Open Access
Add to BibSonomy
Abstract
Monolithic integration of GaN-based phototransistors and light-emitting diodes (LEDs) is reported. Starting with an LED epitaxial wafer, selective Si diffusion was performed to produce an n–p–i–n structure for the phototransistor. A traditional AlGaN bulk electron-blocking layer (EBL) can block electron injection from an emitter to a collector, thereby hindering the photocurrent amplification process. We used an LED wafer with a superlattice EBL; blocking can be removed under a bias of approximately 7 V and above. External quantum efficiencies of more than 100% and 600% at approximately 380 nm and 330 nm, respectively, were achieved at room temperature and a bias of 11 V, corresponding to responsivities of 0.31 and 1.6 A/W, respectively, significantly higher than commercially available ultraviolet (UV) detectors. Furthermore, we demonstrated an integrated operation of the device. UV light was detected using a phototransistor that sent signals to drive an integrated LED as an indicator.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Monolithic integration of optoelectronic components can provide functionality, compactness and sometimes cost effectiveness. Tchernycheva et al. reported integration of GaN-based single-wire LED and photodetectors optically coupled by silicon nitride waveguides [1]. Lau et al. reported monolithic integration of a III-nitride LED, high electron mobility transistor, and photodiode by selective-area epitaxy on sapphire substrate [2,3]. Lee et al. reported monolithic integration of GaN-based LEDs and metal-oxide-semiconductor field-effect transistors [4]. Li et al. reported monolithic integration of photodiodes, LEDs and waveguides selectively detached from a substrate to reduce crosstalk [5,6]. Wang et al. demonstrated in-plane data transmission between a transmitter (LED), waveguide, and receiver (photodiode) that were suspended from a silicon substrate [7,8]. Moreover, a GaN-based edge-emitting laser was monolithically integrated with an electro-absorption modulator and a photodiode grown on a silicon substrate [9]. On the other hand, traditional GaN-based phototransistors were fabricated using an epitaxial wafer with an n–p–n or n–p–i–n phototransistor structure [10–14]. We previously reported the fabrication of phototransistors from an LED epitaxial wafer with a traditional bulk EBL [15]. Silicon diffusion was used to convert a part of p-GaN into n-GaN for forming an n–p–i–n phototransistor structure. Moreover, we added additional deep-level traps in the EBL region to bypass the EBL barrier. However, the use of additional deep-level traps caused persistent photoconductivity that reduced the response speed. In this study, we used an LED wafer with a superlattice EBL that allowed electron tunneling and current amplification under a bias of approximately 7 V or higher. Moreover, we monolithically fabricated both phototransistors and LEDs, and demonstrated an integrated operation. UV light was detected by using a phototransistor that sent signals to drive an integrated blue LED as an indicator.
2. Device structure and fabrication
A schematic of the integrated device structure is displayed in Fig. 1. We began with a commercial LED wafer grown on a c-plane patterned sapphire substrate with the following epitaxial structure from bottom to top: undoped GaN, n-GaN, 15-pair InGaN/GaN multiple quantum wells (MQWs) with a peak wavelength of 448 nm, 4-pair p-AlGaN/GaN superlattice EBL, 150-nm p-GaN and p+-GaN cap layer. First, each device was isolated by a deep etch down to the insulating sapphire substrate through inductively coupled plasma reactive-ion etching (ICP-RIE). Then ICP-RIE simultaneously produced mesas on the LED and phototransistor to provide access to the n-GaN layer. Next, on top of the emitter region of the phototransistor, the p+-GaN cap layer was removed through ICP-RIE, and then a thin silicon film was deposited and in-diffused at 800 °C in ambient N2 for 30 min., corresponding to a diffusion depth of approximately 40 nm [16]. The diffused region was converted into n-type resulting in an n–p–i–n structure. Subsequently, a patterned SiO2 insulation/passivation layer, and an indium-tin oxide layer (for current spreading of LEDs and p-contact for all devices) were deposited. Finally, p/n-metal electrodes were formed. In addition, p–i–n photodiodes were simultaneously fabricated on the same chip with the same design and processing as LEDs, except there was no ITO deposition on the light-receiving area for avoiding UV light absorption by the ITO layer. The purpose of adding this device was to use it as a fair reference in comparison with the phototransistor which has no ITO on the light-receiving area.
