We report the first InGaAsP-based uni-travelling carrier photodiode structure grown by Solid Source Molecular Beam Epitaxy; the material contains layers of InGaAsP as thick as 300 nm and a 120 nm thick InGaAs absorber. Large area vertically illuminated test devices have been fabricated and characterised; the devices exhibited 0.1 A/W responsivity at 1550 nm, 12.5 GHz −3 dB bandwidth and −5.8 dBm output power at 10 GHz for a photocurrent of 4.8 mA. The use of Solid Source Molecular Beam Epitaxy enables the major issue associated with the unintentional diffusion of zinc in Metal Organic Vapour Phase Epitaxy to be overcome and gives the benefit of the superior control provided by MBE growth techniques without the costs and the risks of handling toxic gases of Gas Source Molecular Beam Epitaxy.
©2012 Optical Society of America
Research on sub-millimetre and terahertz (THz) waves has been an area of strong recent interest as the nature of these electromagnetic waves is suited to spectroscopic sensing and ultra-broadband wireless communications. A major obstacle to developing applications of these waves is the lack of portable solid-state signal sources capable of generating high output power and operating at room temperature. For the generation of sub-millimetre and terahertz waves, photonic techniques are considered to be superior to conventional electronic techniques with respect to maximum frequency, tuneability, and stability . Furthermore, the use of optical fibres enables the distribution of such high-frequency signals over long distances. In such a system, achieving high optical-to-electrical (O-E) conversion efficiency plays a critical role. The devices intended to accomplish this conversion (photodiodes) should operate at long optical wavelengths (1.3-1.55 μm) for compatibility with telecommunications wavelength optical sources and optical amplifiers, must be able to generate high output current and operate at high speed (high power and broad-band devices). Among various types of long-wavelength photodiode technologies, the uni-travelling carrier photodiode (UTC-PD) and its derivatives have exhibited the highest output powers at frequencies from 100 GHz to 1.5 THz [1,2] and very high output RF power [3,4], with steady improvement in layer and device structures since their debut in 1997.
All the UTC-PDs exhibiting record breaking performance in terms of bandwidth and output power, for operation at 1.55 μm wavelength, have been fabricated on Al-free InGaAsP based materials: record-high output power of 0.75 W (28.8 dBm) at 15 GHz and a third order output intercept point (OIP3) up to 59 dBm ; 3dB bandwidth of 310 GHz and a pulse width (FWHM) of 0.97 ps ; 5 μW at 1 THz ; maximum saturation output power of 20.8 mW at 100 GHz ; resonant peak exhibiting a maximum (detected) output power of 10.9 µW at 1.04 THz ; 3 dB bandwidth higher than 110 GHz, up to 36 mA photocurrent and a record breaking 10 dBm extracted power at 110 GHz ; record breaking emission of up to 148 µW at 457 GHz and 25 µW at 914 GHz ; record levels of Terahertz figure of merit (PTHz/Popt2 in W−1) ranging from 1 W−1 at 110 GHz to 0.0024 W−1 at 914 GHz . Lattice-matched compounds of InGaAsP allow for composition of absorptive and transparent layers; InGaAsP/InP properties include bandgap variation between 1.65 μm and 0.92 μm depending on the InGaAsP composition and absorption constant for In0.53Ga0.47As of approximately 7,000 cm−1 at 1.55 μm . This material system is also desirable because many types of devices can be constructed without the use of aluminium, an advantage as aluminium containing devices can have reduced life-time due to aluminium oxidation .
Metal Organic Vapour Phase Epitaxy (MOVPE) and Gas Source Molecular Beam Epitaxy (GSMBE) have long been the predominant growth techniques for the production of high quality InGaAsP material . Until the early 1990s, advanced phosphide epitaxy was only possible by means of these two techniques due to the problems associated with the high vapour pressure of white phosphorus, which is unfavourable for conventional Knudsen-type Solid Source Molecular Beam Epitaxy (SSMBE) furnaces . In MOVPE and GSMBE, arsenic and phosphorus are supplied from arsine and phosphine gas; however, because of the high toxicity of arsine and phosphine, special and expensive safety precautions must be taken [13,15]; besides, phosphine is not ideal because it may also introduce water vapour into the reactor chamber . Furthermore zinc is the most frequently used acceptor in InP based compounds grown by MOVPE but it has been shown that very rapid Zn diffusion and other redistribution effects occur at Zn concentrations exceeding the mid 1018 cm−3 range making Zn doping profiles difficult to control in this concentration regime ; displacements of pn-junctions can occur with deleterious impact on initial device characteristics [16,17].
