Charge transport in thin films of molecular semiconductors as investigated by measurements of thermoelectric power and electrical conductivity

ELSEVIER

Thin Solid Films 258 (1995) 317-324

Charge transport in thin films of molecular semiconductors as investigated by measurements of thermoelectric power and electrical conductivity J.-P. MeyeFb, D. Schlettwein”,*, D. W6hrleb, N. I. Jaeger” “Universittit Bremen, Institut fiir Angewandte und Physikalische Chemie, Postfach 330440, D-28334 Bremen, Germany bUniversitiit Bremen, Institut ftir Organische und Makromolekulare Chemie, Postfach 330440, D-28334 Bremen, Germany Received

19 July 1994; accepted

14 September

1994

Abstract

Thin films ( - 100 nm) of molecular semiconductors were prepared by physical vapour deposition on quartz substrates equipped with gold electrodes that allowed simultaneous measurement of the electrical conductivity (specific conductivity a) under an applied electric field and of the thermoelectric power (Seebeck coefficient S) under an applied temperature gradient. The measurements were performed in vacua and during exposure to molecular oxygen. The thin film samples were chosen as representatives of three characteristic groups of molecular semiconducting materials: an unsubstituted phthalocyanine (phthalocyaninatozinc( II) (PcZn)), a modified phthalocyanine with an electron-withdrawing ligand (tetrapyrido( 2,3-b;2’3’-g;2”3”-1;2”‘3”‘-q) tetra-azaporphyrinatozinc( II) (TPyTAPZn)) and a perylene pigment (N,N’-dimethyl-3,4,9, IO-perylenetetracarboxylic acid di-imide ( MePTCDI)). The p-type character of PcZn as reported from a number of studies is confirmed whereas the films of TPyTAPZn and MePTCDI showed n-type behaviour. These results are inferred from the sign of S as well as from the influence of oxygen on cr. Both S and g of the three materials were studied under variation of the film temperature. Activation energies for S and for 0 are derived and lead to a discussion of charge carrier generation and charge carrier transport in each film. It is concluded that both a delocalized charge carrier transport as well as a thermally activated hopping mechanism have to be considered to discuss the electrical properties of the samples. Keywords: Conductivity;

Electrical

properties

and measurements;

1. Introduction

Molecular crystals of organic pigments such as phthalocyanines and di-imides of perylenetetracarboxylic acid have a number of electrical properties in the dark as well as under illumination which are similar to those of classical semiconductors. Some molecular crystals behave like intrinsic semiconductors [ l-31, but most of these materials have to be discussed as extrinsic semiconductors, because of interactions with gaseous dopant molecules or impurities [ 1, 4, 51. For example, oxygen increases the conductivity of molecular p-conductors by several orders of magnitude [6-91. Following

*Corresponding 0040-6090/95/$9.50 SSDI 0040-6090(

author. ((2 1995 ~ 94)06369-9

Elsevier

Science S.A. All rights

reserved

Organic

substances;

Semiconductors

such a treatment these materials lead to cathodic photocurrents in photoelectrochemical experiments [lo], to ohmic contacts with metals of high workfunction (e.g. Au) [8, 11, 121, to blocking contacts (Schottky type) with metals of low workfunction (e.g. Al, In) [ 11, 13- 161 and to a behaviour in organic photovoltaic cells which also shows their p-type characteristics [ 17, 181. These p-type properties are generally observed for unPerylenetetracarboxylic substituted phthalocyanines. acid di-imides and substituted tetra-azaporphyrins with a low electron density in the inner ring n-system on the other hand have to be discussed as n-type molecular materials. Their conductivity decreases when exposed to oxygen [5], anodic photocurrents are observed in photoelectrochemical experiments [ 19, 201, contacts with metals [21-231 of high workfunction lead to blocking

J.-P. Meyer et al. / Thin Solid Films 258 (1995) 317-324

318

to provide a broader basis for this ongoing discussion and to extend it into the groups of n-type materials. A discussion of the present results is started in terms of different models of charge carrier transport in molecular materials, insulators and semiconductors.