Fig. 1. Cross-sectional schematic structure of the integrated LED and phototransistor, where ITO and MQW represent indium tin oxide and multiple quantum wells, respectively.
3. Experimental results and discussion
3.1. Performance of LEDs
Figure 2 shows a charge-coupled device (CCD) image of an LED under probe testing and the optical microscope image of the integrated LED and phototransistor. The voltage-current-light (V-I-L) curves of the LED with a size of 0.10 mm2 are plotted in Fig. 3(a). In addition, the external quantum efficiency (EQE) versus wavelength of the p–i–n photodiode was measured (for comparison with that of a phototransistor later), by employing an incident photon-to-electron conversion efficiency (IPCE) apparatus, which consists of a broadband 300-W Xenon lamp (Newport, model #6258), monochromator (Newport, model #74024), chopper, and lock-in amplifier (Newport, SR830). The results are plotted in Fig. 3(b). The peak EQE was approximately 56% at 384 nm, exhibiting saturation at high bias voltages.
Fig. 2. A CCD image of the LED under probe testing. Also shown below is an optical microscope image of the integrated LED and phototransistor.
Fig. 3. (a) V-I-L characteristics of the LED; (b) EQE versus wavelength of the p–i–n photodiode at various bias voltages.
3.2. Performance of phototransistors
The dark current versus emitter-collector voltage (VEC) curve of the phototransistor in an open-base configuration at room temperature is presented in Fig. 4(a). At voltages of approximately 5 V and 6 V when positive and negative voltages were applied, respectively, the dark current started increasing with voltage, indicating the onset of current tunneling/amplification. Despite the primary current herein was formed by thermally generated electron-hole pairs. EQE versus wavelength at various VCE voltages was measured and plotted in Fig. 4(b). When the VCE was low (5 V or below), the EQE was low and exhibited saturation, similar to the behavior of a p–i–n photodiode. Nevertheless, when the VCE was further increased to 7 V or above, the phototransistor began to exhibit measurable photocurrent gain and achieved more than 100% EQE at 380 nm at 11 V. This phenomenon is illustrated in Fig. 5. For conventional as-grown n–p–n phototransistors with no EBL structure, the injected electrons from the emitter to a thin base partially arrive at the depleted region and are then collected by the collector, resulting in a photocurrent gain. In this study, the LED-converted phototransistor had a superlattice EBL; when the VCE was low, the EBL blocked the diffused electrons and thus no photocurrent amplification occurred. The photogenerated holes recombined with the diffused electrons within the base or joined the diffused hole current into the emitter; therefore, the device in this mode acted as a photodiode combined with a barely forward biased diode in series. Thus, the resultant photocurrent was even lower than that of a simple photodiode under the same bias voltage. However, when the VCE was high, enabling electron tunneling through the superlattice EBL, the amplification mechanism was unlocked, resulting in high EQE. Moreover, a rapid increase in EQE with bias voltage, particularly at short wavelengths, was an indication that the device was operating as a phototransistor. The incident light intensity from the Xenon lamp was not uniform versus wavelength; it increased rapidly with wavelength in the UV wavelength range (i.e., from 60 µW/cm2 at 330 nm to 2.15 mW/cm2 at 400 nm), and then remained at a level of 2.2-2.4 mW/cm2 in the wavelength range of 400 nm to 460 nm. Because the photocurrent gain of a phototransistor is nonlinear (i.e., gain is higher for a small signal), UV wavelengths of a low incident intensity level obtained a higher gain. By contrast, a p–i–n photodiode has no internal gain; its EQE values at short wavelengths remained fairly small regardless of incident intensity level, as shown in Fig. 3(b). (The EQE at wavelengths shorter than 330 nm could not be measured because of a lack of incident light.) Furthermore, we used a continuous-wave UV lamp with a peak wavelength of 254 nm to illuminate our devices. Different incident light intensity was obtained by adjusting the height of the lamp. Photocurrent versus incident light intensity are plotted in Fig. 6, when the phototransistor was biased at 9 V, and the p–i–n photodiode was biased at 5 and 0 V, respectively. Clearly, the phototransistor has significantly higher responsivity than the p–i–n photodiode. And the responsivity of the phototransistor was approximately 0.15 A/W under 1-mW/cm2 level of incident light intensity at 254 nm.
