oleds,the dawn of ultra efficient lighting

OEM Research Group

Prof Monkman, OEM Research Group

Department of Physics, Durham University

http://www.dur.ac.uk/OEM.group [email protected]

OLEDs, the dawn of ultra efficient lighting

Themes

Monkman OEM Research Group

OLEDs, the dawn of ultra efficient lighting

General considerations of making OLED lighting

Problems facing blue phosphorescence

The brave new world of TADF!

Deutsche Bank Berlin

Better design esthetics, better quality of lighting, cheaper to run ‘off grid’ lighting for the third world

We must get away from this

UK strongly engaged with lighting designers and architects and lighting industry in major UK funded R&D projects

TOPLESS and TOPDRAWER projects

What does OLED lighting offer?

1953 2013

Design No No’s

impractical/fanciful

more of the same

welcome to the mental hospital


Better Design

Introducing dimensionality to break up the ceiling space


This is now much better Design Small panels give much higher resolution to the shapes within a small vertical distance. Very important for design vital for production yields. We do not want 1m x 1m panels And it will be hybrid OLED/LED


Fabrication/manufacture minimal patterning required solution/vacuum deposition hybrid processing thus slot die coating for soluble layers pre patterned TCO with bus bars potentially ink jet/electroless plating hybrids enable multilayer and p-i-n structures Small panels, improve yield reduced cost of bus bars on TCO high compatibility with hybrid

What does OLED lighting offer? FP7-224122 – OLED100.eu D5.7 - Second set of psycho-physiological perception case studies

Version 3.0 - 26/11/2010 Consortium Confidential Page 40 of 111

4.2.2.2 “Crater” Gradient

Figure 4.10: Real picture of the light box (right) and luminance distribution pattern with indicated luminance profile cross-sections

Figure 4.11: Transversal (top) and diagonal (bottom) cross-sectional luminance profile

Figure 4.12: Luminance distribution pattern for 8 gradient levels, the mean luminance is varying between 944 and 1029 cd/m!

!

"!!

#!!

$!!

%!!

&!!!

&"!!

&#!!

!"#$%&%

'()$%)'*+#,

-./&0(/-)0/&%12(/1( '(%

'()

'($

'(*

'(#

'(+

'("

!

*!!

&!!!

&*!!

!"#$%&%

'()$%)'*+#3

-./&0(/-)*$&45%&6 '(%

'()

'($

'(*

'(#

'(+


OLED ‘lamps’ being large area surface emitters have lower glare and so do not require a luminaire to control brightness

Fluorescent T5 lamps inside a metal reflector luminaire to control brightness

The luminaire introduces typical 40% loose of light output, so a 90 lm/W tube effectively gives a light sources of less than 55 lm/W

Inside an OLED

Glass

ITO (125 nm)

PEDOT:PSS (ca. 50 nm)

Emissive layer (ca. 70 nm)

LiF (1 nm)

Al (100 nm)

TPBI (30 nm)

Simple OLED pixel structure

low work function composite metal cathode

electron transport layer ‘ETL’ combined emissive and hole transport layer HTL hole injection layer ‘HIL’

transparent anode ‘TCO’

Inside an OLED

Simple OLED pixel structure

hole injection layer ‘HIL’

Schematic energy level diagram of an OLED under bias

Both HTL and ETL are thick layers via the use of doping i.e. p-i-n structure

Inside an OLED

Complex OLED lighting structure

OLED Stack 1

OLED Stack 2

Charge regeneration layer MoO3 or TPBi:Cs3PO4

Two stacked OLEDs running at half effective brightness, thus half drive current required at twice the applied voltage of a single stack. This improves lifetime by a factor of 4 at high brightness.

Best commercial white panels from LG Chem 10 cm x 10 cm, 80 lm/W

Inside an OLED

varying IL thickness were processed as described in Sec. IIA. For comparison a reference device without IL was fabri-cated as well. Fig. 3 shows the j-V characteristics.

It is remarkable that the steepest characteristic is meas-ured for the device with 4 nm CuPc IL and not for the refer-ence device without IL. One would assume intuitively thatby inserting an undoped IL material, the IL should act as aresistor and therefore the voltage drop across the CGLshould be increased. But this is not the case here. The volt-age drop across the CGL can even be reduced by inserting4 nm of the organic IL CuPc. Whereas inserting 4 nm Al2O3

leads to a rise of the applied voltage for a given tunnelingcurrent. The larger the Al2O3 layer thickness the flatter the j-V characteristics become. It can be concluded that all CGLtest devices comprising the Al2O3 IL have lower conductiv-ity compared to the device without IL and the devices withCuPc IL, respectively. Therefore, an Al2O3 IL acts as a car-rier blocking layer in a doped p-n junction.

The stability of the CGL test device comprising theCuPc and the Al2O3 IL was investigated by applying a con-stant current density of 10 mA/cm2 and monitoring the volt-age over time. The experimental data can be found in Figs. 4and 5, respectively. We note that these current densities cor-respond to accelerated testing conditions in OLEDs. For ahigh operational stability in an OLED, it is required that thevoltage remains constant over time. Within the first hour apronounced voltage rise for the device without IL can bemeasured. Clearly, such a behavior would be detrimental fora stacked OLED. At the same time, the devices with theCuPc IL (see Fig. 4) show a voltage drop, which we attributeto a temperature enhanced conductivity of organic materialsdue to heating during the measurement.22

At later times, a significant voltage rise can be seen forthe device with 2 nm CuPc during electrical aging. Only asmall voltage rise can be found for the device with 4 nmCuPc, whereas there is no voltage rise for the device with8 nm CuPc.

The devices with the Al2O3 IL show qualitatively a sim-ilar behavior during electrical aging. With rising interlayerthickness, the voltage stability of the devices is increased.But we have to point out that for an acceptable stability of

FIG. 3. Current density-voltage characteristics of CGL test devices (struc-ture shown in Fig. 1) with a narrow-gap organic and wide-gap oxidic inter-layer material of different thicknesses. (Filled symbols: CuPc, opensymbols: Al2O3).

FIG. 2. Device structure of the non-stacked (a) reference OLED and thestacked (b) OLED without and with an 8 nm CuPc IL. The total organiclayer thickness of all 3 fabricated devices is the same.

FIG. 4. Voltage dependency over time at a constant current density of10 mA/cm2 of CGL test devices. The CuPc IL thickness was varied between0 nm and 8 nm.

