Solid-State Light-Emitting Electrochemical Cells Based on Cationic Transition Metal Complexes for White Light Generation

(1)

Based on Cationic Transition Metal Complexes

for White Light Generation

Hai-Ching Su, Ken-Tsung Wong, and Chung-Chih Wu

Contents

1 Introduction . . . 106

1.1 Features of Light-Emitting Electrochemical Cells . . . 106

1.2 Organization of This Chapter . . . 106

2 Review of Light-Emitting Electrochemical Cells Based on Cationic Transition Metal Complexes . . . 107

2.1 Increasing Device Efficiency . . . 107

2.2 Color Tuning . . . 118

2.3 Lengthening Device Lifetime . . . 120

2.4 Shortening Turn-On Time . . . 121

3 White Light-Emitting Electrochemical Cells Based on Cationic Transition Metal Complexes . . . 127

3.1 Photophysical and Electrochemical Properties of Novel Blue-Green and Red-Emitting Cationic Iridium Complexes . . . 127

3.2 Electroluminescent Properties of White Light-Emitting Electrochemical Cells Based on Host–Guest Cationic Iridium Complexes . . . 131

4 Outlook . . . 133

References . . . 134

Abstract Solid-state light-emitting electrochemical cells (LECs) based on cationic transition metal complexes (CTMCs) exhibit several advantages over conventional light-emitting diodes such as simple fabrication processes, low-voltage operation,

H.-C. Su (*)

Institute of Lighting and Energy Photonics, National Chiao Tung University, No. 301, Gaofa 3rd Road, Guiren Township, Tainan County 711, Taiwan, ROC

e-mail: [email protected] K.-T. Wong

Department of Chemistry, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan, ROC

C.-C. Wu (*)

Department of Electrical Engineering, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan, ROC

e-mail: [email protected]

V.W.W. Yam (ed.),WOLEDs and Organic Photovoltaics, Green Energy and Technology, DOI 10.1007/978-3-642-14935-1_4,# Springer-Verlag Berlin Heidelberg 2010

(2)

and high power efficiency. Hence, white CTMC-based LECs may be competitive for lighting applications. In this chapter, we review previous important works on CTMC-based LECs, such as increasing device efficiency, color tuning, lengthening device lifetime, and shortening turn-on time. Our demonstration of white CTMC-based LECs by using the host–guest strategy is then described.

1 Introduction

1.1 Features of Light-Emitting Electrochemical Cells

White organic light-emitting diodes (OLEDs) based on polymers and small-molecule materials have attracted much attention because of their potential applications in flat-panel displays and solid-state lighting [1–5]. Compared with conventional white OLEDs [6,7], solid-state white light-emitting electrochemical cells (LECs) [8–10] possess several promising advantages. LECs generally require only a single emissive layer, which can be easily processed from solutions, and can conveniently use air-stable electrodes. The emissive layer of LECs contains mobile ions, which can drift toward electrodes under an applied bias. The spatially separated ions induce doping (oxidation and reduction) of the emissive materials near the electro-des, that is, p-type doping near the anode and n-type doping near the cathode [8–10]. The doped regions induce ohmic contacts with the electrodes and conse-quently facilitate the injection of both holes and electrons, which recombines at the junction between p- and n-type regions. As a result, a single-layered LEC device can be operated at very low voltages (close toEg/e, where Egis the energy gap of

the emissive material ande is elementary charge) with balanced carrier injection, giving high power efficiencies. Furthermore, air-stable metals, for example, gold and silver, can be used since carrier injection in LECs is relatively insensitive to work functions of electrodes. Therefore, the power-efficient properties and easy fabrication processes make LECs competitive in solid-state lighting technologies.

In the past few years, cationic transition metal complexes (CTMCs) have also been increasingly studied for solid-state LECs because of the following several advantages over conventional LECs or polymer LECs (a) ion-conducting material is not needed since these metal complexes are intrinsically ionic and (b) higher electroluminescent (EL) efficiencies could be achieved due to the phosphorescent nature of the transition metal complexes. The development of solid-state LECs based on CTMCs thus will be described and discussed in this chapter.

1.2 Organization of This Chapter

This chapter is organized as follows. Section2reviews advances of LECs based on CTMCs, including topics of device efficiency (Sect.2.1), colors (Sect.2.2), device

(3)

lifetimes (Sect.2.3), and turn-on times (Sect. 2.4). Section 3is devoted to white LECs based on CTMCs. In Sect.3.1, photophysical and electrochemical properties of newly developed blue-green and red emitting cationic iridium complexes are discussed. EL properties of white LECs based on host–guest cationic iridium complexes are then discussed in Sect. 3.2. Finally, we give an outlook about white LECs based on CTMCs.

2 Review of Light-Emitting Electrochemical Cells Based

on Cationic Transition Metal Complexes

The first solid-state LEC was demonstrated with a polymer blend sandwiched between two electrodes [8]. The polymer blend was composed of an emissive conjugated polymer, a lithium salt (lithium trifluoromethanesulfonate), and an ion-conducting polymer (polyethylene oxide). The salt provides mobile ions and the ion-conducting polymer prevents this blend film from phase separation that may be induced by polarity discrepancy between the conjugated polymer and the salt. More recently, CTMCs have also been used in LECs, which show several advan-tages over conventional polymer LECs. In such devices, no ion-conducting material is needed since these metal complexes are intrinsically ionic. They generally show good thermal stability and charge-transport properties. Furthermore, high EL effi-ciencies are expected due to the phosphorescent nature of these metal complexes.

2.1 Increasing Device Efficiency

The first solid-state LEC based on transition metal complexes was reported in 1996 [11], where a ruthenium poly-pyridyl complex was utilized as the emissive material. Since then, many efforts have been made to improve performances of the LECs. In 1999, LECs based on low-molecular-weight ruthenium complexes were reported to have external quantum efficiency (EQE), which is defined as photons/electrons, of 1% [12]. In 2000, LECs using [Ru(bpy)3](ClO4)2(where bpy is 2,20-bipyridine) with

the gallium–indium eutectic electrode showed an EQE up to 1.8% [13]. Later, [Ru(bpy)3](PF6)2blended with poly(methyl methacrylate) (PMMA) was reported

to improve the film quality and increase the EQE to 3% [14]. In 2002, a single-crystal LEC was made by repeatedly filling nearly saturated solution of [Ru(bpy)3]

(ClO4)2between two ITO (indium tin oxide) slides, followed by evaporating the

solvent [15]. Such devices possessed low turn-on voltages and exhibited an EQE of 3.4%. Further improvement of performances was achieved by reducing self-quench-ing of excited states in [Ru(bpy)3]2+ with adding alkyl substituents on the bpy

ligands [16], raising the EQE to 4.8% under the DC bias and to 5.5% under the pulsed driving. Yet a further improvement of the [Ru(bpy)3](ClO4)2 device was

(4)

achieved by forming a heterostructure device, thus moving the emission zone away from the electrode and giving an efficiency of 6.4% [17].

More efficient cationic iridium complexes have also been used in LECs. In 2004, LECs based on the yellow-emitting (560 nm) cationic iridium complex [Ir(ppy)2(dtb-bpy)]PF6, where ppy is phenylpyridine and dtb-bpy is 4,40

-di-tert-butyl-2,20-bipyridine, were reported, exhibiting efficiencies of 5% and 10 lm W1 [18, 19]. On the one hand, replacing the ppy ligands by F-mppy ones, where F-mppy is 2-(40-fluorophenyl)-5-methylpyridine, led to green emission (531 nm) and an EL efficiency of 1.8% [19,20]. On the other hand, employing the dF(CF3)ppy

ligands, where dF(CF3)ppy is 2-(2,4-difluorophenyl)-5-trifluoromethylpyridine,

increased the energy gap of the cationic iridium complexes and shifted the EL emission to blue-green (500 nm, with an EQE of 0.75%) [21]. Another series of cationic phenylpyrazole-based iridium complexes were also recently reported to give blue (492 nm), green (542 nm), and red (635 nm) emission [22]. Blue, green, and red devices made of these complexes on poly(3,4-ethylenedioxythiophene):poly (styrene sulfonate) (PEDOT:PSS) coated ITO showed EQEs of 4.7%, 6.9%, and 7.4%, respectively.

Most LEC devices described above were made in a single-layered neat-film structure. In a neat film of an emissive material, interactions between closely packed molecules usually lead to quenching of excited states, detrimental to EQEs of devices. Thus, in addition to tuning of emission colors, the capability of the ligands to provide steric hindrance for suppressing self-quenching must also be carefully considered in designing ligands for emissive cationic metal complexes. In our previous studies of oligofluorenes, we had found that the introduction of aryl substitutions onto the tetrahedral C9 of oligofluorenes can provide effective hin-drance to suppress interchromophore packing and self-quenching yet without perturbing energy gaps of the molecules [23–27], making the photoluminescent quantum yields (PLQYs) of oligo(9,9-diarylfluorene)s in neat films rather close to those in dilute solutions. Thus, we introduced 4,5-diaza-9,90-spirobifluorene (SB) [28] as a steric and bulky auxiliary ligand of cationic iridium complexes, [Ir(ppy)2(SB)]PF6 (1) and [Ir(dFppy)2(SB)]PF6 (2) (Fig. 1), and investigate the

influence of the SB ligand on reducing the self-quenching [29].