Fig. 4. (a)In-darkness I-V characteristics of the phototransistor in an open-base configuration; (b) EQE versus wavelength of the phototransistor at various VCE voltages. The inset shows the complete results.
Fig. 5. Energy band diagram of the phototransistor under a positive VCE bias. The marks indicate: (1) photogeneration of electron-hole pairs; (2) electron-hole recombination via deep level traps; (3) recombination of holes with injected electrons; (4) recombination of diffused holes with majority carriers (electrons); (5) electron tunneling through the superlattice EBL.
Fig. 6. Photocurrent versus incident light intensity at 254 nm for the phototransistor biased at 9 V and p-i-n (PIN) photodiode biased at 5 and 0 V.
The response speed of the phototransistor was characterized by using the IPCE apparatus except that the lock-in amplifier was replaced by a precision high-speed sampling multimeter (Keithley DMM7510) and the incident wavelength was fixed. The phototransistor was emitter-side illuminated with an incident light intensity of 1.53 mW/cm2 at 384 nm. The light-receiving area size was 0.158 mm2. Figure 7 exhibits the temporal response to a 150-Hz optical pulse signal at room temperature. The VCE was set at 9 V and the dark current was approximately 0.68 µA. The dominating factor affecting the response speed of a GaN-based phototransistor was the fall time, which was measured to be approximately 1 ms (the fall time is the time required for the response to decay from the peak value to e−1 of the photocurrent). This fall time was significantly shorter than previously reported [11,15] by two orders of magnitude due to less deep level defects. (Deep level traps can lengthen the recombination time of photogenerated holes, as well as the response time.)
3.3. Integrated operation of the phototransistor and LED
We developed a module that converted UV-light-induced photocurrent signals of phototransistors to voltage signals, which were used to operate an integrated LED indicator. The module was designed for switching on the LED indicator once the UV-light safety level was exceeded. The more the UV-light intensity exceeded the specified safety level, the brighter was the glow of the LED. Figure 8 illustrates the schematic of the module. The module comprises a monolithically integrated phototransistor and LED, and an external circuit. The circuit converted the phototransistor’s current signals to voltage signals (V1) and amplified the voltage signals to drive the LED. The current-to-voltage gain of the transimpedance amplifier was proportional to the feedback resistance R1 at low frequencies, and the circuit bandwidth was inversely proportional to the product of the resistance R1 and capacitor C1. The circuit parameters used in this experiment are listed in Table 1. The circuit bandwidth was 31 kHz, which is considerably larger than the operating frequency of 150 Hz. The voltage VOffset was used to offset the DC component of V1, which was caused by dark current. Assuming that the specified UV safety level was 1.50 mW/cm2 at 384 nm, the associated illuminating current at a UV intensity of 1.53 mW/cm2 should be converted to an output voltage (Vout) marginally larger than the turn-on voltage of the LED. Figure 9 illustrates the resultant voltages V1 and Vout while the current signals were as shown in Fig. 7. And the integrated LED was switched on and off accordingly. We noticed that there would be measurable crosstalk between the phototransistor and LED if their distance is too close or the LED power is large. Light emission from an LED can be partially coupled into neighboring phototransistors through a patterned sapphire substrate. The degree of crosstalk was evaluated by measuring the photocurrent of the phototransistor when only one LED was lit. The photocurrent versus center-to-center distance between the phototransistor and LED is illustrated in Fig. 10, when the phototransistor was biased at 9 V and the surface-emitting power of the LED was continuous-wave 1 mW. By using an LED with a distance of 2.86 mm and operating at a power level of approximately 0.1 mW in the module in Fig. 8, we reduced the crosstalk to a negligible low. Otherwise, the crosstalk would induce a positive feedback to the module that would result in an increase in the glow of the LED. Different designs will be developed in the future to decrease or increase the crosstalk according to the applications of the device.
