- Email: [email protected]
Electroluminescence-, conductivity-, and photoconductivity-detected magnetic resonance study of poly(p-phenylenevinylene)-based light emitting diodes
Synthetic Metals, 55-57 (1993) 241-248
241.
ELECTROLUMINESCENCE-, CONDUCTIVITY-, AND PHOTOCONDUCTIVITY-DETECTED MAGNETIC RESONANCE STUDY OF POLY(P-PHENYLENEVINYLENE)-BASED LIGHT EMITTING DIODES
L. S. SWANSON and J. SHINAR
Ames Laboratory - USDOE and Physics Department, Iowa State University, Ames, IA 50011 A. R. BROWN, D. D. C. BRADLEY, and R. H. FRIEND
Cavendish Laboratory, Cambridge University, Madingley Rd., Cambridge CB3 0HE, UK P. L. BURN, A. KRAFT, and A. B. HOLMES
Chemistry Department, University Chemical Laboratory, Lensfield Rd., Cambridge CB2 1EW, UK
ABSTRACT The strong electroluminescence (EL)-detected magnetic resonance of poly(p-phenylenevinylene) (PPV)-based light emitting diodes is compared to the conductivity (o)-, photoconductivity (Oph)- and photoluminescence (PL)-detected resonances. In contrast to the narrow PL-enhancing resonance assigned to polaron fusion into singlet excitons, strong EL- and o-quenching resonances are attributed to the spin-dependent polaron-to-bipolarondecay. While the half-field PL-detected resonance reveals only one triplet exciton resonance, believed to result from triplet-triplet fusion into singlets, the halffield EL- and o-detected resonances yield two distinct triplets. While both are o-enhancing, one is ELquenching. The nature of the two triplet states is discussed.
KEYWORDS: poly(p-phenylenevinylene), electroluminescence, photoluminescence, electroluminescence-detected magnetic resonance, conductivity-detected magnetic resonance, photoconductivitydetected magnetic resonance, polarons, bipolarons, singlet excitons, triplet excitons
INTRODUCTION Over the past twenty years, optically (i.e., photoluminescence (PL)) detected magnetic resonance (ODMR) has provided invaluable insight into various radiative and nonradiative decay processes in a wide array of semiconductors [1-6]. In an ODMR experiment, microwave-induced changes in the PL are measured as the sample is subjected to a slowly swept dc magnetic field. Light emitting diodes (LEDs) have been available even longer than twenty years, yet the authors are aware of only one brief electroluminescence (EL) detected magnetic resonance study of any semiconducting system [7]. 0379-6779/93/$6.00
© 1993- Elsevier Sequoia. All rights reserved
242
Due to their potential advantages relative to previously studied organic LEDs, conjugated polymer-based diodes in general, and poly(p-phenylenevinylene) (PPV)-based LEDs in particular, are currently drawing intense attention [8-13]. The device structure was initially a ~100 nm thick polymer layer sandwiched between indium/fin oxide (ITO) as positive contact and AI as negative contact. The light generation efficiency of these devices was subsequently raised to -1% photon/electron through selection of Ca for the negative electrode [9], and could also be raised by optimizing the emissive properties of the polymer layer through alterationto form copolymers [10,11]. The electronic processes that govern the operation of these devices are of obvious interest. In nondegenerate ground-state systems such as PPV, these include singly charged spin 1/2 polarons (p÷ and p-), doubly charged spinless bipolarons (bp÷÷ and bp--), and neutral singlet and triplet (polaronic) excitons (S 1 and T 1, resp.) [14-17]. Burroughes et al. [8] proposed that the operation of PPV-based LEDs involved double charge injection of electrons and holes at opposite electrodes, transport through the polymer film as polarons, fusion of oppositely charged polarons to excitons, and radiative decay of the singlets. Significant nonradiative decay processes of singlets include intersystem crossing to lowlying triplet states and direct decay to the ground state. Since previous ODMR studies of poly(3alkylthiophenes) (P3ATs), PPVs, and poly(p-phenyleneacetylenes) (PPAs) [3-6] indicated that the polaron levels are close to the singlet level, fission of singlets to polarons may also be significant. Other decay mechanisms clearly affect the EL and/or o. If, as widely believed [16,17], the mobility of bipolarons is significantly lower than that of polarons, the decay of polarons to bipolarons should decrease both the EL and ~. Charged excitations are also believed to act as effective nonradiative decay sites for singlet and triplet excitons [11,18,19]. Of particular importance are triplet-triplet collisions which generate singlets. These were established to be the source of the delayed fluorescence in, e.g., anthracene and related organic materials [20-23]. They are thus suspected to be the source of the ODMR triplet powder patterns observed in P'3ATs, PPVs, and PPAs [3-6]. The spectral dependence of the half-field triplet ODMR of PPV is nearly identical to the PL spectrum, which is entirely consistent with this scenario. The singlets may, of course, also fission to polarons or decay nonradiatively. Triplet-triplet pairs may obviously also directly decay nonradiatively. It should be noted that the above-mentioned processes do not exhaust the possible reactions and decay modes of singlets, triplets, polarons, and bipolarons.
Others include, e.g., the decay of a bipolaron via collision with a polaron.
Detection of magnetic resonance through modulation of the EL may be particularly revealing since, in stark contrast to conventional ODMR, all the excited states generated by carrier injection are inherently nongeminate, i.e., not created by geminate electron-hole pairs. The EL-detected resonance described in this paper is indeed strikingly different and much (~20 times) more intense than the ODMR. In addition to the ODMR and EL-detected resonance, the conductivity (o)- and photoconductivity (Oph)- detected magnetic resonances are also described. The results are entirely consistent with the above picture of transport of injected carders as polarons and radiative decay via singlet genration. However, they also suggest that decay of polarons to bipolarons is a significant competing process which may impose an upper limit on the efficiency of such LEDs. In addition, the EL- and odetected triplet powder pattern resonances suggest that singlet generation via triplet-triplet fusion may
243
also contribute significantly to the EL. The dependence of the EL- and o-detected magnetic resonance on the injected current and the temperature are consistent with the suggested role of charged excitations in the nonradiative quenching of singlet and triplet excitons.
