221 S. V. Frolov, Z. V. Vardeny, and K. Yoshino, Phys. Rev. B57, 9141 (1998). 222 Q. B. Pei, G. Yu, C. Zhang, Y. Yang, and A. J. Heeger, Science 269, 1086 (1995). 223 j. Gao, G. Yu, and A. J. Heeger, Appl. Phys. Lett. 71, 1293 (1997). 224 S. Tasch, L. Holzer, F. P. Wenzl, J. Gao, B. Winkler, L. Dai, A. W. H. Mau, R. Sotgiu, M. Sampietro, U. Scherf, K. Mullen, A. J. Heeger, and G. Leising, Synthetic Metals 102, 1046 (1999). 225 L. Holzer, B. Winkler, F. P. Wenzl, S. Tasch, L. Dai, A. W. H. Mau, and G. Leising, Synthetic Metals 100, 71 (1999). 226 y. Yang and Q. B. Pei, Appl. Phys. Lett. 70, 1926 (1997). 227 Z. L. Shen, P. E. Burrows, V. Bulovic, S. R. Forrest, and M. E. Thompson, Science 276, 2009 (1997). 228 R. S. Deshpande, V. Bulovic, and S. R. Forrest, Appl. Phys. Lett. 75, 888 (1999). 229 B. Crone, A. Dodabalapur, Y.-Y. Lin, R. W. Filas, Z. Bao, A. LaDuca, R. Sarpeshkar, H. E. Katz, and W. Li, Nature 403, 521 (2000).

116

I . H . C A M P B E L L A N D D. L. S M I T H I

I

I

I

6

8

I

10

i001 0.01 t

,, 2

4

10

12

Bias (V) 80

,

,'

,

/

,

/ /

60

/ / /

40 O

/

/

O

0

0

20

40

60

80

Input Power (W/cm 2) FIG. VI.1. Calculated device current density (solid lines) and recombination current density (crosses) as a function of bias (upper panel), and output power as a function of input power (lower panel) for single-layer LEDs with different mobilities. The LEDs are 100 nm thick, have equal electron and hole mobilities with Eo = 5 x 104 V/cm for all cases, and with #o values of 10 -6, 10 -4, and 10 -2 cm2/Vs.

so that thin films made from it have specific electrical and optical properties. This goal requires a thorough understanding of the properties of the molecule or polymer and how those properties determine the thin-film behavior. An important objective is to increase the carrier mobility of the organic thin films. For example, increasing the carrier mobility would dramatically improve the power efficiency of the LEDs. To illustrate this point, model calculations of the device current density, recombination current density, and input and output power for single-layer LEDs with different mobilities are shown in Fig. VI.1. The LEDs are 100 nm thick and have equal electron and hole mobilities; the carrier mobility has the Poole-Frenkel form with Eo = 5 x 104 V/cm for all cases and with #o values of 10 -6, 10 -4, and 10 .2 cm2/Vs. The diode with a mobility prefactor of

PHYSICS OF ORGANIC ELECTRONIC DEVICES

117

10 -6 cm2/Vs is typical of current LEDs. The upper panel of Fig. VI.1 shows the calculated device current density (lines) and recombination current density (crosses) as a function of bias for the three diodes. In all cases, the recombination current is nearly equal to the device current but the bias required to reach a given current decreases rapidly as the mobility prefactor is increased from 10 -6 to 10 -2 cmZ/Vs. This decrease in bias significantly improves the device power efficiency. The lower panel of Fig. VI.1 shows the output power as a function of input power for the 3 diodes (solid lines); the dashed line represents a device with 100% efficiency, i.e. equal input and output power. At an input power of 10 W/cm 2, the power efficiencies are 20%, 50%, and 95% (increasing with higher mobility). Increasing the mobility to the 10-2 cmZ/rVs range will allow devices with near ideal power efficiency for input powers up to about 20W/cm 2. This could enable the development of high efficiency, low cost, lighting and other high brightness applications. The development of organic electronic materials and devices is at an exciting time, in some ways analogous to the early development of inorganic semiconductor devices, in which the interaction between physical understanding, improved materials, and new device measurements led to rapid progress in both scientific understanding and technological application. Organic electronic materials and devices have many unique properties, such as large area processing, mechanical flexibility, tunable light emission, chemical sensing interactions, and biocompatibility, which make them attractive for a wide range of applications that are largely inaccessible to conventional inorganic semiconductor devices. The next decade is likely to see continuing rapid progress and exciting new developments in this rich area of science and technology. ACKNOWLEDGEMENTS

The authors are grateful to many collaborators who have made major contributions to this work, including Alan Bishop, David Brown, Brian Crone, Paul Davids, Thomas Hagler, Christian Heller, Michael Joswick, Joel Kress, Richard Martin, Duncan McBranch, Avadh Saxena, Zhi Gang Yu, and Thomas Zawodzinski at Los Alamos, and Nikolai Barashkov, John Ferraris, and Charles Neef at the University of Texas at Dallas. This research was supported by the Los Alamos Directed Research and Development Program, the Defense Advanced Research Projects Agency, and the Department of Energy.