FIG. 5. Voltage dependency over time at a constant current density of10 mA/cm2 of CGL test devices. The Al2O3 IL thickness was variedbetween 0 nm and 12 nm.

103107-3 Diez et al. J. Appl. Phys. 111, 103107 (2012)CGL technology

expect an additional voltage drop due to the undoped dielec-tric material acting as a resistor. In fact, this is what weobserve with the Al2O3 IL, but not with CuPc. Instead, forCuPc the voltage drop decreases up to a layer thickness of4–8 nm, corresponding to the formation of a closed layer29–31

and thus resulting in a smaller barrier and hence higher tun-neling current as observed in our experiments.

We now turn to the mechanism of charge generation.Al2O3 is a wide-gap insulating material, whereas CuPc is amaterial with a narrow-gap lying between the HOMO of thep-doped layer and the LUMO of the n-doped layer.23,33

Fig. 11 shows a simplified drawing of the CGL energy-leveldiagram for devices without IL ((a) and (b)), with Al2O3 IL((c) and (d)), and with CuPc IL ((e) and (f)). The energy-level diagram for the device without IL is taken from Krogeret al. Note that in an ideal p-n junction without consideringinterfacial dipoles the depletion zone is in the order of10–20 nm assuming typical free carrier concentrations of10!1–10!19 cm!3. As a matter of fact, Kroger et al. foundthat without considering interfacial dipoles in their system itis not possible to explain the tunneling mechanism. There-fore, Figs. 11(a) and 11(b) have to be seen as a simplifiedsketch. In the case of the wide-gap Al2O3 the interlayer leadsto an additional tunneling barrier (see Fig. 11(d)), resultingin an increased drive voltage as we found in our experiments.By inserting the narrow-gap CuPc interlayer the large bandbending at the interface may result in an additional tunnelingpath from the HOMO of the p-type organic semiconductorvia the LUMO of CuPc to the LUMO of the n-type semicon-ductor as indicated in Fig. 11(f). The sketch of the energy-

level diagram is not correct in a quantitative sense, but itqualitatively explains the observed charge generation. Fur-ther measurements by, e.g., ultraviolet photoelectron spec-troscopy (UPS) would be helpful to clarify the exact valuesof the energy levels but are beyond the scope of thispublication.

Furthermore, the tunneling probability of charge carriersthrough the depletion zone could be further enhanced by thethin CuPc IL within the CGL, because of the formation of anadditional interfacial dipole. Lai et al.36 found that there is aninterfacial dipole between CuPc and F16CuPc. Another impor-tant aspect is that CuPc may introduce additional gap states.Such gap states have been proposed in several reports onenhanced charge carrier injection by inserting CuPc layersbetween the anode and the organic layer sequence.29–31,33

Schobel found by UPS measurements additional gap stateswhich explained the improved charge carrier injection.34 Inaddition, in other CGL systems gap states have been deliber-ately introduced for assisting charge carrier injection byinserting an Li2O interlayer35 or by inserting a combination ofa thin LiF/Al layer.5 In our case, such gap states could assistelectron tunneling from the HOMO of the p-doped layer tothe LUMO of the n-doped layer.

Finally, band-to-band tunneling from the HOMO ofCuPc to the LUMO of CuPc is also likely, if the band bend-ing is large enough. Our experimental findings of theincreased current density for the device with the 4 nm CuPcIL support these hypotheses explaining the very efficientcharge generation.

V. CONCLUSION

We showed that a highly stable CGL based on a doped p-n junction can be formed by using MoO3 as p-type dopant andCs3PO4 as novel n-type dopant. The investigation of two dif-ferent classes of interlayer material inserted within the dopedjunction, on the one hand, the narrow-gap organic materialCuPc, and on the other hand, the wide-gap oxidic materialAl2O3, demonstrated that both types of IL help to stabilize theCGL. Furthermore, the voltage consumption of a CGL com-prising a thin IL of CuPc can be further improved and an opti-mal IL thickness of 8 nm can be found as a trade-off betweenminimal operating voltage and maximum voltage stability ofthe CGL. Luminance-current density-voltage as well as life-time measurements on stacked green OLEDs confirmed thegood functionality and stability of the developed CGL. Thelifetime of the stacked device was enhanced by a factor of 3.5without reduction in efficiency or luminous flux. We dis-cussed the charge generation mechanism of the CGL with dif-ferent types of IL and proposed a model of the energy-levelalignment which describes consistently our experimental find-ings. From our point of view, the IL is needed to preventchemical reactions or dopant interdiffusion at the p-n interfaceleading to a reduced drive voltage and an enhanced stabilityof the devices. By choosing a proper IL material, e.g., CuPc,the energy-level alignment at the interface can be modified byinterfacial dipoles, and gap states can be deliberately intro-duced leading to an enhanced tunneling current and thereforeincreased performance of stacked OLEDs.

(a.) no external bias (b.) under reverse bias

p-doped n-doped p-doped n-doped

(c.) no external bias,with Al O IL2 3

(d.) under reverse bias,with Al O IL2 3

p-doped n-doped

IL

p-doped n-doped

IL

(e.) no external bias,with CuPc IL

(f.) under reverse bias,with CuPc IL

p-doped n-dopedIL

p-doped n-dopedIL

FIG. 11. Simplified model of the energy-level alignment for a CGL devicewithout IL ((a) and (b)), with Al2O3 IL ((c) and (d)) and with CuPc IL ((e)and (f)): (a), (c), and (e) no external bias, (b), (d), and (f.) under reverse bias.

103107-6 Diez et al. J. Appl. Phys. 111, 103107 (2012)

The CGL under reverse bias generates electron hole pairs which are separated by the bias field and moves into the two adjacent OLED stacks.

Diez et al, J. Appl. Phys., 11, 103107, 2012

Anode NPD:MoO3

Cathode BCP:Cs3PO4

What challenges remain in OLED lighting?

There are two major current challenges 1)  Improve the out-coupling of light from the OLED

Only 22-25% of light escapes from the OLED, current out-coupling techniques can increase this to nearly 50%. Thus there is still half the light being lost, getting that light out is a major challenge 2) BLUE – a real materials challenge The current blue emitters are not blue enough or stable enough. Host materials have to be extremely high triplet energy which causes many further problems New emitter materials and new ways of generating light are needed.