Spin-coated neat films of 1 and 2 exhibit PL spectra similar to those observed for their acetonitrile (MeCN) solutions (Fig.2). However, spin-coated neat films of 1 and 2 show higher PLQYs (0.316 for 1, 0.310 for 2) and longer excited-state lifetimes (0.60ms for 1, 0.59 ms for 2) than their MeCN solutions (Table1). Since MeCN is a strongly polar solvent and the emitters are ionic, the photophysical properties of complexes 1 and 2 perhaps are significantly perturbed by strong solute–solvent interaction, rendering the observed PLQYs and lifetimes of 1 and 2 in MeCN less intrinsic. This indeed can be verified by measuring photophysical properties of 1 and 2 in a less polar solvent (yet still with enough solubility for spectroscopic measurements), such as dichloromethane (DCM). In DCM (5 105M), both 1 and 2 exhibit longer excited-state lifetimes (0.79ms for 1, 0.42ms for 2) and higher PLQYs (0.467 for 1, 0.364 for 2) than in MeCN (Table1).

(5)

Thus, to better characterize the intrinsic photophysical properties of complexes 1 and 2 at the room temperature, complexes 1 and 2 were dispersed (with 1.5 mol%) in a large-gap thin-film host ofm-bis(N-carbazolyl)benzene (mCP), which is rather nonpolar and has a large triplet energy. The measured PLQYs and excited-state lifetimes of the dispersed mCP films are (0.667, 0.81ms) for 1 and (0.421, 0.74 ms) for 2 (Table1). Thus 1 and 2 dilutely dispersed in mCP exhibit higher PLQYs and longer excited-state lifetimes than those in neat films. Shorter lifetimes in neat films indicate that interaction between closely packed molecules provides additional deactivation pathways, shortening excited-state lifetimes, and lowering the PLQYs. However, PLQYs of complexes 1 and 2 in neat films still retain ~50% and ~75% of those in mCP blend films. These are indeed rather high retaining percentages for PLQYs in neat films when compared with other phosphorescent molecules. For instance, [Ir(ppy)3] [1.5 mol% dispersed in 4,40-bis(carbazol-9-yl)

200 300 400 500 600 700 800 0 1 2 3 4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Absorbance (a.u.) Wavelength (nm) PL, Abs, Solution PL, Abs, Neat film EL Intensity (a.u.) 200 300 400 500 600 700 800 0 1 2 3 4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Absorbance (a.u.) Wavelength (nm) PL, Abs, Solution PL, Abs, Neat film EL

Intensit

y

(a.u.)

a b

Fig. 2 Absorption and PL spectra in acetonitrile solutions and in neat films of (a) complex 1 and (b) complex 2. Also shown in (a) and (b) are EL spectra of complex 1 and complex 2, respectively. Reproduced with permission from [29]. Copyright 2007, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim N N Ir N N F F F F N N Ir N N + PF6– PF6– + 1 2

(6)

biphenyl (CBP)] and bis[(4,6-difluorophenyl)pyridinato-N,C2](picolinato)iridium (III) [FIrpic] (1.4 mol% dispersed in mCP) films had been reported to have very high PLQYs of 97 2% and 99 1%, respectively [30], while PLQYs of [Ir(ppy)3)] and [FIrpic] neat films are only ~3% and ~15%, respectively. On the

one hand, severe self-quenching in [Ir(ppy)3] and [FIrpic] implies that the ppy,

dFppy, and picolinic acid ligands cannot provide enough hindrance against inter-molecular interactions in neat films. On the other hand, the high retaining percen-tages in neat-film PLQYs of complex 1 (with two ppy ligands and one SB ligand) and complex 2 (with two dFppy ligands and one SB ligand) compared with those in dispersed films indicate that SB ligands in these compounds provide effective steric hindrance and greatly reduce self-quenching. It is noted that self-quenching in neat films of [FIrpic] is not as severe as that in neat films of [Ir(ppy)3]. It is likely that the

fluoro substituents on the ppy ligands somewhat hinder the intermolecular interac-tions [30]. A similar effect is also observed here for complex 2, which exhibits a higher PLQY retaining percentage (74%) than complex 1.

In the operation of LEC devices, when a constant bias voltage is applied, a delayed EL response that is associated with the time needed for counterions in the LECs to redistribute under a bias is typically observed. For the cases of neat films of complexes 1 and 2, the redistribution of the anions (PF6) leads to the formation of

a region of Ir(IV)/Ir(III) complexes (p-type) near the anode and a region of Ir(III)/Ir

Table 1 Summary of physical properties of complexes 1 and 2

Complex lmax, PL(nm)a PLQY, lifetime (ms)b Eox1=2(V)c Ered1=2(V)d DE1/2(V)e Solutionf Neat film Solution Film

1 605 593 0.226, 0.33f 0.316, 0.60g 1.35 1.33 2.64 0.467, 0.79h 0.667, 0.81i –, 4.31j 0.381, 0.72k –, 3.27l 2 535 535 0.278, 0.39f 0.310, 0.59g 1.70 1.26 2.92 0.364, 0.42h 0.421, 0.74i –, 4.49j 0.329, 0.55k –, 4.29l a_{PL peak wavelength}

b_{Photoluminescence quantum yields and the excited-state lifetimes} c_{Oxidation potential}

d_{Reduction potential} e_{The electrochemical gap}_DE

1/2is the difference betweenE1ox=2andE1red=2corrected by potentials of the ferrocenium/ferrocene redox couple

f_{Measured in acetonitrile (5}₁₀5_{M) at room temperature} g_{Neat films}

h_{Measured in dichloromethane (5}₁₀5_{M) at room temperature} i_{1 and 2 were dispersed (1.5 mol%) in mCP films}

j_{Measured in acetonitrile (5}₁₀5_{M) at 77 K} k_{Films with 0.75 mol [BMIM][PF}

6] per mole of 1 and 2 l_{Measured in dichloromethane (5}₁₀5_{M) at 77 K}

(7)

(II) complexes (n-type) near the cathode [31]. With the formation of p- and n-regions near the electrodes, carrier injection is enhanced, leading to a gradually increasing device current and emission intensity. Devices based on neat films of complexes 1 and 2 (with the structure of glass substrate/ITO/complex 1 or 2 (100 nm)/ Ag) exhibited very long response times. Very slow device response (e.g., tens of hours to reach the maximum brightness) had also been observed before for other LEC materials (e.g., [Ru(bpy)3](PF6)2 derivatives with esterified ligands) [12].

Plausibly, bulky side groups on molecules impede the migration of ions. Thus to accelerate formation of the p- and n-doped regions in the emissive layer, 0.75 mol ionic liquid 1-butyl-3-methylimidazolium hexafluorophosphate [BMIM][PF6] per

mole of complex 1 or 2 was added to provide additional anions (PF6) [32].

It had been reported that in polymer LECs, incorporation of polar salts into conjugated polymer films might induce aggregates or phase separation due to discrepancy in polarity [33–35]. Thus, to study the effect of the [BMIM][PF6]

addition on thin-film morphologies of complexes 1 and 2, atomic force microscopy (AFM) of thin films was performed. As shown in Fig.3, the AFM micrographs for films of complexes 1 and 2 with and without [BMIM][PF6] coated on ITO glass

substrates show no significant differences and all give similar root-mean-square (RMS) roughness of ~1 nm. At this concentration of [BMIM][PF6] in complex 1 or 2

(0.75:1, molar ratio), no particular features of aggregation or phase separation were

a b c d 1000 10 800 600 400 200 0 200 400 600 800 1000 0 0 5 10 15 nm nm nm nm 1000 10 800 600 400 200 0 200 400 600 800 1000 0 0 5 10 15 nm nm nm nm 1000 10 800 600 400 200 0 0 200 400 600 8001000 0 5 10 15 nm nm nm nm 1000 10 800 600 400 200 0 0 200 400 600 800 1000 0 5 10 nm nm nm nm

Fig. 3 3D AFM micrographs of (a) neat film of complex 1, (b) blend film of [BMIM][PF6] and complex 1 (0.75:1, molar ratio), (c) neat film of complex 2, and (d) blend film of [BMIM][PF6] and complex 2 (0.75:1, molar ratio). Reproduced with permission from [29]. Copyright 2007, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

(8)

observed, and uniform spin-coated thin films could be routinely obtained. Such characteristics may be associated with the ionic nature of complexes 1 and 2, which may make them more compatible with the added salts. Further, to examine the effects of the [BMIM][PF6] addition on photophysical properties of complexes

1 and 2, PLQYs and excited-state lifetimes of blend films were also measured and are shown in Table1. In general, films of both 1 and 2 containing [BMIM][PF6]

(1:0.75, molar ratio) exhibit PLQYs and excited-state lifetimes comparable to those of neat films, indicating that the addition of [BMIM][PF6] does not induce

particu-lar quenching of emission. Stable operation was also achieved in devices using such a formulation. In the following, device characteristics based on the structure of [glass substrate/ITO/1:[BMIM][PF6] or 2:[BMIM][PF6] (100 nm)/Ag] are

dis-cussed and are summarized in Table2.