EXPERIMENTAL PROCEDURE Two types of EL devices were studied, ITO-coated glass was used as the positive electrode in both cases. In type A, a layer of the methoxy-leaving group PPV precursor was then deposited by spin-coating, and thermally converted to a ,-60 nm thick partially conjugated copolymer, the PL of which is blue-shifted with respect to PPV [10]. AI was then evaporated onto the polymer. Type B LEDs were similarly prepared, but a --600 nm thick emissive layer of PPV was prepared by the standard tetrahydrathiophenium leaving-group precursor [22], and Ca was evaporated onto the PPV. The Ca was protected from oxidation by additional metallic layers. The spectra and resonances of type A LEDs were very similar to those of type B, but of poorer signal-to-noise ratio due to the weaker EL. The home-built ODMR spectrometer has been previously described [3-5]. For EL-, o-, and Ophdetected resonances, thin copper wires were inserted into the quartz dewar of the cryostat to provide bias. The o- and Oph-detected resonances were measured by lockin detection of the microwaveinduced changes in the current across a standard resistor connected in series with the LED. For Ophdetected resonance, the forward bias was below the EL threshold voltage.
RESULTS The low-temperature EL and PL of both LED types are shown in Figure 1. The EL
5 D
TYPE,,EO
~4 m
closely matches the PL measured under the same conditions [4,16], exhibiting strong
PL AND EL OF
._J O.
A
A//,,\/A
3 2
vibrational coupling between the excited and ground states.
1.6 6
The full-field ODMR of a type B LED is shown in Figure 2. The main narrow (AH1/2 = 13 G) PL-enhancing resonance
5
1.8
2.0
PL AND EL OF TYPE B LED
2.2
2.4
2.6
A
D
4 F i g u r e 1. Photoluminescence (PL) and
_J tlJ "1o
electroluminescence (EL) spectra o f type A (PPV copolymer) and type B (PPV) LEDs at 20 K. The EL o f PPV (type B) is redshifted by 100 meV relative to the room temperature spectrum (see ref. 14).
...1 ft.
3 2 1 .--.-
i %
0
1.6
1.8
2.0 Energy (ev)
2.2
2.4
244
was previously assigned to magnetic resonance enhancement of singlet exciton generation by intrachain "distant pair" polaron recombination [3-6]. The full- and half-field triplet powder patterns are also similar to the previously reported spectra [4-61 and are consistent with magnetic resonance enhancement of singlet generation from triplet-
1.6
~ TYPEB LED
splitting parameters [24] D (= 540 G) and 0
1.2
~ RF P0wer=811 mw
< E < D/3 suggest that the triplets in PPV
0.8
= D/3,
0.4
triplet fusion.
The resulting zerofield
and P3ATs, where D -~ 620 G and E
'~
are only slightly larger than a phenylene or thiophene ring.
0.0
a. -0.4 2.6
Figure
2. (a) The photoluminescence
row resonance and the broad triplet powder
4.0 3.5 3.0 2.5 2.0 1.5
pattern underneath. (b) The main narrow
1.0
(i.e., optically) detected magnetic resonance
o
(PLDMR or ODMR) of a type B LED at 20 K, excited at 488 nm. Note the main nar-
ODMR
without bias, and during 70
Figure 2 also shows the effect of a dc seen, it decreases AL/L, in agreement with it/s assignment to fusion of p+ and F, and that of the EL-and o-quenching resonances to the decay of polarons to bipolarons (see below).
2.8
' 3.0
3.2
o-, and Oph- quenching resonances of a type
A O
4.2
v .-I
3.32
3.34 3.36 Magnetic Field (kG)
3.38
2 0
-J LU
n-
-4 ~
9
-6
/ TYPEB LED / RFPower-161mw
W
i=5o~A
T-20K
0.4.
? 0.0
A
o -0.4 -0.8
T = 20 K. The main narrow EL- and o-
~ -1.2
detected resonances of type A LEDs was
n- -1.6
similar; the lineshape and g-value were
-2.0 O -2.4
temperature independent from 20 to 296 K.
4.0
0.0 -0.5 3.30
B LED at J = 50 pA/mm 2, V= 30.7 V, and
The difference between these quenching and
' ' 3.6 3.8
A PLDMROF 1.\ TYPEB LED L;." .:F P:.w.er:~:~::
LU
Figure 3 displays the main narrow EL-,
3.4
0.5
~.
i~A/mm2 current injection.
bias on the main narrow ODMR. As deafly
T=20K
o
/ TYPEBLED / RF Power=l 6! mw
0
the PL-enhancing resonances is obvious and striking.
~0 . -2 <1 3. The main narrow EL-, a-, and
~-3
oph-detected magnetic resonance (ELDMR,
a oD. -4
Figure
CDMR, and PCDMR, resp.) spectra of a type B LED at 20 K and 50 pA/mm 2.
-5 3.30
RF Power-161 mw
3.32
3.34 3.36 Magnetic Field (kG)
3.38
245
However, the g-values and lineshapes of the PL-, EL-, o-, and Oph-detected resonances are all identical within the available experimental precision. The temperature dependence of the relative amplitudes of the resonances differ sharply:
When warming from 20 to 296 K, the PL-detected
resonance decreases by over an order of magnitude, while the EL-detected decreases by only -50%, and IAo/ol remains essentially unchanged. The EL- and o-detected resonances saturate at quite low current densities (-25 ,uA/mm 2) for a type B LED: IAEL/ELI and IAo/ot decrease from 5.7x 103 and 3.5x10 -3, resp., at I < 25/~A (V _<36.6 V) to 1.85x10 -3 and 5.3x10 -4, resp., at I = 803 ,uA (V = 43.5 V). The PL-, EL-, o-, and Oph-detected resonances of type B LEDs also clearly display the half-field Am s = 2 transitions of triplet (polaronic) excitons (Figure 4); the strong main narrow quenching resonance obscures the full-field powder patterns due to the Am s = 1 transitions. Both LED types show similar behavior, but the EL- strikingly differs from the o-detected resonance: Both clearly display two distinct peaks, yet while one is EL-quenching and the other EL-enhancing, both are o-enhancing. Figure 5 shows that the amplitudes of
~-. 2.8 2.4~
PLDMR OF
f % TYPEBLED f ,\Power=8,,mw .
