Large Active Area Organic Photodiodes for Short-Pulse X-ray Detection
Organic thin film light sensors are promising devices for X-ray imaging systems. Compared to crystalline silicon photodiodes (c-Si), organic sensors can be fabricated on large active area at low cost. Furthermore, organic semiconductors have the advantage of low X-ray absorption. Here, we show that despite the high diode capacitance of several nF/cm/sup2sup/, a single X-ray pulse detection as low as ∼14 µGy can be detected in the ms regime. Such device properties match industrial requirements for X-ray sensing.
I. INTRODUCTION
Organic electronics is attractive for fabrication of solution-processed photodiodes at low temperatures with large active areas even on flexible substrates with a variety of geometrical degrees of freedom [1]. In addition, the spectral sensitivities of the devices can be tailored with the organic semiconductor absorber for the respective application [2]. Such low-effort and low-cost processing conditions are of interest for sensing with >1 cm2 organic photodetectors as potential replacement of silicon photodiodes [1] [2] since large active areas of several cm2 are difficult to achieve with solid-state devices. In particular organic photodetectors are of high interest in medical applications due to the low X-ray absorption as a result of the thin film architecture. Due to optical limitations to focus X-ray photons on small areas, devices with large active areas of several cm2 are needed for high sensitivity. High external quantum efficiencies (EQE) and low dark currents are also required to detect extremely low X-ray doses. In addition, devices should show high dynamic responses to detect short X-ray pulses in the ms range and low degradation in order to become a final product. Devices with such characteristics are useful in a variety of areas. For instance, to detect low dose single X-ray pulses in medical sensing with a temporal resolution of less 10 ms, arrays of large active area photodiodes are of interest. To fabricate an organic device covering characteristics such as large active area, low dark current, high EQE, fast dynamic response, high reproducibility and no degradation by the X-ray photons, is challenging. In previous reports bulk heterojunction (BHJ) [3] based organic photodiodes (OPDs) have shown appropriate low dark current densities of ~10-4 mA cm-2 at field strengths > 10 V µm-1, high EQE > 70 % in the visible range, good reproducibility, and sufficient lifetime for industrial products [1] [2] [4]. However, compared to c-Si detectors, dynamic behavior remains a major challenge with organic sensors due to the low mobility of the charge carriers and the high capacitance of the device as result of the thin-film architecture. The large capacitance C of the thin-film devices with thicknesses of less 1 µm is an important limiting parameter of the diode bandwidth. OPDs with a BHJ thickness of ~1 µm and 5 cm2 active areas have capacitances C and shunt resistances RSH of ~23 nF and ~100 MΩ, respectively. The load resistance RL is usually set to 50 Ω for impedance matching to the connected circuit. The maximum cut-off frequency for the OPD only (f = (1/2πCRSHL) with RSHL = RSH//RL) [5] corresponding to a decrease of the voltage amplitude by 2-1/2 is calculated to be ~140 kHz though the actual bandwidth is usually lower due to the low charge carrier mobility, charge trap phenomena, series resistance and capacitances of wire connections. In brief, we demonstrate OPDs on large active areas of 5 cm2 with dark current densities of ~1 10-5 mA cm-2 at -5 V reverse bias, EQE > 70 % in the visible range, fast dynamic response in the ms range under X-ray excitation doses as low as 14 µGy and high reproducibility. We present also a concept for a compact read-out electronic to drive a large active area OPDs array and we demonstrate that despite the large capacitance of each diode, a response time in the ms range can be achieved both under low light illumination condition as low as 50 nW cm-2 and X-ray excitations of a few µGy s-1. Results match for the first time medical requirements for X-ray imaging.
