Effects of systematic methyl substitution of metal (III) tris(n-methyl-8-quinolinolato) chelates on material properties for optimum electroluminescence device performance

Article abstract of DOI:10.1021/ja010120m

We relate the chemical structure of a series of methyl (Me) substituted group III metal tris(8-quinolinolato) chelates (nMeq₃M: n = 0, 3, 4, 5; M = Al³⁺, Ga³⁺) to their photoluminescence (PL), electroluminescence, and thermal properties. Methylation of the 8-quinolinol ligand at the 3 or 4 position (pyridyl ring) results in a factor of 1.4 and 3.0 enhancement of PL quantum efficiency (O_PL), respectively, whereas methylation at the 5 position (phenoxide ring) results in a factor of ～ 3.0 decrease in _PL relative to the unsubstituted analogue. Electroluminescent quantum efficiencies of undoped organic light-emitting devices using the aluminum tris(8-quinolinolato) chelates are 1, 0.45, 1.4, and 0.80% for unsubstituted 5-, 4-, and 3-methyl-8-quinolinol ligands, respectively. Devices made with the latter two ligands have a higher operating voltage to generate the same current density. Similar trends were observed for methylation of gallium tris(8-quinolinolato) chelates. We relate these results to the thermal properties of the compounds measured by simultaneous differential scanning calorimetry and thermal gravimetric analysis. The C-4 methylated derivatives exhibit ～60 °C lower crystalline melting points than all other derivatives, indicating the weakest cohesive forces between molecules. Unlike Alq₃, both the C-4 and C-5 methylated derivatives show no recrystallization of the glassy state below 500 °C and exhibit ～-20-25 °C higher glass transition temperatures. We infer that methylation of the 8-quinolinol ligand reduces intermolecular interactions and consequently impedes charge transport through the film.

Full text of DOI:10.1021/ja010120m

6300
J. Am. Chem. Soc. 2001, 123, 6300-6307
Effects of Systematic Methyl Substitution of Metal (III)
Tris(n-Methyl-8-Quinolinolato) Chelates on Material Properties for
Optimum Electroluminescence Device Performance
Linda S. Sapochak,*,† Asanga Padmaperuma,† Nancy Washton,† Flocerfida Endrino,†
Gregory T. Schmett,† Jeffrey Marshall,† Daniel Fogarty,† Paul E. Burrows,‡,§ and
Stephen R. Forrest‡
Contribution from the Department of Chemistry, UniVersity of NeVada, Las Vegas, NeVada 89154-4003,
and Department of Electrical Engineering, Princeton UniVersity, Princeton, New Jersey 08544
ReceiVed January 16, 2001
Abstract: We relate the chemical structure of a series of methyl (Me) substituted group III metal tris(8-
quinolinolato) chelates (nMeq3M: n ) 0, 3, 4, 5; M ) Al3+, Ga3+) to their photoluminescence (PL),
electroluminescence, and thermal properties. Methylation of the 8-quinolinol ligand at the 3 or 4 position
(pyridyl ring) results in a factor of 1.4 and 3.0 enhancement of PL quantum efficiency (φPL), respectively,
whereas methylation at the 5 position (phenoxide ring) results in a factor of ∼3.0 decrease in φPL relative to
the unsubstituted analogue. Electroluminescent quantum efficiencies of undoped organic light-emitting devices
using the aluminum tris(8-quinolinolato) chelates are 1, 0.45, 1.4, and 0.80% for unsubstituted 5-, 4-, and
3-methyl-8-quinolinol ligands, respectively. Devices made with the latter two ligands have a higher operating
voltage to generate the same current density. Similar trends were observed for methylation of gallium tris(8-
quinolinolato) chelates. We relate these results to the thermal properties of the compounds measured by
simultaneous differential scanning calorimetry and thermal gravimetric analysis. The C-4 methylated derivatives
exhibit ∼60 °C lower crystalline melting points than all other derivatives, indicating the weakest cohesive
forces between molecules. Unlike Alq3, both the C-4 and C-5 methylated derivatives show no recrystallization
of the glassy state below 500 °C and exhibit ∼20-25 °C higher glass transition temperatures. We infer that
methylation of the 8-quinolinol ligand reduces intermolecular interactions and consequently impedes charge
transport through the film.
Introduction
More recently, Kido and Iisumi showed that aluminum tris(4-
methyl-8-quinolinolato) (4Meq3Al) exhibited a larger EL ef-
ficiency10 than that of Alq3.1 This material was later shown to
exhibit both a higher photoluminescence (PL) quantum ef-
ficiency than that of Alq3 in solution and in the solid-state.11
However, despite many experimental and theoretical studies of
the archetypal Alq3,3 systematic analyses of the effects of small
chemical changes are sparse and we lack a road map for
improving on the overall EL device performance of Alq3.
EL occurs when electrons and holes are injected and
transported through the device where they form excitons near
the organic heterojunction, which decay to give light emission.