2. Experimental

PcZn (X = CH) TPyTAPZn (X = N)

MePTCDl

Scheme 1.

contacts (low workfunction leads to ohmic contacts) and in organic double-layer cells they behave as the n-component of an n/p-like heterojunction [24, 251. As the most reliable experiment to assign their conduction type independent of contact phenomena and relaxation steps following illumination, however, thermoelectric power measurements have to be performed [12, 26-281. In this experiment it could be shown that charge transport in unsubstituted phthalocyanines is clearly dominated by defect-electron conduction [29, 301. In first qualitative experiments on compressed powder samples we were able to show that excess electrons dominate the electrical conduction in samples of substituted phthalocyanines with a low electron density in the inner ring rc system as well as in samples of perylene pigments [ 3 11. This classification is important to understand results of experiments which are performed to test the potential of photoelectrochemical, photovoltaic and xerographic devices of this class of materials [32, 331. In addition to this general classification of the conduction type there is also a great interest in the conduction mechanism of molecular organic semiconductors. It is known that molecular orbital (MO) calculations [34, 351 that the bands in these materials are very narrow ( %O.Ol-0.1 eV). Nevertheless many electrical properties in the dark and under illumination can be discussed in terms of the band model [36-381. On the other hand from detailed time-of-flight (TOF) experiments [39] and also from theoretical calculations [40] there are indications that the charge carrier mobility in molecular semiconductors is thermally activated so that a hopping transport has to be presumed. In this study the temperature dependence of the thermoelectric by the Seebeck coefficient S power, characterized as well as the temperature dependence of the specific electrical conductivity 0, is studied for thin films of NJ’-dimethylphthalocyaninatozinc( II) (PcZn), 3,4,9, lo-perylenetetracarboxylic acid di-imide (MePTCDI) and tetrapyrido(2,3-b;2’3’-g;2”3’‘-1;2”’3”’-q) tetraazaporphyrinatozinc( II) (TPyTAPZn) (Scheme 1) as representatives of the three groups of materials in order

details

PcZn and MePTCDI were obtained from Aldrich. TPyTAPZn was synthesised as described in Ref. [41]. All substances were purified by zone sublimation. The films were prepared by vapour deposition on quartz substrates with the dimensions of 20 x 50 x 0.5 mm3 obtained from the Westdeutsche Quarzschmelze. The substrates were carefully cleaned with ethanol and distilled water and dried with gaseous nitrogen. In the first step gold electrodes (3 x 15 mm2, thickness 100 mm) were vapour deposited at spacings of 7, 14 and 28 mm. The gold wire was obtained from Balzers. After preparation of the gold electrodes the sample film was grown ex situ at a rate of 0.3 nm s-’ up to a thickness of 100 nm. The samples were than transferred through air to the analytical chamber and positioned on two copper blocks which were resistively heated. The temperature of each block was controlled independently with a precision of f 1 K by use of two Omron ESCW temperature controllers. All experiments were performed in a high vacuum system (base pressure, 5 x lo-‘Torr). Contacts to the sample were directly made by Ni/CrNi thermocouples attached to the gold electrodes by use of silver paste. Electrical conductivity G, Seebeck coefficient S and temperature T were thereby measured at the same position on the sample. The thermopower was determined by measuring the voltage for a temperature gradient of 5- 10 K over the samples by using a Keithley 610C electrometer. The temperature gradient was adjusted by heating the copper blocks individually to different temperatures. In a second step the polarity of the temperature gradient was switched and the corresponding voltage was determined. This procedure was repeated until contact values were found. Measurements always took place on both poles of each thermocouple. The difference between the measured values of S was almost the expected 40 pV K-’ for Ni/CrNi thermocouples. The small thermopower of the gold electrodes (2-2.5 uV K-‘) can be neglected compared to the measured voltages of 300400 uV K-‘. The specific electrical conductivity was determined independently from the thermopower measurements. Under a constant temperature of the copper blocks and all over the sample the current was measured at different applied voltages. The length of the sample equals the spacing of the gold electrodes, its width equals the

319

J.-P. Meyer et al. 1 Thin Solid Films 258 (1995) 317-324

width of the substrate (1.5 cm) and as its height the average film thickness (100 nm) is assumed. For measurements of the conductivity under exposure to oxygen the chamber was flooded with oxygen ( 299.998 vol.‘%; N,, H,O, Ar, CO, and C,H,, contents < 5 volume per million) up to a pressure of 200 mbar. To control the solid state structure of the samples which also influences the intermolecular interaction of the chromophores and to investigate whether heat treatment led to irreversible changes in the film structure, UV-visible absorption spectra in transmission of the films were collected at room temperature under air using a Perkin-Elmer Lambda 9 spectrophotometer unit.