Charge Carrier Recombination

↑ ↓

25 % Singlet Excitons

↑

↓

75 % Triplet Excitons

recombination

↑ ↓

75 % 3P+P- 25 % 1P+P-

↑

↓

h+ e-

None emissive states in purely organic emitters

Light perception

The ‘average’ human eye has poor sensitivity to blue so it will be hardest to make a ‘bright’ blue-white lamp. The eye response shifts to the blue as intensity diminishes as rods dominate which are responsive to blue-green only

here, much more blue required, much less red (relative) so power efficiency must go down

with FIrpic we will not make 4000K and CRI>80, so its a non starter.

Why is the current blue not good enough?


Even a fluorescent tube looses about 40% efficiency to achieve 6200K because of poor blue eye response



LG Chem say they will mass produce 100x100 mm glass 80 lm/W 4th quarter 2013, 3000K, 75 lm output, 20,000 hrs note typical fluorescent luminaire is a 3500 lm package

Data from UDC

The lighting industry considers LT 90 as standard

Further, FIrpic is unstable to loose of the fluorine atoms, both on deposition and during operation. This is also true for any Ir based complex containing fluorine electron withdrawing groups to blue shift the emission. Also, for high SOC, the HOMO must contain a significant Ir d-orbital content. As the MLCT gap widens (giving bluer emission) it becomes easier to also populate the dd* states which very efficiently quench emission. This there are intrinsic limits on how blue an Ir complex can be pushed whilst retaining high PLQY



What then might be able to replace an Ir based phosphorescent blue emitter?

Usually people only consider emitter degradation but the hosts, especially high triplet energy host are a problem


Host degradation

9

Part II We did measurements of another sample which was evaporated at the rate between 30-70 angstroms per sec i.e. very high evaporation rate. Normalized delayed fluorescence and phosphorescence spectra at various temperatures are shown in figure 15. There are four features of phosphorescence at low temperature and two of them (497nm and 532nm) become more and more insignificant with increase of temperature as crest at 607 steadily peaks out in relation to the crest at 560.

400 500 600 700

0.0

0.4

0.8

Inte

nsity

, A.U

.

Wavelength, nm

d20msi10ms100K d20msi50ms112K d20msi70ms110K d10msi50ms120K d20msi10ms135K d20msi50ms142K d20ms i10ms d20msi70ms270K

14K

not annealed, evaporation rate 30 to 70 a per sec, measured just after evaporation

Figure 15 Then, after couple of weeks (about 9/10 of time sample was held in the compartment near the glove box i.e. it had some exposure to oxygen) the same sample was measure again. The results are shown in fig 16-18, red curves. As well this sample was heated in ambient atmosphere for two hours at 108 degrees Celsius and effect on photophysics of heating (annealing) is as shown in fig 16-18, green curves.

11

400 500 600 700

0.0

0.4

0.8

Inte

nsity

, A.U

.

Wavelength, nm

d200ms i50ms just after evaporation d200ms i300ms couple of weeks after evaporation d200ms i300ms after annealing in air

Figure 18 III Change of spectra without annealing Besides the above changes in phosphorescence spectra we were able to observe the shifted phosphorescence with the peaks only at 560 and 607 without annealing the film just after evaporation (not later than two hours). The results are shown in Figures 19-21. The spectra of this sample is very similar to the spectra at ~16K of heated films (Figure 17,18), at ~16K of films which spectra were recorded after considerable time of evaporation (Figure 12,13,17,18), as well to the spectra of unheated films at room temperature (Figure 15). We think this might have some relation to the conformation of film. The structure (conformation) which is formed on substrate during /after evaporation should depend on the rate material is cooled. This depends on temperature difference between the material and substrate, the latter normally is colder. The only reason for such performance we can think of, might be due to different substrate temperature (higher) than during the evaporations before thus forming similar structure as we are able to get after annealing or after keeping sample at room temperature for a long time.

CBP

1

CBP phosphorescence Structure:

Other info:

Sample Preparation: CBP was evaporated on quartz substrate which before was cleaned in concentrated nitric acid. Then it was rinsed for 6 minutes with soaped water, deionized water, acetone and isopropanol in ultrasonic bath. Then it was cured in UV-ozone atmosphere for 6 minutes. CBP was evaporated using Kurt Lesker OLED evaporator at the controllable rate. Total thickness of the samples was around 250nm. Spectra recording Gated luminescence measurements were made at the temperature of 12K or at room temperature or as indicated in the figures. Samples were excited with YAG laser light at the peak wavelength of 355nm, at 45 degree angle to the substrate plane. Energy was around 70 uJ, the diameter of the beam falling on the sample was about 1 cm. With the help of monochromator and other optics luminescence was collected with the sensitive CCD camera from Stanford Computer Optics. Part I First of all we measured fluorescence and phosphorescence spectra of evaporated 250nm thickness CBP films. The evaporation speed was about 2.5 angstroms per second, and spectra in figures 1-3 were recorded not later than 2 hours after evaporation.

OSA3251 DCBP, CBP Name: 4,4'-Bis(carbazol-9-yl)biphenyl;

sublimed Application: OLED-Hole Transport

Abs.Max: (nm) 318 nm (Methylene Chloride) Melting Point(C): 281 (DSC) Formula: C36H24N2 Appearance: Odorless pale yellow powder Abbreviation: DCBP, CBP

Use triplets because they live so long that they can sample a large volume and find the defects, especially at room temperature

Very commonly used host

DF

Ph

triplet energy reduces markedly over time whilst TF increases showing better hopping

Basic Poly(N-vinylcarbazole) triplet energy measurements

Mw PVK > 106, high purity In film ET ~ 2.9 eV The phosphorescence spectra does not shift with time implying that the triplet does not migrate. Triplet lifetime is also >10 s at 14 K.

Compare to our original pulse radiolysis energy transfer measurements made in benzene in solution, ET ~ 3 eV. Pina et al Chemical Physics letters 400, 2004, 441-445


400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

Intensity,A.U.

Wavelength, nm400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

Temperature dependence of phosphorescence

PHOTONIC MATERIALS CENTRE Monkman and Jankus

14 K

70 ms delay 355 nm excitation 150 ps pulse

ET=2.89 eV

400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

Intensity,A.U.

Wavelength, nm400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

14 K 44 K


ET=2.55 eV



400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

Intensity,A.U.