In dye-doped OLEDs, EL occurs by energy transfer from the
host to the lower energy “guest” dye, which is usually
incorporated by coevaporation.4 In both types of devices, the
design of efficient emitter materials and/or charge transporting
hosts requires the optimization of several properties including
(1) high PL efficiency, (2) sublimation without chemical
degradation, (3) appropriate energy-level alignment with the
cathode and hole-transport layer (HTL) for balanced charge
injection and transport, and (4) the ability to form structurally
stable films. These properties are strongly coupled to the
Vacuum-deposited organic light-emitting device (OLED)
technology has developed rapidly since the first report by Tang
and Van Slyke of low voltage green electroluminescence (EL)
from aluminum tris(8-quinolinolato) (Alq3).1 Since then, many
other organic metal chelates have been demonstrated to be useful
emitter materials or electron transporting hosts in dye-doped
OLEDs.2-5
OLEDs are typically fabricated by vapor deposition of organic
charge-transport and emitter materials on an indium tin oxide
anode, followed by a low work function metal cathode. Many
derivatives of Alq3 have been investigated as emitter materials,
including substitution of the metal ion with other trivalent metals
(M+3 ) Ga, In, and Sc)5 and substitution of the 8-quinolinol
ligand (substituent ) F, Cl, NO2 , CN,7 and alkyl8,9 groups).
6
* To whom correspondence should be addressed.
† University of Nevada.
‡ Princeton University.
§ Present address: Pacific Northwest National Laboratories, Richland,
WA 99352.
(1) Tang, C. W.; VanSlyke, S. A. Appl. Phys. Lett. 1987, 51, 913.
(2) Tang, C. W.; VanSlyke, S. A. J. Appl. Phys. 1989, 65, 3610.
(3) Chen, C. H.; Shi, J. Coord. Chem. ReV. 1998, 171, 161.
(4) Shoustikov, A. A.; You, Y.; Thompson, M. E. IEEE J. Quantum
Electron. 1998, 4(1), 3.
(5) Burrows, P. E.; Sapochak, L. S.; McCarty, D. M.; Forrest, S. R.;
Thompson, M. E. Appl. Phys. Lett. 1994, 64(20), 2718.
(6) Matsumura, M.; Akal, T. Jpn. J. Appl. Phys. 1996, 35, 5357.
(7) Burrows, P. E.; Shen, Z.; Bulovic, V.; McCarty, D. M.; Forrest, S.
R.; Cronin, J. A.; Thompson, M. E. J. Appl. Phys. 1996, 79(10), 7991.
(8) Hamada, Y.; Sano, M.; Fujita, T.; Nishio, Y.; Shibata, K. Jpn. J.
Appl. Phys. 1993, 32(4A), L514.
(9) Kido, J.; Iizumi, Y. Chem. Lett. 1997, 963.
(10) Kido, J.; Iizumi, Y. Appl. Phys. Lett. 1998, 73(19), 2721.
(11) Murata, H.; Merritt, C. D.; Mattoussi, H.; Kafafi, Z. H. Proc. SPIE-
Int. Soc. Opt. Eng. 1998, 3476, 88.
10.1021/ja010120m CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/09/2001

Effects of Systematic Methyl Substitution
J. Am. Chem. Soc., Vol. 123, No. 26, 2001 6301
4MeqM chelates. These materials exhibited extremely high solubility
in most common recrystallization solvents and precluded the practicality
of purification by recrystallization. However, all materials were purified
by high-vacuum gradient-temperature sublimation15 before analysis.
1
Structures of all materials were confirmed by H NMR spectroscopy
reported elsewhere16 and elemental analysis obtained from Nu Mega
Resonances Lab, Inc. Spectroscopic data for the ligands and nMeq3M
chelates match those previously reported.14
Thermal Stability Characterization. Thermal analysis was deter-
mined by differential scanning calorimetry (DSC) and thermal gravi-
metric analysis (TGA) performed simultaneously using a Netzsch
simultaneous thermal analyzer (STA) system. Pure polycrystalline
samples (5-10 mg) were placed in aluminum pans and run at a rate of
20 °C/min under N2 gas at a flow rate of 50 mL/min. The temperatures
of phase transitions were measured including the crystalline melting
point (Tm). Indium metal was used as the temperature standard. The
melted samples were cooled rapidly or by a controlled cooling rate of
20 °C/min to form glasses. The glass transition (Tg) and the crystal-
lization point (Tc1) were measured from a second heating of the glassy
state. The maximum weight loss temperature (decomposition temper-
ature) was determined for each sample run without a lid and reported
as the maximum peak of the DTGA (derivative of the TGA curve).
Electronic Characterization. Absorption and photoluminescence
characterization were performed on dilute (∼10-5 M) dimethyl form-
amide (DMF) and methylene chloride (CH2Cl2) solutions. Absorption
spectra were recorded with a Varian Cary 3Bio UV-vis spectropho-
tometer, and PL spectra were obtained with a SLM 48000 spectro-
fluorimeter. Relative PL quantum efficiencies (φPL) were determined
from degassed DMF solutions by adjusting the concentration of the
sample so that the optical densities at 390 nm (excitation wavelength)
were <0.2 absorption units. PL quantum yields were calculated relative
to the known value for Alq3 in DMF (φPL ) 0.116)17 and are reported
normalized to Alq3.
Electroluminescence Characterization. Organic light-emitting
devices (OLEDs) were fabricated using each of the nMeq3M chelates
as the emissive layer. Devices were grown on glass slides precoated
with indium tin oxide (ITO) with a sheet resistance of 15 Ω/square.
