Article pubs.acs.org/Langmuir
Factors Affecting the Stability and Performance of Ionic Liquid-Based Planar Transient Photodetectors Simon Dalgleish,*,†,∥ Louisa Reissig,†,‡ Laigui Hu,†,⊥ Michio M. Matsushita,† Yuki Sudo,‡,§,# and Kunio Awaga*,†,§ †
Department of Chemistry and Research Centre for Material Science, ‡Division of Biological Science, Graduate School of Science, and §CREST, JST, Nagoya University, Furo-cho, Chikusa, 464-8602 Nagoya, Japan ∥ Institute for Advanced Research, Nagoya University, Furo-cho, Chikusa, 464-8601 Nagoya, Japan ⊥ Department of Applied Physics, Zhejiang University of Technology, Hangzhou 310023, China # Division of Pharmaceutical Sciences, Okayama University, 1-1-1 Tsushima-naka, Kitaku, 700-8530 Okayama, Japan S Supporting Information *
ABSTRACT: A novel planar architecture has been developed for the study of photodetectors utilizing the transient photocurrent response induced by a metal/insulator/semiconductor/metal (MISM) structured device, where the insulator is an ionic liquid (IL-MISM). Using vanadyl 2,3naphthalocyanine, which absorbs in the communicationsrelevant near-infrared wavelength region (λmax,film ≈ 850 nm), in conjunction with C60 as a bulk heterojunction, the high capacitance of the formed electric double layers at the ionic liquid interfaces yields high charge separation efficiency within the semiconductor layer, and the minimal potential drop in the bulk ionic liquid allows the electrodes to be offset by distances of over 7 mm. Furthermore, the decrease in operational speed with increased electrode separation is beneficial for a clear modeling of the waveform of the photocurrent signal, free from the influence of measurement circuitry. Despite the use of a molecular semiconductor as the active layer in conjunction with a liquid insulating layer, devices with a stability of several days could be achieved, and the operational stability of such devices was shown to be dependent solely on the solubility of the active layer in the ionic liquid, even under atmospheric conditions. Furthermore, the greatly simplified device construction process, which does not rely on transparent electrode materials or direct electrode deposition, provides a highly reproducible platform for the study of the electronic processes within IL-MISM detectors that is largely free from architectural constraints.
■
INTRODUCTION The field of light-to-energy conversion by organic materials for application as alternative energy sources or for light detection for imaging or communication is highly active and still growing.1−3 Although inorganic materials, especially silicon technology, still dominate the market for both applications, the benefits of organic materials, such as the possibility for low-cost, lightweight, flexible, and disposable devices as well as the fact that there is still much scope for scientific improvements, have supported their continued study. We have previously reported on the transient photocurrent response of metal/insulator/semiconductor/metal (MISM) structured devices to incident light.4−7 This device architecture, which in its simplest representation, can be modeled as a photodiode in series with a capacitor (cf. Figure 2c), yields a time-varying photocurrent response, comprising an exponential rise (originating from the generation of a photovoltage in the semiconductor layer) and a decay (originating from the charging of the capacitive element by the generated photovoltage). When the light is off, the capacitor discharges, with a © 2015 American Chemical Society
characteristic inversion of the current signal. This time-varying response, although not being suitable for steady-state applications such as photovoltaics, is applicable to communications where information is transferred in the form of a periodic light signal, with the on and off periods being translated by the MISM device to a time-varying current signal of different polarities. Furthermore, it has been shown that as the frequency of the incident light signal is increased, the shape of the current response changes from peak/decay to a shape resembling a square wave and finally decaying once the frequency exceeds the rise time of the photocurrent transient.5 In addition, these studies have also shown that the peak photocurrent response and therefore the device responsivity are directly related to the capacitance of the insulator layer,4 and by choosing an insulator with a high dielectric constant, the photocurrent response can significantly exceed that of convenReceived: December 23, 2014 Revised: March 24, 2015 Published: April 20, 2015 5235
DOI: 10.1021/la504972q Langmuir 2015, 31, 5235−5243
Article
Langmuir