SSMBE is an attractive technique since comparable material quality can be obtained without the cost and difficulty of handling toxic gases . For SSMBE arsenic and phosphorus are supplied by subliming solid arsenic and phosphorus, since toxic solids are inherently easier to control than toxic gases, the safety hazards and costs of solid group V sources are considerably less than those of gas group V sources . Growth of InGaAsP by SSMBE began in the 1970s but did not progress far due to difficulties related to the growth of arsenide phosphide materials . It was not until the introduction of group V valved cracker sources, arsenic  in 1990 and phosphorus  in 1991, that SSMBE research for InGaAsP moved forward once again. After 1991 a number of growth experiments on phosphide epitaxy by SSMBE were reported and high quality InGaAsP materials were achieved  but results on optoelectronic devices were scarce in the literature and device characterisation data only became available in mid 1990s. The first SSMBE grown InGaAsP high quality laser was demonstrated in 1995 for the wavelength of 1.35 µm  followed by other lasers emitting at λ = 0.98 µm, 0.68 µm and 1.5 µm. SSMBE is an attractive growth technique for UTC devices, since the p contact layer can be doped using beryllium at 500 °C for a phosphorus-based UTC, without the drawbacks caused by the use of zinc. In addition, SSMBE allows realisation of doping profiles that would be more difficult to achieve by MOVPE and above all offers an extremely high degree of control over the local composition, nearly on an atomic layer scale, making it possible to realise experimentally almost any band diagram that can be drawn.
However, to the best of the authors' knowledge there is no reported work on InGaAsP-based UTC devices grown by SSMBE.
We report here the SSMBE growth, device fabrication and characterisation of an InGaAsP-based UTC-PD. This work is intended to be the first stage of work aiming to combine and optimise the merits of SSMBE (no zinc diffusion issues, monolayer control, safe & cost effective) and the proven high potential of InGaAsP based materials for UTC-PD fabrication.
2. Structure growth experiment
The UTC-PD epitaxial structure that we have grown by SSMBE for this work, is similar to the one we have reported previously  (grown by MOVPE) but is intended to achieve band gap engineering improvements (more accurately localised transitions) and more precise doping profiles. The detailed epitaxy diagram is shown in Table 1 . The cap layer consists of a thick (200 nm) layer of InGaAsP (λg = 1.3 µm) functioning as both diffusion block and p-contact. For the absorption layer, we have tried to exploit the superior growth control of the MBE by splitting it into 5 different levels with a graded doping concentration, aimed at achieving further electron acceleration. Two 10 nm spacer layers of InGaAsP (λg = 1.3 and λg = 1.1) were inserted between the absorber and the 300 nm thick InP carrier collection layer. Although in this paper we present results for simple, vertically illuminated, test devices, the main goal that we aim to achieve in our future research is the realisation of edge coupled wave guide photodiodes, as these have shown superior performance [2,8–10]; for this reason our structure also includes a waveguide layer consisting of another thick (300 nm) layer of InGaAsP (λg = 1.3) grown on the 700 nm InP n-contact layer.
The phosphorus-based UTC was grown on an Fe-doped SI (100) InP substrate at 490 °C by a Veeco Gen930 solid-source MBE system, which includes Veeco SUMO cells for Indium, Gallium and Aluminum as well as valved cracker cells for Phosphorus and Arsenic. P2 and As4 were used for the growth. An in situ 15 keV RHEED system was used to monitor the growth process and the substrate temperature was measured by a pyrometer. Under P2 rich conditions, the InP substrate was heated to a temperature of 530 °C to deoxidise the surface. Si and Be sources were used for n-type and p-type doping respectively.
The lattice matched quaternary InGaAsP layers require accurate control of the flux ratio for As/P. To calibrate the thickness and composition of the InGaAsP layers, a 300 nm thick InGaAsP layer and subsequent 7-period (10-nm InGaAsP/10-nm InP) superlattice were checked by high-resolution X-ray diffraction (HRXRD). All structures were characterised by HRXRD with lattice mismatch below 10−3, such as the HRXRD diagram of InGaAsP (λg = 1.3 µm) shown in Fig. 1 .