The ITO substrates were degreased in detergent and then boiled in 1,1,1-
trichloroethane, rinsed with reagent grade acetone, and finally rinsed
with methanol before being dried in a stream of pure nitrogen. The
substrates were exposed to UV-ozone for 5 min before being loaded
into a nitrogen glovebox coupled to the vacuum system. A 500 Å thick
layer of the preferentially hole transporting material, N,N′-diphenyl-
N,N′-bis(3-methylphenyl)1,1′-biphenyl-4,4′-diamine (TPD) or N,N′-
diphenyl-N,N′-bis(1-naphthol)1,1′-biphenyl-4,4′diamine (R-NPD), was
deposited on the ITO substrate by thermal evaporation from a baffled
Mo crucible at a nominal rate of 2-4 Å/s under a base pressure of <2
× 10-6 Torr. A 550 Å thick layer of the preferentially electron-
transporting (ETL) metal tris(8-quinolinolato) chelates, also serving as
the emitter layer (EML), were then deposited on the HTL. A top
electrode consisting of circular 1 mm diameter contacts was subse-
quently deposited by thermal evaporation through a shadow mask. Two
different types of cathodes were utilized, LiF/Al and a Mg:Ag alloy.
The LiF/Al cathode consisted of a 5 Å LiF layer deposited on the EML
followed by a 1000 Å layer of Al metal. For the Mg:Ag cathode, a
Mg:Ag alloy layer (1000 Å) was deposited by coevaporation of the
two metals from separate Ta boats in a 10:1 Mg:Ag atomic ratio under
a base pressure of 10-5 Torr, followed by a 300 Å Ag cap. For the
systematic study, all HTL and cathode layers were deposited simulta-
neously and the devices were never exposed to air during fabrication.
Thus, devices produced from different metal chelate materials were
identical in all respects. A quartz crystal oscillator placed near the
substrate was used to measure the thickness of the films. Film thickness
calibration was performed by ellipsometry of films grown on silicon.
Devices were tested in air with an electrical pressure contact made
Figure 1. Metal (M ) Al3+, Ga3+) tris(8-quinolinolato) chelates
(nMeq3M) used in this study.
molecular and electronic structure of the molecule, as well as
the bulk packing of molecules achieved by vapor deposition.
In this paper, we present a study of the photoluminescent,
electroluminescent, and thermal stability properties of aluminum
and gallium tris(8-quinolinolato) (Mq3; M+3 ) Al, Ga) and their
methyl-substituted derivatives, metal tris(4-methyl-8-quinoli-
nolato) (4Meq3M), tris(3-methyl-8-quinolinolato) (3Meq3M),
and tris(5-methyl-8-quinolinolato) (5Meq3M; see Figure 1).
We show that methylation of the 8-quinolinol ligand decreases
the crystalline melting point of the metal tris-chelates and
increases the glass transition temperature of the amorphous
materials compared to the unsubstituted analogues, in both cases
indicative of reduced intermolecular interactions. In the case
of the 3Meq and 4Meq ligands, we observe higher EL device
operating voltages when the metal tris-chelates were used as
the emitter material in OLEDs. We interpret this as evidence
of a decreased overlap between the π-electron systems in the
pyridyl rings of adjacent molecules, which is the site of the
lowest unoccupied molecular orbital (LUMO) and, hence, the
likely location of injected electrons. To our knowledge, this is
the first established link between the electroluminescent and
thermal (physical) properties of a series of electroluminescent
compounds.
Experimental Section
Material Synthesis and Structural Characterization. All reagents,
including Alq3 used as the reference material in these studies were
obtained from Aldrich Chemical Co. The ligands 3Meq and 4Meq were
synthesized by the Doebner-Von Miller reaction starting with the
appropriately substituted o-aminophenol and unsaturated aldehyde or
ketone.12 After extraction of the reaction mixture, the ligands were
isolated in low yields (30%). The 5Meq ligand was prepared in higher
yields (57%) by electrophilic substitution of 8-quinolinol with form-
aldehyde and hydrochloric acid followed by reductive catalytic
hydrogenation.13 All ligands were purified by sublimation. The metal
tris-chelates were prepared by combining either AlCl3 or Ga(NO3)3
hexahydrate salt and the appropriate ligand in a 1:3 molar ratio in
aqueous solution buffered with ammonium acetate.14 The crude
materials were recrystallized from methanol with the exception of the
(15) Forrest, S. R.; Kaplan, M. L.; Schmidt, P. H. Annu. ReV. Mater.
Sci. 1987, 17, 189.
(12) Utermohlen, W. P. J. Org. Chem. 1943, 8, 544.
(13) Buckhalter, J. H.; Ceib, R. I. J. Am. Chem. Soc. 1961, 26, 4078.
(14) Schmidbaur, H.; Lattenbauer, J.; Dallas, L.; Muller, W. G.;
Kumberger, O. Z. Naturforsh. 1991, 46B, 901.
(16) Padmaperuma, A. M.S. Thesis in Chemistry, University of Nevada,
Las Vegas, 2000.
(17) Lytle, F. E.; Story, D. R.; Juricich, M. E. Spectrochim. Acta. 1973,
29A, 1357.

6302 J. Am. Chem. Soc., Vol. 123, No. 26, 2001
Table 1. Thermal Analysis Data
Sapochak et al.
evolved species after the major weight loss (DTGA) was
conducted.
max. wt.
Tm
∆Hfusion
Tg
Tc1 onset loss temp.
All metal tris(8-quinolinolato) materials in our study exhibited
similar weight loss behavior, where the magnitude of the weight
loss was independent of metal ion substitution but dependent
on the position of methylation of the ligand. Further heating to
500 °C resulted in decomposition of the samples.