E 2.0 ~ ~. rr ~ ~-
1.6 1.2 0.8 0.4 0.0
both sharply weaken with increasing
4
tempera-ture, but with distinct behavior.
o 2 "7
The EL-detected resonance of neither is
v
observable above 50 K, while the o -
~0
detected resonance of the broader (higher
...1
~-e
field) peak is still faintly observed at 100K.
rr
a~-4 Figure
4.
_.J UJ
The half-field PLDMR (i.e.,
5
/ / •
~4
ODMR) without bias, and EL-, or-, and
/
Q
~rph-detected
magnetic
resonance
E3
/
(ELDMR, CDMR, and PCDMR, resp.) of a type B LED at 5 K and 100 pA/mm 2.
/
\
CDMR OF TYPEBLED
\ RE Power=811mw
"~
i.loo~
a o 0 DISCUSSION The foregoing results all illustrate the wealth of detailed information on the electronic processes that govern the EL of polymer-based LEDs, that can be obtained from EL- and o-detected magnetic resonance. The first noteworthy observation is that the g-values of the main narrow PL-,
~ 2.4 ,,9 o 2.0
•
E 1.6
PCDMR OF PPV BLEND RF Pow:r=811 mw
i
1.2 0.8 ~ 0.4 ~ 0.0 O
a. -0.4 -0.8 , 1.58 1.60
1.62 1,64 1.66 1.68 Magnetic Field (kG)
1.70
246
Figure 5. The temperature dependence of the intensity
of the half-field triplet exciton EL- and o-detected magnetic resonances (ELDMR and CDMR, resp.). Triangles are the values of the higher-field component at 1.63 kG and circles are those of the lower-field line at 1.61 kG.
4;
~-2
4-4-6
i=I 0011A
-8
EL-, o-, and Oph-detected resonances (Figs. 2 and 3) are
0
I0
all identical. The latter was previously assigned to mag-
20
30 T (K)
40
50
60
6,
netic resonance enhancement of nongeminate intrachaln p+-p- fusion to singlet excitons [3-6]. The narrow EL- and
4
o-detected resonances are then entirely consistent with the picture of EL due to singlet exciton generation via double
_02
.LEDB
charge injection, followed by polaron formation, and nongeminate p+-p- fusion. The essentially identical response of the two types of LEDs fabricated with differ-
0
- - - -
-1
0
20
40 r (K)S0
ao
100
ent emissive layers and different electron-injecting contacts suggests that it is intrinsic to the polymers, and not specific to the nature of defects or to the electrodes. The quenching nature of the narrow o-, Oph-, and EL-detected resonances at g = 2.0023 clearly suggest an intrinsic underlying spin-dependent decay mechanism competing with the transport and EL. Although spin-dependent nonradiative trapping by interface defects would also yield such resonances, they would be expected to vary with the diode structure. In addition, they should also exhibit an ESR or light induced ESR; yet no such signals could be detected from these diodes. It is clear that the sought mechanism must be able to account for the signs which are opposite to that of the PL-detected resonance, since both the EL and PL involve singlet excitons which are produced via p+-p- fusion. An intrinsic process consistent with much of the available results is the spin-dependent fusion of two like-charged spin 1/2 polarons to spinless bipolarons.
Bipolarons are believed to be the
long-lived charged excitations in PPV, of mobility much lower than that of polarons [16,17]. Magnetic resonance enhancement of bipolaron production would then reduce both the current through the polymer and the polaron population available for EL. In addition, we note that charged excitations are believed to act as effective nonradiative decay sites for singlet excitons [11,18,19], and this should also contribute to the EL-quenching effect. The observed saturation of the quenching EL- and o-detected resonances at higher current densities is not clearly understood at present, but three explanations may be considered: (i) the imbalance of electron and hole injection is probably reduced at higher current densities, thus weakening these resonances; (ii) a higher frequency of like-charged polaron-polaron scattering events reduces the spin dependence of bipolaron formation at high current densities; (iii) higher charge-injection rates may be saturating bipolaron generation, if it preferentially occurs at charge-trapping sites. The weak temperature dependence of the EL- and o-quenching resonances contrasts sharply with the strong decay of the PL-detected resonance upon warming to room temperature. It was previously suggested that the latter may reflect the temperature dependence of the polaron spin-lattice relaxation rate Tl-1 [3,4]. This interpretation would imply a similarly strong temperature dependence of the EL-
247
and o-detected resonances. We are thus compelled to reexamine the source of the temperature dependence in both cases. In an alternative picture, consistent with the behavior of all resonances, the observed dependence is governed by the dynamics of polarons and mobile triplet excitons. As the temperature increases, their diffusivity increases (due, e.g., to thermally activated hopping), resulting in a weaker spin-dependence of the PL. In the diodes, however, polaron motion is determined by the product of their drift mobility and the electric field, while that of the neutral excitons is limited by their diffusivity, as in ODMR measurements. These two very different conditions may be responsible for the distinct temperature dependence of the ODMR as compared to the EL- and o-detected resonances. The strong temperature dependence of the EL- and o-detected half-field triplet resonances (Fig. 7) is entirely consistent with this interpretation. The appearance of two EL- and o-detected half-field triplet resonances (Fig. 6) is also striking. The lower field implies a more strongly localized triplet than the higher-field one. These more localized triplets may possibly decay by triplet-triplet fusion near the electrode-film interfaces. As they fuse, the resulting singlets either rapidly decay or separate into p + and p-. However, the presumably higher density of bipolarons and proximity of the electrode, which sweeps the appropriate polars from this region, would reduce the radiative yield from these singlets [11,18,19]. The overall effect of these processes occurring near the electrodes would be to reduce the contribution of triplets to the EL but add to their contribution to a.