Wavelength, nm400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

400 450 500 550 600 650 700

0.0

0.2

0.4

0.6

0.8

1.0

14 K 44 K 85 K


ET=2.41 eV



14 K 44 K

14K - 44K

Monomeric ET=2.88 eV Dimeric ET=2.4 eV

Pina et al Chemical Physics letters 400, 2004, 441-445



Jankus and Monkman Adv Func Mat 21, 3350, 2011

dilute solution state phosphorescence

Is this a ground state or excited state species?

350 400 450 500

0.0

0.1

0.2

Opt

i

cald

ensity

W avelength, nm

0.04 mg/m l0.4 mg/m l4 mg/m l38 mg/m l

Clear concentration dependent ground state absorption.

Emission at 12us delay, 450 nm excitation at 14 K, identical to dimeric species emission NB, good vibronic structure

ET=2.41 eV

400 450 500 550 600 650

0.2

0.4

0.6

0.8

1.0

Intens

ity,A.U

.

W aveleng th, nm

450 nm excitation in toluene


0 10000 20000 30000

0.01

0.1

1

Intensity,A.U.

Time, ns

14 K44 K85 K100 K295 K

6650 ns and 22520 ns4980 ns and 8850 ns2640 ns2190 ns980 ns

Evidence for two ground state species

Excitation 450 nm

Two distinct lifetime components which are highly temperature dependent

Very clearly these species are monomer and dimer they are NOT ‘excimer’


What form has the dimer?

Benten et al, J.Phys.Chem.B 2007, 111, 10905

The closer the co-facial interaction between neighboring carbazole units the larger the shift in the phosphorescence. Further, the shift is so large that these dimers must have substantial charge transfer character.

½ co-facial dimer

Full co-facial dimer


What could replace Ir ?


We need new blue emitters that combine the stability of a fluorescence molecule with the an ability to ‘harvest’ triplet excitons very efficiently

Is this fundamentally impossible?

E-type delayed fluorescence to harvest triplet excitons


Ground State

Triplet State

Singlet State

ΔEST

Phosphorescence

Fluorescence

DelayedFluorescence

(TADF)

75%

25%

e- h+ : TADF = E-type DF !"##$%& $'( )"*+,,-

!"#$

$%&'

()*+),

+)

-.'*/.'()*+),+)

0%)+1(2+3%04+21312'

,56

57

89:

;9:!6

!" #"$%&'"# (' "))%*%"'+ ,-./ 0(+"1%(2 3(4%'&./!012345 1"$62+"# %' +7+(2 89 "))%*%"'*: 7) ;<=>

?'#"1 "2"*+1%*(2 "@*%+(+%7'5 56789*: ;<4 A($#"07'$+1(+"#>

B%'*" +3" 072"*62(1 #"$%&' %$ 6'2%0%+"#5 A" "@C"*+ +3(+)61+3"1 %0C174"0"'+ A%22 $61"2: 1"(2%D" =5.) >$? 5!@52"(#%'& +7 E<<= F9 "))%*%"'*:>

!A

!3

'% % '?B

Excitation

Excitation

%7@CDE:

<=5>?677:@

=5>?677:@

@A

!"#$%&'#(%)*+,-./012+345675+8#%9':6%(4;

Thermally activated delayed fluorescence

K. Goushi, K. Yoshida, K. Sato, and C. Adachi, "Organic light-emitting diodes employing efficient reverse intersystem crossing for triplet-to-singlet state conversion," Nature Photonics, vol. 6, pp. 253-258, Apr 2012.

Introduction

!"#$!%&!'()*+!,#$!%&!'(&*

!EST=2KEH :HOMO Energy EL : LUMO Energy K :Exchange Energy

・ !"#$$%!EST

! ! ! ! " "= ## 1 212

1(1) (2) (2) (1)L U L UK d dr

(LUMO)(HOMO) (LUMO)(HOMO)

&'()*+%,-$'%./,%("#$$%!EST%

!  Donor-Acceptor backbone !  X:Introduction of steric hindrance !  Rather high radiative decay rate

Donor Acceptor X

・0#,*'%!EST

!"#$!%&!'()*+!,#$!%&!'(&*

!EST=2KEH :HOMO Energy EL : LUMO Energy K :Exchange Energy

・ !"#$$%!EST

! ! ! ! " "= ## 1 212

1(1) (2) (2) (1)L U L UK d dr

(LUMO)(HOMO) (LUMO)(HOMO)

&'()*+%,-$'%./,%("#$$%!EST%

!  Donor-Acceptor backbone !  X:Introduction of steric hindrance !  Rather high radiative decay rate

Donor Acceptor X

・0#,*'%!EST

Basic theory of CT states K, the exchange energy dependents on the spatial separation of the orbitals and the overlap of the HOMO and LUMO in this case

The trade off, if HOMO and LUMO have zero overlap the radiative decay rate becomes very small so the CT state has low PLQY unless the lifetime is long i.e. IC and NR decay are even weaker.


New family of D-A-D molecules incorporating para and meta D-A coupling so we call systems containing 1 ‘linear’ and 2 ‘bent’ based on the electron transport unit dibenzothiophene-S,S-dioxide developed in Durham

ST energy splitting from 0.35 to 0.9 eV

Dias and Monkman et al., Adv. Mater. 2013, 25, 3707


TADF

1d

Dias et al., J.Phys.Chem.B. 110, 19329, 2006

K. Moss et al., J. Org. Chem. 2010, 75, 6771–6781 Dias et al., J. Phys. Chem. B 2006, 110, 19329-19339

HOMO LUMO hν

Intramolecular Charge Transfer States (ICT)

Small singlet-triplet energy splitting in CT states

Singlet state

Triplet state

1ICT 3ICT

ΔEST∼ 0.3 to 0.9 eV

D D

A

ΔE (1CT-3CT)<100 meV

TADF


carbazole donors dibenzothiophene-S,S-dioxde acceptor

Ground and excited state orbitals are nearly independent with very little orbital overlap. This is important.

But also gives low radiative decay rate

2b

150 ps Nd:YAG (70 mJ @ 355 nm)

200 ps FWHM gated iCCD HP pulse generator FWHM 10 ns

Rothe, Monkman Phys.Rev.B (2002), Rothe et al Chem.Phys. (2002), Sinha et al Phys.Rev.Lett (2003)

Delayed Fluorescence measurements


100 101 102 103 104 105 106 107 108 109

10-1110-1010-910-810-710-610-510-410-310-210-1100

Inte

nsity

, A.U

.

