Enhancing device efficiencies of solid-state near-infrared light-emitting electrochemical cells by employing a tandem device structure

(1)

Enhancing device efﬁciencies of solid-state near-infrared

light-emitting electrochemical cells by employing a tandem

device structure

Chia-Lin Lee, Chia-Yu Cheng, Hai-Ching Su

⇑

Institute of Lighting and Energy Photonics, National Chiao Tung University, Tainan 71150, Taiwan

a r t i c l e

i n f o

Article history:

Received 15 November 2013

Received in revised form 25 December 2013 Accepted 1 January 2014

Available online 17 January 2014 Keywords:

Light-emitting electrochemical cells Organic light-emitting devices Near-infrared

a b s t r a c t

Compared to near-infrared (NIR) organic light-emitting devices, solid-state NIR light-emitting electrochemical cells (LECs) could possess several superior advantages such as simple device structure, low operating voltages and balanced carrier injection. However, intrinsically lower luminescent efficiencies of NIR dyes and self-quenching of excitons in neat-film emissive layers limit device efficiencies of NIR LECs. In this work, we demonstrate a tandem device structure to enhance device efficiencies of phosphorescent sensitized fluorescent NIR LECs. The emissive layers, which are composed of a phosphorescent host and a fluorescent guest to harvest both singlet and triplet excitons of host, are connected vertically via a thin transporting layer, rendering multiplied light outputs. Output electro-luminescence (EL) spectra of the tandem NIR LECs are shown to change as the thickness of emissive layer varies due to altered microcavity effect. By fitting the output EL spectra to the simulated model concerning microcavity effect, the stabilized recombination zones of the thicker tandem devices are estimated to be located away from the doped layers. Therefore, exciton quenching near doped layers mitigates and longer device lifetimes can be achieved in the thicker tandem devices. The peak external quantum efficiencies obtained in these tandem NIR LECs were up to 2.75%, which is over tripled enhancement as compare to previously reported NIR LECs based on the same NIR dye. These efficiencies are among the highest reported for NIR LECs and confirm that phosphorescent sensitized fluoresce combined with a tandem device structure would be useful for realizing highly efficient NIR LECs.

1. Introduction

Near-infrared (NIR) organic light-emitting devices (OLEDs) would be promising NIR light sources offering advantages of light weight, low power consumption and compatibility with large area and ﬂexible substrates. Therefore, they have received much attention due to their potential applications in telecommunications, displays and

bio-imaging [1]. However, sophisticated multilayer structures and low-work-function cathodes are generally required for NIR OLEDs to optimize device efﬁciencies, inﬂuencing their competitiveness with other solid-state NIR emitting technologies. In contrast with conventional NIR OLEDs, solid-state NIR light-emitting electrochemical cells (LECs) could possess several superior advantages. Generally, LECs require only a single emissive layer, which can be easily processed from solutions, and can conve-niently use air-stable electrodes. The emissive layer of LECs contains mobile ions, which can drift toward electrodes under an applied bias. These spatially separated ions

⇑Corresponding author. Tel.: +886 6 3032121x57792; fax: +886 6 3032535.

E-mail address:[email protected](H.-C. Su).

Contents lists available atScienceDirect

Organic Electronics

(2)

consequently induce doping (oxidation and reduction) of the emissive materials near the electrodes, i.e. p-type doping near the anode and n-type doping near the cathode

[2,3]. The doped regions induce ohmic contacts with the electrodes and facilitate carrier injection, giving low oper-ation voltages and high power efﬁciencies.

A few literatures about solid-state NIR LECs based on small-molecule cationic transition metal complexes (CTMCs) [4–7] and polymers containing pendant CTMC groups [8] have been reported. Nevertheless, compared with visible light-emitting CTMCs[9–24], deteriorated pho-toluminescence quantum yields (PLQYs) of NIR light-emit-ting CTMCs [4,5,7] are generally measured due to the energy gap law, which states that the nonradiative decay rates of CTMCs increase as the energy gaps decrease[11]. Furthermore, self-quenching of excitons in neat films also limits the device efficiency of NIR LECs based on neat films of CTMCs. Therefore, neat-film CTMC-based NIR LECs typi-cally exhibited external quantum efficiencies (EQEs) lower than 0.1% photon/electron[4–8]. To improve device effi-ciencies of NIR LECs, NIR LECs based on phosphorescent sen-sitized fluorescence have been reported [25]. In phosphorescent sensitized fluorescence[25–28], both sin-glet and triplet excitons in the host could be harvested due to effective Förster energy transfer[29]from triplet excitons of the phosphorescent host to singlet excitons of the fluores-cent guest. With well developed CTMC hosts possessing good carrier balance, commercially available efficient fluo-rescent ionic NIR laser dyes[30]can be conveniently utilized as guests to achieve efficient NIR electroluminescence (EL). In addition, self-quenching of excitons could be mitigated in host–guest emissive layers. As a result, the reported phos-phorescent sensitized fluorescent NIR LECs exhibited EQEs of ca. 1%[25]. However, these efficiencies are still not en-ough for practical applications and further improving device efficiencies of NIR LECs would be required.