SUMMARY In summary, X-band photoluminescence (PL)-, electroluminesence (EL)-, conductivity (o)- and photoconductivity (apO-detected magnetic resonance of polyparaphenylenevinylene-based light emitting diodes were described. The optically (i.e., PL)-detected resonance (ODMR) is identical to that of films. It includes a main narrow PL-enhancing resonance at g = 2.0023 attributed to fusion of polarons to singlet excitons which then decay radiatively, and PL-enhancing full- and half-field triplet powder patterns attributed to singlet exciton generation via triplet-triplet fusion. The main EL-, o-, and oph-detected magnetic resonances are intense narrow quenching lines at the same g-value. Their temperature and current-injection dependence are consistent with an assignment to decay of polarons to bipolarons. The half-field EL- and o-detected resonances yield two distinct triplet lines. While both are o-enhancing, the less symmetric is EL-quenching. Their temperature dependence is consistent with a tentative assignment to bulk- and near-interface triplet-fusion.
ACKNOW1 .F.DGEMENTS Ames Laboratory is operated by Iowa State University for the US Department of Energy under Contract W-7405-Eng-82. The work at Ames was supported by the Director for Energy Research, Office of Basic Energy Sciences, USDOE, and at Cambridge by the UK Science and Engineering Research Council. PLB thanks Christ's College (Dow Research Fellowship) and AK thanks the Deutsche Forschungsgemainschaft for financial support.
248
REFERENCES 1
B.C. Cavenett, Adv. Phys., 30 (1981) 475.
2
F. Boulitrop, Phys. Rev. B, 28 (1983) 6192.
3
L . S . Swanson, J. Shinar, and K. Yoshino, Phys. Rev. Lett., 65 (1990) 1140.
4
L.S. Swanson, P. A. Lane, J. Shinar, and F. Wudl, Phys. Rev. B 44 (1991) 10617.
5
L.S. Swanson, Ph.D. Thesis, Iowa State University, Ames, IA 50011, 1991.
6
J. Shinar and L. S. Swanson, Syn. Met., 50 (1992) 621; L. S. Swanson, J. Shinar, P. A. Lane, B. C. Hess, and F. Wudl, ibid. 481; L. S. Swanson, P. A. Lane, J. Shinar, K. Yoshino, and F. Wudl, ibid. 473.
7
K . P . Homewood, B. C. Cavenett, I. G. Austin, T. M. Searle, W. E. Spear, and P. G. LeComber, J. Phys. C, 17 (1984) L103.
8
J.H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. MacKay, R. H. Friend, P. L. Burn, and A. B. Holmes, Nature, 347, 539 (1990).
9 10
D. Braun and A. J. Heeger, Appl. Phys. Lett., 58 (1991) 1982. P . L . Bum, A. B. Holmes, A. Kraft, D. D. C. Bradley, A. R. Brown, and R. H. Friend, .1. Chem. Soc., Chem. Comm. (1992) 32.
11
D . D . C . Bradley, P. L. Burn, R. H. Friend, A. B. Holmes, and A. Kraft, in H. Kuzmany (ed.), Electronic Properties of Conjugated Polymers, Proc. Int. Winter School, Kirchberg, Austria, 1991 (Springer Series on Solid State Sciences, Springer Vedag, NY, 1991), in press.
12
P.L. Burn, A. B. Holmes, A. Kraft, D. D. C. Bradley, A. R. Brown, R. H. Friend, and R. W.
13
G. Gustafsson, Y. Cao, G. M. Treacy, F. Klavetter, N. Colaneri, and A. J. Heeger, Nature,
Gymer, Nature, 356 (1992) 47. 357 (1992) 477. 14
A..1. Heeger, S. Kivelson, J. R. Schrieffer, and W.-P. Su, Rev. Mod. Phys., 60 (1988), 781.
15
K. Fesser, A. R. Bishop, and D. K. Campbell, Phys. Rev. B, 27 (1983) 4804.
16
R.H. Friend, D. D. C. Bradley, and P. D. Townsend, J. Phys. D, 20 (1987) 1367.
17
N . F . Colaneri, D. D. C. Bradley, R. H. Friend, P. L. Burn, A. B. Holmes, and C. W.
18
D . D . C . Bradley and R. H. Friend, J. Phys. Condensed Matter, 1 (1989) 3671.
19
K . E . Ziemelis, A. T. Hussain, D. D. C. Bradley, R. H. Friend, J. Ruhe, and G. Wegner, Phys. Rev. Lett., 66 (1991) 2231.
20
C.E. Swenberg and N. E. Geacintov, in J. B. Birks (ed.), Organic Molecular Photophvsics, J.
21
H.P. Schwob and D. F. Williams, Chem. Phys. Lett., 13 (1972).
22
Y a. B. Zeldovich et al., Sov. Phys. Uzp., 31 (1988) 385.
Spangler, Phys. Rev. B, 42 (1990) 11671.
Wiley, NY, 1973, chap. 10.
23
D.W. Pratt, in R. H. Clarke (ed.), Triplet State ODMR Spectroscopy, John Wiley, NY, 1982, chap. 3; K. P. Dinse and C. J. Winscom, ibid., chap. 4.
24
A. Carrington and A. D. McLachlan, Introduction to Magnetic Resonance (Harper and Row, NY, 1967), Chap. 8, N. M. Atherton, Electron Spin Resonance, (Halstead, NY, 1973); D.R. Torgeson et al., J. Mag. Res. 68, 85 (1986); P. C. Taylor et al., Chem. Rev. 75, 203 (1975).