A distinct characteristic of LECs is that they can be operated under a bias voltage close toEg/e. As shown in Table1, the electrochemical gaps (DE1/2), which were

derived from the difference betweenEox 1=2andE

red

1=2corrected with potentials of the ferrocenium/ferrocene redox couple, for complexes 1 and 2 are 2.64 eV and 2.92 eV, respectively. The devices based on complexes 1 and 2 were thus first tested under the biases of 2.6 V and 2.9 V, respectively, although the energy gaps in films are usually smaller than those in solutions because of the environmental polarization. EL spectra of the devices based on complexes 1 and 2 (added with [BMIM][PF6] are shown in Figs.2aand2b, respectively, for comparison with their

PL spectra. EL spectra are basically similar to PL spectra, indicating similar emission mechanisms. Commission internationale de l’Eclairage (CIE) coordinates for the EL spectra of complexes 1 and 2 are (0.51, 0.48) and (0.35, 0.57), respec-tively. Time-dependent brightnesses and current densities of the devices operated under bias conditions described above are shown in Figs.4aand4bfor complexes 1 and 2, respectively. Both devices exhibited similar electrical characteristics. On the one hand, the currents of both devices first increased with time after the bias was applied and then stayed at a constant level after 350–400 min. On the other hand,

Table 2 Summary of the LEC device characteristics based on complexes 1 and 2 Complex Bias (V) lmax, EL(nm)a tmax(min)b Lmaxcd m2 ext, max,p, max

(%, lm W1)d Lifetime (h)e 1 2.6 580 170 330 6.2, 19.0 26 2.5 150 100 7.1, 22.6 54 2 2.9 535 85 145 6.6, 23.6 6.7 2.8 90 52 7.1, 26.2 12 a_{EL peak wavelength}

b_{Time required to reach the maximal brightness} c_{Maximal brightness achieved at a constant bias voltage}

d_{Maximal external quantum efficiency and maximal power efficiency achieved at a constant bias} voltage

e_{The time for the brightness of the device to decay from the maximum to half of the maximum} under a constant bias voltage

(9)

the brightness first increased with the current and reached the maximum of 330 cd m2 for complex 1 at ~170 min and of 145 cd m2 for complex 2 at ~85 min. The brightness then decreased with time even though the device current stayed rather constant. The decrease in brightness was irreversible, that is, the maximum brightness obtained in the first measurement could not be fully recovered in the followed measurements even under the same driving conditions. It is ratio-nally associated with the degradation of the emissive material during the LEC operation, which was commonly seen in LEC devices [36].

Time-dependent EQEs and corresponding power efficiencies of the complex 1 device under the 2.6-V bias and the complex 2 device under the 2.9-V bias are shown in Figs.5aand5b, respectively. Both devices exhibited similar time evolu-tion in EQE. When a forward bias was just applied, the EQE was rather low due to unbalanced carrier injection. During the formation of the p- and n-type regions near electrodes, the balance of the carrier injection was improved and the EQE of the device thus increased rapidly. The peak EQE and the peak power efficiency are

0 2 4 6 8 0 4 8 12 16 20 24 2.6 V 2.5 V

External Quantum Efficiency (%)

Time (min) Power Efficiency (lm /W) 0 100 200 300 400 500 600 0 0 100 200 300 400 500 600 2 4 6 8 0 5 10 15 20 25 2.9 V 2.8 V

Time (min)

Power Efficiency (lm

/W)

b a

Fig. 5 The time-dependent EQE and the corresponding power efficiency of the single-layered LEC device for (a) complex 1 driven at 2.6 or 2.5 V and (b) complex 2 driven at 2.9 or 2.8 V. Reproduced with permission from [29]. Copyright 2007, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 0 50 100 150 200 250 300 350 0.0 0.5 1.0 1.5 2.0 2.5 2.5 V 2.6 V Brightness (cd /m 2) Brightness (cd /m 2) Time (min)

Current Density (mA

/cm

2)

Current Density (mA

/cm 2) 0 100 200 300 400 500 600 0 0 100 200 300 400 500 600 25 50 75 100 125 150 0.00 0.25 0.50 0.75 1.00 1.25 2.8 V 2.9 V Time (min) b a

Fig. 4 The time-dependent brightness and current density of the single-layered LEC device for (a) complex 1 driven at 2.6 or 2.5 V and (b) complex 2 driven at 2.9 or 2.8 V. Reproduced with permission from [29]. Copyright 2007, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

(10)

(6.2%, 19.0 lm W1) for the complex 1 device under the 2.6-V bias and (6.6%, 23.6 lm W1) for the complex 2 device under the 2.9-V bias. One notices that the peak efficiencies occurred before the currents reached their final maximal values. Such a phenomenon may be associated with two factors. First, although with the formation of the p- and n-type regions near electrodes, both contacts are becoming more ohmic and the carrier injection at both electrodes are becoming more bal-anced; however, the carrier recombination zone may consequently move toward one of the electrodes because of discrepancy in electron and hole mobilities. The recombination zone moving to the vicinity of an electrode may cause exciton quenching such that the EQE of the device would decrease with time while the current and the brightness are still increasing. Besides, degradation of the emissive material under a high field would also contribute to the decrease in the EQE when the recombination zone is still moving or when the recombination zone is fixed.

Both LEC devices driven under 0.1-V lower biases exhibit higher peak EQEs and lower degradation rates. As shown in Figs.5aand5b, the peak EQE and the peak power efficiency are (7.1%, 22.6 lm W1) for the complex 1 device under the 2.5-V bias and (7.1%, 26.2 lm W1) for the complex 2 device under the 2.8-V bias. It is interesting to note that the peak EQEs of the single-layered devices (no PEDOT:PSS is used) based on complexes 1 and 2 approximately approach the upper limits that one would expect from the PLQYs of their neat films, when considering an optical out-coupling efficiency of ~20% from a typical layered light-emitting device structure. To our knowledge, these EQEs are among the highest values reported for orange-red (or yellow) and green solid-state LECs based on CTMCs. Such results indicate that CTMCs with superior steric hindrance are essential and useful for achieving highly efficient solid-state LECs.

Since these newly developed cationic complexes (1 and 2) are intrinsically efficient, to further reduce self-quenching and increase EL efficiency, one possible approach is to spatially disperse the emitting complex (guest) into a matrix complex (host), as previously reported for conventional OLEDs and solid-state LECs [37–39]. The mixed host–guest films (~100 nm) for PL studies were spin-coated onto quartz substrates using mixed solutions of various ratios. Since in LECs, a salt [BMIM][PF6] of 19 wt% (where BMIM is 1-butyl-3-methylimidazolium) was also

added to provide additional mobile ions and to shorten the device response time [29,32], PL properties of the host–guest–salt three-component system were also characterized.

Figure6ashows the absorption spectrum of the guest and the PL spectra of the host–guest two-component systems having various guest concentrations. With the increase of the guest concentration, a gradual red shift from the host-like emission to the guest-like emission is observed. As shown in Fig.6b, the highest PLQY of ~37% (vs. ~31% of neat host and guest films) is obtained at the guest concentration of 25 wt%, at which the emission is nearly completely from the guest. Accompany-ing this enhanced PLQY is the longer excited-state lifetime (0.69ms) when com-pared with those of neat films (0.59ms for the host and 0.60 ms for the guest), indicating the effectiveness of the dispersion in suppressing quenching mechanisms of guest molecules. Less complete transfer is observed at lower guest concentrations,

(11)

0.00 0.04 0.08 0.12 a 0.0 0.2 0.4 0.6 0.8 1.0 A b so rb an ce Wavelength (nm) Guest Absorption Guest Concentration 0 wt.% 5 wt.% 15 wt.% 25 wt.% 50 wt.% 100 wt.% P L In te ns ity ( a . u .) 30 35 40 45 50 b Phot ol um inescence Q u an tum Yie ld (% ) Guest Concentration (wt.%) 400 500 600 700 800 0 20 40 60 80 100 0 20 40 60 80 100 0.60 0.65 0.70 0.75 0.80 0.85 E x ci te d-Sta te L if e ti m e (μ s) Without BMIM+(PF6–) With BMIM+_(PF 6–) Without BMIM+(PF6–) With BMIM+_(PF 6–) Guest Concentration (wt.%) c

Fig. 6 (a) The absorption spectrum of the neat guest film and PL spectra of host–guest films with various guest concentrations (without [BMIM][PF6]. (b) Photoluminescence quantum yields and (c) excited-state lifetimes as a function of the guest concentration for host–guest films without and with [BMIM][PF6] (19 wt%). Reproduced with permission from [49]. Copyright 2006, American Institute of Physics

(12)

which may be associated with the less extensive overlap between the host emission and the weak guest absorption. A calculation of the F€orster radius for the host–guest energy transfer gives a small value of <16 A˚ [40] indicating inefficient energy transfer at lower guest concentrations. With the addition of [BMIM][PF6] (19 wt%),

the trend in PL properties [spectra (not shown), PLQYs (Fig. 6b), and lifetimes (Fig.6c)] as a function of the guest concentration is similar to those of the two-component system. Yet, an even higher PLQY of ~50% (about 1.6 enhancement compared with PLQYs of neat host and guest films) and longer excited-state lifetime of ~0.82 ms are observed around the guest concentration of 25 wt%. It appears that [BMIM][PF6] not only provides additional mobile ions but is also

effective in suppressing interchromophore quenching.