3. Results All samples were investigated in the linear portion of the current-voltage curve under applied electrical fields ranging from 0.13 to 13 V cm-‘. Higher fields have not been applied in order to avoid charging of the sample due to space charge limitation of the current. Measurements of the conductivity were made for different sample geometries to ensure that the observed phenomena characterize the sample material. The conductivity was found to be inversely proportional to the electrode distance (sample length) and proportional to the film thickness (sample diameter) at average thicknesses above 4 nm [ 51. Measurements of S were only possible for samples with specific conductivities higher than about lo-’ R-’ cm-‘, because of the low signal-to-noise ratio for lower conductivities. Therefore the thermoelectric power for TPyTAPZn and also for PcZn could only be measured for temperatures higher than 340 K. Only for MePTCDI could it be determined over the whole temperature range. The thermovoltage of the samples was always linearly correlated to the applied temperature difference as also reported for nickel phthalocyanine [30] and for metal-free phthalocyanine [37]. For this reason it was possible to determine S for only one temperature difference. After contact with air the conductivity of thin films prepared from MePTCDI or TPyTAPZn could not be measured using our experimental set-up (specific conductivity lower than lo-” 0-l cm-‘). After heating to lOO- 150 “C the conductivity increased by several orders of magnitude and could therefore be studied in detail after this treatment. A constant specific conductivity was reached only after annealing at these temperatures for some hours. Fig. 1 shows the the conductivity of MePTCDI after heating in vacuum and after supply of oxygen. In both cases the measured values are understood by simple thermal activation as seen from the Arrhenius plot. The

-200

.3 -300.

z

.5

*

1 -6004

.7

0.0030

0.0026 l/T

Fig. the 100 and

b $

0.0034

.9

[l/K]

1. Temperature dependence of the thermoelectric power ( 0) and specific conductivity (*) of a thin film of MePTCDI (thickness, nm) as reached after annealing the film in vacua at 383 K for 3 h specific conductivity after its exposure to oxygen ( x ).

initial increase of CJduring the annealing step is reversed by exposure to pure oxygen. The specific conductivity decreases again by orders of magnitude. The amount of impurities in the oxygen supply (N,, Ar, CO, and C, H,) is negligible. By exposing the film to 0.1 Torr H,O in a separate experiment it could be shown that the only other component which should be discussed, H,O, has no influence on the conductivity of the film. Therefore oxygen is the part of the air which interacts with the organic molecules. The activation energy EA of 0.2 eV as measured for the film after annealing at 383 K for 3 h increases to 0.42 eV when the film is exposed to 200 mbar oxygen. Although TPyTAPZn belongs to a different class of pigments with no structural analogy to MePTCDI, it behaves very similarly in conductivity and in its interaction with oxygen. As in the case of MePTCDI the films have to be annealed before the conductivity can be measured. Only then does the specific conductivity increase above the detection limit of our experiment. After exposure to oxygen it drops again by orders of magnitude [5]. Obviously it is typical for n-type molecular semiconductors that the electron acceptor oxygen reduces the number of charge carriers probably by compensation of donors which are present also in the as-prepared film. In MePTCDI and TPyTAPZn the size of the thermoelectric power is also very similar and the sign of S confirms that the charge carriers in both molecular semiconductors are electrons (Figs. 1 and 2). The plot of S versus 1/T is linear as expected from models for transport properties of semiconducting materials as discussed below. The activation energy as determined from the temperature dependence of S for both materials is smaller than the activation energy of 0. This gives a strong hint towards different mechanisms involved in the generation of the two effects.

320

J.-P. Meyer et al. / Thin Solid Films 258 (1995) 317-324 -200I-4.2

-3.5 I

-. x

-4.6 P 5 3

-300

D -5

B

v)

-4.5 i D

B -5.5.

-5.4

-400

-5.6

0.003

0.0026 l/T

““x1”’ x

xi

*i

,Q

E,t,=0.33eV

-4

_

D $

--4.5

AE = 0.31 e V 1601

0.0032

0.0026

l/T

0.0026

0.0032 l/T

(l/K]

Fig. 2. Temperature dependence of the thermoelectric power ( q) and the specific conductivity (*) of a thin film of TPyTAPZn (thickness, 80 nm) as reached after annealing the film in vacua at 433 K for 24 h.

360

-6.5

I-5

[l/K]

Fig. 3. Temperature dependence of the thermoelectric power ( 0) and the specific conductivity (*) of a thin film of PcZn (thickness, 100 nm) as reached without annealing in vacua.

