Open Access
Add to BibSonomy
Abstract
In this work, by comparing and analyzing dynamic biasing InGaAs/InAlAs avalanche photodiodes(APDs) with different active areas, it is found that they have different noise suppression frequency ranges. The upper limit frequency(defined as the frequency at which the noise suppression effect begins to fail) of InGaAs/InAlAs APDs with active area diameter of 50 µm, 100 µm and 200 µm are 2400 MHz, 1990MHz and 1400 MHz respectively. In addition, for InGaAs/InAlAs APDs with an active area diameter of 50 µm, 100 µm and 200 µm, their optimal frequencies of dynamic biasing (defined as the frequency corresponding to the optimal SNR) are 1877MHz, 1670 MHz and 1075 MHz respectively. At last, applying dynamic biasing technology, it achieves a useful gain of 6698.1, which is much greater than that of DC bias (47.2), and this technology has the potential to be applied in high sensitivity laser radar receivers.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
In the eye-safety short-wave infrared (SWIR) waveband, InGaAs/InP and InGaAs/InAlAs APDs are the most widely used due to their high sensitivity [1,2]. For InGaAs/InP and InGaAs/InAlAs APDs operating in linear mode, the excess noise of APDs will increase as the gain increases, which ultimately limits both the signal-noise ratio (SNR) and the maximum useful gain(the gain corresponding to the optimal SNR, i.e., the maximum useful gain). For linear mode APDs operating in SWIR waveband, the maximum useful gain is generally less than 50, usually at M = 10 to 20 [3]. Therefore, in order to improve the SNR of the light detection and ranging (lidar) receiver, it is necessary to increase the maximum useful gain of the APDs or reduce the noise of the receiver. In order to improve the sensitivity of APDs, they can be operated as single photon avalanche diodes (SPADs) in Geiger mode, which can reach a gain of ∼105 [4]. However, in Geiger mode, after-pulse and dark count will be generated, and it takes a long integration time to eliminate the influence of after-pulse and dark count [5,6]. Actually, there exists a region between the linear mode and Geiger mode that is insufficient for single photon detection and indicates the dark counts can not be generated. In contrast to the linear mode, this region displays a greater gain and higher excess noise. If a method can be found to keep the high gain of this region while suppressing the increase of excess noise, it will effectively improve SNR of the lidar receiver.
The excess noise of APDs is largely generated in the tail of collision ionization process [7] and by applying a high-frequency dynamic biasing, the avalanche process can be quenched before the establishment of the long time tail [8], so the high frequency dynamic biasing has the potential to keep high gain while simultaneously suppressing excess noise. Some excellent work about avalanche photodiodes with dynamic biasing technology have been reported [9–16]. Dynamic biasing technique can be used to improve the gain bandwidth product of APDs [9,10]. The principle of this approach is to suppress the ionizations near the end of the optical pulse and promote strong impact ionizations in the early phase of the pulse. The sensitivity of dynamically biased Ge APDs at 3Gbps is 4.2 dB higher than that of DC biased APDs [11]. In [12], using the commercial dynamic biasing InGaAs/InP APDs, the receiver sensitivity is improved by 5.2 dB and the bit error rate is reduced by 10,000 times at the rate of 25Gbps. In [16], the concept of transition zone is proposed, and the gain of 620 is achieved in the transition zone by applying dynamic biasing technology, which is greatly improved compared with the DC bias mode (M = 17.2) and improves the SNR of the lidar receiver. However, no one has studied the dynamic biasing technology for InGaAs/InAlAs APDs of different active area with different capacitance.
This paper extends the previous work, RMS noise and SNR measurements of the dynamic biasing InGaAs/InAlAs avalanche photodiodes with different active areas are investigated in this work. By studying InGaAs/InAlAs APDs of different active area with dynamic biasing technology, it is found that there are different noise suppression frequency ranges for InGaAs/InAlAs APDs with different capacitances. RC limit bandwidth is the dominant factor limiting its upper limit frequency for the InGaAs/InAlAs APD with active area diameter of 200 µm, but not for InGaAs/InAlAs APDs with active area diameters of 50 µm and 100 µm. In addition to this, for APDs with different capacitance, there are different optimal frequency of dynamic biasing. At the same time, an available gain of 6698.1 is achieved using dynamic biasing technology, which is much greater than the maximum useful gain (47.2) in DC bias.