Recently, we have demonstrated tandem LECs to reach over doubled light outputs and device efﬁciencies by stack-ing two sstack-ingle-layered LECs via a thin connectstack-ing layer

[31]. A similar approach employing a thin metal interlayer to produce tandem white light emitting devices was also reported recently [32]. It would be a simple way to en-hance device efficiencies of phosphorescent sensitized fluorescent NIR LECs. In this work, we report tandem phos-phorescent sensitized fluorescent NIR LECs based on emis-sive layers containing a phosphorescent CTMC host doped with a fluorescent ionic NIR dye. To clarify effects of emis-sive-layer thickness on device performance, tandem NIR LECs with various thicknesses are studied. Output EL spec-tra of tandem NIR LECs can be modified by adjusting the thicknesses of the emissive layers due to altered microcav-ity effect. Furthermore, improved device lifetimes are found in thicker devices due to reduced exciton quenching near electrodes. The peak EQE obtained in these tandem NIR LECs is up to 2.75%, which is over triple of that achieved in single-layered NIR LECs based on the same NIR dye[25]. The device efficiencies achieved are among the highest reported for NIR LECs and thus confirm that phosphorescent sensitized fluoresce combined with a tan-dem device structure would be useful for realizing highly efficient NIR LECs.

2. Experiment section 2.1. Materials

The host complex (1) used in the emissive layer of the NIR LECs was Ru(dtb-bpy)3(PF6)2 (where dtb-bpy is 4,40

-ditertbutyl-2,20_-bipyridine) _[33]_{. The ionic NIR dye 3,3}0

-diethyl-2,20_{-oxathiacarbocyanine iodide (DOTCI), which}

has been reported as an active material in efﬁcient NIR dye lasers[30], was utilized as the guest material doped in the host. Ru(dtb-bpy)3(PF6)2and DOTCI were purchased

from Luminescence Technology Corp. and Sigma–Aldrich, respectively, and they were used as received.

2.2. Photoluminescent characterization

Photoluminescence (PL) characteristics of DOTCI in eth-anol were recorded at room temperature using 105_M

solutions. The neat ﬁlm of complex 1 for PL studies was spin-coated at 3000 rpm onto a quartz substrate using ace-tonitrile solution with a concentration of 80 mg mL1_{. The}

thickness of the spin-coated neat host film was ca. 250 nm, as measured using profilometry. UV–Vis absorption and PL spectra were obtained with a Princeton Instruments Acton 2150 spectrophotometer. The exciting wavelengths of DOTCI in ethanol and neat film of complex 1 are 650 and 450 nm, respectively.

2.3. LEC device fabrication and characterization

Indium tin oxide (ITO)-coated glass substrates were cleaned and treated with UV/ozone prior to use. A thin poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) layer (30 nm) was spin-coated at 4000 rpm onto the ITO substrate in air and was then baked at 150 °C for 30 min. For the single-layered devices (S1, S2 and S3), the emissive layers were then spin-coated at 3000 rpm from the mixed acetonitrile solutions of complex 1 and DOTCI (weight ratio of complex 1 and DOTCI = 99:1). The concentrations of the solutions used for spin coating of the emissive layers of devices S1, S2 and S3 are 80, 150 and 220 mg mL1_{, respectively. The thicknesses of the emissive}

layers for devices S1, S2 and S3 are 250, 440 and 660 nm, respectively. For the tandem devices, the lower emissive layers were ﬁrst spin-coated at 3000 rpm from the mixed acetonitrile solutions of complex 1 and DOTCI (weight ra-tio of complex 1 and DOTCI = 99:1) with concentrara-tions of 80, 150 and 220 mg mL1 _{for devices T1, T2 and T3,}

respectively. The thicknesses of the lower emissive layers for devices T1, T2 and T3 are 250, 440 and 660 nm, respec-tively. The connecting PEDOT:PSS layers were then spin-coated onto the lower emissive layers at 3500 rpm. The thickness of the connecting PEDOT:PSS layers is ca. 55 nm. Finally, the upper emissive layers were spin-coated by the same procedures used for spin-coating of the lower emissive layers. After spin coating, the samples were then baked at 70 °C for 10 h in a nitrogen glove box, followed by thermal evaporation of a 100-nm Ag top contact in a vac-uum chamber (106_{torr). Thicknesses of thin ﬁlms were}

(3)

characteristics of LEC devices were measured using a source-measurement unit and a Si photodiode calibrated with the Photo Research PR-650 spectroradiometer. All device measurements were performed under constant bias voltages in a nitrogen glove box. The EL spectra were taken with a calibrated CCD spectrograph.

3. Results and discussions

PL spectra of the neat ﬁlm of complex 1 and absorption/ PL spectra of DOTCI in ethanol solution (105_{M) are shown}

inFig. 1. Neat ﬁlms of complex 1 exhibit red PL centered at 635 nm. DOTCI in ethanol solution exhibits concentrated NIR PL spectra centered at 720 nm. DOTCI shows intense absorption (molar extinction coefﬁcient > 104_M1_cm1₎

at the emission band of neat films of complex 1 and conse-quently efficient energy transfer between complex 1 and DOTCI would be expected to be feasible. Calculated Förster radius of complex 1/DOTCI host–guest system is ca. 4.8 nm. Large Förster radius confirms effective host–guest energy transfer due to high molar extinction coefficients of guests (DOTCI) and significant spectral overlap of host (complex 1) emission and guest absorption. Efficient host–guest energy transfer is essential for quenching resid-ual host emission at lower guest doping concentrations, which reduces self-quenching of excitons on guest mole-cules and enhances device efficiencies consequently.