Contrasting behaviour was found in the case of PcZn. The film could also be measured before the initial annealing step. Data for S and CTas obtained in vacuum under increasing temperatures are depicted in Fig. 3. The sign of S confirms defect electrons to be the majority charge carriers. At moderate temperature a linear plot is obtained of S vs. l/T with a slope corresponding to an activation energy of 0.31 eV. The thermoelectric power starts to increase at temperatures above ~350 K. The conductivity was measured simultaneously and starts to drop with increasing temperature at ~350 K. Decreasing D leads to the increase of S as expected from the band model [38]. Heating the sample in vacuum leads to a strongly decreasing (T below the limit necessary for the measurement of S. An activation energy for CJof 0.33 eV is obtained from the linear portion up to 355 K. Obviously desorption of oxygen starts and 0 decreases. This behaviour could be monitored until 0 dropped below the detection limit. An intermediate situation is depicted in Fig. 4. Linear

[l/K]

I 0.00 I36

Fig. 4. Temperature dependence of the specific conductivity of a thin film of PcZn as reached after annealing the film in vacua (*) and after its exposure to oxygen ( x).

behaviour is observed. Although g has dropped by an order of magnitude the same activation energy of 0.31 eV is obtained. The rather high conductivity indicates that the film is still doped with oxygen under these conditions. The original situation could be re-established by subsequent exposure to pure oxygen (Fig. 4). r~increases to nearly the same values as obtained before the thermal treatment and the film again shows clear deviations from the Arrhenius behaviour. The activation energy of 0.32 eV as determined from the linear part of the plot still coincides with the other values. This had not been observed in earlier studies in which the activation energy either decreased [l] or increased [5] after treatment with oxygen. Obviously quite different trap distributions can be obtained in these materials even if the same method for film preparation is employed. A similar conclusion was drawn from photoelectrochemical experiments on vapour-deposited films of PcGaCl in which even a different direction of the photopotential could be established [42]. The effect of O2 could be compensated only by annealing in a reducing H2 atmosphere. Only under such conditions a doping situation close to the intrinsic material was established in phthalocyanine crystals and an activation energy for the electrical conductivity of 1.07 eV was obtained [ 361. The temperatures which were applied during the initial annealing steps and during the measurements clearly exceeded the deposition temperatures and reached a range in which changes in the crystal structure of the samples should be considered [43]. Visible absorption spectra are a sensitive tool to monitor this kind of change. The size and the pattern of splitting for the Q band of phthalocyanines is a good indicator of the crystal structure. In the present experiments no changes were observed for any of the investigated films. Absorption spectra before and after the heat treatment showed no changes in either band positions or absolute

321

J.-P. Meyer et nl. 1 Thin Solid Films 258 (1995) 317-324

Table 1 Thermoelectric power S,, determined at 340 K and specific electrical conductivities at room temperature after annealing oz9s, a and after exposure to oxygen gz9s, o2 and the thermal activation energies AE of S and E,,, = AE + E, of 0 as obtained from temperature-dependent measurements

MePTCDI TPyTAPZn PcZn

s34”

AE

(~IV K-l)

(eV)

- 394 -351 +285

0.14 0.11 0.31

1.3 X 1o-4 2.5 x 1o-6 8.1 x lo-’

or relative intensities. As expected for the applied conditions, films of PcZn are characterized by spectra which are typical for the cc-polycrystalline modification [ 11.

4. Discussion The variation of the sample geometry has shown that bulk properties of the materials were measured and that contact phenomena controlled by injection of carriers from the metal electrodes are excluded to cause the observed properties. Further, no changes in the intermolecular arrangement are seen for any of the sample materials. Therefore the observed results have to be discussed as arising from changes within the bulk of the films which have a constant crystal structure. Properties which are characteristic for the different materials are obtained. The results are summarized in Table 1. A general expression for the thermoelectric power has been given by Fritzsche [44]:

&_k e

(1)

where k is Boltzmann’s constant, e the elemental charge, E the respective energy, EF the Fermi energy, T the temperature, c the specific conductivity of the material and o(E) the portion of the overall conductivity provided by the state with the energy E. This expression describes the thermopower of metals, semiconductors and insulators as well as that of crystalline and noncrystalline materials. For extrinsic semiconductors, conduction has to be considered only in one band and the familiar expressions [44] can be obtained from Eq. (1):