Upon methyl substitution of Alq3, Tm increases in the order
4Meq3Al , 3Meq3Al ∼ 5Meq3Al < Alq3, and for the gallium
chelate series, 4Meq3Ga , 5Meq3Ga < Gaq3< 3Meq3Ga. In
addition, multiple phase transitions similar to Alq3 were
observed for Gaq3 and 3Meq3Al, suggesting that these materials
also exhibit polymorphism (see Figure 2). Although crystalline
melting points of the Mq3 chelates are lowered upon methyla-
tion, the thermal decomposition temperatures were higher than
the unsubstituted analogues with the exception of the 3Meq3M
chelates. The trend in thermal decomposition temperatures were
similar for both metal chelate series, with stability increasing
in the order 3Meq3M < Mq3 < 4Meq3M < 5Meq3M.
metal chelate (°C) (kJ/mol)a (°C)b
(°C)b
(°C)
Alq3
419
406
345
401
409
414
352
407
n.d.
n.d.
177
n.o.
194
198
n.o.
204
192
202
251
n.o.
n.o.
n.o.
244
269
n.o.
n.o.
487
465
494
495
472
454
473
480
3Meq3Al
4Meq3Al
5Meq3Al
Gaq3
3Meq3Ga
4Meq3Ga
5Meq3Ga
43.0
53.2
n.d.
56.7
40.8
52.9
a n.d. ) not determined because of overlapping peaks. b n.o. ) not
observable.
by means of a 25 µm diameter Au wire. Current-voltage characteristics
were measured with a Hewlett-Packard HP4145 semiconductor pa-
rameter analyzer, and the EL intensity was measured with a Newport
835 optical power meter with a large area photodetector placed directly
below the glass substrate. Thus, all efficiencies reported are measured
directly from the back of the device with no corrections or assumptions
regarding the angular dependence of the EL or the relationship between
luminance and quantum efficiency.10 Spectra were recorded with an
EG&G optical multichannel analyzer on a 0.25 focal length spec-
trograph.
The glassy state of Alq3 was difficult to form as discussed
by Naito19 but was achieved by cooling at a rate of 20 °C/min
of the crystalline material immediately following the melting
transition in the first heating cycle. A second heating cycle of
Alq3 revealed a glass transition (Tg) at 177 °C and a recrystal-
lization exotherm (Tc1) at 251 °C, followed by melting. These
values are slightly higher than previously reported19 because
of the faster heating rate used here. Glass formation was also
difficult for Gaq3 and 3Meq3Al, preventing determination of
Tg. However, 3Meq3Ga and the 4Meq3M and 5Meq3M chelates
formed glasses facilely and showed 15-20 °C higher Tg than
that of Alq3. A comparison of the second heating cycles for
Alq3 and 4Meq3Al glasses is shown in Figure 3. Unlike Alq3,
the 4Meq3M and 5Meq3M chelates showed no recrystallization
of the glassy state even after heating beyond Tg and up to
decomposition near 500 °C. This suggests that the entropic
difference between the crystalline and amorphous states is
smaller for the C-4 and C-5 methylated derivatives compared
to those of the C-3 methylated and unsubstituted analogues.
Electronic Properties. The optical transition responsible for
photoluminescence in metal (8-quinolinolato) chelates is cen-
tered on the organic ligand. This transition is due to a π-π*
charge transfer from the electron rich phenoxide ring (location
of the highest occupied molecular orbital, or HOMO) to the
electron deficient pyridyl ring (location of the LUMO).20
Therefore, as reported previously,3 substitution of an electron
donating group on the pyridyl ring raises the energy of the
LUMO and on the phenoxide ring raises the energy of the
HOMO, resulting in a blue and red-shift of the absorption
energies, respectively. The absorption spectra of the nMeq3M
chelates in DMF are shown in Figure 4. Upon methyl substitu-
tion (weakly electron donating) of aluminum tris(8-quinolino-
lato), the absorption energy shifts were largest for C-5 meth-
ylation of the phenoxide ring (red-shift, >1000 cm-1) and very
small for C-3 and C-4 methylation of the pyridyl ring (blue-
shift, < 350 cm-1). Substitution of the heavier metal ion Ga
causes a red-shift of absorption compared to those of the nMeq3-
Al chelates because of the heavy atom effect discussed previ-
ously,5 but the trends in absorption shifts upon methylation of
the 8-quinolinol ligand were identical to those of the aluminum
series. The polarity of the solvent had almost no effect on
absorption energies as shown in Table 2 for CH2Cl2 and Table
3 for the more polar DMF solvent.
Results
Thermal Stability. The morphological stability of Alq3,
desirable for practical OLED applications, has been attributed
to its intrinsic polymorphic, racemic nature, and strong dipolar
interactions18 giving rise to a high Tg (175 °C).19 The thermal
properties of Alq3 and Gaq3 and their methyl-substituted
derivatives were evaluated by simultaneous thermal analysis
(STA), where TGA and DSC experiments were run simulta-
neously; therefore, thermal events observed in DSC can be
directly correlated with weight loss events. Results are presented
in Table 1.