The EL characteristics of the single-layered and tandem NIR LECs were measured and summarized inTable 1for comparison. The emissive layers of single-layered devices S1, S2 and S3 (weight ratio of complex 1 and DOTCI = 99:1) exhibit thicknesses of 250, 440 and 660 nm, respectively. The tandem devices are composed of two identical emis-sive layers vertically connected with a thin PEDOT:PSS layer (55 nm). The thickness of each emissive layer of tan-dem devices T1, T2 and T3 is 250, 440 and 660 nm, respec-tively. All tandem devices were operated under 5 V and single-layered devices S1, S2 and S3 were operated under 2.47, 2.45 and 2.45 V, respectively, to reach similar current densities as compared to their tandem counterparts. Time-dependent EL spectra of single-layered devices S1 (2.47 V), S2 (2.45 V) and S3 (2.45 V) are shown inFig. 2a–c, respec-tively. As shown inFig. 2a, predominant NIR EL emission centered at ca. 730 nm accompanied by some residual host emission was measured for device S1. The relative

intensity of the residual host emission increased with time and remained almost unchanged after ca. 2 h. It can be explained by the energy level alignments of the host and guest molecules [34] depicted in the inset of Fig. 2a. In host–guest LECs, doped layers result in ohmic contact with electrodes and consequently facilitate carrier injection onto both the host and the guest. Thus, both exciton forma-tion on the host followed by host–guest energy transfer and direct exciton formation on the guest induced by charge trapping contribute to the guest emission. At the early stage of operation, the doped layers have not yet well established and higher carrier injection barrier is present. Such energy level alignments favor carrier injection and trapping on the smaller-gap guests, resulting in direct car-rier recombination/exciton formation on the guest. There-fore, larger fractions of guest emission were observed initially. When the doped layers are getting well estab-lished, carrier injection onto the host and subsequent host–guest energy transfer would be facilitated, resulting in enhanced host emission. More significant residual host emission was observed in thicker device S2 (Fig. 2b). Car-rier injection onto the host and subsequent host–guest energy transfer would be less preferred in a thicker device under a slightly lower bias. Therefore, significantly enhanced residual host emission in a thicker device may result from modified output EL spectrum due to microcav-ity effect[34,35]. Constructive interference took place at the host emission band and the increasing temporal evolu-tion of host emission was amplified in consequence. For further thicker device S3 (Fig. 2c), the measured EL spectra were relatively less time-dependent. In addition, the host emission was blue-shifted and the guest emission was red-shifted, resulting in a deep valley at ca. 700 nm. The spacing between the cavity modes in a rather thick device would be smaller than that in a thinner device. If the emis-sive layer is thick enough, it would be possible for two or more cavity modes, at which constructive interference oc-curs, to appear at the visible spectral region. Shifts of emis-sion peak wavelengths and reshaping of the emisemis-sion spectra confirm that the output EL spectra are modified by microcavity effect (Fig. 2c). Evolution of EL spectra due to carrier trapping effect (Fig. 2a) would not result in such wavelength shift and spectral reshaping. Since both host and guest emissions are modified by interference, the temporal evolution of the EL spectra mainly comes from moving of recombination zone in the LEC device, which affects the relative amplitudes and spectral posi-tions of these two emission bands. Effect of carrier trap-ping on temporal evolution of the EL spectra is relatively minor in thicker device S3.

Time-dependent EL spectra of tandem devices T1 (5 V), T2 (5 V) and T3 (5 V) are shown inFig. 3a–c, respectively. The NIR EL emission of device T1 resembled that of its sin-gle-layered counterpart (device S1) while device T1 exhib-ited blue-shifted and narrowed residual host emission as compared to device S1 (cf.Figs. 2a and3a). Furthermore, enhancement of residual host emission with time was more signiﬁcant in device T1. It reveals that for device T1, constructive interference occurred at ca. 600 nm and thus resulted in blue-shifted, narrowed and enhanced EL emission there. For device T2, the NIR EL emission Fig. 1. PL spectra of the neat ﬁlm of complex 1 and absorption/PL spectra

of DOTCI in ethanol solution (105

(4)

blue-shifted with time and the residual host emission was signiﬁcantly reduced as compared to its single-layered

counterpart (device S2) (cf. Figs. 2b and 3b). It indicates that the recombination zones of device T2 moved with Table 1

Summary of EL characteristics of tandem NIR LECs. Devicea

Bias (V)b

Jmax(mA cm2)c tmax(min)d Lmax(lW cm2)e gext, max(%)f gp, max(mW W1)g

S1 2.47 0.36 91 5.83 1.04 6.99 S2 2.45 0.18 210 2.69 0.86 6.43 S3 2.45 0.14 402 2.56 1.14 8.16 T1 5.00 0.37 45 14.90 2.75 9.37 T2 5.00 0.17 303 6.82 2.19 8.02 T3 5.00 0.15 219 5.66 2.40 8.63 a

Device structures of devices S1, S2 and S3: ITO (120 nm)/PEDOT:PSS (30 nm)/emissive layer (weight ratio of complex 1 and DOTCI = 99:1)/Ag (100 nm), where the thicknesses of the emissive layers for devices S1, S2 and S3 are 250, 440 and 660 nm, respectively. Device structures of devices T1, T2 and T3: ITO (120 nm)/PEDOT:PSS (30 nm)/emissive layer (weight ratio of complex 1 and DOTCI = 99:1)/PEDOT:PSS (55 nm)/emissive layer (weight ratio of complex 1 and DOTCI = 99:1)/Ag (100 nm), where the thicknesses of the upper and lower emissive layers for devices T1, T2 and T3 are 250, 440 and 660 nm, respectively.

b

Bias voltage for each device to reach a similar current density.

c

Maximal current density.

d

Time required to reach the maximal light output.

e

Maximal light output achieved at a constant bias voltage.

f

Maximal external quantum efﬁciency achieved at a constant bias voltage.

g

Maximal power efﬁciency achieved at a constant bias voltage.