II. Organic Photodiodes
Fig. 1(a) shows schematically the layer stack of an OPD. The absorption of an incident photon in the semiconductor results in the formation of an exciton that dissociates in the semiconductor layer at the donor/acceptor interfaces. The generated charges are collected at the electrodes. Holes are extracted at the anode and electrons at the cathode. To achieve a high rectification ratio of the photodiode together with a low dark current, electrodes of asymmetrical work function are needed [6]. The most common used anode material for organic photodetectors and organic solar cells is the indium-tin-oxide (ITO) with high work function of ~4.7 eV [7] while as interlayer (IL) is prevalently used PEDOT:PSS (work function of ~5.1 eV [8]). In 2009 Liang et al. reported that the use of poly(3-hexylthiophene) (P3HT) as IL results in a device with lower dark current and higher quantum efficiency compared to a reference device [9]. They use a multi-spin coating method to fabricate the P3HT/semiconductor layers. Although this method is very simple, it seems to be a non-quantitative coating method because the precoated P3HT film can be dissolved by the co-solvent of the photoactive solution [10]. Recently Oh et al. [10] reported an improvement of the stack were to prevent the dissolution of the pre-coated P3HT layer by the co-solvent, the semiconductor film is deposited onto the P3HT layer by a tricky transfer printing method. We use P3HT as IL due to the low dark current of the resulting device. OPDs are fabricated on 5x5 cm2 glass substrates (with high transmittance of ~90 %) coated by ITO. A substrate with low surface roughness is necessary to avoid short circuits between electrodes due to spikes. In order to further reduce the dark current value, ITO rims are passivated by ~1 µm SU-8 photoresist and a square-shaped photoactive area of 5 cm2 of the anode is define by photolithography. After photoresist hard backing (200 °C for 10 min on a hotplate) substrate is cleaned in acetone, isopropanol and deionized water and a RIE plasma is applied to activate the ITO surface prior to the spray-coating [4] of the interlayer and the semiconductor layers. First, a solution of 1% wt. P3HT in xylene is spray-coated on the ITO surface as electron blocking interlayer for the OPD. Second, a BHJ layer is spray-coated on top of the interlayer from a xylene solution of P3HT and [6,6]-phenyl C61 butyric acid methyl ester (PCBM) with a weight ratio of 1:0.75. Both IL and BHJ are deposited in ambient conditions. To obtain a low intermixing between interlayer and semiconductor we set-up the spray-coating gun to generate aerosol droplets with diameter < 10 µm and we verify that semiconductor particles arrived to the substrate dried. Surface profilometry is used to characterize the average layer thicknesses of ~70 nm for the P3HT interlayer and ~1 µm for the BHJ. Third, samples are annealed at 140 °C for 5 min in nitrogen atmosphere before the thermal evaporation of 100 nm of aluminium as top electrode (low work function of ~4.1 eV [11]). Last, diodes are encapsulated with transparent solvent-free UV hardening epoxy glue and a 100 µm thick transparent glass slide. Encapsulation is necessary to protect the diode from degradation, for instance due to semiconductor oxidation. In Fig. 1(b) an exemplary current-voltage measurement (IV) of a 5 cm2 diode is shown with dark current and photocurrent density under green illumination (532 nm wavelength). The photocurrent is measured through the transparent ITO electrode using a high power LED with an irradiation intensity of 780 µW cm-2. Despite the high surface roughness of the spray-coated semiconductor [4] and the large active area of 5 cm2, the diode shows at -5 V reverse bias a low dark current density of ~1 10-5 mA cm-2 or absolute current of ~53 nA, corresponding to a shunt resistance ~100 MΩ and a rectification ratio ~2.6 105 at + / -1 V. The series resistance of the diode is ~200 Ω at +2 V and an EQE of ~70% in the visible range is observed, Fig. 1(c), which is comparable to reported values [4]. The EQE reduction at ~500 nm wavelength is related to both the absorption of the P3HT IL and the ~1 µm BHJ thickness [12]. Array fabrications require a high reproducibility of the thin film detectors. We fabricated several OPDs on different substrates to characterize the reproducibility of the process. All devices showed comparable IV and EQE. As an example Fig. 1(d) shows dark current density statistics at -2 V and -5 V reverse bias of twelve OPDs with an active area of 5 cm2 each. Finally, in Fig. 2(a) OPD dark current densities with varying temperature are presented showing low dark currents at working temperatures (< 30 °C). The forward and backward current density variation with temperature is show in Fig. 2(b) for -3 V reverse bias.
