Triplet exciton formation and decay in polyfluorene light-emitting diodes
H. H. Liao,1,2H. F. Meng,1,*S. F. Horng,2 J. T. Shy,3K. Chen,2and C. S. Hsu4
1Institute of Physics, National Chiao Tung University, Hsinchu 300, Taiwan, Republic of China 2Department of Electric Engineering, National Tsing Hua University, Hsinchu 300, Taiwan, Republic of China
3Department of Physics, National Tsing Hua University, Hsinchu 300, Taiwan, Republic of China 4Department of Applied Chemistry, National Chiao Tung University, Hsinchu 300, Taiwan, Republic of China
共Received 27 May 2005; published 30 September 2005兲
We study the triplet excitons in poly共9,9-dioctylfluorene-2,7-diyl兲 light-emitting diode using infrared in-duced absorption. The infrared absorption is exclusively due to the triplet excitons and there is no spectral overlap with any other species. A strong suppression of the triplet exciton density relative to the singlet by voltage is observed. Through an unique independent measurement on the triplet exciton lifetime it is shown that the suppression solely comes from triplet exciton quenching by current injection. The triplet-to-singlet exciton formation ratio is independent of voltage as well as temperature, implying a spin-independent exciton formation.
DOI:10.1103/PhysRevB.72.113203 PACS number共s兲: 78.60.Fi, 72.80.Le
In organic semiconductors the lowest excited states are one spin singlet exciton and three spin triplet excitons. One of the most important questions in the electroluminescence 共EL兲 of organic semiconductors is whether the formation rates of all four excitons depend on their total spin. If the formation is spin independent, the triplet-to-singlet formation ratio will be three based simply on multiplicity. Since only the singlet excitons decay radiatively, the theoretical limit for the internal quantum efficiency共photon per electron兲 will be 1 / 4 for spin-independent exciton formation. For small-molecule organic light-emitting diodes 共LED兲 it has been established that the triplet-to-singlet ratio is indeed three.1 The key consequence is that the LED efficiency can be raised four times by adding phosphorescent dopants.2On the other hand for conjugated polymer LED the situation is much less clear. Many theoretical models have been proposed for spin-dependent exciton formation, including interchain charge transfer effect3,4 and phonon bottleneck.5,6 There are also many experimental works suggesting that the exciton forma-tion ratio is not three in polymer LED.7,8However none of these experiments are done in a working LED with typical electroluminescent polymer, so they are unable to give a de-cisive answer to this question. Recently triplet excitons in a working poly共phenylene vinylene兲 共PPV兲 LED is directly measured,9–13 the triplet signal is however mixed with po-laron signal. In this work we perform an unambiguous deter-mination of the formation and decay of the triplet excitons in poly共9,9-dioctylfluorene-2,7-diyl兲 共PFO兲 LED. It is found that there is a strong suppression of the triplet exciton rela-tive to the singlet as the voltage increases. This suppression turns out to be completely due to the quenching of the triplet excitons by polarons. The formation process itself might be in fact spin independent. Our result indicates that there is actually no difference between small molecule LED and polymer LED in terms of the exciton formation process and the importance of harvesting triplet energy by phosphores-cent dopants.
The triplet excitons in conjugated polymers are detected by their infrared induced absorption under photoexcitation 共PA兲 or electrical excitation 共EA兲.14In addition to triplet
ex-citons, polarons also contribute to the infrared absorption. Moreover, there is an interaction between these two species in a working polymer LED.12,13Unfortunately, for PPV the induced absorption spectra for triplet exciton, electron po-laron, and hole polaron are both peaked around 1.47 eV and there is a strong overlap among these signals.12,15PA and EA for PPV become a superpositon of triplet exciton and polaron spectra. Even though these overlapping signals can be sepa-rated in principle by the difference in frequency responses,14 the coincidence in the infrared spectrum implies that PPV is not a good material to start with. It turns out that the other commonly used polymer polyfuorene is a much better can-didate for this purpose due to the absence of overlap.16PA of PFO film is shown in Fig. 1. The spectral shape and the frequency response are identical to the previous report on the PFO triplet exciton.16The same spectrum is also observed in EA of 50 nm thick PFO LED after the turn-on voltage. The results at 14 K are shown in Fig. 2. As the voltage increases the induced absorption spectrum maintains a fixed line shape with increasing amplitude. The spectra for various voltages are normalized in the inset of Fig. 2. They are identical as a
FIG. 1. Relative transmission spectrum for photoinduced ab-sorption共PA兲 of PFO film is shown at 130 K with 100 Hz lock-in modulation. The dependence on modulation frequency is shown in the inset. The peak at 1.44 eV is due to triplet exciton. The power of the 405 nm CW exitation laser is 15 mW with focus area around 4 mm2.