Fig. 8. Schematic diagram of the module consisting of a monolithically integrated phototransistor and LED, and an external circuit, where OPA represents operational amplifier. The external circuit includes a transimpedance preamplifier and a non-inverting amplifier.
Fig. 9. Temporal response of voltages V1 and Vout to a 150-Hz optical pulse signal at 384 nm. The inset photos to the right exhibit the integrated LED being switched on and off, respectively.
Fig. 10. Photocurrent versus center-to-center distance between the phototransistor and LED, when the phototransistor was biased at 9 V and the surface-emitting power of the LED was 1 mW.
Table 1. Circuit parameters used in this experiment
4. Conclusion
Monolithic integration of GaN-based phototransistors and LEDs was demonstrated. Based on an LED epitaxial wafer with a superlattice EBL, selective silicon diffusion converted part of the p-GaN into n-GaN, resulting in an n–p–i–n phototransistor structure. Under a VCE bias of approximately 7 V or above, electron tunneling through the superlattice EBL occurred, thereby enabling photocurrent amplification process. At a bias of 11 V, EQE of more than 100% and 600% at approximately 380 nm and 330 nm, respectively, were achieved at room temperature, corresponding to responsivities of 0.31 and 1.6 A/W, respectively. Higher responsivity is possible at a higher bias voltage or when a superlattice EBL of fewer pairs is used. Moreover, the phototransistor’s response speed was measured to be approximately 1 ms, faster than previously reported by two orders of magnitude. Furthermore, we demonstrated an integrated operation of this device. UV light was detected using a phototransistor that sent signals to drive an integrated blue LED as an indicator or warning light.
Funding
Ministry of Science and Technology, Taiwan (106-2221-E-011-109, 107-2218-E-011-020).
Disclosures
The authors declare no conflicts of interest.
References
1. M. Tchernycheva, A. Messanvi, A. de Luna Bugallo, G. Jacopin, P. Lavenus, L. Rigutti, H. Zhang, Y. Halioua, F. H. Julien, J. Eymery, and C. Durand, “Integrated photonic platform based on InGaN/GaN nanowire emitters and detectors,” Nano Lett. 14(6), 3515–3520 (2014). [CrossRef]
2. Z. Liu, J. Ma, T. Huang, C. Liu, and K. M. Lau, “Selective epitaxial growth of monolithically integrated GaN-based light emitting diodes with AlGaN/GaN driving transistors,” Appl. Phys. Lett. 104(9), 091103 (2014). [CrossRef]
3. C. Liu, Y. Cai, H. Jiang, and K. M. Lau, “Monolithic integration of III-nitride voltage controlled light emitters with dual-wavelength photodiodes by selective-area epitaxy,” Opt. Lett. 43(14), 3401–3404 (2018). [CrossRef]
4. Y.-J. Lee, Z.-P. Yang, P.-G. Chen, Y.-A. Hsieh, Y.-C. Yao, M.-H. Liao, M.-H. Lee, M.-T. Wang, and J.-M. Hwang, “Monolithic integration of GaN-based light emitting diodes and metal-oxide-semiconductor field-effect transistors,” Opt. Express 22(S6), A1589–A1595 (2014). [CrossRef]