studies on the properties of the photocurrent response. This study provides a platform for future detailed analysis of the photocurrent waveform from MISM devices and shows ionic liquids to be promising candidates for MISM photodetectors due to the high responsivities achieved without external bias and a greatly simplified device fabrication process. For the purposes of this study, vanadyl naphthalocyanine (VONPc) was chosen to serve as the chromophore (in conjunction with C60, thereby setting up a bulk heterojunction) due to its high absorptivity and high stability to dissolution. In addition, its peak absorption (λmax = 850 nm, Figure 2a) is significantly red-shifted from the intense Q band of conventional phthalocyanines (λmax = 600−700 nm)11,12 on account of its greater degree of conjugation and its nonplanar structure, caused by the large vanadyl group in the porphyrazine pocket.13−15 This makes VONPc an attractive target for communication applications, such as for short-range fiberoptic data transmission, which is currently operated in the first telecommunications window (800−900 nm), where the signal attenuation in the fibers is acceptably low so as to allow for the use of established and high-performance silicon devices. Additionally, near-infrared (NIR) chromophores are of interest for photovoltaic application, where extending light harvesting further into the red should reap significant gains in cell efficiency due to better spectral matching with incident solar light.16,17
tional metal/semiconductor/metal (MSM) devices. To this end, ionic liquids (ILs) have shown good applicability to MISM devices6,7 on account of the high capacitance of the electric double layers formed at the interfaces with the ionic liquid.8−10 Although the above model can help to rationalize these trends, it does not account for the various nonidealities in the generation of the photovoltage, such as the balance of carrier mobilities, and contact effects at the metal/semiconductor interface, which should cause significant deviations from a single-exponent rise. To elucidate the origins of these deviations and thereby further improve the performance of such devices in terms of responsivity and speed, it is necessary to develop a platform for the analysis of the waveform where aspects such as contact resistance and the nature and composition of the semiconductor layer can be controlled independently of architectural constraints. In this article, we build on our previous results to show that IL-MISM photodetectors can also be fabricated in a planar architecture (Figure 1a), where the electrodes can be
■
MATERIALS AND METHODS
VONPc and C60 (99.9%) (Figure 2b) were purchased from SigmaAldrich and were used as received. All the ionic liquids used were purchased at ≥98% purity and used as received: 1-ethyl-3methylimidazolium bis(trifluoromethylsulfonyl)imide (EMIM-TFSI, ET) (Kanto), 1-butyl-3-methylimidazolium tetrafluoroborate (BMIM-BF4, BB) (Wako), 1-methyl-3-octylimidazolium hexafluorophosphate (OMIM-PF6, OP) (Wako), EMIM-BF4 (EB) (Kanto), BMIM-PF6 (BP) (TCI), BMIM-TFSI (BT) (Wako), and N,N-diethylN-methyl-N-(2-methoxyethyl)ammonium TFSI (DEME-TFSI, DT) (Kanto). Device Fabrication. Devices were prepared on glass slides with dimensions of 26 × 18 × 1 mm3, precleaned by mild bath ultrasonication (45 W, isopropanol, acetone, chloroform, 10 min each). The slides were mounted on a shadow mask, defining electrodes of 2 × 26 mm2 with 4 mm separation and inserted into a physical vapor deposition (PVD) apparatus (ULVAC). The chamber was evacuated to an initial vacuum of 4 × 10−4 Pa prior to deposition. Silver (Ag, 200 nm) and gold (Au, 200 nm + 25 nm chromium adhesion layer) electrodes were deposited sequentially at rates of 0.5 and 0.2 Å s−1 (measured by quartz crystal microbalance (QCM)) for Ag and Cr/Au, respectively, alternately masking one electrode with Kapton tape for each electrode deposition. For devices of electrode separations of 1, 3, and 7 mm, the substrates were repositioned on the mask before the deposition of the second electrode. The substrates were used immediately for active-layer deposition without removal from the shadow mask. In each case, the active layer was deposited onto the gold electrode at an initial vacuum of 2 × 10−4 Pa. Because the window for efficient sublimation of VONPc was narrow (380−430 °C; prolonged heating or increased temperature caused a change in morphology of the material, which sublimed poorly, though the chemical identity by FTIR was the same; see ESI Figure S1), fresh material (40 mg) was used for each deposition to ensure good thermal contact with the ceramic boat used. The narrow window for sublimation limited the deposition rate to 0.1 Å s−1, which slowed slightly during the course of the deposition. For the codepositions of VONPc and C60, each material was loaded into adjacent launchers, with their deposition rates monitored independently by different QCMs. Because of the slow rate of deposition (being at the resolution limit of the QCMs),
Figure 1. (a) Profile view of the novel planar device architecture and experimental setup. (b) Dependence of the total amplitude (max− min) of the initial photocurrent response on interelectrode spacing for EMIM-BF4 device (inset) transient photocurrent waveforms at different electrode spacings, measured at 100 Hz.