Fig. 2. Time-dependent EL spectra of single-layered devices (a) S1 (2.47 V), (b) S2 (2.45 V) and (c) S3 (2.45 V). Inset of (a) energy level diagram of the host and guest molecules in the NIR LECs.

Fig. 3. Time-dependent EL spectra of tandem devices (a) T1 (5 V), (b) T2 (5 V) and (c) T3 (5 V).

(5)

time such that the center wavelength of constructive inter-ference was blue-shifted. Since the blue-shifted center wavelength of constructive interference was moving toward the host emission band, the residual host emission was suppressed due to destructive interference taking place nearby the spectral region of constructive interfer-ence. For the thickest device T3, both the host and guest emissions were narrowed and blue-shifted with time (Fig. 3c). In addition, the relative amplitudes of the two emission bands also varied with time. Initially, both emis-sions exhibited comparable amplitudes (30 min, Fig. 3c) while the host emission dominated ﬁnally (600 min,

Fig. 3c). In thick devices, the spacing between two cavity modes is reduced and thus both the host and guest emis-sions are significantly modified by interference effect. Moving of the recombination zones in device T3 lead to shifting of cavity modes and resulted in spectral shift and alternation of relative amplitudes of the host and guest emissions consequently. These results show that the output spectra of the tandem NIR LECs are significantly influenced by the microcavity effect from device structures. The thicknesses of tandem NIR LECs should be properly chosen to obtain desired EL spectra.

Time-dependent light outputs of the single-layered and tandem NIR LECs with various thicknesses are compared in

Fig. 4a–c. The peak light outputs (

l

W cm2_{) of tandem}

de-vices T1, T2 and T3 were enhanced by factors of 2.56, 2.54 and 2.21, respectively, when compared to those of their single-layered counterparts (devices S1, S2 and S3, respec-tively) under similar current densities (Table 1). These re-sults confirm that both the upper and lower cells connected vertically in the tandem NIR LECs contributed to output EL emission. It is noted that all of the enhance-ment ratios were higher than 2, which reveals that at least one of the cells in the tandem NIR LECs exhibited higher efficiencies than the single-layered device. It would arise from improved carrier balance of the lower cell when the connecting PEDOT:PSS layer was used. In previously re-ported time-of-flight experiments, 2,20_{-bipyridine based}

ionic ruthenium complexes showed higher electron mobil-ities than hole mobilmobil-ities[36]. The upper cell of the tandem device possesses the same device structure as the single-layered device, e.g., PEDOT:PSS and Ag are used for anode and cathode, respectively. When the electrochemical doped layers facilitate balanced carrier injection, the number of electrons would be higher than that of holes. Furthermore, the doped guest DOTCI, which exhibits a much lower ionization potential than the host complex 1 (the inset ofFig. 2a), would act as a hole trapper and thus further reduces the number of holes. However, when a PEDOT:PSS layer is used as the cathode, electron injection into the lower cell of the tandem device would be impeded and carrier balance is improved consequently. As shown in

Fig. 5a–c, all tandem devices exhibited over doubled EQEs as compared to their single-layered counterparts. The enhancement ratios of EQEs in tandem devices T1, T2 and T3 are 2.64, 2.55 and 2.11, respectively (Table 1). These results conﬁrm that utilizing a connecting layer with a high electron injection barrier improves carrier balance of the lower cell of the tandem device and thus the overall device efﬁciency of the tandem device can be higher than twice of

the corresponding single-layered device. Modifying device efficiencies of LECs by tailoring carrier injection efficiencies has been reported previously[37]. Even with electrochem-ically doped layers, the ohmic contacts for carrier injection could be formed only when the carrier injection barriers were relatively low[37]. For LEC devices with a high carrier injection barrier, e.g., electron injection at the complex 1/ PEDOT:PSS interface, the doped layers reduce carrier injec-tion barrier but not result in a perfect ohmic contact. Therefore, the work function of the carrier injection layer would still affect carrier balance and device efficiencies.

The light outputs of the tandem NIR LECs are relatively low as compared to other reported NIR LECs. To further in-crease the light outputs, the constant current driving mode was tested in the tandem NIR LECs. When a constant cur-rent, which is the maximum value achieved in the constant voltage mode, was applied, the peak light outputs were enhanced by ca. 40% while the peak EQEs were reduced by ca. 15%. Furthermore, the device was turned on imme-diately under a constant current driving due to a high volt-age applied initially. The light output increased rapidly and Fig. 4. Light output as a function of time for devices (a) S1 (2.47 V) and T1 (5 V), (b) S2 (2.45 V) and T2 (5 V) and (c) S3 (2.45 V) and T3 (5 V).

(6)

reached the maximum within 1 min, suffering less material degradation. However, the slow increasing of light output of the LECs under constant voltage driving was accompanied by continuous material degradation and thus the peak light output was lower. Both the EQEs of the LECs under the two driving modes took place shortly after the driving was performed. Therefore, the peak EQEs of the LECs under constant voltage driving were higher due to lower initial current densities. When the tandem NIR LECs were operated under a constant current, rapid device response and higher light outputs would be obtained at the expense of lower device efﬁciencies.