When we ramp the temperature up and down we observe an exponential correlation between temperature and dark current. We operate at temperatures close to room temperature to ensure minimum dark current densities of ~1 10-5 mA cm-2. To perform single pulse X-ray detection a conversion layer of terbium-doped gadolinium oxysulfide (Gd2O2S:Tb, GOS) of ~12 µm thickness (~4 µm net thickness due to fill factor of ~33 %) is used, Fig. 1(a). The GOS scintillator particles are sedimentated on a transparent glass from an isopropanol dispersion and coupled to the transparent anode of the encapsulated OPD with poly-dimethyl siloxane (PDMS). The green fluorescence emission at 545 nm of the GOS film matches the spectral sensitivity of the photodiode. To irradiate the OPD X-ray pulses are generated with a commercial X-ray source. A variation of the X-ray generator anode current from 150 mA to 2 mA corresponds to a generated X-ray dose rates from ~90 µGy s-1 to ~3 µGy s-1. The GOS conversion layer at such low X-ray radiation level emits a green fluorescence in the low nW cm-2 range. Diodes are reverse biased and current vs. time acquired with a Keithley 2400 SMU. Fig. 3 shows exemplary photoresponse of an OPD reverse biased and irradiated by X-ray photons at 70 kV energy with dose rates ranging from 90 µGy/s to 3 µGy/s. Note that the OPD shows low current drift and the current noise can be neglected. The lowest detectable dose is ~3 µGy s-1. Due to hardware limitation to acquire more than ~12 datapoints/sec (distance between datapoints of ~80 ms), from these measurements photocurrent rise/fall time cannot be measure.
III. Read-out electronic
In order to fully characterize the response of the OPDs to short-pulse X-ray, a read-out electronic circuit to read the OPD signals with high sampling rate is required. An important problem to solve is the high dark current load of 10 – 100 nA (depending on the reverse polarization) that restricts the output signal swing. We decided to compensate the dark current value to zero implementing a negative feedback amplifier in closed loop configuration. The principle of the negative feedback is that the output signal is fed-back to the input and combined with the input signal in order to control the output of the loop. In this way the board output is null when no X-ray pulse is applied and we can use the complete output signal ranging from zero to the voltage supply for the photocurrent measurement. The dark current compensation has been implemented both analog and digital. A user defined polarization voltage between -2 V and +2 V, Vp in Fig. 4(a) and (b), is applied to the cathode electrode and the diode current signal is read from the anode.
Current coming from the diode, ID in Fig. 4(a) and (b), is amplified and converted to a voltage signal by a transimpedance operational amplifier (OP-amp1 in Fig. 4(a) and Fig. 4(b)). The OP-amp1 feedback resistance is chosen by a switch resulting in a variable gain ranging from 103 to 106 V A-1. For the digital solution, Fig. 4(a), we first measure the output signal of the OP-amp2 (LMP7721) with a standard 20 Bit-ADC (ADS1248IPWR), scale it with the CPU and than we apply the result to a 16 Bit-DAC (DAC8164IAPW). The result is sent to the non-inverting input of the difference amplifier OP-amp2 through OP-amp3. Operation is iterated until the output of OP-amp2 is smaller than 0.3 mV. A reference signal (Ref, Fig 4(a)) is used to keep the unipolar output of the DAC (0 to 2.0 V) to a bipolar value (-2.0 V to +2.0 V). Digital board is equipped with a CPU from Philips (ARM-Cortex-M3 LPC1764FBD100) with 120 MHz clock frequency. For the analog solution, Fig. 4(b), we short the output of the OP-amp4 with its positive input through two additional operational amplifiers (OP-amp5 and OP-amp6). Due to the high current load, a large capacitor of 47 µF is used as feedback for OP-amp6. resulting in a RC time constant for the circuit of ~155 s. Important for the analog solution is Switch2 (ADG801BRT), which is close for dark current compensation and open during X-ray exposure. To drive 3 OPDs at the same time, in order to test the possibility to measure the data from an array, we replicate the circuit 3 times. Organic diodes are connected with shorted cathodes, hence the polarization bias is the same for all OPDs. In