(2)

(AE -t 4, hvx.a (eV)

~298.02 (W’ cm-‘)

0.20 0.33 0.32

1.0 x 10-x

0.42

1.8 x 1O-5 1.4 x IO-5

0.33 0.31

ing the energy dependence of the relaxation time for holes and A, a constant describing the energy dependence of the relaxation time for electrons. For intrinsic semiconductors with conduction in two bands and the Fermi level being in the gap centre the integration of Eq. (1) yields

with b being the ratio pL,/pr, of mobilities for electrons in the conduction band and defect electrons in the valence band and A a weighted average of A, and A,. The thermoelectric power of intrinsic semiconductors generally is smaller than that of doped materials and depends on the mobility ratio 6. In inorganic semiconductors which are described by the band model the temperature dependence of p cc T-” generally is negligible compared to the thermal activation of the charge carrier concentration and therefore does not influence the temperature dependence of a(E) nor that of (T. If, however, delocalization of energy levels and thereby charge carrier mobilities become rather small a thermally activated mobility has to be assumed, as for instance in the case of polaron hopping [45]: P = p.

EH

kT

with p. being the prefactor mobility and EH the activtion energy of the mobility. The exponential factor for p appears in CT(E)as well as in o of Eq. (1). Even in this case the temperature dependence of the mobility therefore does not influence the temperature dependence of the thermopower. However, the temperature dependence of the mobility has to be considered in the exponential term of the specific conductivity: a=c,exp(-

with S, being the Seebeck coefficient for n-type semiconductors, S, the Seebeck coefficient for p-type semiconductors, EC the position of the conduction band edge, E, the valence band edge, A, a constant describ-

( 1

f-w -

$$)~Oexp(-~)=Coexp(-

AE$EH)

(6) with co being the degenerated concentration of charge carriers, rro the pre-exponential factor of the conductiv-

322

J.-P. Meyer et al. / Thin Solid Films 258 (1995) 317-324

ity and AE = EC - EF for n-type materials and AE = EF -E, for p-type materials. Differences in the temperature dependence of (T (EA = AE + EH) and S (AE) as observed for MePTCDI and TPyTAPZn in the present experiments can principally be discussed in two ways. If the material under investigation is considered as intrinsic the observed differences in activation energies would have to be understood as arising from a different mobility of electrons and holes, because the factor (b + l)/(b - 1) is always smaller than 1. This would lead to an activation of S being smaller than EC - E,. Experiments from the literature as described above and the conductivity measurements on the films before and after exposure to oxygen as investigated in this study, however, have shown that the conductivity in the materials is of extrinsic character. It is known that organic pigments not only are very sensitive to gaseous dopants and also impurities but their extrinsic character is also seen in photoelectrochemical experiments [ 461. The assumption of an extrinsic material leads to an alternative approach. The difference in activation energies would then be explained by a thermally activated mobility leading to AE + EH > AE. Conduction is dominated by charge carriers of one sign only, defect electrons caused by oxygen in the case of PcZn and excess electrons caused by donor impurities in the case of MePTCDI and TPyTAPZn. The validity of Eqs. (2) and (3) can be assumed, however taking into account a possible temperature dependence of p (Eq. (5)). The validity of this approach is further confirmed by a comparison of the frontier orbital gap (“bandgap”) obtained from optical emission spectroscopy [5] with the activation energy of CJas derived from the present experiments. The bandgap of MePTCDI is 2.2 eV. The highest activation energy of 0 indicates a difference of only 0.2 eV from the Fermi energy to the respective frontier orbital which is much smaller than half the gap. For TPyTAPZn a value of 1.6 eV has been obtained from optical data. The highest activation energy derived from conductivity is only 0.3 eV. These differences of the values show the presence of doping levels in these materials which are responsible for the observed conduction behaviour. Activation energies of the conductivity were also measured for the films prepared in situ under ultrahigh vacuum conditions [5]. It was found in these experiments that MePTCDI and TPyTAPZn behave as n-type materials even without any contact to the atmosphere or other doping surroundings. Contamination with impurities from synthesis or from the deposition step has to be assumed. In the case of MePTCDI it was found that even after zone sublimation 2% of monoimide-monoanhydride remains in the crystals [4]. Intrinsic properties as found for p-type organic pigments [47-491 could not be investigated for the n-type materials yet.