Several phase transitions were observed for Alq3 in addition
to a large melting transition (Tm) at 419 °C. A small broad
transition beginning at 320 °C and a coupled endotherm (393
°C) and exotherm (396 °C) were observed prior to Tm. These
additional phase transitions are attributed to polymorphism of
the crystalline material as discussed by Brinkmann18 and were
nearly unobservable at a slower heating rate (5 °C/min). The
thermal transitions were accompanied by a ∼15% weight loss
measured immediately following Tm (endpoint ) 422 °C). This
weight loss is attributed to sublimation of the sample, which
occurs simultaneously with melting of the different polymorphic
phases. This is supported by Brinkman’s assignment of the
exothermic transition observed by DSC (see Figure 3a) as a
phase transition from R-Alq3 to the γ-Alq3 phase at ∼395 °C,
where further heating of the material (>410 °C) was shown to
result in sublimation (∆H ) 49 kJ/mol) of the R-Alq3 phase,
confirmed by X-ray diffraction. We observed similar thermal
behavior for Alq3 and, thus, strongly believe that the nature of
the ∼15% weight loss was due to sublimation. Furthermore,
when stopping the thermal experiment just past the major
endothermic transition (Tm) a film was observed on the inside
lid of the sample pan for all materials studied. The material left
in the pan was analyzed by 1H NMR and FT-IR, and no
decomposition products were observable by those techniques.
In a separate experiment, DTGA, and thus percent weight loss,
was determined without a lid and no attempt to analyze the
(18) Brinkmann, M.; Gadret, G.; Muccini, M.; Taliani, C.; Masciocchi,
N.; Sironi, A. J. Am. Chem. Soc. 2000, 122, 5147.
(19) Naito, K.; Muira, J. Phys. Chem. 1993, 97, 6240.
(20) Curioni, A.; Boero, M.; Andreoni, W. Chem. Phys. Lett. 1998, 294,
263.

Effects of Systematic Methyl Substitution
J. Am. Chem. Soc., Vol. 123, No. 26, 2001 6303
Figure 3. DSC traces of the second heating after melting and controlled
cooling to the glassy state for (a) Alq3 and (b) 4Meq3Al showing Tg’s
of 177 and 194 °C, respectively.
Figure 2. DSC and TGA traces of metal tris(8-quinolinolato) chelates
showing multiple thermal transitions (°C): (a) Alq3, 354, 393, 396
(exotherm), 419; (b) Gaq3, 369, 387, 396 (exotherm), 409; and (c)
3Meq3Al, 233, 275, 355, 358 (exotherm), 370, 377 (exotherm), 406.
Figure 4. Absorption spectra run in DMF showing the energy shifts
upon methylation of the ligand for (a) nMeq3Al and (b) nMeq3Ga
chelates.
Figure 5 shows the photoluminescence (PL) spectra obtained
in DMF solution. The effect of methyl substitution on the PL
emission energies were similar to those observed for absorption.
The Franck-Condon (F-C) shifts (energy difference between
absorption and emission maxima; ∆) and the fwhm of emission
are also reported. The smallest F-C shifts and the narrowest
fwhm were observed for 4Meq3Al (5912 cm-1) and 4Meq3Ga
(6280 cm-1) compared to those of the unsubstituted analogues
Alq3 (6623 cm-1) and Gaq3 (7026 cm-1).

6304 J. Am. Chem. Soc., Vol. 123, No. 26, 2001
Table 2. Photophysical Data in CH2Cl2 Solution
Sapochak et al.
abs.
energy
shift
emission
energy
shift
abs.
emission
CH2Cl2
λmax
CH2Cl2 from
from
metal
chelate
λmax
(nm)
Mq3
Mq3
fwhm
∆
(cm-1
)
(nm)
(cm-1
)
(nm) (cm-1
)
Alq3
388
386
381
405
392
386
385
408
514
520
495
547
543
540
514
568
102
92
92
109
110
110
102
112
6318
6776
6045
6410
7162
7388
6519
6904
3Meq3Al
4Meq3Al
5Meq3Al
Gaq3
3Meq3Ga
4Meq3Ga
5Meq3Ga
+133
+473
-224
+747
-1082
-1174
+396
+464
-1000
+102
+1039
-810
The PL quantum yields are reported relative to the known
quantum yield for Alq3 in DMF and normalized to φPL(Alq3)
) 1.00 for clarity in Table 3. The highest φPL is exhibited by
4Meq3Al (3 times Alq3). This is higher than those in reports by
Murata et al.11 in CHCl3 (2 times Alq3) and in the solid-state
(1.6 times Alq3). The higher polarity of the DMF solvent
enhances the quantum efficiency more for 4Meq3Al relative to
Alq3 but has almost no effect on the PL λmax which is similar
to that reported by Kido and Iizumi for the solid-state film.10
The 3Meq3Al chelate also showed an increase in φPL compared
to Alq3 but of smaller magnitude, whereas the 5Meq3Al
derivative exhibited a significant decrease in PL quantum
efficiency (approximately 3 times less than that of Alq3). The
same trends in PL efficiencies upon methyl substitution were
observed when Al was replaced with Ga but with a correspond-
ing 4 times decrease in φPL as shown previously for Gaq3.5,21
In general, methylation of the pyridyl ring enhances φPL and
methylation of the phenoxide ring decreases it.