PHYSICAL REVIEW B 72, 113203共2005兲
proof that there is only one species in this spectral region, and this species has been confirmed to be the triplet exciton in PA. Such identical normalized EA curves can never be observed in PPV. Below we take EA as its peak value at 1.47 eV. The line shape remains the same for whatever voltage, modulation frequency, and temperature.
The most striking feature of Fig. 2 is that the growth of the amplitude of the spectrum with voltage is strongly sup-pressed at higher voltage. The amplitude of the spectrum is proportional to the average triplet exciton density NTin the LED. In Fig. 3 NT is plotted against voltage V. After an initial rapid growth NTstarts to saturate. It is to be compared with the average singlet exciton density NS and the current density J. NSis proportional to the EL intensity emitted from the LED. All the curves are normalized to one at their re-spective maxima. NS and J are almost proportional to each other meaning a fixed conversion efficiency共CE兲 from elec-trical current to singlet excitons as is also shown in the inset of Fig. 3. Unlike the singlet exciton, the density of the triplet exciton is not correlated with the injected current at all. At high voltage while the current increases fast with voltage the
triplet exciton is suppressed. In order to stress this remark-able difference between the voltage dependence of the sin-glet and triplet excitons, their ratio NT/ NS is shown as a function of voltage. The ratio drops for more than an order of magnitude as the voltage increases. Such a dramatic varia-tion of the ratio between excitons with different spin would lead to a possible interpretation as the strong suppression of triplet exciton formation cross section by the electric field. If this were true due to whatever mechanism, the exciton for-mation would be spin dependent and the triplet-to-singlet exciton formation rate must not be a fixed number of three in polymer LED. However, the suppressed formation is not the only possibility to explain the behaviors in Fig. 3. The triplet exiciton density NTis a product of its formation rate and its lifetime. Either the suppression of the formation rate or the lifetime can lead to the drop of NT/ NSwith voltage. Assum-ing␥is the triplet-to-singlet exciton formation ratio and G is the average singlet exciton formation rate per unit volume, the exciton densities are
NS= GS, NT=␥GT. 共1兲
Dividing these two relations we have
NT
NS =␥T
S
⬀␥T. 共2兲
S andT are the singlet and triplet exciton lifetimes. Here we assume thatSis independent of voltage and other con-ditions in the experiment because of the fast radiative decay. The similarity of NSand J curves in Fig. 3 suggests that G is in fact roughly proportional to the current density. NT/ NScan be experimentally measured as shown in Fig. 3 as a function of voltage V. Now we need to find out whether its strong V dependence comes from ␥ or T according to Eq.共2兲. If it comes from␥, field dissociation of triplet exciton is probably the main reason. Exciton-polaron interaction might be the cause if it turns out to come fromT.
Since EA can only determine the variation of the product of␥andT, it will be helpful to study a situation where one of the two are known to be fixed. For this purpose we per-form an unique PA measurement for a LED under constant voltage bias. In the EA measurement discussed above, the voltage is square-wave modulated at a frequency which is used as the reference frequency for the lock-in amplifier for the infrared induced absorption. In the PA for LED, the LED is simultaneously excited by a modulated pump laser 共405 nm兲 and a dc voltage bias. The pump laser modulation fre-quency is used as the lock-in reference while the voltage does not change with time. Even though both the photoexci-tation and electrical current injection cause triplet exciton formation, only the photoexcited triplet excitons are detected by the lock-in amplifier as the PA signal. The basic idea here is that for photoexcitation the triplet exciton completely comes from the intersystem crossing from the photoexcited singlet exciton, so the triplet exciton formation is indepen-dent of the voltage V. The triplet exciton density NTPAin this measurement is therefore proportional to T only as V FIG. 2. Electroluminescence-induced absorption共EA兲 spectrum
for PFO LED is shown for various voltage at 14 K. The modulation frequency is 1 kHz. The spectra at different voltages can be normal-ized to one single line shape as shown in the inset. The LED struc-ture is indium-tin-oxide/ PEDOT: PSS/ PFO/ Ca/ Al. PEDOT:PSS is poly共3,4-ethylenedioxythiophene兲 doped with polystyrene sulpho-nated acid
FIG. 3. Voltage dependence of EA共proportional to triplet exci-ton density NT兲, EL 共proportional to singlet exciton density NS兲, and the current density J are plotted. The ratio NT/ NSare also shown. The EL conversion efficiency共CE兲 NS/ J are shown in the inset. For absolute scale, the maximal current density J at 14 V is 966 mA/ cm2.