Both n-type materials show a significant difference in the activation energy of 0 and AE obtained from the temperature dependence of S so that a thermally activated mobility has to be assumed for these materials. For MePTCDI the difference of activation energies is in the order of 0.06 eV. In the case of TPyTAPZn the difference EA - AE is 0.22 eV. In both cases the activation energy of p is a relevant portion of the overall EA as obtained in the conductivity. The temperature dependence of charge transport is dominated by the temperature dependence of the mobility in TPyTAPZn and is at least strongly influenced in MePTCDI. A significant thermal activation of a charge carrier’s mobility which in turn leads to the difference in activation energies of thermopower and conductivity in a material is not expected from the classical semiconductor band model and localized mechanisms of charge carrier transport have to be considered. There exist at least three alternative models which can be used to explain the thermal activation of the mobility in these molecular materials. Most studies, especially of ionic crystals with strong interactions between the carrier and the crystal lattice, use the phonon-assisted hopping model of small polarons as mentioned above (Eq. (5)) to explain a significant thermal activation of p [44, 45, 501. For molecular crystals it is not proven whether the polaron binding energy is strong enough to produce relevant polaronic effects. In the model of small polarons the activation energy is given by [ 511 E,,+_-/

(7)

with d, being the polaron binding energy and J the transfer integral. As transfer integrals in molecular solids are typically 0.01 eV [52] the polaronic binding energies of MePTCDI and TPyTAPZn are in the order of 0.15 eV and 0.46 eV. These values are rather large compared to the polaron binding energy of 0.016 eV suggested for molecular crystals of naphthalene [53] and we doubt that they are realistic. Furthermore thermally activated mobility as given in Eq. (5) was explained by multiple shallow trapping due to the presence of a large number of chemical impurities or structural defects [54]. In this case a small effective trap-limited mobility is observed instead of the real microscopic mobility [55]. These trap-limited mobilities would not be considered inconsistent with the band model [38] and the activation energy of the mobility is given by the energy of the trap levels. To check the validity of this model for the present experiments the energy of the traps should be compared to trap levels determined from measurements of space-chargelimited currents (SCLCs). In such experiments trap levels are found 0.7 eV below the conduction band in

J.-P. Meyer et al. / Thin Solid Films 258 (1995) 317-324

MePTCDI [ll]. This value is much bigger than the activation energy of the mobility found in this study which stands in contradiction to the model of trap-limited mobilities and hence to the band model. Trap levels very close to the conduction band should not strongly influence the mobility. Also for this reason the activation energy of 0.06 eV for p cannot be explained by trap limitation. On the other hand the SCLC can also be explained by an exponential fit as assumed for the polaron hopping model [ 111. Alternatively it has been worked out that a temperature dependence of the mobility can also be explained by a hopping model based on a gaussian energy distribution of localized hopping sites. This so-called disorder formalism leads to the following temperature dependence of the mobility [ 561:

(8) with A, being the width of the hopping site manifold. In TOF experiments with N,h”-bis( 2-phenethyl)perylene-3,4,9, lo-bisdicarboximide which is very similar to MePTCDI it has been shown that fluctuations in the energy of the hopping sites are the principal components of the activation energy of the mobility as descibed by the disorder formalism. However, an additional source of activation was also assumed by polaron formation or trapping [39]. If the mobility behaves as described by the disorder formalism the measured values of g especially for TPyTAPZn should deviate systematically from the Arrhenius plot (lower values for high and also low temperatures) and should be better plotted vs. T-* as given in Eq. (8). A significant deviation has not been found in the present experiments but the present results should not be discussed in too much detail as a small but permanent change of the electrical properties of the samples caused by desorption of gaseous dopants as explicitly seen for the film of PcZn cannot be excluded either for the other samples. Furthermore the conductivity data would have to to be plotted in a mixed mode of T-’ and Tp2 and this further complicates the distinction between disorder formalism and polaron hopping. In conclusion it can be pointed out that charge transport of the explored n-type molecular materials cannot be explained by a trap limitation within the band model but that a hopping mechanism has to be assumed. However, to decide whether the disorder formalism or the polaron model supplies the better model for the mechanism of charge transport in these materials we need more precise data. The results for the p-type PcZn are principally different from the results for the n-type materials. Constant activation energies of S and 0 indicate that for this material there is no significant thermal activation of p (E, = 0) and that therefore the band model gives an