addition, a switch in every channel allows for diode selection. Signals are added and converted in voltages by a summing operational amplifier with unity gain OP-amp7, Fig. 3(c). For the feedback loop of OP-amp7 a parallel of 10 kΩ resistor and 1 nF capacitor is used resulting in a fast circuit with RC time constant of ~10-5 s. Finally, signals are filtered in both boards. Low pass filtering is implemented using three ADG509 and a TC1563-2. With a USB connection and a PC software we set the filter cut-off frequencies to 0, 1, 5 or 10 kHz and pole numbers to 1 or 2, which correspond to 20 and 40 dB dec-1, respectively. For an output signal ranging from 0 to +10 V we need to amplify with an additional OP-amp. To prevent OPD damage and to shield from noise sources, the input stage of the read-out electronic is galvanically separated from the rest of the circuit and the electronic board is placed in a Faraday cage. The digital solution needs less than 2 seconds for a stable signal. This time period is relative short and fits to the ramp-up time of the X-ray source. Instead, the analog board needs about 20 min to stabilize the dark current before the first X-ray signal can be acquired due to the high value of the RC time constant. However, we suppose that with appropriate engineering the dark current compensation could be reduced to < 20 s. Note, diodes are kept in dark condition until the current is completely compensated. Both boards are compact and of portable size.
IV. Results
Fig. 5(a) and (b) show the output signal generated by the digital electronic board. Three diodes are connected to the digital board inputs and all switches are set to on. Diodes are reversed biased at -1 V and illuminated with a 1 ms length single green light pulse (532 nm wavelength) with intensity of ~50 nW cm-2, Fig. 5(a) black line. The single pole low-pass filter of the board is set to 5 kHz. The electronic board reads with a high sampling rate the photocurrent signal from the diodes and generated as output a voltage pulse of ~0.76 V corresponding to a photocurrent value of ~152 nA with a high signal to noise ratio (S/N), Fig. 5(a) red line. Note, that the dark current level of the baseline is correctly compensated.
Finally, with a dedicated electronic capable of reading signals from the large active area OPDs with ms temporal resolution we are able to characterize our OPD response under short-pulse X-ray excitation. OPD is reverse biased at -2 V, coupled to a 12 µm thick GOS scintillator and irradiated by a single-short X-ray pulse with 13.7 µGy dose at 60 kV. Fig. 5(b) shows the output of the digital board with no filter applied. A voltage pulse with amplitude of ~3.23 V corresponding to a photocurrent of ~646 nA is read-out. The rise and fall time are ~1.8 ms and ~1.4 ms, respectively. We performed several X-ray measurements on this diode changing the dose and the X-ray pulse length (from 3 to 30 ms) and we could not observe any variation of the rise time (not shown). From these measurements we calculate a mean OPD sensitivity of ~0.1 mV µGy-1 s-1 cm-2. However, this value can be enhanced by increasing the gain of the amplification stage to match the appropriate requirements of a medical or industrial application.
We fully characterize our OPD by varying the GOS scintillator total thicknesses from ~4 µm to ~28 µm. Figure 6 shows the photocurrent of an OPD to detect dose rates from 3 to 90 µGy s-1 with varying GOS thicknesses.
Direct BHJ X-ray conversion without scintillator layer (Fig. 6, points next the origin of the graph) shows 10-fold less X-ray sensitivity than a device with 8 μm GOS indicating that the influence of energy absorption by the P3HT:PCBM blend can be neglected. The photocurrents scale linearly with increasing GOS thickness up 28 µm and linear dependence between photocurrent and dose rates up to 2 mGy s-1 is observed (data not shown). In addition, we tested the aging of the OPDs due to X-ray absorption. With an accumulated dose of 15 Gy we observe a slight increase of the dark current from ~2 10-5 mA cm-2 to ~2.6 10-5 mA cm-2 at -5 V reverse bias, Fig. 7a. The diode rectification ratio remains almost constant (~1.5 105 at + / - 2 V) and the OPD shows the same X-ray sensitivity after 15 Gy accumulated dose, Fig. 7b. Results are consistent with reported values [13].