BRIEF REPORTS PHYSICAL REVIEW B 72, 113203共2005兲
changes. Because there might be some slight dependence of the photoexcited singlet exciton density NSPAon V, it is more precise to write NTPA NSPA ⬀T. 共3兲 NS PA
is determined by the photoluminescence共PL兲 intensity. We are now free from the coupling of␥ andT occurred in EA. Since the LED is under voltage bias as in EA, the pho-toexcited triplet exciton will experience the same environ-ment as in EA in terms of field and polaron density, etc. The lifetime should therefore be the same function of V. PA/PL, i.e., NTPA/ NSPA⬀T, is plotted versus V in Fig. 4共a兲 together with EA/EL, i.e., NT/ NS⬀␥T. Equations 共2兲 and 共3兲 are used. It turns out that they are the same function of V. It implies that␥TandThave the same V dependence. In other words, the triplet-to-singlet exciton formation ratio␥ is in-dependent of V. The suppressed EA in Fig. 3 comes solely from the suppression of T by voltage which occurs after about 6 V. Interestingly this onset of suppression coincides with the onset of the large current injection as shown in Fig. 4. This implies that the suppression of the triplet lifetime is due to polarons instead of electric field field. We also mea-sure NT
PA / NS
PA
in the reverse bias and found that there is no significant voltage dependence. Since the field is large in the reverse bias, this confirms that field alone is not the reason for the suppression. Current injection and polarons must be involved in the mechanism.13Independent evidence for
trip-let lifetime suppression by voltage is provided by measuring the frequency response of LED PA under various dc volt-ages. The results are shown in Fig. 4共b兲. PA turns from con-stant to inverse frequency behavior at higher frequency as V increases. The fitted lifetimeT
f is also shown in Fig. 4共a兲. As expected it has the same voltage dependence asTmeasured from PA. LED with thickness 100 nm shows the same fea-tures.
After knowing that␥ is independent of the voltage, the next question is whether it depends on another basic param-eter for a working LED, i.e., temperature T. So far T is fixed at 14 K. Here we repeat all the measurements above except T is changed and voltage fixed at 11 V. In this way we can measure the product␥TandTitself as functions of T. One would be curious to know in addition to same V dependence if they have the same T dependence as well. NT/ NS共⬀␥T兲,
NTPA/ NSPA共⬀T兲, andT f
from frequency response are all plot-ted against T in Fig. 5共a兲. They all have the same T depen-dence. The discrepancy at low temperature is due to non-steady state caused by long triplet lifetime.T
f
is fitted from Fig. 5共b兲. These results show that␥is neither a function of V nor T, i.e., a true constant. The suppression of Tat higher temperature also correlates well with the increasing current density J due to higher polaron mobility, consistent with the above interpretation of triplet exciton-polaron interaction.
Even though the absolute value of triplet-to-singlet forma-tion rate␥is not determined, the fact that␥is a true constant independent of LED operation conditions strongly suggests that the exciton formation is spin independent and the con-stant is in fact three. ␥ is not three would imply that the capture cross sections for singlet and triplet electron-hole pairs are different. The electron-hole pair in spin configura-tion with smaller capture cross secconfigura-tion, say triplet, would FIG. 4. Triplet lifetimeTis obtained from PA/PL关see Eq. 共3兲兴,
measured at 14 K with 100 Hz optical pump modulation.␥T is obtained from EA/EL关see Eq. 共2兲兴. The frequency response of PA due to photoexcited triplet excitons measured in LED is shown to depend on the applied dc voltage in共b兲. The fitted lifetimeTf from 共b兲 is plotted as a function of voltage V in 共a兲. Note that␥T,Tand T
f
share the same voltage dependence. The fitting is done by draw-ing a line with slope minus one to fit the high-frequency part and a horizontal line共slope zero兲 from the low frequency limit. The cross-ing of these two lines determines the lifetime.
FIG. 5. In共a兲␥TandT, both measured at 100 Hz modulation, are plotted as functions of T at 11 V. The EL conversion efficiency 共CE兲 NS/ J is shown to be independent of T. The LED PA frequency response at difference T in共b兲 are measured without bias. Note that ␥T,T andT
f
share the same temperature dependence.