2. Experiment setup
The dynamic biasing added by APDs can be expressed as follows:
where ω and φ are the angular frequency and initial phase for the cosine wave signal, respectively. Vpp is the peak-peak voltage of the cosine wave signal. VDC is the DC bias and f is the frequency.The experiment setup is schematically demonstrated in Fig. 1(a). Electrical pulses with the repeat frequency (ffre) of 10 MHz and full width at half maximum (tw) of 500 ps was used to trigger the laser. The wavelength of the optical pulse signal is 1550 nm. The variable optical attenuator (VOA) attenuated the optical pulse signal and then the coupler divided it into two parts. One part was coupled to the APD by using a single-mode fiber (SMF) for the SNR measurements, and the other was monitored by an optical power meter(the mean optical power was measured). Resistors R2 and capacitor C2 formed a low-pass filter to filter the ripple of DC power supply. Resistor R1 and capacitor C1 formed a high-pass filter to filter the low-frequency signal that may be generated by the cosine wave signal generator to reduce the signal interference. To eliminate the influence of the cosine wave on optical signal, the low-pass filters(LPFs) were used in the experiment, and the cutoff frequency of the LPFs is 755 MHz. The avalanche signal was further amplified by a transimpedance amplifier (TIA) and connected to an oscilloscope. Figure 1(b) shows the operating voltage range of dynamic biasing APDs. The operating voltage range is located in the transition region defined in Ref. [16]. Where Vl denotes the voltage with the optimum SNR in linear mode and Vg represents the voltage at which the dark counts begin to be generated in Geiger mode.
The SNR can be expressed as [17]:
where, Vs-peak and VRMS are the peak voltage of signal and the root mean square voltage of total noise respectively. The VRMS and Vs-peak are measured by the oscilloscope. In the SNR measurement, the oscilloscope only counts the parameters of the waveform displayed on the screen, so the pulse signal can be moved out of the screen by adjusting the horizontal position knob of the oscilloscope, thereby extracting the VRMS from the oscilloscope. The measured gain of the InGaAs/InAlAs APDs can be calculated by: The optical signal in this paper is a pulse signal, and its peak optical power can be expressed by the following formula: where Ppeak and Pm are the peak optical power and mean optical power of the optical pulse signal respectively. $\Re \textrm{ and }{R_{TIA}}$ are the responsivity for the APD and the transimpedance of the TIA, respectively.3. Results and discussion
Figure 2 (a),(b) and (c) shows the photocurrent (1550 nm light), dark current and gain of InGaAs/InAlAs APD with active area diameter of 50 $\mathrm{\mu}\textrm{m}$, 100 $\mathrm{\mu}\textrm{m}$, 200 $\mathrm{\mu}\textrm{m}$ versus reverse voltage as the optical power is 1 µW. The measured responsivity (M = 1) of InGaAs/InAlAs APDs with active area diameter of 50 $\mathrm{\mu}\textrm{m}$, 100 $\mathrm{\mu}\textrm{m}$, 200 $\mathrm{\mu}\textrm{m}$ are both 1 A/W. Figure 2(d) shows the data of capacitance of InGaAs/InAlAs APDs with active area diameter of 50 $\mathrm{\mu}\textrm{m}$,100 $\mathrm{\mu}\textrm{m}$,200 $\mathrm{\mu}\textrm{m}$ change with reverse voltage. Their capacitance are 0.64 pF, 0.83 pF, 2.1 pF at high voltage respectively. It should be noted that at high reverse voltage the capacitance does not scale quadratically and similar results have been found in [18]. So there are other factors that affect its capacitance besides its diameter.This suggests that there are potential limitations to this comparison, so the comparison we are making now is a qualitative illustration of a changing trend.
Fig. 2. (a) The photocurrent (1550 nm light), dark current and gain of InGaAs/InAlAs APDs with active area diameter of 50 µm versus reverse voltage as the optical power is 1 µW. (b) The photocurrent (1550 nm light), dark current and gain of InGaAs/InAlAs APDs with active area diameter of 100 µm versus reverse voltage as the optical power is 1 µW. (c) The photocurrent (1550 nm light), dark current and gain of InGaAs/InAlAs APDs with active area diameter of 200 µm versus reverse voltage as the optical power is 1 µW. (d) The data of capacitance of InGaAs/InAlAs APDs with active area diameter of 50 µm, 100 µm, 200 µm change with reverse voltage at room temperature.