V. Conclusion
In summary, we discussed OPDs with an active area of 5 cm2 for X-ray sensing applications. We show that with an optimum layer stack a device with dark current density of ~1 10-5 mA cm-2 and EQE > 70 % can be achieved with high reproducibility. We demonstrated that despite the large OPD capacitance and the high current load a single light or X-ray short-pulse detection at extremely low light levels can be achieved with a rise/fall time in the ms range. The read-out circuit here presented can be easily modified to manage OPDs with even larger active area or an array of diodes. The millisecond resolution and the high sensitivity observed in the organic diodes match the requirements for medical and industrial X-ray sensing applications.
References
[1] T. N. Ng, W. S. Wong, M. L. Chabinyc, S. Sambandan, R. A. Street, “Flexible image sensor array with bulk heterojunction organic photodiode,” Appl. Phys. Let., Vol. 92, 213303, 2008.
[2] T. Rauch et al, “Near-infrared imaging with quantum-dot-sensitized organic photodiodes,” Nature Photon., Vol. 3, pp. 332-336, 2009.
[3] N. S. Sariciftci, L. Smilowitz, A. J. Heeger, F. Wudl, “Photoinduced Electron Transfer from a Conducting Polymer to Buckminsterfullerene,” Science, Vol. 258, pp. 1474-1476, 1992.
[4] S. F. Tedde et al, ”Fully Spray Coated Organic Photodiodes”, Nano Letters, vol. 9(3), pp. 980-983, 2009.
[5] M. Fukuda, “Optical semiconductor device,” 1st ed. Wiley-Interscience, J. Willey & Sons Inc. New York, 1999.
[6] E. S. Zaus, S. Tedde, T. Rauch, J. Furst, G. H. Dohler, “Design of highly transparent organic photodiodes,” IEEE Transactions on Electron Devices, Vol. 55 (2), pp. 681-684, 2008
[7] Y. Park, V. Choong, Y. Gao, B. R. Hsieh, C. W. Tang, “Work function of indium tin oxide transparent conductor measured by photoelectron spectroscopy,” Appl. Phys. Lett., Vol. 68(19), pp. 2699-2701, 1996.
[8] A.M. Nardes, M. Kemerink, M.M. de Kok, E. Vinken, K. Maturova, R.A.J. Janssen, “Conductivity, work function, and environmental stability of PEDOT:PSS thin films treated with sorbitol,” Organic Electronics, Vol. 9 pp. 727–734, 2008.
[9] C. W. Liang, W. F. Su, L. Wang, “Enhancing the photocurrent in poly(3-hexylthiophene)/[6,6]-phenyl C61 butyric acid methyl ester bulk heterojunction solar cells by using poly(3-hexylthiophene) as a buffer layer,” Appl. Phys. Let., Vol. 95, 133303, 2009.
[10] J. Y. Oh, W. S. Jang, T. I. Lee, J. M. Myoung, H. K. Baik, “Driving vertical phase separation in a bulk-heterojunction by inserting a poly(3-hexylthiophene) layer for highly efficient organic solar cells,” Appl. Phys. Let., Vol. 98, 023303, 2011.
[11] A. E. J. Lim et al, “Effectiveness of aluminum incorporation in nickel silicide and nickel germanide metal gates for work function reduction,” Japanese Journal of Applied Physics, Vol. 47(4), 2008.
[12] E. Chen et al, “Polymer photodetector with voltage-adjustable photocurrent spectrum“, Appl. Phys. Let., vol. 96, 043507, 2010.
[13] P. E. Keivanidis et al. “X-ray stability and response of polymeric photodiodes for imaging applications,” Appl. Phys. Lett., vol. 92, 023304, 2008.