Time, ns

emission measured at 450 nm

prompt

delayed

400 450 500 550 600 650

0.0

0.4

0.8

Nor

mal

ized

inte

nsity

, A.U

.

Wavelength, nm

3.6 µJ 0.1 µJ

b)

0.1 1

105

106

107

108

Inte

nsity

, A.U

.

Laser pulse energy, µJ

DF - slope 2

PF - slope 1

PH-slope 1

a)

HOMO

LUMO

D D

A

hν

2c

0.4

0.8

1.2

0.0 0.1 0.2 0.3 0.4

0.4

0.8

1.2

0.0 0.1 0.2 0.3 0.4

ΔE /Δf= 0.91

ΔE (eV

)

1e ΔE /Δf= 1.01

1b

Δf

2e

ΔE /Δf= 1.55 ΔE /Δf= 2.47

2d

Enhanced CT character for bent materials (2’s) TADF


Lippert-Mataga plots

1d 2d

4a

5d RT Phosphorescence

400 450 500 550 6000.0

0.4

0.8

1.2

1.6

2.0

N o O2

with O2In

tens

ity x10

6 (cps

)

waveleng th (nm)

2d in z eonex

300 400 500 6000

40

80

120

160

200

Intens

ityx1

06 (cps

)

Waveleng th (nm)

T 290K T 270K T 250K T 200K T 150K T 100K T 80K

400 500 6000

20

40

60

80

290K 100K

Intens

ityx1

06 (cps

)

0 2 4 6 80

2

4

6

8

τ1: 1.60 ns

Cou

ntsx

103

T ime (ns )

hexaneλx : 363 nm λem: 410 nm

2d

IR F : 21 ps

0 1 2 3 4 5 60

2

4

6

8

τ: 3.08 ns

Cou

ntsx

103

T ime (ns )

hexaneλx :363nm λx :420nm

1d

IR F : 21 ps

The CT character is not responsible for the observation of strong Phosphorescence

ΔEST=0.46 eV

ΔEST=1.1 eV

TADF


Fluorescence lifetimes

Lone pairs are important Dias and Monkman et al., Adv. Mater. 2013, 25, 3707

4a

3.0 3.2 3.4 3.6 3.8 4.0 4.2100

101

102

DF in

tegralx1

06 (cou

nts)

103/T

E a=0.16±0.05 eV

4a

ΔEST=1.1 eV

1b

ΔEST=0.84 eV

3.0 3.5 4.0 4.5 5.0101

102

103

DF in

tegralx1

06 (cou

nts)

103/T (K -‐1)

E a=0.18±0.05 eV

1b

1ππ*

1CT*

3ππ*

3.26 eV

3.12 eV

2.28 eV

ΔEST=0.84 eV

3CT*

10 -‐10 10 -‐8 10 -‐6 10 -‐4 10 -‐210 -‐11

10 -‐9

10 -‐7

10 -‐5

10 -‐3

10 -‐1

101

T AD F (bi exponentia l) 378 µs (67% ), 1.8 ms (33% )

Norm. Inten

sity A

.U.

T ime (ns )

P rompt (7 .41 ns )

1b

TADF Properties of delayed fluorescence (DF)


barrier to viscous flow in ethanol is 0.15 eV

Pure Triplet Fusion

DF∝I2


400 500 600 700

Normalized

Intens

ity (a.u.)

waveleng th (nm)

S S -‐R T T AD F -‐300K T AD F -‐200K

1e

100 101 102 103100

101

102

103

Emission

Integralx1

06 (cou

nts)

P ower (µJ )

1eD F integ ra l vs exc ita tion dos eR T , e tO H

linea r fit with s lope fixed a t 2(coef. corr. 0 .995)

ΔES T= 0.63±0.06 eV

1e

2.5 3.0 3.5 4.0 4.5 5.0 5.5101

102

103

DF-‐in

tegralx1

06 (cou

nts)

103/T (K -‐1)

E a=0.26±0.02eV1e

240K

ΔES T=0.63 eV

1ππ*

1CT*

3ππ*

2.95 eV

2.74 eV

2.11 eV

ΔEST=0.63 eV

3CT*

ΔEST=0.63 eV

10 -‐1010 -‐910 -‐8 10 -‐7 10 -‐610 -‐5 10 -‐4 10 -‐310 -‐2 10 -‐110 -‐11

10 -‐9

10 -‐7

10 -‐5

10 -‐3

10 -‐1

101

T AD F (487 µs )

Norm. Inten

sity A

.U.

T ime (s )

P rompt (8.66 ns )

1e

D F /P F : 0 .0028

TADF Properties of delayed fluorescence (DF) Mixed Triplet Fusion TADF


We observe TADF with a ST energy gap of 0.63 eV!

2e

100 101 102 103

101

102

103

104

ΔES T= 0.48±0.06 eV

Emission

Integralx1

06 (cou

nts)

P ower (µJ )

2e

linea r fit with s lope fixed a t 1coef corre l: 0 .996

D F Integ ra l a s a func tionof exc ita tion powerR T , e tO H

400 500 600 7000.0

0.4

0.8

1.2 S teady-‐S ta te T AD F

Normalized

Intens

ity (a.u.)

Waveleng th (nm)

2e1ππ*

1CT*

3ππ*

3.15 eV

2.87 eV

2.39 eV

ΔEST=0.48 eV

3CT*

2.5 3.0 3.5 4.0 4.5 5.0 5.5101

102

103

104

DF-‐Integ

ralx10

6 (cou

nts)

103/T (K -‐1)

E a=0.28±0.02eV2eΔE

S T=0.48 eV

240K

ΔEST=0.48 eV

10 -‐1010 -‐910 -‐8 10 -‐7 10 -‐610 -‐5 10 -‐4 10 -‐310 -‐2 10 -‐110 -‐10

10 -‐8

10 -‐6

10 -‐4

10 -‐2

100

102

T AD F (818 µs )

Norm. Inten

sity A

.U.

T ime (s )

P rompt (3.07 ns )

D F /P F :0 .026

2e

TADF Properties of delayed fluorescence (DF) Dominant TADF


We observe strong TADF with an energy gap of 0.48 eV!