Since the operation voltages of the tandem NIR LECs are higher than those of their single-layered counterparts, we have to examine the enhancements in power efﬁciencies for the tandem devices. In principle, both the light output and the operation voltage are doubled in a tandem device and thus the power efﬁciency (mW/W) is unchanged. However, the operation voltages of the tandem devices to

achieve similar current densities are usually higher than twice of those of the single-layered devices due to the additional voltage drop across the connecting layer. It would result in reduced power efficiencies in tandem devices consequently. As shown in Table 1, we can estimate a voltage of ca. 0.1 V across the connecting PED-OT:PSS layer in the NIR tandem LECs. Nevertheless, the peak power efficiencies obtained in the tandem NIR LECs are higher than those of their single-layered counterparts (Table 1). It can be attributed to that enhanced power effi-ciencies resulting from additionally increased (higher than twice) EQEs owing to improved carrier balance are higher than reduced power efficiencies due to slightly increased operation voltages. The additionally increased ratios in EQEs, which are defined as the measured EQEs of the tandem devices normalized to twice of the EQEs of their single-layered counterparts, are 1.32, 1.27 and 1.05 for devices T1, T2 and T3, respectively. These values are close to the enhancing ratios of power efficiencies of the tandem deices in comparison of their single-layered counterparts (1.34, 1.25 and 1.06, for devices T1, T2 and T3, respec-tively), confirming enhanced power efficiencies due to additionally increased EQEs.

The tandem NIR LECs are designed for NIR light sources, however, some residual red host emission is present in the output EL spectrum (Fig. 3). For device T1, the percentage of the NIR guest emission in the total EL emission is ca. 80% (Fig. 3a). With a proper dichroic filter to selectively pass the NIR EL emission, device T1 are still capable of offering efficient NIR EL with a peak EQE of 2.2%, which is 2.1 times the peak EQE of device S1 and 2.75 times the peak EQE of the previously reported single-layered NIR LEC based on the same NIR dye[25]. Therefore, over dou-bled enhancement in device efficiency can still be achieved in a tandem device structure when only NIR emission is considered.

It is interesting to note that the device lifetime of the thinner tandem device (T1) was significantly shorter than that of the thicker devices (T2 and T3) (cf.Fig. 4a–c). After ca. 9 h, tandem device T1 even showed lower light outputs than single-layered device S1 (Fig. 4a). Base on neat films of complex 1 (250 nm), deteriorated device lifetimes of tandem LECs as compared to single-layered LECs were also observed[31]. Such deteriorated device lifetimes in thin-ner tandem device (T1) would not likely result from degra-dation of the emissive material since no significant degradation was observed for all single-layered devices and thicker tandem devices under similar current densities during 10-h operation (Fig. 4a–c). Compared to devices T2 and T3, device T1 exhibited a higher current density (Table 1). However, no significant decay in light output un-der 10-h continuous operation was observed in device S1 (Fig. 4a), which showed a similar current density as device T1. It reveals that degradation of the emissive material is not significant under a low current density (ca. 0.36 mA cm2_{). Furthermore, as shown in} _{Fig. 6}_{, ca. 80%}

current density of device T1 retained after 10-h continuous operation while only 33% light output left ﬁnally. Hence, electrical degradation would not totally account for the decay in light output. Main degradation of light output would be related to moving of recombination zone and Fig. 5. External quantum efﬁciency as a function of time for devices (a) S1

(2.47 V) and T1 (5 V), (b) S2 (2.45 V) and T2 (5 V) and (c) S3 (2.45 V) and T3 (5 V).

(7)

subsequent exciton quenching near doped regions. Fitting of output EL spectra to simulated model concerning micro-cavity effect has been reported to extract temporal evolu-tion of recombinaevolu-tion zone in LECs under a constant bias

[34]. By employing this technique, a thicker LEC was shown to exhibit larger distances between the recombina-tion zone and the doped layers, rendering mitigated exci-ton quenching and enhanced device efficiency [24]. Similar methods can be adopted to study the effects of de-vice thickness on dede-vice lifetimes of tandem LECs. The emission properties of the emissive layer can be modified in a microcavity structure and the output EL spectrum of a bottom emitting tandem LEC can be calculated approxi-mately by using the following modified equation from Ref.[35]:

where R1and R2are the reﬂectance from the cathode and

from the glass substrate, respectively,

u

1and

u

2are the

phase changes on reﬂection from the cathode and from the glass substrate, respectively, T2 is the transmittance

from the glass substrate, L1, L2and L3are the optical

thick-nesses of the upper cell, connecting layer and lower cell, respectively, jEintðkÞj2 is the emission spectrum of the

organic materials without alternation of the microcavity effect, jEextðkÞj2is the output emission spectrum from the

glass substrate, ziand zjare the optical distances between

the emitting sublayer i and the cathode for the upper cell and between the emitting sublayer j and the connecting layer for the lower cell, respectively, c1and c2are the

emis-sion contribution weighting factors of the upper and lower cells, respectively. The emitting layers of the upper cell and lower cell are divided into N and M sublayers, respectively, and their contributions are summed up. Since the width of p–n junction estimated by capacitance measurements when p- and n-type layers were fully established was shown to be ca. 10% of the thickness of the active layer of LECs[38,39], the emitting layer width is estimated to be tenth of the active layer thickness. Thickness of each emitting sublayer of 1 nm was used and thus N

(M) = thickness of the upper cell (lower cell)/10. The PL spectrum of the structure of a tandem LEC coated on a quartz substrate was used as the emission spectrum with-out alternation of the microcavity effect since no highly reflective metal layer is present in this sample. Since the EL spectra of the NIR tandem LECs are composed of tempo-ral evolution of host and guest emission due to carrier trapping effect under electrical driving (Fig. 3a), it cannot be predicted by the model of microcavity effect and would affect fitting the simulated and measured EL spectra to extract recombination zone positions. Hence, tandem LECs based on neat films of complex 1 were employed alterna-tively to study thickness-dependent device lifetimes.