Figure7ashows the time-dependent brightness and current density under con-stant biases of 2.5–2.7 V for the LEC using the mixture giving the highest PLQY (i.e., with host, guest, and [BMIM][PF6] concentrations of 56 wt%, 25 wt%, and

19 wt%, respectively). After the bias was applied, on the one hand, the current first increased and then stayed rather constant. On the other hand, the brightness first increased with the current and reached the maxima of 10, 30, and 75 cd m2at<1 h under biases of 2.5 V, 2.6 V, and 2.7 V, respectively. The brightness then dropped with time with a rate significantly depending on the bias voltage (or current). Corresponding time-dependent EQEs and power efficiencies of the same device are shown in Fig.7b. When a forward bias was just applied, the EQE was rather low due to poor carrier injection. During the formation of the p- and n-type regions near electrodes, the capability of carrier injection was improved and the EQE thus rose rapidly. The peak EQE, current, and peak power efficiencies at 2.5 V, 2.6 V, and 2.7 V are (10.4%, 29.3 cd A1, 36.8 lm W1), (9.9%, 27.9 cd A1, 33.7 lm W1), and (9.4%, 26.5 cd A1, 30.8 lm W1), respectively.

The maximum current density vs. voltage characteristics of LECs with various guest concentrations are shown in Fig. 8a. The bias voltage required for same

0 20 40 60 80 0.0 0.1 0.2 0.3 0.4 0.5 2.5 V 2.6 V 2.7 V Brightness (cd /m 2) Time (min)

Current Density (mA

/c m 2) 0 100 200 300 400 500 600 0 0 100 200 300 400 500 600 2 4 6 8 10 12 0 5 10 15 20 25 30 35 2.5 V 2.6 V 2.7 V

Time (min)

Power Efficiency (lm

/W)

a b

Fig. 7 (a) Brightness (solid symbols) and current density (open symbols) and (b) external quantum efficiency (solid symbols) and power efficiency (open symbols) as a function of time under a constant bias voltage of 2.5–2.7 V for the host–guest LEC with host, guest and [BMIM] [PF6] concentrations of 56 wt%, 25 wt%, and 19 wt%, respectively. Reproduced with permission from [49]. Copyright 2006, American Institute of Physics

(13)

0.0 0.5 1.0 1.5 2.0 2.5 – 3.1 eV Host Guest – 3.0 eV – 6.0 eV – 5.7 eV Guest (wt.%) 0 16 25 35 81

Maximum Current Density (mA/cm

2) Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 Guest Concentration 0 wt.% 7 wt.% 25 wt.% 81 wt.% EL Intensity (a.u.) Wavelength (nm) 2.5 2.6 2.7 2.8 2.9 400 500 600 700 800 0 20 40 60 80 7 8 9 10 11 16 20 24 28 32 36 Quantum Efficiency Peak External Quantum Efficiency (%) Guest Concentration (wt.%) Power Efficiency

Peak Power Efficiency (lm

/W

)

a

b

c

Fig. 8 (a) Maximum current density vs. voltage characteristics for LECs with various guest concentrations. (b) EL spectra (at 2.8 V) for LECs with various guest concentrations. (c) Peak external quantum efficiencies and peak power efficiencies (at current densities<0.1 mA cm2) of host–guest LECs as a function of the guest concentration.Inset of (a): the energy level diagram of the host and guest molecules. Reproduced with permission from [49]. Copyright 2006, American Institute of Physics

(14)

current drops as the guest concentration is raised. This may be understood by the energy level diagram obtained from cyclic voltammetry [29] (inset of Fig.8a). Such an energy level alignment favors carrier injection/transport (at least for holes) through the smaller-gap guest and direct carrier recombination/exciton formation on the guest (rather than host–guest energy transfer) if the guest concentration is high enough. EL spectra of host–guest LECs with various guest concentrations (Fig. 8b) indeed support the mechanism of direct exciton formation on guest molecules, which would greatly reduce host emission due to incomplete energy transfer. Compared with PL spectra (e.g., Fig. 6), EL spectra are much less dependent on the guest concentration. Even with a guest concentration as low as 7 wt%, the EL spectrum is almost the same as that of the pure guest device [i.e., the LEC with 81 wt% of guest and 19 wt% of [BMIM][PF6], indicating predominant

guest emission.

Figure8cshows peak EQEs and peak power efficiencies for host–guest LECs as a function of the guest concentration (at current densities<0.1 mA cm2). The peak EQEs and power efficiencies roughly follow the trend of the PLQYs. The highest peak EQE and power efficiency of 10.4% and 36.8 lm W1are achieved at the guest concentration of 25 wt%, coincident with the concentration giving the highest PL efficiency. Such an EQE represents a 1.5 enhancement compared with those of pure host and guest devices. It is also worth noting that such a peak EQE approximately approaches the upper limit that one would expect from the PLQY of the host–guest emissive layer (~50%), when considering an optical out-coupling efficiency of ~20% from a typical layered light-emitting device structure. Such EQEs and power efficiencies also are among the highest reported for solid-state LECs based on CTMCs. Such results indicate that the host–guest system is essential and useful for achieving highly efficient solid-state LECs.

2.2 Color Tuning

Early developments of CTMCs for LECs were focused on ruthenium-based com-plexes, which typically emit in the red and orange–red part of the spectrum due to low metal-to-ligand charge transfer (MLCT) energies. Ester-substituted ligands have been found to red-shift the emission of [Ru(bpy)3](PF6)2[12,13]. Ng et al.

[41] published a series of polyimides containing a Ru complex on the main chain. The device showed deep-red EL centered at about 650 nm and near infrared (IR) emission centered near 750–800 nm. However, the efficiency of such device is rather low (0.1%). Bolink et al. [42] also have achieved a deep red emission, using bis-chelated Ru complex [Ru(tpy)(tpy-CO2Et)](PF6)2, where tpy is terpyridine.

The ester substituents were found to significantly improve the device efficiency over the unsubstituted terpyridyl complex, but the device efficiency is still very low. Low efficiency is a general phenomenon of terpyridyl Ru complexes since the luminescent3MLCT state is quenched by low-lying3MC levels due to poor bite angles of the tpy ligands [43].

(15)

Iridium was shown to have promise for color tuning over ruthenium complexes because of increased ligand-field splitting energies. The pioneering work of Ir-based complex ([Ir(ppy)2(dtb-bpy)]PF6 for LECs was reported by Slinker et al.

[18]. Subsequent color tuning of CTMCs has centered on appropriate selection of ligand or transition metal core. Complexes based on fluorine substituents on the phenylpyridine ligands were reported to cause a blue shift of the emission spectrum relative to unfluorinated complexes. For example, ([Ir(ppy)2(dtb-bpy)]PF6and [Ir

(dF(CF3)ppy)2(dtb-bpy)]PF6showed EL centered at 542 nm and 500 nm,

respec-tively [20,21]. The increasing of bandgap of complexes results from the fact that metal-centered highest occupied molecular orbital (HOMO) level is highly stabi-lized by introduction of the fluorine atoms [21]. We also studied the electrochemi-cal characteristics of complexes 1 and 2 via cyclic voltammetry. As shown in Fig.9, complex 2 exhibits a reversible oxidation peak at 1.70 V vs. Ag/AgCl, which is significantly higher than that of complex 1 (1.35 V). The HOMO level of ([Ir (ppy)2(dtb-bpy)]PF6had been reported to be a mixture of the d-orbitals of iridium

and thep-orbitals of the phenyl ring of the ppy ligand [21]. Since 1 and 2 are also cationic ppy-based Ir(III) complexes, their HOMO distributions should be similar to that of ([Ir(ppy)2(dtb-bpy)]PF6. The higher oxidation potential of complex