3 4 53 4 5101

102

103

104

3 4 5

2e

ΔE S T =0.48±0.05 eV

E a=0.28±0.02eV

240 K

DF-‐in

tegralx1

06 (cou

nts)

103/T (K -‐1)

E a=0.26±0.02eV

ΔE S T =0.63±0.06 eV1e

240 K

2d

E a =0.28±0.02 eVΔE S T =0.35±0.04 eV

240 K

2e 1e 2d

TADF Constant acceptor


Activation energy is independent of the ST gap and is not due to viscous flow


3 4 5101

102

103

104

3 4 5 3 4 5

DF-‐in

tegralx1

06 (cou

nts)

2d

E a =0.28±0.02 eV

ΔE S T =0.35±0.04 eV

240 K

3d

103/T (K -‐1)

E a=0.47±0.03eV

ΔE S T =0.57±0.05eV

260 K

4d

E a=0.83±0.02 eV

ΔE S T =0.87±0.04 eV

280 K

2d 3d 4d

TADF Constant donor


Activation energy is dependent on the donor structure ie. lone pairs


1e 2e

1ππ*

1CT*

3ππ*

2.95 eV

2.74 eV

2.11 eV

ΔEST=0.63 eV

3CT*

ΔEST=0.28 eV

1ππ*

1CT*

3ππ*

3.15 eV

2.87 eV

2.39 eV

ΔEST=0.48 eV

3CT*

ΔEST=0.28 eV

ΔE=0.35 eV

ΔE=0.2 eV

Large gap TF dominates

Small gap TADF dominates

We require another energy level to explain the temperature dependent behaviour of the DF

TADF



TADF We introduce the “little known” nπ* orbital


400 500 600 7000

2

4

6

8

200 220 240 260 280 300 320 3400

200

400

600

800

1000

200K

Intens

ityx1

06 (cps

)

wave leng th (nm)

2d

330K

k -‐T

is c= (5±1)x109 s -‐1

ΦT=0.05±0.01

ΔETADF

=0.38±0.05 eV

Emission

Integralx1

06 (cou

nts)

T empera ture (K )

Triplet yield ∼ 5% DF/PF ratio∼ 5%

2d

100 101 102 103100

101

102

103

linea r fit with s lope fixed a t 1.coef. corre l.: 0.995

2d

Emission

Integral (co

unts)

power (µJ )400 500 600 700

Normalized

Intens

ity (a.u,)

waveleng th (nm)

T AD F s teady-‐s ta te emis s ion

2d

10 -‐10 10 -‐9 10 -‐8 10 -‐7 10 -‐6 10 -‐5 10 -‐4 10 -‐3 10 -‐210 -‐10

10 -‐8

10 -‐6

10 -‐4

10 -‐2

100

102

T AD F (230 µs )

Norm In

tens

ity A

.U.

T ime (s )

2d

P rompt (4.7 ns )

DF /P F : 0.048

TADF How efficient is TADF even with a gap of 0.36 eV


100% efficient

TADF

Dias and Monkman et al., Adv. Mater. 2013, 25, 3707 M. N. Berberan-Santos, J. M. M. Garcia, JACS. 1996, 118, 9391

TADF in solid state

Critical role of host triplet energy


In TAPC no 1,3CT host quenching TADF PLQY 0.5 In TPBi 2d TF dominates with a small TADF component

Jankus et al,. PRL in press

350 400 450 500 550 600 650 700-0.10.00.10.20.30.40.50.60.70.80.91.01.1

Excitation pulse energy in microjouls

Nor

mal

ized

inte

nsity

, A.U

.

Wavelength, nm

0.130mkJ 1.061mkJ 2.32mkJ 7.36mkJ 7.94mkJ

2d in TPBi

350 400 450 500 550 600 650 700-0.10.00.10.20.30.40.50.60.70.80.91.01.1 Excitation pulse

energy in microjouls

Nor

mal

ized

inte

nsity

, A.U

.

Wavelength, nm

0.063mkJ 0.205mkJ 1.09mkJ 4.16mkJ

2d in TAPC

400 450 500 550 600 650 7000.0

0.2

0.4

0.6

0.8

1.0

Nor

mal

ized

inte

nsity

, A.U

.

Wavelength, nm

0.170mkJ 0.776mkJ 1.2mkJ 4.75mkJ 9.1mkJ 15.6 50mkJ

Excitation pulseenergy in microjouls

Neat 2d Film

10-2 10-1 100 101 102

10-610-510-410-310-210-1100101102

neat 2d red edge TADF dominated

2d in TPBi red edge- smaller TF influence

Inte

grat

ed in

tens

ity, A

.U.

Pulse fluence, µJ

1.04

1.87

1.171.29

1.05

1.09

2d in TAPC - TADF dominated 2d in TPBi blue edge- strong TF influence

neat 2d blue edge TF dominated

400 500 600 7000.0

0.4

0.8

Nor

mal

ized

inte

nsity

, A. U

.

Wavelength, nm

neat in TPBi in TAPC

PLQY14 %8%50%

2d

TADF in solid state Critical role of host triplet energy


In TAPC no 1,3CT host quenching TADF PLQY 0.5 So TAPC is special as all excitations remain trapped on the 2d thus we find much longer lifetimes of excited states 433 ns for the prompt emission 25 us for TADF and 165 us for phosphorescence We also find that the first and last decay components decrease with increasing temperature whereas the middle one increases with increasing temperature

100 101 102 103 104 105 106

10-8

10-6

10-4

10-2

100

DF/PF~0.082DF/PF~1.8 DF/PF~0.15DF/PF=0.048

Nor

mal

ized

inte

nsity

, A.U

.

Time, ns

2d neat film 7pc 2d in TAPC 7pc 2d in TPBi 2d in solution

DF/PF ratio of 1.8 the triplet yield is 64% and the fluorescence yield is 50% This signifies long CT lifetimes yielding high triplet CT production, energy cycling between the iso energetic CT states and thermal activation from the ππ* triplet to the 3CT

TADF devices


How do OLEDs performed based on TADF emitters?