As shown in Fig. 3 (a), the variation of RMS noise of InGaAs/InAlAs APDs with active area diameter of (VDC = 50.9 V, Vpp = 2.24 V), 100 µm (VDC = 50.8 V, Vpp = 2.24 V) and 200 µm (VDC = 50.8 V, Vpp = 2.24 V) as a function of f is shown in the absence of incident light. 50.9 V, 50.8 V and 50.8 V are the voltage with the optimum SNR in linear mode for InGaAs/InAlAs APDs with active area diameters of 50 µm, 100 µm and 200 µm respectively. Within a certain frequency range, dynamic biasing has an effect on noise suppression. However, when the frequency of dynamic biasing increases to a certain extent, its suppression on excess noise will start to fail, that is, the frequency of dynamic biasing has an upper limit frequency(defined as the frequency at which the noise suppression effect begins to fail). This indicates that when the upper limit frequency is exceeded, the noise suppression effect of APDs is limited to some extent, which leads to the weakening of its noise suppression effect. The upper limit frequency of InGaAs/InAlAs APDs with active area diameter of 50 µm, 100 µm and 200 µm are 2400 MHz,1990MHz,1400 MHz respectively. These experimental results indicate that there are different upper limit frequencies for InGaAs/InAlAs APDs with different capacitances.
Fig. 3. (a) The data of RMS noise of InGaAs/InAlAs APDs with active area diameter of 50 µm (VDC = 50.9 V, Vpp = 2.24 V), 100 µm (VDC = 50.8 V, Vpp = 2.24 V) and 200 µm (VDC = 50.8 V, Vpp = 2.24 V) change with frequency of dynamic biasing. (b) The data of upper limit frequency and RC limited bandwidth change with the capacitance.
In order to better illustrate the influence of capacitance, the RC limited bandwidths of InGaAs/InAlAs APDs with active area diameters of 50 µm, 100 µm and 200 µm are calculated using the device capacitance. As shown in Fig. 3 (b), the RC limited bandwidths of InGaAs/InAlAs APDs are 4976 MHz, 3837 MHz and 1517 MHz respectively. It can be seen that the RC limited bandwidth decreases as the capacitance increases, which is the similar trend as the upper limit frequency changes with the capacitance. For the InGaAs/InAlAs APD with active area diameter of 200 µm, the upper limit frequency of 1400 MHz is consistent with the RC limited bandwidth of 1517 MHz well, which indicates that its frequency of dynamic biasing is mainly limited by the RC limited bandwidth. This is because when the frequency of dynamic biasing increases to a certain extent, it is difficult for the charging and discharging of InGaAs/InAlAs APD to respond quickly, resulting in the electric field of the multiplication layer can not be quickly transformed, and the avalanche process will not be effectively quenched, so RMS noise tend to stabilize with the change of frequency of dynamic biasing. For InGaAs/InAlAs APDs with active area diameters of 50 µm and 100 µm, the RC limited bandwidth is much higher than the upper limit frequency, indicating that the RC time constant is not the dominant factor, and other factors may influence the upper limit frequency of APDs, including avalanche build up time and total carrier transit time, which is worthy of further study in the future.
As mentioned above, the optimal frequency of dynamic biasing is defined as the frequency corresponding to the optimal SNR. As shown in Fig. 4 (a), there are different optimal frequency of dynamic biasing for InGaAs/InAlAs APDs with different active area diameters of 50 µm (VDC = 50.9 V, Vpp = 5.97 V), 100 µm (VDC = 50.8 V, Vpp = 5.97 V) and 200 µm (VDC = 50.8 V, Vpp = 6.32 V) as the peak optical power is 3.5 nW. As shown in Fig. 2(d), the capacitances of APDs with active area diameters of 50 µm, 100 µm and 200 µm are 0.64 pF, 0.83 pF and 2.1 pF respectively, and their corresponding optimal frequencies of dynamic biasing are 1877MHz, 1670 MHz and 1075 MHz respectively. Experiments show that InGaAs/InAlAs APDs with different capacitances have different optimal dynamic biasing frequencies. The smaller the capacitance of APDs is, the larger the optimal frequency of dynamic biasing is. When the frequency of dynamic biasing surpasses the optimal level, because electrons and holes persist in the avalanche zone of APDs [8] and the avalanche process are consequently quenched by the rapid dynamic biasing signal when the gain of the APDs has not yet reached a sufficiently high value, which leads to a decrease in the SNR improvement at higher frequencies. To this end, the optimal frequency may be related to the transit time of electrons and holes.
Fig. 4. (a) The data of normalized SNR of InGaAs/InAlAs APDs with active area diameter of 50 µm (VDC = 50.9 V,Vpp = 5.97 V), 100 µm (VDC = 50.8 V,Vpp = 5.97 V) and 200 µm (VDC = 50.8 V, Vpp = 6.32 V) change with frequency of dynamic biasing as the peak optical power is 3.5 nW. (b) The data of optimal frequency of dynamic biasing changes with capacitance of InGaAs/InAlAs APDs.