241.
ELECTROLUMINESCENCE-, CONDUCTIVITY-, AND PHOTOCONDUCTIVITY-DETECTED MAGNETIC RESONANCE STUDY OF POLY(P-PHENYLENEVINYLENE)-BASED LIGHT EMITTING DIODES
L. S. SWANSON and J. SHINAR
Ames Laboratory - USDOE and Physics Department, Iowa State University, Ames, IA 50011 A. R. BROWN, D. D. C. BRADLEY, and R. H. FRIEND
Cavendish Laboratory, Cambridge University, Madingley Rd., Cambridge CB3 0HE, UK P. L. BURN, A. KRAFT, and A. B. HOLMES
Chemistry Department, University Chemical Laboratory, Lensfield Rd., Cambridge CB2 1EW, UK
ABSTRACT The strong electroluminescence (EL)-detected magnetic resonance of poly(p-phenylenevinylene) (PPV)-based light emitting diodes is compared to the conductivity (o)-, photoconductivity (Oph)- and photoluminescence (PL)-detected resonances. In contrast to the narrow PL-enhancing resonance assigned to polaron fusion into singlet excitons, strong EL- and o-quenching resonances are attributed to the spin-dependent polaron-to-bipolarondecay. While the half-field PL-detected resonance reveals only one triplet exciton resonance, believed to result from triplet-triplet fusion into singlets, the halffield EL- and o-detected resonances yield two distinct triplets. While both are o-enhancing, one is ELquenching. The nature of the two triplet states is discussed.
KEYWORDS: poly(p-phenylenevinylene), electroluminescence, photoluminescence, electroluminescence-detected magnetic resonance, conductivity-detected magnetic resonance, photoconductivitydetected magnetic resonance, polarons, bipolarons, singlet excitons, triplet excitons
INTRODUCTION Over the past twenty years, optically (i.e., photoluminescence (PL)) detected magnetic resonance (ODMR) has provided invaluable insight into various radiative and nonradiative decay processes in a wide array of semiconductors [1-6]. In an ODMR experiment, microwave-induced changes in the PL are measured as the sample is subjected to a slowly swept dc magnetic field. Light emitting diodes (LEDs) have been available even longer than twenty years, yet the authors are aware of only one brief electroluminescence (EL) detected magnetic resonance study of any semiconducting system [7]. 0379-6779/93/$6.00
© 1993- Elsevier Sequoia. All rights reserved
242
Due to their potential advantages relative to previously studied organic LEDs, conjugated polymer-based diodes in general, and poly(p-phenylenevinylene) (PPV)-based LEDs in particular, are currently drawing intense attention [8-13]. The device structure was initially a ~100 nm thick polymer layer sandwiched between indium/fin oxide (ITO) as positive contact and AI as negative contact. The light generation efficiency of these devices was subsequently raised to -1% photon/electron through selection of Ca for the negative electrode [9], and could also be raised by optimizing the emissive properties of the polymer layer through alterationto form copolymers [10,11]. The electronic processes that govern the operation of these devices are of obvious interest. In nondegenerate ground-state systems such as PPV, these include singly charged spin 1/2 polarons (p÷ and p-), doubly charged spinless bipolarons (bp÷÷ and bp--), and neutral singlet and triplet (polaronic) excitons (S 1 and T 1, resp.) [14-17]. Burroughes et al. [8] proposed that the operation of PPV-based LEDs involved double charge injection of electrons and holes at opposite electrodes, transport through the polymer film as polarons, fusion of oppositely charged polarons to excitons, and radiative decay of the singlets. Significant nonradiative decay processes of singlets include intersystem crossing to lowlying triplet states and direct decay to the ground state. Since previous ODMR studies of poly(3alkylthiophenes) (P3ATs), PPVs, and poly(p-phenyleneacetylenes) (PPAs) [3-6] indicated that the polaron levels are close to the singlet level, fission of singlets to polarons may also be significant. Other decay mechanisms clearly affect the EL and/or o. If, as widely believed [16,17], the mobility of bipolarons is significantly lower than that of polarons, the decay of polarons to bipolarons should decrease both the EL and ~. Charged excitations are also believed to act as effective nonradiative decay sites for singlet and triplet excitons [11,18,19]. Of particular importance are triplet-triplet collisions which generate singlets. These were established to be the source of the delayed fluorescence in, e.g., anthracene and related organic materials [20-23]. They are thus suspected to be the source of the ODMR triplet powder patterns observed in P'3ATs, PPVs, and PPAs [3-6]. The spectral dependence of the half-field triplet ODMR of PPV is nearly identical to the PL spectrum, which is entirely consistent with this scenario. The singlets may, of course, also fission to polarons or decay nonradiatively. Triplet-triplet pairs may obviously also directly decay nonradiatively. It should be noted that the above-mentioned processes do not exhaust the possible reactions and decay modes of singlets, triplets, polarons, and bipolarons.
Others include, e.g., the decay of a bipolaron via collision with a polaron.
Detection of magnetic resonance through modulation of the EL may be particularly revealing since, in stark contrast to conventional ODMR, all the excited states generated by carrier injection are inherently nongeminate, i.e., not created by geminate electron-hole pairs. The EL-detected resonance described in this paper is indeed strikingly different and much (~20 times) more intense than the ODMR. In addition to the ODMR and EL-detected resonance, the conductivity (o)- and photoconductivity (Oph)- detected magnetic resonances are also described. The results are entirely consistent with the above picture of transport of injected carders as polarons and radiative decay via singlet genration. However, they also suggest that decay of polarons to bipolarons is a significant competing process which may impose an upper limit on the efficiency of such LEDs. In addition, the EL- and odetected triplet powder pattern resonances suggest that singlet generation via triplet-triplet fusion may
243
also contribute significantly to the EL. The dependence of the EL- and o-detected magnetic resonance on the injected current and the temperature are consistent with the suggested role of charged excitations in the nonradiative quenching of singlet and triplet excitons.