We speculate that the large enhancement in φPL by C-4
methylation of Al and Ga tris (8-quinolinolato) chelates is due
to reduced energy loss in the excited-state vibrational manifold,
as indicated by the smaller ∆’s and narrower fwhm’s. This is
supported by a recent study of the infrared absorption and
photoluminescence of matrix-isolated and thin-film samples of
Alq3 and 4Meq3Al compared to vibrational spectra based on
density functional calculations.22 There, the PL spectra of matrix-
isolated molecules revealed vibronic bands characterized as a
combination of ring breathing and metal-ligand stretching
modes. Kushto et al. noted that there was a lack of metal-
ligand character in the vibrations of 4Meq3Al as compared to
Alq3, suggesting that the stronger coupling of the metal-ligand
stretching coordinates to the electronic transition in Alq3 may
provide additional paths for nonradiative decay that are not
available to 4Meq3Al, giving rise to a higher φPL for the latter.
Electroluminescence Properties. EL spectra are shown in
Figure 6. The EL emission energy shifts are consistent with
those observed in the PL spectra. However, the 5Meq3Ga gave
rise to an additional lower energy peak in the EL spectrum that
was not observed in the solution PL spectrum (Figure 6b). The
origin of this peak has not been elucidated but was observed as
a very weak shoulder in the PL spectrum of a vapor-deposited
film, thus excluding the origin due to exciplex formation at the
interface with the HTL material as reported for other materials.23
However, this does not exclude the possibility of excimer
Figure 5. Photoluminescence spectra run in DMF for (a) nMeq3Al
and (b) nMeq3Ga chelates.
formation, which may be one of the reasons for the substantially
lower PL quantum efficiencies for the C-5 methylated deriva-
tives. The EL emission spectrum for the 3Meq3Al derivative is
not shown but was identical to that of Alq3.
The current versus voltage characteristics of OLEDs com-
posed of ITO/NPD/nMeq3M/LiF-Al are shown in Figure 7.
The device operating voltages (at a fixed current) increases in
the order Mq3 < 5Meq3M < 4Meq3M for both aluminum and
gallium tris-chelates. Figure 8 shows the dependence of the
optical output power on drive current for each of the OLEDs.
At a current of 100µA, the EL output increases in the order
5Meq3M < Mq3 < 4Meq3M. Identical devices prepared with a
Mg:Ag cathode gave similar results with the exception of
4Meq3Ga which required a lower voltage compared to that of
5Meq3Ga.
The EL efficiencies (φEL) relative to Alq3 ) 1.0 are reported
in Table 4. For both series of nMeq3M chelates, the C-4
methylated derivatives exhibited the highest efficiencies. 4Meq3-
Al and 4Meq3Ga showed approximately a 1.4 times increase
in ηEL compared to Alq3 and Gaq3 respectively, for both types
of cathodes utilized. However, the ηEL for 4Meq3Al was smaller
than previously reported for a similar device using TPD as the
HTL and Mg:Ag as the cathode.9,10 Therefore, a second series
of devices were prepared by varying the HTL (R-NPD versus
(21) Sapochak, L. S.; Burrows, P. E.; Garbuzov, D.; Ho, D. M.; Forrest,
S. R.; Thompson, M. E. J. Phys. Chem. 1996, 100, 17766.
(22) Kushto, G. P.; Iizumi, Y.; Kido, J.; Kafafi, Z. H. J. Phys. Chem. A
2000, 104, 3670.
(23) Ogawa, H.; Okuda, R.; Shirota, Y. Mol. Cryst. Liq. Cryst. Sci.
Technol. 1998, 315, M489.

Effects of Systematic Methyl Substitution
J. Am. Chem. Soc., Vol. 123, No. 26, 2001 6305
Table 3. Photophysical Data in DMF Solution
metal
chelate
abs. DMF
abs. energy shift
emission
DMF λmax (nm)
emission energy shift
fwhm
(nm)
∆
relative φPL
(Alq3 )1.0)
λ
max (nm)
from Mq3 (cm-1
)
from Mq3 (cm-1
)
(cm-1
)
Alq3
388
387
383
405
392
392
388
409
521
521
494
546
541
543
513
558
105
106
92
113
110
111
103
119
6623
6668
5912
6376
7026
7094
6280
6529
1.0
1.4
3.0
0.29
0.28
0.38
0.79
0.08
3Meq3Al
4Meq3Al
5Meq3Al
Gaq3
3Meq3Ga
4Meq3Ga
5Meq3Ga
+66
+336
0
+1049
-879
-1082
0
+263
-1060
-68
+1009
-563
Figure 7. Current vs voltages plots for Alq3, 4Meq3Al, and 5Meq3Al
OLEDs.
Figure 6. EL spectra obtained from LiF-Al/nMeq3M/NPD OLEDs
for (a) nMeq3Al and (b) nMeq3Ga chelates.
Figure 8. Light output vs current plots for Alq3, 4Meq3Al, and 5Meq3-
Al OLEDs.