To illustrate the advantages of dynamic biasing technology, we discuss InGaAs/InAlAs APDs with active area diameter of 50 µm as a representative example. As shown in Fig. 6, the variation of SNR and gain of InGaAs/InAlAs APDs with active area diameter of 50 µm as a function of Vpp is shown. In the experiment, the VDC of InGaAs/InAlAs APDs is 50.9 V. As shown in Fig. 5, it should be noted that 50.9 V is the DC voltage at which the optimal SNR is located, and the corresponding optimal gain is 47.2. The frequency of dynamic biasing of InGaAs/InAlAs APDs is 1.877 GHz. When Vpp is greater than 5.97 V, the dark counts is generated, which will disturb the signals. According to [16], the transition zone of the InGaAs/InAlAs APDs is from 50.9 V to 53.885 V. As shown in Fig. 6, with the increase of Vpp, SNR and gain of InGaAs/InAlAs APDs increase gradually. When the Vpp = 1 V, the SNR and gain of InGaAs/InAlAs APDs are 5.1 and 358.5, respectively. When the Vpp is increased to 5.97 V, the SNR and gain of InGaAs/InAlAs APDs are increased to 58.7 and 6698.1. The maximum useful gain for dynamic biasing mode (M = 6698.1) is much larger than that in DC bias (M = 47.2).
Fig. 6. (a) The SNR changes with Vpp as the peak optical power is 3.3 nW. (b) The data of gain changes with Vpp.
As shown in Fig. 7 (a), when the SNR is 10, compared with the DC bias, the peak optical power of dynamic biasing mode (0.4 nW) has a 17.7 dB reduction than that of the DC bias (23.8 nW). This indicates that if dynamic biasing technology is used in a lidar system, the ranging sensitivity of lidar will be improved. As shown in Fig. 7(b), the rise time demonstrates a relatively stable behavior when there is a variation in peak optical power, thereby proving advantageous in enhancing the ranging precision of lidar systems with a leading-edge timing discriminator [3].
Fig. 7. (a) The SNR versus the peak optical power. (b) The data of rise time of signal for dynamic biasing mode changes with peak optical power.
However, it is worth noting that the calculation of M in this study is based on pulse mode, and the gain improvement for non-pulse detection may differ, warranting further research in the future. In addition, the gigantic gain in this paper is measured at optical power of the nW order. As shown in the Fig. 8, it is found that the gain tends to decrease as the optical power increases, so it can be expected that the gain improvement will weaken as the optical power is further increased, which can be attributed to space charge effect [19]. The hole with positive charge may accumulate in the vicinity of the charge layer and cancel out the negative charge of the charge layer, which may weaken the control of the charge layer on the electric field and lead to the decrease of the electric field of the multiplication layer, thus reducing the gain. In addition, this effect becomes even more critical when the input optical power is higher. Therefore, at high optical power, the gigantic gain may not be obtained.
Figure 9 shows the time domain diagram of the signal pulse in DC bias and dynamic biasing modes. As can be seen from the Fig. 8, the amplitude of the signal in dynamic biasing mode is much larger than that in DC bias, which means that there is a higher gain and SNR in dynamic biasing mode.
Fig. 9. The signal pulse of dynamic biasing mode (VDC = 50.9 V, Vpp = 4.48 V) and DC bias (VDC = 50.9 V) as the peak optical power is 13.2 nW.
4. Conclusion
The results of this paper experimentally demonstrates that there are different upper limit frequency for InGaAs/InAlAs APDs with different capacitances. The upper limit frequency of InGaAs/InAlAs APD with active area diameter of 200 µm is mainly limited by RC limited bandwidth. However, the RC limited bandwidth is much higher than the upper limit frequency for InGaAs/InAlAs APDs with active area diameters of 50 µm and 100 µm, indicating that the RC time constant is not the dominant factor, but that other factors are involved which is worthy of further study in the future. In addition, it is also found that there are different optimal frequency of dynamic biasing for APDs with different capacitances. Finally, with the dynamic biasing technology, the gain of 6698.1 is realized, which is much higher than the maximum useful gain (47.2) under DC bias in the order of tens.
Funding
National Key Research and Development Program of China (2018YFB2200204); Science and Technology Project from Wuhan City (2020010601012162).
Disclosures
The authors declare no conflicts of interest.
Data availability
All the data supporting this study are available in the paper and Supplementary Information. Additional data related to this paper are available from the corresponding authors upon request.