2 compared with that of complex 1 indicates that the two electron-withdrawing fluoro substituents on the phenyl ring of the ppy ligand possibly reduce the electron density on the Ir metal center and consequently stabilize the HOMO level. Accord-ing to the previous studies, the lowest unoccupied molecular orbital (LUMO) of ([Ir(ppy)2(dtb-bpy)]PF6is located mainly on the dtb-bpy ligand [21], and therefore,

for complexes 1 and 2, reduction is expected to take place on the auxiliary SB ligand. Similar reduction potentials for complex 1 (–1.33 V) and complex 2 (–1.26 V) confirm that this common ligand is responsible for reduction of both compounds. Still, the reduction potential of complex 2 is slightly less negative than

2.0 1.5 1.0 0.5 0.0 – 0.5 –1.0 –1.5 – 0.01 0.00 0.01 _{Complex 1} Complex 2 Current (mA) Potential (V vs. Ag / AgCl)

Fig. 9 Cyclic voltammograms of complexes 1 and 2. Potentials were recorded vs. the reference electrode Ag/AgCl (sat’d). Reproduced with permission from [29]. Copyright 2007, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

(16)

that of complex 1 perhaps because the fluoro substituents make complex 2 more electrophilic and thus easier to reduce [21]. With the HOMO level being signifi-cantly shifted and the LUMO level being only slightly changed by fluoro substitu-tion, the PL emission of complex 2 is significantly blue shifted compared with PL of complex 1, reflecting a characteristic of Ir complexes that their energy gaps can be tuned through introducing substituents of different electronic properties on the ppy ligands.

Tamayo et al. [22] demonstrated tuning of the emission color across a large portion of the visible part of the spectra by independent tuning of the HOMO and LUMO levels of the parent complex [Ir(ppy)2(bpy)]PF6, where ppz is

1-phe-nylpyrazolyl. Highly efficient red (lmax¼ 635 nm), green (lmax ¼ 542 nm), and

blue (lmax ¼ 492 nm) EL emissions were obtained in LECs based on [Ir

(tbppz)2(biq)]PF6, [Ir(ppy)2(bpy)]PF6, and [Ir(dF-ppz)2(dtb-bpy)](PF6), where tb

is 50-tert-butyl and biq is 2,20-biquinoline [22].

Color tuning can also be achieved by destabilization of the LUMO rather than stabilization of the HOMO. Nazeeruddin et al. [44] achieved blue-green EL emis-sion (lmax ¼ 520 nm) in LECs based on [Ir(ppy)2(dma-bpy)]PF6, where dma-bpy is

4,40-(dimethylamino)-2,20-bipyridine. Introduction of the electron-donating dimethylamino groups was demonstrated to significantly destabilize the LUMO relative to the HOMO when compared with the parent complex [Ir(ppy)2(dtb-bpy)]

PF6, resulting in blue-shifted emission.

Attaching a charged side group on the periphery of the cyclometalating ligands proposed by Bolink et al. [45] converted a neutral complex into a CTMC. A blue-green emission (lmax ¼ 487 nm) was observed for [Ir(ppy-PBu3)3](PF6)3. Such a

blue shift in emission, compared with the [Ir(ppy)3] complex, was attributed to

the electron-withdrawing nature of the tri-butylphosphonium group on the phenyl-pyridine ligand. However, EL was found to change from blue-green to yellow (lmax ¼ 570 nm) after 100 s of operation at 4 V.

Recently, we have also developed a novel blue-green emitting ppz-based Ir-complex (stable EL lmax¼ 488 nm) and a saturated red emitting ppy-based

Ir-complex (PL lmax¼ 672 nm). Details of photophysical and EL properties of

these two complexes will be described in Sect.3.

2.3 Lengthening Device Lifetime

The lifetimes of the devices defined as the time it takes for the brightness of the device to decay from the maximum to half of the maximum under a constant bias are important figure-of-merit when evaluating the durability of devices for display or lighting applications. Much effort has been made to lengthen device lifetimes of LECs.

The rates of electrochemical reactions of emissive CTMCs under electrical driving are bias dependent. In general, running LEC devices at a higher luminance

(17)

leads to shorter lifetimes, so long lifetime can often be claimed at the expense of luminance [12,29,46–49]. The lifetimes for complex 1 under the 2.6-V driving and for complex 2 under the 2.9-V driving are ~26 h (extrapolated) and 6.7 h, respec-tively (Fig.4). Although operated under a lower current density, the lifetime of the complex 2 device is significantly shorter than that of the complex 1 device. Irreversible multiple oxidation and subsequent decomposition under a high electric field had been proposed as the mechanism for degradation of LECs based on cationic iridium complexes [32, 50]. Since a higher bias voltage is required for operating the device based on complex 2 due to its larger energy gap, the higher electric field in the emissive layer of the complex 2 device perhaps accelerated the degradation. To mitigate the device degradation, slightly lower bias voltages of 2.5 V and 2.8 V were then applied for testing the devices based on complex 1 and complex 2, respectively. As shown in Figs. 4a and 4b, slight reduction of the applied bias by 0.1 V had led to greatly reduced degradation rates for brightnesses of both devices. The lifetimes of the complex 1 device under the 2.5-V driving and the complex 2 device under the 2.8-V driving are ~54 and ~12 h (both extrapo-lated), roughly two times longer than those driven at 0.1-V higher bias voltages. The longer device lifetimes, however, are achieved at the expense of reduced brightness. Therefore, emissive materials with high quantum yields are essential for LECs to be operated at a practical brightness yet using a lowest possible bias, which would ensure long-lifetime operation of LECs.

Blending CTMCs in an inert polymer can also improve lifetime [14,16]. It may result from lowered current density caused by separation of conducting chromo-phores in LEC devices. The choice of metal contacts was found to influence the lifetime, even for devices stored in the off state because reactive metal (e.g., Al) may activate some electrochemical reactions with CTMCs and lead to degradation [14, 51]. Degradation of [Ru(bpy)3]2þ-based device involves the water-induced

formation of photoluminescence quenchers, which were suggested to be complexes such as [Ru(bpy)2(H2O)2]2þand [Ru(bpy)2(H2O)]2O4þ[36,52,53]. Thus, CTMCs

with phenanthroline ligands credited with improved hydrophobicity increased resistance toward water-induced substitution reactions, leading to the higher device stability. [Ir(ppy)2(dp-phen)]PF6 device, where dp-phen is

4,7-diphenyl-1,10-phenanthroline, was found by Bolink et al. to have a lifetime of 65 h, which is the highest reported for a neat-film Ir-based LEC to date [54].

2.4 Shortening Turn-On Time

Turn-on time is defined as the time required for LEC devices to reach maximum emission under dc bias. For practical applications, turn-on times must be signifi-cantly reduced; however, most schemes for improving turn-on time come at a cost to stability.

Higher applied biases can lead to faster turn-on, but results in shorter lifetimes [12,48]. To achieve faster turn-on without losing the stability, a pulsed-biasing

(18)

scheme – a short pulse at a higher voltage was used to turn the device on, followed by a lower voltage that was used to drive the device for extended periods, was proposed by Handy et al. [12]. However, such technique complicates the driving circuits and the integration for applications. Reducing the thickness of the emissive layer of LECs also can reduce the turn-on time, but leads to a decrease in the device efficiency due to exciton quenching near the electrodes [55]. Smaller counterions such as BF4or ClO4lead to faster turn-on times than larger one, e.g., PF6, but

also deteriorate device lifetimes [13, 16]. Increasing ionic conductivity of the emissive layer by addition of an electrolyte to the CTMC layer has been shown to improve device lifetimes due to additional ions [20,32,56,57]. However, device lifetimes were deteriorated as well.

Chemically attaching ionically charged ligands to metal complexes represents an alternative and promising approach to increase the ionic conductivity and to improve the turn-on speed of LECs, since the phase compatibility is less an issue in such configurations. Bolink et al. reported a homoleptic iridium complex containing ionically charged ligands, giving a net charge ofþ3 for each complex [45]. The LEC based on such complex showed a much improved turn-on time, yet also suffered other issues, e.g., color shift during operation and low efficiencies, which are relevant to practical applications as well. Zysmen-Colman et al. recently also reported homoleptic ruthenium-based and heteroleptic iridium-based CTMCs con-taining tethered ionic tetraalkylammonium salts [58]. Similarly, LECs constructed using such complexes also exhibited reduced turn-on times; yet EL efficiencies from these complexes are not satisfactory, compared with general EL efficiencies achievable with CTMCs. Hence, further studies on CTMCs containing ionically charged ligands that can give improved turn-on times, stable operation, and also high device efficiency are still highly desired.

In one of our previous works, we demonstrated the reduction in turn-on times of single-component Ir-based LECs with tethered imidazolium moieties. The new CTMC is achieved by fusing two imidazolium groups at the ends of the two alkyl side chains of a more conventional complex [Ir(ppy)2(dC6-daf)]PF6 (3)

(where dC6-daf is 9,90-dihexyl-4,5-diazafluorene), forming the new complex [Ir (ppy)2(dCMIM-daf)](PF6)3(4) (where dC6MIM-daf is

9,9-bis[6-(3-methylimida-zolium)hexyl]-1-yl-4,5-diazafluorene) [59]. Molecular structures of complexes 3 and 4 are shown in Fig.10.