EXPERIMENTAL PROCEDURE Two types of EL devices were studied, ITO-coated glass was used as the positive electrode in both cases. In type A, a layer of the methoxy-leaving group PPV precursor was then deposited by spin-coating, and thermally converted to a ,-60 nm thick partially conjugated copolymer, the PL of which is blue-shifted with respect to PPV [10]. AI was then evaporated onto the polymer. Type B LEDs were similarly prepared, but a --600 nm thick emissive layer of PPV was prepared by the standard tetrahydrathiophenium leaving-group precursor [22], and Ca was evaporated onto the PPV. The Ca was protected from oxidation by additional metallic layers. The spectra and resonances of type A LEDs were very similar to those of type B, but of poorer signal-to-noise ratio due to the weaker EL. The home-built ODMR spectrometer has been previously described [3-5]. For EL-, o-, and Ophdetected resonances, thin copper wires were inserted into the quartz dewar of the cryostat to provide bias. The o- and Oph-detected resonances were measured by lockin detection of the microwaveinduced changes in the current across a standard resistor connected in series with the LED. For Ophdetected resonance, the forward bias was below the EL threshold voltage.
RESULTS The low-temperature EL and PL of both LED types are shown in Figure 1. The EL
5 D
TYPE,,EO
~4 m
closely matches the PL measured under the same conditions [4,16], exhibiting strong
PL AND EL OF
._J O.
A
A//,,\/A
3 2
vibrational coupling between the excited and ground states.
1.6 6
The full-field ODMR of a type B LED is shown in Figure 2. The main narrow (AH1/2 = 13 G) PL-enhancing resonance
5
1.8
2.0
PL AND EL OF TYPE B LED
2.2
2.4
2.6
A
D
4 F i g u r e 1. Photoluminescence (PL) and
_J tlJ "1o
electroluminescence (EL) spectra o f type A (PPV copolymer) and type B (PPV) LEDs at 20 K. The EL o f PPV (type B) is redshifted by 100 meV relative to the room temperature spectrum (see ref. 14).
...1 ft.
3 2 1 .--.-
i %
0
1.6
1.8
2.0 Energy (ev)
2.2
2.4
244
was previously assigned to magnetic resonance enhancement of singlet exciton generation by intrachain "distant pair" polaron recombination [3-6]. The full- and half-field triplet powder patterns are also similar to the previously reported spectra [4-61 and are consistent with magnetic resonance enhancement of singlet generation from triplet-
1.6
~ TYPEB LED
splitting parameters [24] D (= 540 G) and 0
1.2
~ RF P0wer=811 mw
< E < D/3 suggest that the triplets in PPV
0.8
= D/3,
0.4
triplet fusion.
The resulting zerofield
and P3ATs, where D -~ 620 G and E
'~
are only slightly larger than a phenylene or thiophene ring.
0.0
a. -0.4 2.6
Figure
2. (a) The photoluminescence
row resonance and the broad triplet powder
4.0 3.5 3.0 2.5 2.0 1.5
pattern underneath. (b) The main narrow
1.0
(i.e., optically) detected magnetic resonance
o
(PLDMR or ODMR) of a type B LED at 20 K, excited at 488 nm. Note the main nar-
ODMR
without bias, and during 70
Figure 2 also shows the effect of a dc seen, it decreases AL/L, in agreement with it/s assignment to fusion of p+ and F, and that of the EL-and o-quenching resonances to the decay of polarons to bipolarons (see below).
2.8
' 3.0
3.2
o-, and Oph- quenching resonances of a type
A O
4.2
v .-I
3.32
3.34 3.36 Magnetic Field (kG)
3.38
2 0
-J LU
n-
-4 ~
9
-6
/ TYPEB LED / RFPower-161mw
W
i=5o~A
T-20K
0.4.
? 0.0
A
o -0.4 -0.8
T = 20 K. The main narrow EL- and o-
~ -1.2
detected resonances of type A LEDs was
n- -1.6
similar; the lineshape and g-value were
-2.0 O -2.4
temperature independent from 20 to 296 K.
4.0
0.0 -0.5 3.30
B LED at J = 50 pA/mm 2, V= 30.7 V, and
The difference between these quenching and
' ' 3.6 3.8
A PLDMROF 1.\ TYPEB LED L;." .:F P:.w.er:~:~::
LU
Figure 3 displays the main narrow EL-,
3.4
0.5
~.
i~A/mm2 current injection.
bias on the main narrow ODMR. As deafly
T=20K
o
/ TYPEBLED / RF Power=l 6! mw
0
the PL-enhancing resonances is obvious and striking.
~0 . -2 <1 3. The main narrow EL-, a-, and
~-3
oph-detected magnetic resonance (ELDMR,
a oD. -4
Figure
CDMR, and PCDMR, resp.) spectra of a type B LED at 20 K and 50 pA/mm 2.
-5 3.30
RF Power-161 mw
3.32
3.34 3.36 Magnetic Field (kG)
3.38
245
However, the g-values and lineshapes of the PL-, EL-, o-, and Oph-detected resonances are all identical within the available experimental precision. The temperature dependence of the relative amplitudes of the resonances differ sharply:
When warming from 20 to 296 K, the PL-detected
resonance decreases by over an order of magnitude, while the EL-detected decreases by only -50%, and IAo/ol remains essentially unchanged. The EL- and o-detected resonances saturate at quite low current densities (-25 ,uA/mm 2) for a type B LED: IAEL/ELI and IAo/ot decrease from 5.7x 103 and 3.5x10 -3, resp., at I < 25/~A (V _<36.6 V) to 1.85x10 -3 and 5.3x10 -4, resp., at I = 803 ,uA (V = 43.5 V). The PL-, EL-, o-, and Oph-detected resonances of type B LEDs also clearly display the half-field Am s = 2 transitions of triplet (polaronic) excitons (Figure 4); the strong main narrow quenching resonance obscures the full-field powder patterns due to the Am s = 1 transitions. Both LED types show similar behavior, but the EL- strikingly differs from the o-detected resonance: Both clearly display two distinct peaks, yet while one is EL-quenching and the other EL-enhancing, both are o-enhancing. Figure 5 shows that the amplitudes of
~-. 2.8 2.4~
PLDMR OF
f % TYPEBLED f ,\Power=8,,mw .