The simulated and measured EL spectra of tandem LECs possessing thinner (265 nm) and thicker (440 nm) thick-nesses of upper/lower emissive layers are compared in

Figs. 7 and 8, respectively. Bias voltages of 5.15 and 5.2 V were chosen for the thinner and thicker tandem devices, respectively, to achieve similar current densities. The times at which the EL spectra were recorded and the correspond-ing extracted recombination zone positions are labeled on the subﬁgures ofFigs. 7 and 8. z1is the distance between

the cathode and the recombination zone of the upper cell while z2is the distance between the connecting layer and

the recombination zone of the lower cell. Moving of recom-bination zones for the thinner and thicker tandem devices are schematically depicted inFig. 9a and b, respectively. For each emissive layer in the thinner tandem devices, the recombination zone was initially located closer to the cathode and moved toward the anode when the doped

layers were well established. The injection barrier for hole is much lower than that for electron for both the upper and lower cells of the tandem device (cf. the energy levels shown in the inset ofFig. 2a). After a bias was applied, the required number of accumulated mobile ions near the anode to achieve ohmic contact for hole was smaller than that required to achieve ohmic contact for electron at the cathode. Therefore, the hole injection efficiency was higher at the early stage of formation of electrochem-ically doped layers and the recombination zone would be relatively closer to the cathode consequently. When the doped layers were getting well established, balanced car-rier injection could be achieved and the recombination zone would move toward the center of the active layer. After the doped layers were well established, the positions of recombination zones were relatively fixed since a stable p–i–n structure has been formed[39]. It is noted that the distances between the recombination zones and the p-type doped layers for the thinner tandem LEC reduced with time and were less than 90 nm finally (Fig. 9a). Exciton quench-ing in the recombination zone close to the doped layer[39]

took place and resulted in deteriorated light output with time. As shown inFig. 10, the device lifetime of the thinner Fig. 6. Light output (solid symbol) and current density (open symbol) as a

function of time for device T1 (5 V).

jEextðkÞj2¼ T2 c11N XN i¼1 1 þ R1þ 2 ffiffiffiffiffiR1 p cos 4p k ðziÞ þ

u

1 þ c2M1 XM j¼1 1 þ R1þ 2 ffiffiffiffiffiR1 p cos 4p kðL1þ L2þ zjÞ þ

u

1 ( ) 1 þ R1R2 2 ffiffiffiffiffiffiffiffiffiffiR1R2 p cos 4p k ðL1þ L2þ L3Þ þ

u

1þ

u

2 jEintðkÞj2

(8)

tandem LEC, defined as the time it takes for the light out-put of the device to decay from the maximum to half of the maximum, is only ca. 160 min. For the thicker tandem LEC under a similar bias voltage, the electric field inside the device reduced, leading to lower speed for formation of the doped layers and thus slower device response (Fig. 10). Furthermore, the number of accumulated ions near elec-trodes under a lower electric field would be reduced and enhancing of electron injection from cathode would be sig-nificantly affected due to a high potential barrier (the inset ofFig. 2a). Hence, the recombination zone in the upper cell of the thicker tandem LEC was relatively closer to the cath-ode as compared to that in the upper cell of the thinner tandem LEC (Fig. 9a and b). For the lower cell of the thicker tandem LEC, an even higher potential barrier for electron injection at the PEDOT:PSS/complex 1 interface may im-pede formation of the n-type doped layer under a lower bias. Slow formation and reduced thickness of the n-type doped layer near high-work-function gold cathode was also observed in planar LECs[40]. A thicker intrinsic layer,

where a lower electric field is present, could thus be expected in the lower cell of the thicker tandem LEC. As the intrinsic layer shrank with time due to extension of the doped layers [39], field-dependent carrier mobilities would alter the recombination zone positions in the lower cell of the thicker tandem LEC. The electron mobilities of Ru complexes have been shown to decrease with increas-ing electric field[36]. Shrinking of the intrinsic layer under a constant bias resulted in increased electric field and decreased electron mobility consequently. Therefore, the recombination zone in the lower cell of the thicker tandem LEC moved toward the cathode with time (Fig. 9b). The sta-bilized recombination zone positions of the upper and low-er cells in the thicklow-er tandem LEC showed larglow-er distances away from the doped layers (>170 nm, Fig. 9b). Hence, exciton quenching would be mitigated and the device life-time of the thicker tandem LEC (ca. 270 min) was signifi-cantly longer than that of the thinner tandem LEC (ca. 160 min) under similar current densities (Fig. 10). With doped low-gap guest molecules (1 wt.%), which preferably trap holes (the inset ofFig. 2a), the recombination zones would be more closer to the anode as compared to those estimated in the host-only tandem LECs (Fig. 9). Exciton quenching near the p-type doped layers would be more se-vere in the thinner tandem host–guest devices due to short distances between the recombination zones and the p-type doped layers (Fig. 9a). For the thicker tandem host–guest devices, relatively longer distances between the recombi-nation zones and the p-type doped layers would reduce exciton quenching and longer device lifetimes would be rationally expected.