Homogeneous thin films of both complexes 3 and 4 can be obtained by spin-casting from CH3CN solution. We also tried to prepare a physical blending film of

3 and ionic liquid [BMIM][PF6] for comparative studies; however, the resulting

cloudy films indicated strong phase separation and incompatibility between com-plex 3 with the long alkyl chains and the ionic liquid with high polar character. The UV-visible absorption and PL spectra of 3 and 4 in CH2Cl2solution (105M) and

in neat film (~100 nm) are shown in Figs. 11a and 11b, respectively. Similar absorption features are observed in either solutions or films for both complexes. Solution PL emission is in the yellow range (~570 nm) for both complexes 3 and 4, indicating tethering imidazolium groups at the ends of the alkyl chains in complex 3 has no significant effects on modulating energy gaps.

(19)

The PLQYs of complexes 3 and 4 in CH2Cl2, as determined with a calibrated

integrating sphere, are 0.59 and 0.47, respectively (Table 3). Room-temperature transient PL of complexes 3 and 4 in CH2Cl2, as determined by the time-correlated

single photon counting technique, exhibits the single-exponential decay behavior, and the extracted excited-state lifetimes for 3 and 4 are 1.12 ms and 0.94 ms, respectively (Table3). Spin-coated neat films of complexes 3 and 4 exhibit PL spectra similar to those observed for their solutions (Fig.11). The measured PLQYs and excited-state lifetimes of the neat films are (0.33, 0.66 ms) for 3 and (0.35, 0.75ms) for 4 (Table3). Lower PLQYs and shorter excited-state lifetimes in neat

0.0 0.4 0.8 1.2 1.6 2.0 a b 0.0 0.2 0.4 0.6 0.8 1.0 Absorbance (a.u.) Wavelength (nm) PL, Abs, Solution PL, Abs, Neat film EL Intensity (a.u.) 200 300 400 500 600 700 800 0.0200 300 400 500 600 700 800 0.4 0.8 1.2 1.6 2.0 0.0 0.2 0.4 0.6 0.8 1.0 Absorbance (a.u) Wavelength (nm) PL, Abs, Solution PL, Abs, Neat film EL

Intensity (a.u.)

Fig. 11 Absorption and PL spectra in CH2Cl2solution (105M) and in neat film along with EL spectra under a forward bias voltage of 2.8 V for (a) complex 3 and (b) complex 4, respectively. Reproduced with permission from [59]. Copyright 2008, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim N Ir N N N PF6 3 N Ir N N N N N _N N 3PF6 4

(20)

films indicate that interaction between closely packed molecules provides addi-tional deactivation pathways. However, PLQYs of complexes 3 and 4 in neat films still retain ~56% and ~75% of those in solutions. Highly retained PLQYs in neat films of complexes 3 and 4 are likely to be associated with their sterically bulky dC6-daf and dC6MIM-daf ligands. It is noted that self-quenching in neat films of complex 4 is not as severe as that in neat films of complex 3 perhaps due to better steric hindrance provided by imidazolium groups in complex 4.

Figure 12 depicts the electrochemical characteristics of complexes 3 and 4 probed by cyclic voltammetry and the measured redox potentials are summarized in Table3. Complexes 3 and 4 exhibit reversible redox peaks at similar potential

1.5 1.0 0.5 0.0 – 0.5 –1.0 –1.5 – 0.015 – 0.010 – 0.005 0.000 0.005 0.010 0.015 Current /m A E (V vs. Ag / AgCl) / V 3 4

Fig. 12 Cyclic voltammogram of complexes 3 and 4. All potentials were recorded vs. Ag/AgCl (sat’d) as a reference electrode. Reproduced with permission from [59]. Copyright 2008, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Table 3 Summary of physical properties of complexes 3 and 4

Complex lmax, PL(nm)a PLQYb t (ms)c Eox1=2

(V)d Ered 1=2 (V)e DE1/2 (V)f Solutiong Neat film Solutiong Neat film Solutiong Neat film 3 568 562 0.59 0.33 1.12 0.66 +1.33 1.47 2.80 4 570 562 0.47 0.35 0.94 0.75 +1.31 1.48 2.82 a_{PL peak wavelength}

b_{Photoluminescence quantum yields} c_{Excited-state lifetimes}

d_{Oxidation potential} e_{Reduction potential} f_{The electrochemical gap}_DE

1/2is the difference betweenEox1=2andEred1=2 g_{Measured in CH}

2Cl2(105M) at room temperature

(21)

(vs. Ag/AgCl); one reversible oxidation potential occurs at 1.33 V and 1.31 V for 3 and 4, respectively, while one reversible reduction potential occurs at –1.47 V and –1.48 V for 3 and 4, respectively. The HOMO of [Ir(ppy)2(N^N)]PF6complexes

has been reported to be a mixture of the iridium and the phenyl group of the C^N ligand, while the LUMO mostly localizes on the N^N ligand [21]. Since the structures of complexes 3 and 4 differ only in the terminal alkyl chain of N^N ligand, their HOMO and LUMO distribution should be similar and give similar redox potentials. The presence of imidazolium moieties in 4 exhibited limited influence on the electrochemical properties of the Ir-center, especially the frontier molecular orbitals.

Device characteristics based on the structure of [glass substrate/ITO/complex 3 or 4 (100 nm)/Ag] are discussed and are summarized in Table4. EL spectra of the devices based on complexes 3 and 4 are shown in Figs.11aand11b, respectively, for comparison with their PL spectra. EL spectra are basically similar to PL spectra, indicating similar emission mechanisms. Time-dependent brightness and current densities of the LEC devices based on complex 3 operated under 2.7 and 2.8 V are shown in Fig.13a. The currents of devices increased slowly with time after the bias was applied and kept increasing even after 10-h operation. The turn-on time of the device (tturn-on), defined as the time to achieve brightness of 1 cd cm2, was 65 min

under 2.8 V. The brightness increased with the current and reached the maximum of 105 cd m2 at ~500 min under 2.8 V. For devices under 2.7 V, the brightness reached 56 cd m2and was still increasing at 600 min. Comparatively, the LECs based on complex 4 showed much faster response. As shown in Fig.13b, this device turned on sharply in 12 min under 2.8 V, it took only ~200 min for the devices to reach the maximal brightness of 79 cd m2and 31 cd m2under 2.8 V and 2.7 V, respectively. Such improved turn-on times confirm that the chemically tethering imidazolium groups onto complex 4 increase the density of mobile counterions (PF6) in neat films and consequently increase the neat-film ionic conductivity.

Table 4 Summary of the LEC device characteristics based on complexes 3 and 4

Complex Bias (V) lmax, EL(nm)a tturn-on(min)b tmax(min)c Lmax(cd m2)d ext, max,p, max (%, lm W1)e 3 2.8 566 65 500 105 5.5, 18.7 2.7 84 >600 >56f 5.5, 19.2 4 2.8 577 12 200 79 4.0, 14.7 2.7 16 200 31 4.5, 17.1 a_{EL peak wavelength}

b_{Time required reaching the brightness of 1 cd cm}2 c_{Time required to reach the maximal brightness} d_{Maximal brightness achieved at a constant bias voltage}

e_{Maximal external quantum efficiency and maximal power efficiency achieved at a constant bias} voltage

f_{Maximal brightness was not reached during a 10-h measurement}

(22)

Efficient LECs are essential for operation at a practical brightness yet using a lowest possible bias, which would ensure stable operation of LECs. Hence, it is important to examine the effect of chemical tethering of imidazolium groups with an ionic iridium complex on the device efficiencies of LECs. Time-dependent EQEs and corresponding power efficiencies of the complex 3 and 4 devices under biases of 2.7 V and 2.8 V are shown in Figs.14aand14b, respectively. When a forward bias was just applied, the EQE was rather low due to unbalanced carrier injection. During the formation of the p- and n-type regions near electrodes, the balance of the carrier injection was improved and the EQE of the device thus increased rapidly. The peak EQE and the peak power efficiency are (5.5%, 18.7 lm W1) and (5.5%, 19.2 lm W1) for the complex 3 device under biases of 2.8 V and 2.7 V, respectively. The time-dependent evolution trend in device efficiency of the complex 4 device is similar to that of the complex 3 device, yet it took much less time (within 100 min) for the complex 4 device to reach the peak device efficiency. The peak EQE and the peak power efficiency are (4.0%,

0 20 40 60 80 100 0.0 0.2 0.4 0.6 0.8 1.0 2.7 V 2.8 V Brightness (cd /m 2) Time (min)

Current Density (mA

/cm 2) 0 100 200 300 400 500 600 0 0 100 200 300 400 500 600 20 40 60 80 0.0 0.2 0.4 0.6 0.8 2.7 V 2.8 V Brightness (cd /m 2) Time (min) Current Densit y ( mA /c m 2) a b