BRIEF REPORTS PHYSICAL REVIEW B 72, 113203共2005兲
have a higher chance to be dissociated by thermal fluctuation or electric field than the other spin configuration. In such a spin-dependent exciton formation process we would there-fore expect a dependence of the formation rate on tempera-ture, field, or current density. Since none of these parameters affect ␥, we believe that the capture cross section for the singlet and triplet electron-hole pairs are the same, i.e., spin-independent exciton formation. The strong suppression of the triplet exciton lifetime is probably due to the Auger recom-bination where the energy of the triplet exciton is given to the kinetic energy of a free polaron accompanied with a spin flip through exchange Coulomb interaction. In addition to the triplet-polaron interaction, triplet-triplet annihilation may also contribute to the suppression ofTat high voltage. Even though an interesting topic by itself, the mechanism of triplet quenching by the polarons and its bimolecular rate coeffi-cient is not the main focus of this work. The turn-on of triplet exciton in EA requires lower voltage than the singlet共Fig. 3兲, presumably because the singlet lifetime is suppressed by the
cathode quenching at low voltage. What is most important is that, while the decay is spin dependent, the formation rates of the four excitons are all simply proportional to the in-jected current density regardless of their spin. Apparently heavy-metal phosphorescent dopants are needed in order to turn the 3 / 4 of triplet energy into light.
In conclusion, through the direct measurement of density and lifetime of triplet exciton in a working polymer LED we show that there is a strong quenching of triplet exciton by the injected polarons at higher voltage. However the triplet-to-singlet exciton formation ratio is neither a function of volt-age nor temperature. This suggests a spin-independent exci-ton formation in polymer LED, similar to the situation of small-molecule organic LED.
This work was supported by the National Science Council and the Excellence Project of the Ministry of Education of the Republic of China.
*Corresponding author. Electronic address: [email protected]
1M. A. Baldo, D. F. O’Brien, M. E. Thompson, and S. R. Forrest,
Phys. Rev. B 60, 14422共1999兲.
2C. Adachi, M. A. Baldo, M. E. Thompson, and S. R. Forrest, J.
Appl. Phys. 90, 5048共2001兲.
3Z. Shuai, D. Beljonne, R. J. Silbey, and J. L. Brédas, Phys. Rev.
Lett. 84, 131共2000兲.
4K. Tandon, S. Ramasesha, and S. Mazumdar, Phys. Rev. B 67,
045109共2003兲.
5T. M. Hong and H. F. Meng, Phys. Rev. B 63, 075206共2001兲. 6C. H. Chen and H. F. Meng, Phys. Rev. B 68, 094112共2003兲. 7J. S. Wilson, A. S. Dhoot, A. J. A. B. Seeley, M. S. Khan, A.
Köhler and R. H. Friend, Nature共London兲 413, 828 共2001兲.
8M. Wohlgenannt, K. Tandon, S. Mazumdar, S. Ramasesha, and Z.
V. Vardeny, Nature共London兲 409, 494 共2001兲.
9L. C. Lin, H. F. Meng, J. T. Shy, S. F. Horng, L. S. Yu, C. H.
Chen, H. H. Liaw, C. C. Huang, K. Y. Peng, and S. A. Chen, Phys. Rev. Lett. 90, 036601共2003兲.
10A. S. Dhoot and N. C. Greenham, Phys. Rev. Lett. 91, 219702
共2003兲.
11R. Österbacka, Phys. Rev. Lett. 91, 219701共2003兲
12A. S. Dhoot, D. S. Ginger, D. Beljonne, Z. Shuai, and N. C.
Greenham, Chem. Phys. Lett. 360, 195共2002兲.
13A. S. Dhoot and N. C. Greenham, Adv. Mater.共Weinheim, Ger.兲
14, 1834共2002兲.
14A. R. Brown, K. Pichler, N. C. Greenham, D. D. C. Bradley, and
R. H. Friend, Chem. Phys. Lett. 210, 61共1993兲.
15H. D. Burrows, M. da G. Miguel, A. P. Monkman, I. Hamblett,
and S. Navaratnam, J. Mol. Struct. 563-564, 41共2001兲.
16A. J. Cadby, P. A. Lane, M. Wohlgenannt, C. An, Z. V. Vardeny,
and D. D. C. Bradley, Synth. Met. 111-112, 515共2000兲.
BRIEF REPORTS PHYSICAL REVIEW B 72, 113203共2005兲