323

adequate description (EA = AE). This result is rather surprising as it is not expected from the narrow bands as reported from MO calculations [34, 351. Measurements of the temperature dependence of the mobility in phthalocyanine crystals as reported in the literature do not resolve this contradiction. The temperature dependence as expected from the band model (p cc T-“) was found in photoconductivity measurements on single crystals in vacuum [38, 571 as well as thermally activated mobilities with an activation energy of 0.1 eV [ 581. Further measurements of the thermoelectric power of annealed or even Hz-treated PcZn as well as of the n-type materials after exposure to 0, would help to answer the question whether hopping has to be assumed for samples without larger amounts of gaseous dopants and the band model is applicable to doped molecular semiconductors or whether there is a general difference in the charge carrier transport of the investigated p- and n-type molecular semiconductors.

5. Conclusions Temperature-dependent measurements of the thermoelectric power combined with simultaneous measurements of the electrical conductivity provide a valuable tool to study the charge transport mechanism in films of molecular organic materials. Different conduction types are observed in different molecular semiconductors and seem to be accompanied by different charge transport mechanisms. Unsubstituted PcZn as a p-type material behaves as expected from the band model. However, the investigated n-type pigments have to be described by a hopping model.

Acknowledgments

The authors are grateful to Karsten Koblitz (University of Bremen) or technical assistance, Wolfgang Bijhm (University of Dresden) for fruitful discussions, Tatjana Bogdan-Raj (University of Bremen) for the synthesis of the TPyTAPZn and the Senator fur Bildung und Wissenschaft of the State of Bremen for financial support. (55-2/l 1)

References Molecular Semiconductors, Springer, Berlin, 1985. C. Hamann, Phys. Status Solidi, 20 ( 1967) 481. C. Clarisse, M. T. Riouand S. Robinet, Synth. Met.. 38( 1990) 121. J. B. Whitlock, P. Panayotatos, G. D. Sharma, M. D. Cox, R. R. Sauers and G. R. Bird, Opt. Eng., 32 (1993) 1921. D. Schlettwein, N. R. Armstrong, P. A. Lee and K. W. Nebesny, Mol. Cryst. Liq. Cryst., in press.

[l] J. Simon and J. J. Andre, [2] [3] [4] [5]

324

J.-P. Meyer et al.

/ Thin Solid Films 2.58 (1995) 317-324

[6] H. Yasunaga, K. Kojima, H. Yohda and K. J. Takeya, J. Phys. Sot. Jpn., 37 (1974) 1024. [7] A. Ahmad and R. A. Collins, Phys. Status Sohdi A, 123 (1991) 201. [S] A. K. Hassan and R. D. Gould, Int. J. Electron., 73 (1992) 1047. [9] R. D. Gould and A. K. Hassan, Thin Solid Films, 223 (1993) 334. [lo] D. Schlettwein, M. Kaneko, A. Yamada, D. Wiihrle and N. I. Jaeger, J. Phys. Chew., 95 (1991) 1748. [ 1 l] S. Siebentritt, Disserfation, Universitat Hannover, 1992. [ 121 T. G. Abdel-Malik and G. A. Cox, J. Phys. C., 10 (1977) 63. [ 131 A. K. Hassan and R. D. Gould, Int. J. Electron., 69 (1990) 11. [ 141 F. R. Fan and L. R. Faulkner, J. Chem. Phys., 69 (1978) 3334. [ 151 M. Martin, J. J. Andre and J. Simon, J. Appl. Phys., 54 (1983) 2792. [16] Y. I. Verzimacha, A. V. Kovalchuk, M. V. Kurik, C. Hamann and A. Mrwa, Phys. Status Solidi A, 82 (1984) Kl 11. [ 171 Y. Harima, K. Yamashita and H. Suzuki, Appl. Phys. Lett., 45 (1992) 1144. [ 181 S. Antohe and L. Tugulea, Phys. Scams Solidi A, 128 (1991) 253. [ 191 J. Danziger, J. P. Dodelet and N. R. Armstrong, Chem. Mater., 3 (1991) 812. [20] J. Danziger, J. P. Dodelet, P. Lee, K. W. Nebesny and N. R. Armstrong, Chem. Mater., 3 (1991) 821. [21] L. Kreienhoop, Disserfafion, Universitit Bremen, 1994. [22] W. Kowalsky and C. Rompf, Teubner-Texte zur Physik, Vol. 27, Integrated Optics and Microoptics, Teubner, Stuttgart, 1993, p. 76. [23] M. Hiramoto, H. Fujiwara and M. Yokoyama, J. Appl. Phys., 72 (1992) 3781. [24] S. Giinster, S. Siebentritt, J. Elbe, L. Kreienhoop, 9. Tennigkeit, D. Wohrle, R. Memming and D. Meissner, Mol. Cryst. Liq. Cryst., 218 (1992) 253. [25] C. W. Tang, Appl. Phys. Lett., 48 (1979) 183. [26] L. Atkinson, P. Day and M. G. Price, Phys. Status Solidi, 2 (1970) K157. [27] A. K. Ray, S. A. James, S. C. Thorpe and M. J. Cook, Thin Solid Films, 226 ( 1993) 3. [28] I. Caplanus and G. I. Rusu, Rev. Roum. Phys., 33 (1988) 1117. [29] C. Hamann, Phys. Sfafus Solidi, 12 (1965) 483. [30] W. Waclawek and M. Zabkowska, J. Phys. E., 14 (1981) 618. [31] D. Schlettwein, D. Wohrle, E. Karmann and U. Melville, Chem. Mater., 6 (1994) 3. [32] D. Wiihrle and D. Meissner, Adu. Mater., 3 (1991) 129.