5. K. H. Li, W. Y. Fu, Y. F. Cheung, K. K. Y. Wong, Y. Wang, K. M. Lau, and H. W. Choi, “Monolithically integrated InGaN/GaN light-emitting diodes, photodetectors, and waveguides on Si substrate,” Optica 5(5), 564–569 (2018). [CrossRef]
6. K. H. Li, Y. F. Cheung, W. Y. Fu, K. K. Y. Wong, and H. W. Choi, “Monolithic integration of GaN-on-sapphire light-emitting diodes, photodetectors, and waveguides,” IEEE J. Sel. Top. Quantum Electron. 24(6), 1–6 (2018). [CrossRef]
7. Y. Wang, G. Zhu, W. Cai, X. Gao, Y. Yang, J. Yuan, Z. Shi, and H. Zhu, “On-chip photonic system using suspended p-n junction InGaN/GaN multiple quantum wells device and multiple waveguides,” Appl. Phys. Lett. 108(16), 162102 (2016). [CrossRef]
8. Z. Shi, X. Gao, J. Yuan, S. Zhang, Y. Jiang, F. Zhang, Y. Jiang, H. Zhu, and Y. Wang, “Transferrable monolithic III-nitride photonic circuit for multifunctional optoelectronics,” Appl. Phys. Lett. 111(24), 241104 (2017). [CrossRef]
9. M. Feng, J. Wang, R. Zhou, Q. Sun, H. Gao, Y. Zhou, J. Liu, Y. Huang, S. Zhang, M. Ikeda, H. Wang, Y. Zhang, Y. Wang, and H. Yang, “On-chip integration of GaN-based laser, modulator, and photodetector grown on Si,” IEEE J. Sel. Top. Quantum Electron. 24(6), 1–5 (2018). [CrossRef]
10. W. Yang, T. Nohava, S. Krishnankutty, R. Torreano, S. McPherson, and H. Marsh, “High gain GaN/AlGaN heterojunction phototransistor,” Appl. Phys. Lett. 73(7), 978–980 (1998). [CrossRef]
11. R. Mouillet, A. Hirano, M. Iwaya, T. Detchprohm, H. Amano, and I. Akasaki, “Photoresponse and defect levels of AlGaN/GaN heterobipolar phototransistor grown on low-temperature AlN interlayer,” Jpn. J. Appl. Phys. 40(Part 2), L498–L501 (2001). [CrossRef]
12. M. L. Lee, J. K. Sheu, and Y.-R. Shu, “Ultraviolet bandpass Al0.17Ga0.83N∕GaN heterojunction phototransistors with high optical gain and high rejection ratio,” Appl. Phys. Lett. 92(5), 053506 (2008). [CrossRef]
13. S.-C. Shen, T.-T. Kao, H.-J. Kim, Y.-C. Lee, J. Kim, M.-H. Ji, J.-H. Ryou, T. Detchprohm, and R. D. Dupuis, “GaN/InGaN avalanche phototransistors,” Appl. Phys. Express 8(3), 032101 (2015). [CrossRef]
14. T.-T. Kao, J. Kim, T. Detchprohm, R. D. Dupuis, and S.-C. Shen, “High-responsivity GaN/InGaN heterojunction phototransistors,” IEEE Photonics Technol. Lett. 28(19), 2035–2038 (2016). [CrossRef]
15. P. S. Yeh, T.-P. Hsu, Y.-C. Chiu, S. Yang, C.-Y. Wu, and J.-S. Liou, “III-nitride phototransistors fabricated on a light-emitting-diode epitaxial wafer,” IEEE Photonics Technol. Lett. 29(19), 1679–1682 (2017). [CrossRef]
16. P. S. Yeh, C.-C. Chang, Y.-T. Chen, D.-W. Lin, J.-S. Liou, C. C. Wu, J. H. He, and H.-C. Kuo, “GaN-based vertical-cavity surface emitting lasers with sub-milliamp threshold and small divergence angle,” Appl. Phys. Lett. 109(24), 241103 (2016). [CrossRef]