Fig. 13 Time-dependent brightness and current density under 2.7 V and 2.8 V for the single-layered LEC device based on (a) complex 3 and (b) complex 4, respectively. Reproduced with permission from [59]. Copyright 2008, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0 1 2 3 4 5 6 0 4 8 12 16 20 2.8 V 2.7 V

Time (min) Power Efficiency (lm /W ) 0 100 200 300 400 500 600 0 0 100 200 300 400 500 600 1 2 3 4 5 0 4 8 12 16 20 2.8 V 2.7 V

Time (min) Power Efficienc y (lm /W) b a

Fig. 14 Time-dependent EQE and the corresponding power efficiency under 2.7 V and 2.8 V for the single-layered LEC device based on (a) complex 3 and (b) complex 4, respectively. Repro-duced with permission from [59]. Copyright 2008, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

(23)

14.7 lm W1) and (4.5%, 17.1 lm W1) for the complex 4 device under 2.8 V and 2.7 V, respectively. EL efficiencies of devices based on both complexes 3 and 4 are comparable, indicating chemical tethering of imidazolium groups with an ionic iridium complex benefits faster device response yet does not influence EL efficien-cies significantly. Hence, the technique of integrating ionic liquid moieties onto the CTMC directly serves as a useful and promising alternative to improve the turn-on speed of LECs and to realize single-component CTMC LECs compatible with simple driving schemes.

3 White Light-Emitting Electrochemical Cells Based

on Cationic Transition Metal Complexes

Although solid-state LECs possess several advantages attractive for lighting appli-cations, rare demonstration of white-emitting solid-state LECs had been reported. The only previous report of solid-state white LECs was based on phase-separated mixture of a polyfluorene derivative and polyethylene oxide (PEO) [10]. However, the fluorescent nature of conjugated polymers would limit the eventual EL effi-ciency. Recently, CTMCs have also been used in solid-state LECs [11–22,31,32,

36,38,41,42,44–49,51–60], which show two major advantages over conventional polymer LECs (a) no ion-conducting material (e.g., PEO) is needed since these metal complexes are intrinsically ionic; and (b) higher EL efficiencies could be achieved due to the phosphorescent nature of the transition metal complexes. Inspired by previous works regarding energy transfer between ionic complexes reported by Malliaras [38] and De Cola [60], we had previously demonstrated highly efficient solid-state LEC devices by adopting host–guest cationic metal complexes [49]. Yet to our knowledge, there is no white LECs based on CTMCs ever being reported to date, despite their high potential.

Efficient white light emission may be most easily achieved by mixing two comple-mentary colors, such as blue-green and red emission. Thus, the development of efficient blue-green [21,22,44,60] and red-emitting [11–17,22,31,38,42,48,56] CTMCs is highly desired. In this section, we report the characterization of efficient blue-green and red-emitting cationic iridium complexes and their successful applica-tion in white LECs with adopting the effective host–guest strategy [38,49,61].

3.1 Photophysical and Electrochemical Properties of Novel

Blue-Green and Red-Emitting Cationic Iridium Complexes

Molecular structures of complexes 5–8 are shown in Fig.15. The emission proper-ties of complexes 5–8 in solutions (DCM, 105M) or in thin films are summarized in Table5. For reducing the response time of LECs, in this work the ionic liquid

(24)

[BMIM][PF6] was introduced in the emissive layer to provide additional mobile

anions [29, 32, 49]. Thus, emission properties of 5–7 films in the presence of [BMIM][PF6] (19 wt%) were also examined. In general, complexes 5–7 show PL

peaks at 491–499 nm and rather high PLQYs of 0.46–0.66 in solutions. High PLQYs of these complexes indicate that the rigidity of the daf ligand is beneficial for reducing nonradiative (e.g., vibration) deactivation processes. Among the three blue-green complexes 5–7, 7 exhibits shortest emission wavelengths of 488–491 nm and highest PLQYs of 0.28–0.30 in thin films (neat or dispersed with [BMIM][PF6], and thus was chosen as the host material for white LEC studies.

The absorption and PL spectra of blue-green complex 7 and red-emitting complex 8 in solutions and in neat films (~100 nm) are explicitly shown in Fig.16. Interest-ingly, complex 7 with the alkyl substitution on daf exhibits nearly same emission wavelengths in solutions and in solid films. Complex 8 exhibits more saturated red emission (with PL peaks of 656 and 672 nm in solutions and in neat films, respectively) compared with that of [Ir(ppyz)2(biq)]PF6 [22]. The energy gaps

estimated by cyclic voltammetry for complex 8 (2.23 eV, Table5) and [Ir(ppyz)2

(biq)]PF6(2.45 eV) [22] are consistent with the photophysical observation. For [Ir

(ppyz)2(biq)]PF6, the HOMO and the LUMO are predominantly localized on the

ppz and biq ligand, respectively [22]. The smaller energy gap of complex 8, which also contains biq, is thus associated with destabilization of HOMO in replacing ppz with ppy.

Cyclic voltammetry was used to probe the electrochemical properties of these complexes. The results are summarized in Table 5. For complexes 5–7, one reversible oxidation potential was detected, which can be attributed to the oxidation that occurred on the Ir center. Interestingly, the electronic nature of ancillary daf ligands plays a subtle role on the oxidation potential. For example, complex 7 containing a more electro-rich dedaf ligand shows a lower oxidation potential (1.20 V) when compared with those of complexes 5 (1.29 V) and 6 (1.28 V), both

N N F F Ir N N F F N N _PF 6– PF6– R R

5: [Ir(dfppz)2 (dasb)]+ (PF6–), R, R = 2, 2′-biphenyl

6: [Ir(dfppz)2 (bmpdaf)]+ (PF6–), R, R = 4-CH3OPh, 4-CH3OPh 7: [Ir(dfppz)2 (dedaf)]+(PF6–), R, R = C2H5, C2H5 N Ir N N N 8 + +

(25)

containing aryl substitutions on C9 of the diazafluorene ligand. On the contrary, the reduction capability of the diazafluorene ligand is presumably responsible for observed reversible reduction of complexes 5–7. Thus, the Ir-complex coordinated

200 300 400 500 600 700 800 900 0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Absorbance (a.u.) Wavelength (nm) 7, neat film 8, solution 7, neat film 8, solution

PL Intensity (a.u.)

Fig. 16 Absorption (left) and PL spectra (right) of complexes 7 and 8 in dichloromethane solutions (105M) and in neat films. Reproduced with permission from [61]. Copyright 2008, American Chemical Society

Table 5 Summary of physical properties of complexes 5–8

Complex lmax, PL(nm),F, t (ms)a Eox1=2 (V)b Ered 1=2 (V)b DE1/2 (V) Solutionc Neat film or

host–guest film Film with [BMIM][PF6]d 5 499, 0.46, 0.71 513, 0.20, 0.36 504, 0.22, 0.42 +1.29e 1.70b 2.99 6 497, 0.66, 0.85 507, 0.22, 0.51 498, 0.26, 0.56 +1.28e 1.72f 3.00 7 491, 0.54, 0.73 491, 0.28, 0.55 488, 0.30, 0.74 +1.20g 1.79f 2.99 8 656, 0.20, 0.75 672, 0.09, 0.33 – +0.86f 1.37f 2.23 8 (0.2 wt %) : 7 – (493, 606), 0.26, (0.49h, 1.54i) (488, 603), 0.29, (0.54h, 1.68i) – – – 8 (0.4 wt %) : 7 – (493, 614), 0.27, (0.38h_{, 1.52}i₎ (488, 612), 0.28, (0.47h_{, 1.68}i₎ – – – a_{At room temperature}

b_{Potential vs. ferrocene/ferrocenium redox couple} c_{Measured in CH}

2Cl2(105M)

d_{Films containing 19 wt% [BMIM][PF}

6]

e_{0.1 M tetra-n-butylammonium hexafluorophosphate (TBAPF}

6) in acetonitrile f_{0.1 M TBAP} 6in acetonitrile g_{0.1 M TBAPF} 6in CH2Cl2 h_{Measured at 480 nm} i_{Measured at 650 nm}

(26)

to an electron-rich dedaf ligand (complex 7) shows a more negative reduction potential (1.79 V) when compared with those complexed to the aryl-substituted daf ligand [complexes 5 (1.70 V) and 6 (1.72 V)]. These results thus suggest that the energies of frontier orbitals in these complexes can be subtly manipulated by tailoring the electronic properties of the auxiliary daf ligands. For the red complex 8, one reversible oxidation (0.86 V) and one reversible reduction (1.37 V) were detected. The electron-rich character of the ppy ligand and the more p-delocalized biq ligand contribute to the higher ease of oxidation and reduction, respectively, thus leading to a small energy gap of complex 8.