[33] K.-Y. Law, Chem. Rev., 93 (1993) 449. [34] E. Orti, R. Crespo and M. C. Piqueras, Synth. Met., 41-43 (1991) 2647. [35] E. Orti and J. L. Bredas, J. Chem. Phys., 92 (1990) 1228. [36] G. H. Heilmeier and S. E. Harrison, Phys. Rev., 132 (1963) 2010. [37] C. Hamann and M. Starke, Phys. Status Solidi, 4 (1964) 509. [38] H. Meier, Organic Semiconductors, Verlag Chemie, Weinheim, 1974. [39] E. H. Magin and P. M. Borsenberger, J. Appl. Phys., 73 (1993) 787. [40] Z. Gu, Y. Wei and J. Liu, Mater. Sci. Eng. B, 20 (1993) 298. [41] D. Wiihrle, J. Gitzel, I. Okura and S. J. Aono, J. Chem. Sot. Perkin Trans. 2, 136(1985) 1171. [42] T. J. Klofta, T. D. Sims, J. W. Pankow, J. Danziger, K. W. Nebesny and N. R. Armstrong, J. Phys. Chem., 22 ( 1987) 5651. [43] C. Hamann, Organische Festkiirper und organische diinne Schichten, Geest & Portig, Leipzig, 1978. [44] H. Fritzsche, Solid State Commun., 9 (1971) 1813. [45] D. Emin, in L. L. Kazmerski (ed.), Material Science Series: Polycrysrailine and Amorphous Thin Films, Academic Press, London, 1980. [46] D. Schlettwein, N. 1. Jaeger and D. Wiihrle, Ber. Bunsenges. Phys. Chem., 95 (1991) 1526. [47] Y. Sadaoka, Y. Sakai, T. A. Jones and W. Gopel, J. Mater. Sci., 25 ( 1990) 3024. [48] H. Laurs and G. Heiland, Thin Solid Films, 149 (1987) 129. [49] T. G. Abdel-Malik, R. M. Abdel-Latif, M. El-Shabasy and M. Abdel-Hamid, Indian J. Phys. A, 62 (1988) 17. [50] P. Nagels, Top. Appl. Phys., 36 (1979) 113. [51] L. 9. Schein, Philos. Mug. B, 65 (1992) 795. [52] E. A. Silinsh, Organic Molecular Crystals, Their Electronic Stares, Springer, Berlin, 1980, p. 36. [53] V. M. Kenkre, J. D. Andersen, D. H. Dunlap and C. 9. Duke, Phys. Rev. Lett., 62 (1989) 1165. [ 541 N. Karl, Mol. Cryst. Liq. Cryst., 171 ( 1989) 157. [55] E. A. Silinsh, G. A. Shlihta and A. J. Jurgis, Chem. Phys., 155 (1991) 389. [56] H. Bassler, Phys. Status Solidi B, 175 (1993) 15. [57] G. A. Cox and P. C. Knight, Phys. Status Solidi A, 50 (1972) K135. [58] K. Yoshino, K. Kaneto, K. Tatsuno and Y. Inuishi, Technol. Rep. Osaka Univ., 22 (1972) 585.