E 2.0 ~ ~. rr ~ ~-
1.6 1.2 0.8 0.4 0.0
both sharply weaken with increasing
4
tempera-ture, but with distinct behavior.
o 2 "7
The EL-detected resonance of neither is
v
observable above 50 K, while the o -
~0
detected resonance of the broader (higher
...1
~-e
field) peak is still faintly observed at 100K.
rr
a~-4 Figure
4.
_.J UJ
The half-field PLDMR (i.e.,
5
/ / •
~4
ODMR) without bias, and EL-, or-, and
/
Q
~rph-detected
magnetic
resonance
E3
/
(ELDMR, CDMR, and PCDMR, resp.) of a type B LED at 5 K and 100 pA/mm 2.
/
\
CDMR OF TYPEBLED
\ RE Power=811mw
"~
i.loo~
a o 0 DISCUSSION The foregoing results all illustrate the wealth of detailed information on the electronic processes that govern the EL of polymer-based LEDs, that can be obtained from EL- and o-detected magnetic resonance. The first noteworthy observation is that the g-values of the main narrow PL-,
~ 2.4 ,,9 o 2.0
•
E 1.6
PCDMR OF PPV BLEND RF Pow:r=811 mw
i
1.2 0.8 ~ 0.4 ~ 0.0 O
a. -0.4 -0.8 , 1.58 1.60
1.62 1,64 1.66 1.68 Magnetic Field (kG)
1.70
246
Figure 5. The temperature dependence of the intensity
of the half-field triplet exciton EL- and o-detected magnetic resonances (ELDMR and CDMR, resp.). Triangles are the values of the higher-field component at 1.63 kG and circles are those of the lower-field line at 1.61 kG.
4;
~-2
4-4-6
i=I 0011A
-8
EL-, o-, and Oph-detected resonances (Figs. 2 and 3) are
0
I0
all identical. The latter was previously assigned to mag-
20
30 T (K)
40
50
60
6,
netic resonance enhancement of nongeminate intrachaln p+-p- fusion to singlet excitons [3-6]. The narrow EL- and
4
o-detected resonances are then entirely consistent with the picture of EL due to singlet exciton generation via double
_02
.LEDB
charge injection, followed by polaron formation, and nongeminate p+-p- fusion. The essentially identical response of the two types of LEDs fabricated with differ-
0
- - - -
-1
0
20
40 r (K)S0
ao
100
ent emissive layers and different electron-injecting contacts suggests that it is intrinsic to the polymers, and not specific to the nature of defects or to the electrodes. The quenching nature of the narrow o-, Oph-, and EL-detected resonances at g = 2.0023 clearly suggest an intrinsic underlying spin-dependent decay mechanism competing with the transport and EL. Although spin-dependent nonradiative trapping by interface defects would also yield such resonances, they would be expected to vary with the diode structure. In addition, they should also exhibit an ESR or light induced ESR; yet no such signals could be detected from these diodes. It is clear that the sought mechanism must be able to account for the signs which are opposite to that of the PL-detected resonance, since both the EL and PL involve singlet excitons which are produced via p+-p- fusion. An intrinsic process consistent with much of the available results is the spin-dependent fusion of two like-charged spin 1/2 polarons to spinless bipolarons.
Bipolarons are believed to be the
long-lived charged excitations in PPV, of mobility much lower than that of polarons [16,17]. Magnetic resonance enhancement of bipolaron production would then reduce both the current through the polymer and the polaron population available for EL. In addition, we note that charged excitations are believed to act as effective nonradiative decay sites for singlet excitons [11,18,19], and this should also contribute to the EL-quenching effect. The observed saturation of the quenching EL- and o-detected resonances at higher current densities is not clearly understood at present, but three explanations may be considered: (i) the imbalance of electron and hole injection is probably reduced at higher current densities, thus weakening these resonances; (ii) a higher frequency of like-charged polaron-polaron scattering events reduces the spin dependence of bipolaron formation at high current densities; (iii) higher charge-injection rates may be saturating bipolaron generation, if it preferentially occurs at charge-trapping sites. The weak temperature dependence of the EL- and o-quenching resonances contrasts sharply with the strong decay of the PL-detected resonance upon warming to room temperature. It was previously suggested that the latter may reflect the temperature dependence of the polaron spin-lattice relaxation rate Tl-1 [3,4]. This interpretation would imply a similarly strong temperature dependence of the EL-
247
and o-detected resonances. We are thus compelled to reexamine the source of the temperature dependence in both cases. In an alternative picture, consistent with the behavior of all resonances, the observed dependence is governed by the dynamics of polarons and mobile triplet excitons. As the temperature increases, their diffusivity increases (due, e.g., to thermally activated hopping), resulting in a weaker spin-dependence of the PL. In the diodes, however, polaron motion is determined by the product of their drift mobility and the electric field, while that of the neutral excitons is limited by their diffusivity, as in ODMR measurements. These two very different conditions may be responsible for the distinct temperature dependence of the ODMR as compared to the EL- and o-detected resonances. The strong temperature dependence of the EL- and o-detected half-field triplet resonances (Fig. 7) is entirely consistent with this interpretation. The appearance of two EL- and o-detected half-field triplet resonances (Fig. 6) is also striking. The lower field implies a more strongly localized triplet than the higher-field one. These more localized triplets may possibly decay by triplet-triplet fusion near the electrode-film interfaces. As they fuse, the resulting singlets either rapidly decay or separate into p + and p-. However, the presumably higher density of bipolarons and proximity of the electrode, which sweeps the appropriate polars from this region, would reduce the radiative yield from these singlets [11,18,19]. The overall effect of these processes occurring near the electrodes would be to reduce the contribution of triplets to the EL but add to their contribution to a.