TADF devices


LETTERdoi:10.1038/nature11687

Highly efficient organic light-emitting diodes fromdelayed fluorescenceHiroki Uoyama1, Kenichi Goushi1,2, Katsuyuki Shizu1, Hiroko Nomura1 & Chihaya Adachi1,2

The inherent flexibility afforded by molecular design has accele-rated the development of a wide variety of organic semiconductorsover the past two decades. In particular, great advances have beenmade in the development of materials for organic light-emittingdiodes (OLEDs), from early devices based on fluorescent mole-cules1 to those using phosphorescent molecules2,3. In OLEDs, elec-trically injected charge carriers recombine to form singlet andtriplet excitons in a 1:3 ratio1; the use of phosphorescent metal–organic complexes exploits the normally non-radiative triplet exci-tons and so enhances the overall electroluminescence efficiency2,3.Here we report a class of metal-free organic electroluminescentmolecules in which the energy gap between the singlet and tripletexcited states is minimized by design4, thereby promoting highlyefficient spin up-conversion from non-radiative triplet states toradiative singlet states whilemaintaining high radiative decay rates,of more than 106 decays per second. In other words, these mole-cules harness both singlet and triplet excitons for light emissionthrough fluorescence decay channels, leading to an intrinsic fluor-escence efficiency in excess of 90 per cent and a very high externalelectroluminescence efficiency, of more than 19 per cent, which iscomparable to that achieved in high-efficiency phosphorescence-based OLEDs3.The recombination of holes and electrons can produce light, in a

process referred to as electroluminescence. Electroluminescence inorganic materials was first discovered in 1953 using a cellulose filmdoped with acridine orange5, and was developed in 1963 using ananthracene single crystal connected to high-field carrier injection elec-trodes1. Electrical charge carriers of both polarities were injected intothe organic layers, and the subsequent carrier transport and recom-bination produced blue electroluminescence originating from singletexcitons; that is, fluorescence. According to spin statistics, carrierrecombination is expected to produce singlet and triplet excitons ina 1:3 ratio6,7, and this ratio has been examined for many molecularsystems8–12. The singlet excitons produced decay rapidly, yieldingprompt electroluminescence (fluorescence). Two triplet excitons cancombine to form a singlet exciton through triplet–triplet annihilation,which results in delayed electroluminescence (delayed fluorescence).Direct radiative decay of triplet excitons results in phosphorescence,but usually occurs only at very low temperatures in conventionalorganic aromatic compounds. The first demonstration of phospho-rescent electroluminescence using ketocoumarin derivatives in 199013.However, the very faint electroluminescence was observed only at77K, and with difficulty, and was assumed to be virtually useless evenif included in rare-earth complexes, which should also involve bothsinglet and triplet excitons in electrical excitation14. In 1999, efficientelectrophosphorescence was first demonstrated using iridium phenyl-pyridine complexes that achieve an efficient radiative decay rate of,106 s21 by taking advantage of the strong spin–orbit coupling ofiridium2. An internal electroluminescence efficiency of almost 100%was achieved3, providing convincing evidence that OLED technologycan be useful for display and lighting applications.

In the work reported here, we used a novel pathway to attain thegreatest possible electroluminescence efficiency from simple aromaticcompounds that exhibit efficient thermally activated delayed fluo-rescence (TADF) with high photoluminescence efficiency. Figure 1ashows the energy diagramof a conventional organicmolecule, depictingsinglet (S1) and triplet (T1) excited states and a ground state (S0). It waspreviously assumed that the S1 level was considerably higher in energythan the T1 level, by 0.5–1.0 eV, because of the electron exchangeenergy between these levels. However, we found that careful designof organic molecules can lead to a small energy gap (DEST) between S1and T1 levels4,15. Correspondingly, a molecule with efficient TADFrequires a very small DEST between its S1 and T1 excited states, whichenhances T1R S1 reverse intersystem crossing (ISC). Such excitedstates are attainable by intramolecular charge transfer within systemscontaining spatially separated donor and acceptor moieties4. The cri-tical point of this molecular design is the combination of a smallDEST,of= 100meV, with a reasonable radiative decay rate, of.106 s21, toovercome competitive non-radiative decay pathways, leading to highlyluminescent TADF materials. Because these two properties conflictwith each other, the overlap of the highest occupied molecular orbitaland the lowest unoccupied molecular orbital needs to be carefully

1Center for Organic Photonics and Electronics Research, Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan. 2International Institute for Carbon Neutral Energy Research (WPI-I2CNER),Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan.

a

b

Fluorescence

Phosphorescence

Electrical excitationS1

S0

25%

75%

T1~0.5–1.0 eV

TADF

e h

4CzPN: R = carbazolyl2CzPN: R = H

4CzIPN 4CzTPN: R = H4CzTPN-Me: R = Me4CzTPN-Ph: R = Ph

NC CN

RR

NN

N

NN

N

NC CN

R

R

R

R

R

R

R

R

CN

CN

N

N N

N

Figure 1 | Energy diagram and molecular structures of CDCBs. a, Energydiagramof a conventional organicmolecule. b, Molecular structures of CDCBs.Me, methyl; Ph, phenyl.

2 3 4 | N A T U R E | V O L 4 9 2 | 1 3 D E C E M B E R 2 0 1 2

Macmillan Publishers Limited. All rights reserved©2012

that kRISC is higher than the non-radiative rate constant of the T1 state.It should be noted that for temperatures less than 200K the decaycurves do not agree with the double-exponential decay model, butare described well by a multi-exponential decay model. This can beexplained by the widenedDEST distribution caused by inhomogeneousmolecular environments at lower temperatures.We then evaluated the performance of OLEDs containing the

CDCB derivatives 4CzIPN (green emission), 4CzTPN-Ph (orangeemission) and 2CzPN (sky-blue emission) as emitters. Figure 5 showsthe external electroluminescence quantum efficiency of OLEDs con-taining the CDCB derivatives. To achieve high electroluminescenceefficiency in these OLEDs, the T1 state of the CDCB derivatives mustbe confined using a host material with a higher triplet energy level.Therefore, we used CBP as a host material in the green and orangeOLEDs and used 2,8-bis(diphenylphosphoryl)dibenzo[b,d]thiophene(PPT) as a host material in the sky-blue OLED. The structures of theOLEDs were composed of multiple layers of indium tin oxide (ITO),4,4-bis[N-(1-naphthyl)-N-phenylamino]-biphenyl (a-NPD, 35 nm),56 1wt% 4CzIPN or 56 1wt% 4CzTPN-Ph:CBP (15 nm), TPBi(65 nm), LiF (0.8 nm), Al (70 nm) and ITO (100 nm), a-NPD(40 nm), 1,3-bis(9-carbazolyl)benzene (mCP, 10 nm), 56 1wt%2CzPN:PPT (20nm), PPT (40nm), LiF and Al. For the green OLED, avery high external electroluminescence quantum efficiency, of19.36 1.5%, was achieved, which is equivalent to an internal electro-luminescence quantum efficiency of 64.3–96.5% assuming a light out-coupling efficiency of 20–30% (refs 25, 26). The orange and sky-blueOLEDs had external electroluminescence quantum efficiencies of