prefabricated on the same substrate and separated over a macroscopic distance of more than 7 mm (Figure 1b). Such devices benefit from inherently improved reproducibility and are largely independent of architectural constraints, such as electrode choice and active-layer thickness. For such devices, the operational stability is shown to be solely dependent on the solubility of the semiconductor layer in the ionic liquid, and even molecular systems, which have the benefit of being able to be fabricated as thin films of high purity by vapor methods, with precise control of thickness and composition, can show a stable and reproducible photoresponse over days under constant operation. Additionally, in some cases, the thin layer of ionic liquid can help to prevent atmospheric degradation of the active layer without any additional encapsulation, allowing extensive 5236
DOI: 10.1021/la504972q Langmuir 2015, 31, 5235−5243
Article
Langmuir
Figure 2. (a) Absorption spectra of VONPc and C60 as pure films (40 nm), a blend (80 nm, 60% C60), and a bilayer film (30 nm VONPc, 40 nm C60); (b) molecular structure of VONPc and C60; (c) schematic of the energy levels of VONPc and C60 in the device and simplified equivalent circuit; (d) output characteristics of a bottom-contact (Au), bottom-gate (Si/SiO2) VONPc transistor, showing p-type operation. methanol washing, the C60 film showed some deterioration following the methanol wash. However, priority was given to the data of VONPc because the role of C60 is the formation of a heterojunction and not light-to-energy conversion and could thus be replaced by another, more stable, n-type material in the future. To isolate the treated region of the film for analysis, absorption measurements were recorded through a perforated plate (ϕ = 6.5 mm) fixed in the path length. Device Characterization. The time-dependent transient photocurrent measurements were performed using a fiber-coupled LED light source (λmax = 850 nm, Pmax,incident ≈ 2.2 mW cm−2) (Figure 1a), powered by a home-built driver circuit (circuit diagram available upon request) and modulated by a function generator (Tektronix AFG320). The devices were connected to an inverting transimpedance amplifier (Keithley 428) in which the transient short circuit current signal was amplified and converted into a voltage signal, which was visualized with an oscilloscope (Tektronix TDS5104B). The actual light power density was determined before each data set using a silicon photodiode (FID08T13TX), calibrated by an optical power meter (Ophir NovaII). Per convention, all data were not inverted: a positive peak corresponds to holes being extracted from the active layer at the gold electrode, and the first peak in a photocurrent plot corresponds to light on. The error in establishing the amplitude of the response was less than 5−10% of the initial value. However, the poor reproducibility of the film composition increases the error to about 15% when comparing different batches. The error bars were omitted from the figures for clarity. For the frequency response measurements, the devices were connected to a home-built high-bandwidth current amplifier (circuit diagram available upon request) (active chip: Texas Instruments OPA657, 1.6 GHz bandwidth) and converted to a voltage signal (gain = 104), which was recorded and averaged by the oscilloscope. Electronic Characterization. Transistor measurements were performed on 50 nm films of VONPc deposited onto bottom-contact, bottom-gate (BCBG) interdigitated Au array electrodes (source and drain) with a channel width of W = 4 cm and a channel length of L = 50 μm, which were prepared by standard photolithographic methods
reproducibility between batches was difficult to achieve (vide infra). Each device was coupled with a quartz substrate for composition analysis by UV/vis absorption spectroscopy and thickness analysis because the homogeneity of film deposition within a batch was not ideal. To compare the reproducibility of device performance and the effects of different electrode separations, i.e., where identical devices were necessary, a home-built rotating substrate holder (1 rpm) was employed. The densities of the materials used for rate and thickness estimation were determined from the QCM readout, calibrated against thickness measurements performed using surface profilometry (Dektak 150). Accurate thickness and composition analyses were measured ex situ following each deposition, with the relative amounts of the components in the films additionally estimated by the rough deconvolution of their thin film absorption spectra (vide infra). In general, relative amounts of over 40% VONPc content could not be obtained under the above conditions because of the slow deposition rate of VONPc. Following active-layer deposition, the devices were completed with a glass coverslip, fixed with a thermally sealed 60 μm Surlyn spacer (Solaronix). The void was filled with ionic liquid by capillary force through predrilled holes in the coverslip and tested immediately (Figure 1a, Figure S2 ESI). Active-Layer Characterization. Ultraviolet−visible absorption spectra of all films were recorded on a Shimadzu UV-3100PC spectrophotometer. Measured films were fabricated under conditions identical to those for the devices on precleaned quartz substrates (ultrasonication sequence the same as above). For the dissolution experiments, the active layer was deposited on small quartz substrates (10 × 10 × 1 mm3) to a thickness of