In our previous published work[34], we estimated the temporal evolution of recombination zone in single-lay-ered LECs by employing microcavity effect and attributed the reduced device efficiency to exciton quenching when the recombination zone is approaching the doped layers. By utilizing similar procedures, we studied the temporal evolution of recombination zone in tandem LECs and found that deteriorated device lifetimes also result from exciton quenching in the recombination zone near the doped lay-ers. For the first time, these results offer direct evidence for correlating deteriorated device lifetimes with exciton quenching due to moving of recombination zone toward the doped layers. This finding reveals that brightness decay of LECs would not totally result from chemical degradation of the emissive material and proper control of recombina-tion zone posirecombina-tions in LECs would be a feasible way to improve device lifetime.

Previously reported literatures about NIR LECs mainly focused on developing new NIR materials [4–8], but the device efficiencies were moderate. The main contribution of this work is proposing a combinational strategy employ-ing existemploy-ing technologies, i.e., phosphorescent sensitized fluorescence and tandem device structure, to achieve tripled enhancement in device efficiencies of the NIR LECs. Although both approaches are not new, the combination of them is undoubtedly the first demonstration. No other NIR-LEC technologies ever reported so far could offer such a high efficiency enhancement. This work is especially valuable for NIR LECs since ionic NIR materials generally suffer low efficiencies due to energy gap law. An Fig. 7. Simulated (solid symbol) and measured (open symbol) EL spectra

of the thinner tandem LEC (265 nm for the upper and lower cells) at (a) 66, (b) 132 and (c) 263 min after a bias of 5.15 V was applied. The extracted recombination zone positions are shown in each subﬁgure.

(9)

alternatively way to enhance device efﬁciencies by employing device technologies is necessary for this type of LECs.

4. Conclusions

In summary, we have demonstrated phosphorescent sensitized fluorescent NIR LECs in a tandem device struc-ture. Two emissive layers containing a phosphorescent CTMC host doped with a fluorescent ionic NIR dye were connected vertically via a thin PEDOT:PSS layer. The output EL spectrum of the tandem NIR LEC was found to change as the thickness of emissive layer varies due to altered micro-cavity effect. In addition, by fitting the output EL spectra to the simulated model concerning microcavity effect, tem-poral evolution of recombination zones in the tandem LECs can be extracted. When the doped layers were well devel-oped, the recombination zones in a thicker tandem device were closer to the center of the emissive layers while those in a thinner tandem device approached the p-type doped layers. Thus, improved device lifetimes in thicker tandem devices can be rationally attributed to reduced exciton quenching near the doped layers. The peak EQEs obtained in these tandem NIR LECs were up to 2.75%, which are over doubled as compared to those of their single-layered coun-terparts due to improved carrier balance induced by the connecting layer with a high electron injection barrier. Compared to previously reported single-layered NIR LECs based on the same NIR dye[25], over tripled enhancement in device efficiency was achieved in this work. These results confirm that phosphorescent sensitized fluoresce combined with a tandem device structure would be useful for realizing highly efficient NIR LECs.

Fig. 8. Simulated (solid symbol) and measured (open symbol) EL spectra of the thicker tandem LEC (440 nm for the upper and lower cells) at (a) 63, (b) 189 (c) 315 and (d) 505 min after a bias of 5.2 V was applied. The extracted recombination zone positions are shown in each subﬁgure.

Fig. 9. Schematic illustration of moving of recombination zones in (a) the thinner tandem LEC (265 nm for the upper and lower cells) under 5.15 V and (b) the thicker tandem LEC (440 nm for the upper and lower cells) under 5.2 V.

Fig. 10. Light output as a function of time for the thinner (265 nm for the upper and lower cells) and thicker (440 nm for the upper and lower cells) LECs under 5.15 and 5.2 V, respectively.

(10)

Acknowledgement

The authors gratefully acknowledge the ﬁnancial sup-port from the National Science Council of Taiwan. References

[1]G. Qian, Z.Y. Wang, Chem. Asian J. 5 (2010) 1006.

[2]Q. Pei, G. Yu, C. Zhang, Y. Yang, A.J. Heeger, Science 269 (1995) 1086. [3]Q. Pei, Y. Yang, G. Yu, C. Zhang, A.J. Heeger, J. Am. Chem. Soc. 118

(1996) 3922.

[4]A.R. Hosseini, C.Y. Koh, J.D. Slinker, S. Flores-Torres, H.D. Abruña, G.G. Malliaras, Chem. Mater. 17 (2005) 6114.

[5]H.J. Bolink, L. Cappelli, E. Coronado, P. Gaviña, Inorg. Chem. 44 (2005) 5966.

[6]S. Xun, J. Zhang, X. Li, D. Ma, Z.Y. Wang, Synth. Met. 158 (2008) 484. [7]H.J. Bolink, E. Coronado, R.D. Costa, P. Gaviña, E. Ortí, S. Tatay, Inorg.

Chem. 48 (2009) 3907.

[8]S. Wang, X. Li, S. Xun, X. Wan, Z.Y. Wang, Macromolecules 39 (2006) 7502.