TPD) with the nMeq3Al chelates as the emitter material and
Mg:Ag as cathode. Results are shown in Table 5. In general,
we observed similar trends in ηEL with a slightly higher
efficiency for TPD-based devices because of better energy level
alignment at the ETL/HTL interface. The voltage trend did not
change with the composition of the HTL, suggesting that the
source of higher voltages (and hence the overall lower emission
power efficiencies) for the methylated derivatives is mainly due
to reduced charge transport efficiency of these latter compounds
as compared to that of Alq3.
caused by methylation of the ligand. The substantially larger
PL efficiencies for C-4 methylated derivatives contributed to
their larger ηEL. However, for all methylated derivatives, if
charge injection and transport were similar and φPL was the
dominating factor for high EL efficiencies, then the ratio of
relative EL and PL efficiencies (ηEL/φPL) should all be close to
1.0. However, this ratio is <1.0 for 4Meq3Al and 3Meq3Al but
>1.0 for 5Meq3Al. For the gallium series, all ratios are >1.0
with 4Meq3Ga showing the smallest value. This suggests that
charge injection and/or transport efficiencies are not the same
for these materials and are reduced by methylation of the pyridyl
ring for both Alq3 and Gaq3, compared to the other methylated
Discussion
The differences in EL performance of the nMeq3M chelates
in this study can be related to the changes in material properties

6306 J. Am. Chem. Soc., Vol. 123, No. 26, 2001
Table 4. EL Device Data Utilizing Different Cathodes
Sapochak et al.
LiF-Al/EML/NPD device
Mg:Ag/EML/NPD device
ηEL
voltage
η
EL/φPL
(Alq3)1.0)
ηEL
voltage
ηEL/φPL
EML
Alq3
4Meq3Al
5Meq3Al
Gaq3
(Alq3)1.0)
(at 13mA/cm2)
(Alq3)1.0)
(at 13mA/cm2)
(Alq3)1.0)
1.0
1.4
0.45
0.63
1.0
5.3
7.8
7.4
5.7
7.6
5.8
1.0
0.47
1.5
2.2
1.3
2.6
1.0
1.3
0.49
0.65
0.97
0.24
5.6
8.2
7.3
6.4
6.5
8.0
1.0
0.45
1.7
2.3
1.2
3.0
4Meq3Ga
5Meq3Ga
0.21
Table 5. EL Device Data Utilizing Different HTLs
voltage ηEL
Thermal analysis showed that the 4Meq3M chelates exhibited
the lowest (∼60 °C) melting points and smaller enthalpies of
fusion (∆Hfus) of the series, also clearly indicative of weaker
intermolecular interactions. Glass transition temperatures were
∼20 °C higher than those of the unsubstituted analogues with
no recrystallization of the amorphous state, suggesting that the
entropic difference between the crystalline and amorphous states
is small compared to that of Mq3 materials. These differences
in thermal properties result because C-4 methyl substitution of
the ligand acts to eliminate one of the H-bonding interactions,
as well as decrease the overlap between pyridyl rings of adjacent
molecules. Although methyl substitution at the C-3 position
could also act to prevent efficient overlap of pyridyl rings, the
C-4 and C-5 positions of 3Meq3M chelates are still available
for H-bonding interactions. The 3Meq3M chelates do not show
a significantly lower crystalline melting point and do not form
stable glasses. For both 4Meq3M and 3Meq3M chelates,
therefore, a decrease in pyridyl ring overlap leads to inefficient
electron transport properties because the pyridyl ring of the
8-quinolinol ligand is the site of the LUMO and, hence, the
location of injected electrons. This is the likely source of the
higher operating voltages exhibited by 4Meq3Al- and 3Meq3-
Al-based OLEDs.
η
EL/φPL
EML/ HTL (at 13 mA/cm2) (Alq3/NPD ) 1.0) (Alq3/NPD ) 1.0)
Alq3/NPD
6.3
7.2
7.1
6.9
7.1
7.7
6.2
1.0
0.80
1.2
0.39
1.1
1.6
1.0
3Meq3Al/NPD
4Meq3Al/NPD
5Meq3Al/NPD
Alq3/TPD
4Meq3Al/TPD
5Meq3Al/TPD
0.64
0.40
1.3
1.1
0.52
1.7
0.48
derivatives, but less so for the latter. This is consistent with the
need for higher operating voltages for OLEDs utilizing 4Meq3-
Al and 3Meq3Al.
Because methylation of the pyridyl ring (C-3 or C-4 positions)
causes only very small shifts in the absorption energy and thus
does not significantly shift the energy of either the HOMO or
LUMO compared to those of the unsubstituted analogues, it
seems likely that the energy level matching with the metal
cathode and HTL are similar to the unsubstituted analogues.
As a result, charge-injection efficiency is unlikely to be
significantly affected by C-4 or C-3 methylation, and indeed,
we observe power law, interface limited charge transport in all
cases.24 Therefore, the reason for the higher operating voltages
is due to poorer electron transport through the metal/organic
interface region.
The efficiency of charge-transport in organic molecular
crystals is enhanced by strong intermolecular interactions of the
polarizable π-system and is dependent on the molecular packing
in the solid-state. Although vapor-deposited films of Alq3 and
Gaq3 are amorphous, crystallographic data for single crystals
provides important information about their molecular packing
preferences. For example, Alq3 and Gaq3 show strong dipolar
π-π stacking interactions (3.5-3.9 Å) of the 8-quinolinol
ligands between adjacent molecules with preferential overlap
of the pyridyl rings.14,18,25 Because the pyridyl ring is the location
of the LUMO, this explains the higher electron verses hole
mobility in Alq3 and related metal(8-quinolinolato) chelates.