SUMMARY In summary, X-band photoluminescence (PL)-, electroluminesence (EL)-, conductivity (o)- and photoconductivity (apO-detected magnetic resonance of polyparaphenylenevinylene-based light emitting diodes were described. The optically (i.e., PL)-detected resonance (ODMR) is identical to that of films. It includes a main narrow PL-enhancing resonance at g = 2.0023 attributed to fusion of polarons to singlet excitons which then decay radiatively, and PL-enhancing full- and half-field triplet powder patterns attributed to singlet exciton generation via triplet-triplet fusion. The main EL-, o-, and oph-detected magnetic resonances are intense narrow quenching lines at the same g-value. Their temperature and current-injection dependence are consistent with an assignment to decay of polarons to bipolarons. The half-field EL- and o-detected resonances yield two distinct triplet lines. While both are o-enhancing, the less symmetric is EL-quenching. Their temperature dependence is consistent with a tentative assignment to bulk- and near-interface triplet-fusion.
ACKNOW1 .F.DGEMENTS Ames Laboratory is operated by Iowa State University for the US Department of Energy under Contract W-7405-Eng-82. The work at Ames was supported by the Director for Energy Research, Office of Basic Energy Sciences, USDOE, and at Cambridge by the UK Science and Engineering Research Council. PLB thanks Christ's College (Dow Research Fellowship) and AK thanks the Deutsche Forschungsgemainschaft for financial support.
248
REFERENCES 1
B.C. Cavenett, Adv. Phys., 30 (1981) 475.
2
F. Boulitrop, Phys. Rev. B, 28 (1983) 6192.
3
L . S . Swanson, J. Shinar, and K. Yoshino, Phys. Rev. Lett., 65 (1990) 1140.
4
L.S. Swanson, P. A. Lane, J. Shinar, and F. Wudl, Phys. Rev. B 44 (1991) 10617.
5
L.S. Swanson, Ph.D. Thesis, Iowa State University, Ames, IA 50011, 1991.
6
J. Shinar and L. S. Swanson, Syn. Met., 50 (1992) 621; L. S. Swanson, J. Shinar, P. A. Lane, B. C. Hess, and F. Wudl, ibid. 481; L. S. Swanson, P. A. Lane, J. Shinar, K. Yoshino, and F. Wudl, ibid. 473.
7
K . P . Homewood, B. C. Cavenett, I. G. Austin, T. M. Searle, W. E. Spear, and P. G. LeComber, J. Phys. C, 17 (1984) L103.
8
J.H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. MacKay, R. H. Friend, P. L. Burn, and A. B. Holmes, Nature, 347, 539 (1990).
9 10
D. Braun and A. J. Heeger, Appl. Phys. Lett., 58 (1991) 1982. P . L . Bum, A. B. Holmes, A. Kraft, D. D. C. Bradley, A. R. Brown, and R. H. Friend, .1. Chem. Soc., Chem. Comm. (1992) 32.
11
D . D . C . Bradley, P. L. Burn, R. H. Friend, A. B. Holmes, and A. Kraft, in H. Kuzmany (ed.), Electronic Properties of Conjugated Polymers, Proc. Int. Winter School, Kirchberg, Austria, 1991 (Springer Series on Solid State Sciences, Springer Vedag, NY, 1991), in press.
12
P.L. Burn, A. B. Holmes, A. Kraft, D. D. C. Bradley, A. R. Brown, R. H. Friend, and R. W.
13
G. Gustafsson, Y. Cao, G. M. Treacy, F. Klavetter, N. Colaneri, and A. J. Heeger, Nature,
Gymer, Nature, 356 (1992) 47. 357 (1992) 477. 14
A..1. Heeger, S. Kivelson, J. R. Schrieffer, and W.-P. Su, Rev. Mod. Phys., 60 (1988), 781.
15
K. Fesser, A. R. Bishop, and D. K. Campbell, Phys. Rev. B, 27 (1983) 4804.
16
R.H. Friend, D. D. C. Bradley, and P. D. Townsend, J. Phys. D, 20 (1987) 1367.
17
N . F . Colaneri, D. D. C. Bradley, R. H. Friend, P. L. Burn, A. B. Holmes, and C. W.
18
D . D . C . Bradley and R. H. Friend, J. Phys. Condensed Matter, 1 (1989) 3671.
19
K . E . Ziemelis, A. T. Hussain, D. D. C. Bradley, R. H. Friend, J. Ruhe, and G. Wegner, Phys. Rev. Lett., 66 (1991) 2231.
20
C.E. Swenberg and N. E. Geacintov, in J. B. Birks (ed.), Organic Molecular Photophvsics, J.
21
H.P. Schwob and D. F. Williams, Chem. Phys. Lett., 13 (1972).
22
Y a. B. Zeldovich et al., Sov. Phys. Uzp., 31 (1988) 385.
Spangler, Phys. Rev. B, 42 (1990) 11671.
Wiley, NY, 1973, chap. 10.
23
D.W. Pratt, in R. H. Clarke (ed.), Triplet State ODMR Spectroscopy, John Wiley, NY, 1982, chap. 3; K. P. Dinse and C. J. Winscom, ibid., chap. 4.
24
A. Carrington and A. D. McLachlan, Introduction to Magnetic Resonance (Harper and Row, NY, 1967), Chap. 8, N. M. Atherton, Electron Spin Resonance, (Halstead, NY, 1973); D.R. Torgeson et al., J. Mag. Res. 68, 85 (1986); P. C. Taylor et al., Chem. Rev. 75, 203 (1975).