[9]J.D. Slinker, D. Bernards, P.L. Houston, H.D. Abruña, S. Bernhard, G.G. Malliaras, Chem. Commun. (2003) 2392.

[10]J.D. Slinker, A.A. Gorodetsky, M.S. Lowry, J. Wang, S. Parker, R. Rohl, S. Bernhard, G.G. Malliaras, J. Am. Chem. Soc. 126 (2004) 2763. [11]A.B. Tamayo, S. Garon, T. Sajoto, P.I. Djurovich, I.M. Tsyba, R. Bau,

M.E. Thompson, Inorg. Chem. 44 (2005) 8723.

[12]Q. Zhang, Q. Zhou, Y. Cheng, L. Wang, D. Ma, X. Jing, F. Wang, Adv. Funct. Mater. 16 (2006) 1203.

[13]H.-C. Su, F.-C. Fang, T.-Y. Hwu, H.-H. Hsieh, H.-F. Chen, G.-H. Lee, S.-M. Peng, K.-T. Wong, C.-C. Wu, Adv. Funct. Mater. 17 (2007) 1019. [14]H.-C. Su, C.-C. Wu, F.-C. Fang, K.-T. Wong, Appl. Phys. Lett. 89 (2006)

261118.

[15]J.D. Slinker, J. Rivnay, J.S. Moskowitz, J.B. Parker, S. Bernhard, H.D. Abruña, G.G. Malliaras, J. Mater. Chem. 17 (2007) 2976.

[16]H.-C. Su, H.-F. Chen, F.-C. Fang, C.-C. Liu, C.-C. Wu, K.-T. Wong, Y.-H. Liu, S.-M. Peng, J. Am. Chem. Soc. 130 (2008) 3413.

[17]H.J. Bolink, E. Coronado, R.D. Costa, N. Lardiés, E. Ortí, Inorg. Chem. 47 (2008) 9149.

[18]L. He, J. Qiao, L. Duan, G.F. Dong, D.Q. Zhang, L.D. Wang, Y. Qiu, Adv. Funct. Mater. 19 (2009) 2950.

[19]H.-C. Su, H.-F. Chen, Y.-C. Shen, C.-T. Liao, K.-T. Wong, J. Mater. Chem. 21 (2011) 9653.

[20] C.-T. Liao, H.-F. Chen, H.-C. Su, K.-T. Wong, J. Mater. Chem. 21 (2011) 17855.

[21]H.-B. Wu, H.-F. Chen, C.-T. Liao, H.-C. Su, K.-T. Wong, Org. Electron. 13 (2012) 483.

[22]T. Hu, L. He, L. Duan, Y. Qiu, J. Mater. Chem. 22 (2012) 4206. [23]R.D. Costa, E. Ortí, H.J. Bolink, F. Monti, G. Accorsi, N. Armaroli,

Angew. Chem. Int. Ed. 51 (2012) 8178.

[24]Y.-P. Jhang, H.-F. Chen, H.-B. Wu, Y.-S. Yeh, H.-C. Su, K.-T. Wong, Org. Electron. 14 (2013) 2424.

[25]C.-C. Ho, H.-F. Chen, Y.-C. Ho, C.-T. Liao, H.-C. Su, K.-T. Wong, Phys. Chem. Chem. Phys. 13 (2011) 17729.

[26]M.A. Baldo, M.E. Thompson, S.R. Forrest, Nature (London) 403 (2000) 750.

[27]H.-C. Su, Y.-H. Lin, C.-H. Chang, H.-W. Lin, C.-C. Wu, F.-C. Fang, H.-F. Chen, K.-T. Wong, J. Mater. Chem. 20 (2010) 5521.

[28]H.-C. Su, H.-F. Chen, P.-H. Chen, S.-W. Lin, C.-T. Liao, K.-T. Wong, J. Mater. Chem. 22 (2012) 22998.

[29]T. Förster, Discuss. Faraday Soc. 27 (1959) 7.

[30] P.F. Aramendia, R.M. Negri, E.S. Roman, J. Phys. Chem. 98 (1994) 3165.

[31]J.-S. Lu, J.-C. Kuo, H.-C. Su, Org. Electron. 14 (2013) 3379. [32] T. Akatsuka, C. Roldán-Carmona, E. Ortí, H.J. Bolink, Adv. Mater.

http://dx.doi.org/10.1002/adma.201303552.

[33]S. Bernhard, J.A. Barron, P.L. Houston, H.D. Abruña, J.L. Ruglovksy, X. Gao, G.G. Malliaras, J. Am. Chem. Soc. 124 (2002) 13624. [34]T.-W. Wang, H.-C. Su, Org. Electron. 14 (2013) 2269.

[35]X. Liu, D. Poitras, Y. Tao, C. Py, J. Vac. Sci. Technol. 22 (2004) 764. [36]W.K. Chan, P.K. Ng, X. Gong, S. Hou, Appl. Phys. Lett. 75 (1999) 3920. [37]C.-T. Liao, H.-F. Chen, H.-C. Su, K.-T. Wong, Phys. Chem. Chem. Phys.

14 (2012) 9774.

[38]I.H. Campbell, D.L. Smith, C.J. Neef, J.P. Ferraris, Appl. Phys. Lett. 72 (1998) 2565.

[39]M. Lenes, G. Garcia-Belmonte, D. Tordera, A. Pertegás, J. Bisquert, H.J. Bolink, Adv. Funct. Mater. 21 (2011) 1581.