11.26 1%and 8.06 1%, respectively, which are also higher than thoseof conventional fluorescence-based OLEDs.Finally, we consider themechanism that drives such efficient reverse

ISCwithout heavymetals. It is generally accepted that the introductionof spin–orbit coupling provided by heavy atoms is required for bothISC and reverse ISC to be efficient. Thus, metal complexes containingheavy metals and aromatic compounds with halogens and carbonylscan promote efficient spin conversion. However, our novel moleculardesign produces highly efficient spin conversion without needingsuch atoms. This is because the first-order mixing coefficient betweensinglet and triplet states (l) is inversely proportional to DEST (ref. 27):

l!HSO

DEST!2"

Here HSO is the spin–orbit interaction. It follows from equation (2)that heavy atoms are not required to achieve efficient spin conversionwhen a molecule possesses a small DEST and HSO is not vanishinglysmall. This broadens the scope for the molecular design of TADFmaterials. A detailed strategy for large l will be clarified on the basisof correlated quantum-chemical calculations28.

METHODS SUMMARYSynthesis of carbazolyl dicyanobenzene. The synthesis of CDCB derivatives wasperformed according to the synthetic method reported previously for pyrrole andfluorobenzene derivatives29,30. CDCBswere synthesized by reaction of a carbazolylanion with a fluorinated dicyanobenzene at room temperature (300K) for 10 hunder a nitrogen atmosphere. CDCBs were purified by column chromatographyon silica gel, by reprecipitation or both. CDCBs were further purified by sublima-tion before photoluminescence and electroluminescence spectra were measured.Photoluminescence measurements. Organic films for optical measurementswere fabricated by thermal evaporation onto clean quartz and silicon substratesunder high vacuum (,73 1024 Pa). The photoluminescence spectra of thesefilms were recorded with a spectrofluorometer (FluoroMax-4, Horiba JobinYvon), and the photoluminescence quantum efficiencies were measured usingan absolute photoluminescence quantum yield measurement system (C9920-02,Hamamatsu Photonics). The transient photoluminescence characteristics weremeasured (in terms of photon number) under vacuum using a streak camera(C4334, Hamamatsu Photonics). A nitrogen gas laser with a wavelength of337nm and a pulse width of approximately 500 ps (MNL200, Lasertechnik) wasused as an excitation source. Low-temperature measurements were made using acryostat (CRT-006-2000, Iwatani Industrial Gases). InGa alloy was applied as anadhesive to ensure good thermal conductivity between the silicon substrate andsample holder.

10–3 10–2 10–1 100 101 102

Current density (mA cm–2)

4CzIPN4CzTPN-Ph2CzPN

101

100

10–1

10–2

Exte

rnal

ele

ctro

lum

ines

cenc

e qu

antu

m e

f!ci

ency

(%)

400 500 600 700Wavelength (nm)

1

0

Nor

mal

ized

ele

ctro

lum

ines

cenc

ein

tens

ity (a

.u.)

Figure 5 | Performance of OLEDs containing CDCB derivatives. Externalelectroluminescence quantum efficiency as a function of current density forOLEDs containing 4CzIPN (green circles; error within 1.5%), 4CzTPN-Ph (redtriangles; error within 1.0%) and 2CzPN (blue triangles; error within 1.0%) asemitters. Inset, electroluminescence spectra of the same OLEDs (colouredaccordingly) at a current density of 10mAcm22.

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0Pho

tolu

min

esce

nce

quan

tum

ef!

cien

cy

0 50 100 150 200 250 300Temperature (K)

CombinedPromptDelayed

0 10 20 30 40Time (!s)

300 K200 K100 K

100

10–1

10–2

10–3

Nor

mal

ized

pho

tolu

min

esce

nce

inte

nsity

PromptDelayed

400 500 600

1

0

3.5 4.0 4.5 5.010–3/T (K–1)

14.2

14.0

13.8

13.6

13.4

13.2

13.0

12.8

12.6

12.4

In(k

RIS

C)

a b

c d

300 K

"EST = 83 meV

Figure 4 | Temperature dependence of photoluminescence characteristicsof a 56 1wt% 4CzIPN:CBP film. a, Photoluminescence decay curves of a6wt% 4CzIPN:CBP film at 300K (black line), 200K (red line) and 100K (blueline). The photoluminescence decay curves show integrated 4CzIPN emission.The excitation wavelength of the films was 337nm. b, Photoluminescencespectrum resolved into prompt and delayed components. c, Temperaturedependence of photoluminescence quantum efficiencies (errors are within 2%)for combined (prompt plus delayed; black squares), prompt (red circles) anddelayed (blue triangles) components of 4CzIPN emission for 56 1wt%4CzIPN:CBP film. The straight lines are guides for the eye. d, Arrhenius plot ofthe reverse ISC rate from the triplet state to the singlet state of 4CzIPNwith kISCset to 43 107 s21. The straight line (least-squares regression) is used todetermine the activation energy. The ln(kRISC) errors are within 0.2.

LETTER RESEARCH

1 3 D E C E M B E R 2 0 1 2 | V O L 4 9 2 | N A T U R E | 2 3 7

Macmillan Publishers Limited. All rights reserved©2012

For a green device using 4CzTPN devices having EQE of 19% have been reported. This equates to an internal QE approaching 85%, as good as phosphorescent OLED all from TADF

Adachi et al, Nature, 492, 237, 2012

Cost no object

Designed by Kardorff

Deutsche Bank Berlin

OLED Lighting, a bright future

Thank you

http://www.dur.ac.uk/OEM.group [email protected]

The TOPLESS lamp.

OEM Research Group

Blue is always a problem for any system Phosphorescent emitters are difficult to push blue enough High triplet hosts suffer major degradation problem Triplet Fusion not good enough, 62.5% maximum not attainable TADF promising, true 100% triplet harvesting ICT again has host problem DA exciplex being a self ambipolar host is really exciting

oleds,the dawn of ultra efficient lighting

Technology