Emission properties of the 7:8 host–guest films are also summarized in Table5. Figure 17 depicts the PL spectra of the host–guest films with different guest 8 concentrations (0.4 and 0.2 wt%) with or without the presence of [BMIM][PF6]

(19 wt%). In general, with the presence of [BMIM][PF6], both the host and guest

exhibit longer excited-state lifetimes and the PLQYs are slightly raised (Table5), indicating the role of [BMIM][PF6] in suppressing intermolecular interactions [49].

As shown in Fig.17, white emission of different blue-green and red compositions can be achieved by adjusting the guest concentration. Raising the guest concentra-tion effectively enhances the host–guest energy transfer, accompanied by shorter lifetimes of the host emission (Table5). The host–guest film containing 0.4 wt% of 8 (guest) shows white emission having CIE coordinates of (x, y)¼ (0.34, 0.37), which is rather close to the ideal equal-energy white (x, y)¼ (0.33, 0.33), and therefore was subjected to further EL studies.

400 500 600 700 800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Without BMIM+_PF 6– 8 (0.4 wt.%) 8 (0.2 wt.%) With BMIM+PF6– 8 (0.4 wt.%) 8 (0.2 wt.%) PL Intensity (a.u.) Wavelength (nm)

Fig. 17 PL spectra of host–guest films containing different guest concentrations (0.4 and 0.2 wt %) without and with [BMIM][PF6] (19 wt%). Reproduced with permission from [61]. Copyright 2008, American Chemical Society

(27)

3.2 Electroluminescent Properties of White Light-Emitting

Electrochemical Cells Based on Host–Guest Cationic

Iridium Complexes

The LECs have the structure of ITO/emissive layer (100 nm)/Ag (150 nm), where the emissive layer contains host 7 (80.5 wt%), guest 8 (0.4 wt%), and [BMIM][PF6]

(19.1 wt%). Their EL properties are summarized in Table6. The EL spectra of the white LECs under various biases, along with the PL spectra for comparison, are shown in Fig.18a. It is noted that the peak wavelength of the blue component in EL spectra is 488 nm, which is one of the shortest EL wavelengths reported to date for LECs based on CTMCs. Compared with PL, the relative intensity of the red emission with respect to the blue emission is larger in EL and increases as the bias decreases. Furthermore, the maximum current density of the host–guest device (with 0.4 wt%

Table 6 Summary of white LEC device characteristics Bias (V) CIE (x, y)a _CRIa _t

maxb(min) Lmaxc (cd m2)

ext, max,L, max,p, maxd

(%, cd A1, lm W1)

t1/2e(h)

2.9 (0.45, 0.40) 81 240 2.5 4.0, 7.2, 7.8 8.9

3.1 (0.37, 0.39) 80 60 18 3.4, 6.1, 6.2 1.3

3.3 (0.35, 0.39) 80 30 43 3.3, 5.8, 5.5 0.4

a_{Evaluated from the EL spectra}

b_{Time required to reach the maximal brightness} c_{Maximal brightness achieved at a constant bias voltage}

d_{Maximal external quantum efficiency, current and power efficiencies achieved at a constant bias} voltage

e_{The time for the brightness of the device to decay from the maximum to half of the maximum} under a constant bias voltage

0.0 0.4 0.8 1.2 1.6 PL EL 3.3 V EL 3.1 V EL 2.9 V Intensity (a.u.) Wavelength (nm) 400 500 600 700 800 0 2.9 3.0 3.1 3.2 3.3 1 2 3 4 5 6 Guest 0.4 wt.% Guest 0 wt.% Guest – 5.66 eV – 3.43 eV Host – 6.00 eV – 3.01 eV

Maximum Current Density

(m A/c m 2) Voltage (V) b a

Fig. 18 (a) The bias-dependent EL spectra of the white LECs compared with the PL spectrum. (b) Maximum current density vs. voltage characteristics for the host–guest (0.4 wt% doping concentration) and the host-only LECs.Inset: the energy level diagram of the host and guest molecules. Reproduced with permission from [61]. Copyright 2008, American Chemical Society

(28)

guest) is lower than that of the host-only device under same bias conditions (Fig. 18b). These results could be understood by energy level alignments of the host and guest (inset of Fig.18b). At lower biases, such energy level alignments favor carrier injection and trapping on the smaller-gap guest, resulting in direct carrier recombination/exciton formation on the guest (rather than host– guest energy trans-fer). Therefore, larger fractions of guest emission are observed at lower biases. Nevertheless, white emission with CIE coordinates of (0.35, 0.39) could be achieved at higher biases and brightness. Also, the present white LECs exhibit a rather high color rendering index (CRI) (up to 80), which is an important characteristic for solid-state lighting.

Figure 19a shows the time-dependent brightness and current density under constant biases of 2.9–3.3 V for the white LEC. After the bias was applied, the current first rose and then stayed rather constant. On the contrary, the brightness first increased with the current and reached the maxima of 2.5 cd m2, 18 cd m2, and 43 cd m2under biases of 2.9 V, 3.1 V, and 3.3 V, respectively. The brightness then dropped with time with a rate depending on the bias voltage (or current). Corresponding time-dependent EQEs and power efficiencies of the same device are shown in Fig.19b. When a forward bias was just applied, the EQE was rather low due to poor carrier injection. During the formation of the doped regions near electrodes, the capability of carrier injection was improved and the EQE thus rose rapidly. The peak EQEs, current and power efficiencies at 2.9 V, 3.1 V, and 3.3 V are (4.0%, 7.2 cd A1, 7.8 lm W1), (3.4%, 6.1 cd A1, 6.2 lm W1), and (3.3%, 5.8 cd A1, 5.5 lm W1), respectively.

Peak brightness and turn-on time (the time required to reach the maximal brightness) as a function of bias voltage for white LECs are shown in Fig. 20a. An electrochemical junction between p- and n-type doped layers of LECs is formed during device operation. As revealed in previous studies [31], as bias voltage increases, the junction width decreases due to extension of doped layers, conse-quently leading to higher current density (Fig.19a) and higher brightness (Fig.20a). The higher electric field in the device also accelerates redistribution of mobile ions,

0 10 20 30 40 50 0.01 0.1 1 10 2.9 V 3.1 V 3.3 V Brightness (cd /m 2) Time (min)

Current Density (mA

/c m 2) 0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 1 2 3 4 5 0 2 4 6 8 2.9 V 3.1 V 3.3 V External Quantum Efficiency (%) Time (min) Power Efficiency (lm /W ) b a

Fig. 19 (a) Brightness (solid symbols) and current density (open symbols) and (b) external quantum efficiency (solid symbols) and power efficiency (open symbols) as a function of time under a constant bias voltage of 2.9–3.3 V for the white LEC. Reproduced with permission from [61]. Copyright 2008, American Chemical Society

(29)

which facilitates formation of ohmic contacts with the electrodes. Thus, operation of LECs under higher bias voltages speeds up the device response (Fig. 20a). However, higher brightness and faster response are obtained at the expense of device stability. As shown in Fig.20b, peak EQE and device lifetime (the time for the brightness of the device decaying from the maximum to half of the maximum under a constant bias voltage) of the white LECs deteriorate under higher bias voltages. It may be associated with the fact that the higher electric field or current density accelerates degradation (multiple oxidation and subsequent decom-position) [32] of the emissive CTMCs. Previous studies showed that addition of materials altering the electronic properties of the emissive layer of LECs also influence the device lifetimes [14, 56]. Blending an inert polymer [14] in the emissive layer of LECs improves device lifetimes, while addition of an electrolyte [56] shortens device lifetimes. Blending of an inert polymer reduces the electronic conductivity of the emissive layer and thus lowers the current density. On the contrary, additional mobile ions provided by the electrolyte increase the doping level of the doped region, resulting in higher current density due to increased electronic conductivity. These results also imply that higher current leads to shorter device lifetimes. Recently, two reports [52,53] have revealed that electrical driving of LECs based on CTMCs induces oxo-bridged dimers, which effectively quench EL. However, detailed degradation mechanisms of the LECs based on CTMCs remain unclear and further studies are still needed to achieve practical device lifetimes.

4 Outlook

Solid-state CTMC-based LECs exhibit advantages of simple fabrication processes, low-voltage operation, and high power efficiency, which thus may be competitive for lighting applications. However, improvements will be required in our prototype

0 10 20 30 40 50 0 40 80 120 160 200 240 Brightness (cd /m 2) Voltage (V) Brightness

Turn-on Time (min)

Turn-on Time 2.9 3.0 3.1 3.2 3.3 3.2 2.9 3.0 3.1 3.2 3.3 3.4 3.6 3.8 4.0 4.2 0 2 4 6 8 10 External Quantum Efficiency (%) Voltage (V) External Quantum Efficiency Lifetime (hr ) Lifetime b a

Fig. 20 (a) Peak brightness (solid symbols) and turn-on time (open symbols) and (b) peak external quantum efficiency (solid symbols) and lifetime (open symbols) as a function of bias voltage for the white LECs. Reproduced with permission from [61], American Chemical Society