References
1. J. C. Campbell, “Recent advances in telecommunications avalanche photodiodes,” J. Lightwave Technol. 25(1), 109–121 (2007). [CrossRef]
2. J. C. Campbell, “Recent advances in avalanche photodiodes,” J. Lightwave Technol. 34(2), 278–285 (2016). [CrossRef]
3. G. M. Williams, “Optimization of eyesafe avalanche photodiode lidar for automobile safety and autonomous navigation systems,” Opt. Eng. 56(3), 031224 (2017). [CrossRef]
4. J. Zhang, M. A. Itzler, H. Zbinden, and J. W. Pan, “Advances in InGaAs/InP single-photon detector systems for quantum communication,” Light: Sci. Appl. 4(5), e286 (2015). [CrossRef]
5. Y. Liang, B. Xu, Q. Fei, W. Wu, X. Shan, K. Huang, and H. Zeng, “Low-Timing-Jitter GHz-Gated InGaAs/InP Single-Photon Avalanche Photodiode for LIDAR,” IEEE J. Sel. Top. Quantum Electron. 28(2: Optical Detectors), 1–7 (2022). [CrossRef]
6. A. McCarthy, X. Ren, A. Della Frera, N. R. Gemmell, N. J. Krichel, C. Scarcella, A. Ruggeri, A. Tosi, and G. S. Buller, “Kilometer-range depth imaging at 1550 nm wavelength using an InGaAs/InP single-photon avalanche diodedetector,” Opt. Express 21(19), 22098–22113 (2013). [CrossRef]
7. M. M. Hayat and B. E. A. Saleh, “Statistical Properties of the Impulse Response Function of Double-Carrier Multiplication Avalanche Photodiodes Including the Effect of Dead Space,” J. Lightwave Technol. 10(10), 1415–1425 (1992). [CrossRef]
8. D. C. Herbert and E. T. R. Chidley, “Very low noise avalanche detection,” IEEE Trans. Electron Devices 48(7), 1475–1477 (2001). [CrossRef]
9. M. M. Hayat and D. A. Ramirez, “Multiplication theory for dynamically biased avalanche photodiodes: new limits for gain bandwidth product,” Opt. Express 20(7), 8024–8040 (2012). [CrossRef]
10. G. El-Howayek and M. M. Hayat, “Error Probabilities for Optical Receivers That Employ Dynamically Biased Avalanche Photodiodes,” IEEE Trans. Commun. 63(9), 3325–3335 (2015). [CrossRef]
11. M. M. Hayat, P. Zarkesh-Ha, G. El-Howayek, R. Efroymson, and J. C. Campbell, “Breaking the buildup-time limit of sensitivity in avalanche photodiodes by dynamic biasing,” Opt. Express 23(18), 24035–24041 (2015). [CrossRef]
12. P. Zarkesh-Ha, R. Efroymson, E. Fuller, J. C. Campbell, and M. M. Hayat, “5.2 dB Sensitivity Enhancement in 25Gbps APD-Based Optical Receiver Using Dynamic Biasing,” in Optical Fiber Communication Conference (2020).
13. M. M. Hayat, J. P. David, S. Krishna, L. F. Lester, D. A. Ramirez, and P. Zarkesh-Ha, “Impact ionization devices under dynamic electric fields,” U.S. Patent 9,354,113 (2016).
14. M. M. Hayat and P. Zarkesh-Ha, “Resonance avalanche photodiodes for dynamic biasing,” U.S. Patent 10, 777,698 (2018).
15. P. Zarkesh-Ha, M. M. Hayat, and R. Efroymson, “Control circuits for dynamically biased avalanche photodiodes,” U.S. Patent 9,997,644 (2018).
16. Y. Tian, W. Q. Ding, X. Y. Feng, Z. B. Lin, and Y. L. Zhao, “The high signal-noise ratio avalanche photodiodes with dynamic biasing technology for laser radar applications,” Opt. Express 30(15), 26484–26491 (2022). [CrossRef]
17. T. Fersch, R. Weigel, and A. Koelpin, “Increasing the signal-to-noise ratio in APD-based scanning LiDAR sensors onthe transmission side,” in 2016 IEEE Conference on Recent Advances in Lightwave Technology (CRALT) (2016).
18. S. An, W. R. Clark, M. Jamal Deen, A. S. Vetter, and M. Svilans, “Ionization coefficient measurements in InP by using multiplication noise characteristics of InP/InGaAs separate absorption, grading, charge, and multiplication (SAGCM) avalanche photodiodes (APDs),” Proc. SPIE 3287, 48–59 (1998). [CrossRef]
19. Y. Jiang and J. Chen, “Optimization of the Linearity of InGaAs/InAlAs SAGCM APDs,” J. Lightwave Technol. 37(14), 3459–3464 (2019). [CrossRef]