Additionally, hydrogen-bonding interactions between the phe-
nolic oxygen with H-atoms at the C-5 and C-4 positions on
adjacent molecules have been identified for Gaq3.25 Therefore,
it is not unexpected that a decrease in π-π stacking due to
steric factors, changes in polarity (via changes in the dipole
moment of the ligand),26 as well as loss of hydrogen-bonding
interactions of the metal chelates will result upon methylation
of the 8-quinolinol ligand. These changes will reduce the
cohesive forces between molecules, and analysis of the series
of methyl-substituted derivatives by thermal analysis reveals
these differences.
The device operating voltage for 5Meq3Al was also increased
compared to that of Alq3 but less so than the 3- and 4Meq3Al
chelates. The crystalline melting point of 5Meq3Al was slightly
lower than those of 3Meq3Al and Alq3, but it exhibits a high
glass transition temperature with no recrystallization of the
amorphous state similar to that of 4Meq3Al. C-5 methylation
of the phenoxide ring will also result in decreased π-π stacking
compared to that of Alq3 with loss of the C-5 H-bonding
interaction but will not prevent preferential overlap of the pyridyl
rings. Furthermore, unlike the 4Meq3M and 3Meq3M chelates,
in an OLED, hole-injection may be improved because the
HOMO energy of Alq3 is raised on C-5 methylation, and
therefore, 5Meq3Al may also have closer energy level alignment
with the HOMO of the HTL. We speculate, therefore, that both
charge transport and hole injection efficiencies may be higher
in 5Meq3Al OLEDs explaining why ηEL/φPL is much larger for
5Meq3Al compared to those of 4Meq3Al and 3Meq3Al.
The effect of methylation of Gaq3 on the thermal properties
was similar to that of the aluminum series. However, the EL
properties cannot be explained in the same manner as those of
Alq3 and its methylated derivatives. Although the gallium series
exhibited an approximately four times decrease in relative PL
quantum efficiencies compared to those of the nMeq3Al
analogues, ηEL/φPL ratios are substantially larger for all nMeq3-
Ga chelates. The gallium series are red-shifted in absorption
energy compared to that of the aluminum chelate analogues
because of a higher HOMO energy resulting in improved energy
level alignment and more efficient hole injection at the HTL/
nMeq3Ga interface. Because the operating voltage is dependent
on both charge injection and charge transport efficiencies, the
(24) Baldo, M. A.; Forrest, S. R. 10th Int. Workshop on Inorganic and
Organic Electroluminescence (EL’00); Hamazuatsu, Dec 4-7, 2000; Paper
Mo-08, p 53.
(25) Wang, Y.; Zhang, W.; Li, Y.; Ye, L.; Yang, G. Chem. Mater. 1999,
11, 530.
(26) Bader, M. J. Am. Chem. Soc. 1992, 114, 6575.

Effects of Systematic Methyl Substitution
J. Am. Chem. Soc., Vol. 123, No. 26, 2001 6307
Conclusions
enhancement of the former may outweigh the effects of reduced
charge transport caused by methylation of the ligand.
We are currently investigating the effect of methylation of
zinc bis(8-quinolinolato) chelate materials. In those materials,
the trends in absorption and emission energy shifts, PL quantum
efficiencies, and thermal properties upon methyl substitution
are similar to those of the tris-chelate series. This suggests that
the dipolar interaction and preferential overlap of pyridyl rings
of adjacent molecules may be similar for the tetracoordinate
zinc bis-chelates, and therefore, similar trends in EL device
characterization are expected.
One way to test if preferential overlap of pyridyl rings is
important for electron transport in metal (8-quinolinolato) chelate
materials is to compare the effect substituents with different
geometries have on the resulting EL properties. For example,
comparison of metal (8-quinolinolato) derivatives with C-4
substitution of the ligand with a trifluoromethyl group, having
approximately tetrahedral geometry, versus a cyano group with
linear geometry will result in different bulk packing character-
istics. It is expected that the linear cyano group will be less
prone to preventing overlap of pyridyl rings and, thus, should
not inhibit charge transport. However, unlike the methyl group,
both the trifluoromethyl and cyano groups are strongly electron
withdrawing, which will cause a lowering of the LUMO energy
of resulting metal (8-quinolinolato) chelates. Therefore, the
resulting EL efficiencies will also depend on how the C-4
substitution influences the HOMO-LUMO overlap within the
ligand itself.
A systematic study of the effect of methylation of aluminum
and gallium tris(8-quinolinolato) chelates on the PL, EL, and
thermal properties was presented. We showed that, although
φPL and ηEL were increased by C-4 methylation of the 8-quino-
linol ligand for both series of metal chelates, a higher device
operating voltage was required because of a decrease in pyridyl
ring overlap, thus leading to poorer electron transport efficiency.
These results were explained by correlating the thermal proper-
ties of the methylated metal tris-chelates and the OLED device
performance characteristics. Our results suggest that it may be
possible to improve overall EL efficiency of metal tris(quino-
linolato) chelates by a judicious choice of both the type of
substituent group and the position on the 8-quinolinol ligand in
order to achieve a balance between high φPL, charge injection
efficiency, and charge transport ability. The conclusions drawn
from the thermal analysis data are important for understanding
how methylation affects the physical properties of the metal
chelates. However, such an interpretation is only valid when
comparing a series of chemically similar compounds, where sim-
ilar comparisons of unrelated materials would not be as useful.
Acknowledgment. The authors gratefully acknowledge NSF
(DMR-9874765 Grant), Research Corporation, DOE, DARPA,
AFOSR, and Universal Display Corporation for financial support
of this research.
JA010120M