3. Test-device fabrication
Large area vertically-illuminated test devices were fabricated to characterise the new material. To the best of our knowledge these are the first UTC-PDs fabricated from an InGaAsP-based structure grown by solid source MBE. The fabricated device is shown in Fig. 2(a) , while Fig. 2(b) gives a schematic cross-section of the structure.
The fabrication process we used is based on that previously employed by us to realise Multiple-Quantum-Well Asymmetric Fabry-Perot Modulators . Some changes in processing the UTC-PDs are required due to differences in the layer structure.
The p-contact pads, the 5 µm wide air bridges and the mesa-rings marking out the 20 µm diameter optical window, were obtained by standard photolithography, metal evaporation (20 nm of chrome and 300 nm of gold) and lift-off technique with acetone. The complete removal of excess metal from the optical window is essential for efficient device operation. Therefore two slots were included in the mesa-rings to favour a successful lift-off that would otherwise be extremely difficult inside closed rings.
Prior to metal evaporation (20 nm of chrome and 300 nm of gold) to make the n-contact pads, the n-contact area was etched 940 nm down to the n++-InP layer (n-contact layer). Because of the relatively limited thickness of this layer (700 nm) the non-selective etching, employed in , was not considered a suitable option for this case as precise control of the etch rate and the etched profile would have been necessary to obtain a floor entirely inside the n-contact layer. Instead, three successive selective etchants were employed as follows: 1) H2SO4:H2O2:H2O (1:1:8) for preferential etching of InGaAsP + InGaAs (340nm) on InP [24,25]; 2) HCl:H3PO4 (1:1) for preferential etching of InP (300 nm) on InGaAsP [24,26]; 3) H2SO4:H2O2:H2O (1:1:8) for preferential etching of InGaAsP (300 nm) on InP. This selective procedure made it possible to stop precisely on the n-contact layer and provided a flat floor for the n-contact pad metal deposition. The drawback found was the significant undercut of the sulphuric acid solution on InGaAs.
The rectangular area between the gold ring and the n-contact pad, shown in Fig. 2(a), was then etched down to the same depth as the n-contact area with the same selective process. This step opened up part of the mesa side-wall.
In the last stage of the fabrication process the two contact pads, the bridges, the mesa area and the optical window were protected by resist and a deep etch was carried out down to the semi-insulating InP substrate with a non-selective Adachi etch HBr:CH3COOH:K2Cr2O7 (1:1:1) . This etching step was long enough to allow the undercutting action of Adachi to completely clear the material underneath the bridges and thus form air-bridges. Individual UTCs were electrically isolated and the remaining mesa sidewalls were opened up.
4. Device characterisation
The photodiode I-V and C-V characteristics, measured with a KEITHLEY 4200 probe station, are displayed in Fig. 3 and Fig. 4 . The device exhibited 70 mA of current under 3 V of forward bias, 24 Ω series resistance, 23 nA of dark current under a reverse bias of 5 V and a capacitance of 200 fF; the measured capacitance is in good agreement with the theoretical value calculated considering that the whole mesa area (radius ≈15 µm) contributes to the junction capacitance. Transmission line model measurements we performed returned relatively high values of specific contact resistance, which account for the high value (24 Ω) of series resistance shown by the photodiode.
The device was then mounted on a temperature controlled brass block (maintained at 22 °C) and illuminated with a 10° lensed fibre for all the following measurements. Uniform illumination of the optical window can minimise the space charge effect by reducing the density of the photogenerated carriers; it was demonstrated [28,29] that the compression current can double for the same applied voltage if the Gaussian beam is expanded so that 5-10% of the light misses a circular absorbing region. Our measurements were made without optimisation of the beam width for our 20 µm diameter optical window in this way.
Silver epoxy was employed to connect the n-contact pad to the block, which acted as the signal ground, and the p-contact pad to the centre of a 2 cm long 50 Ω characteristic impedance microstrip transmission line. SMA connectors were then attached to both ends of the microstrip line with one end matched in a 50 Ω termination through a bias-T blocking the DC and the other connected to the measuring instrument (Lightwave component analyser or electrical spectrum analyser) and the ammeter, through a second bias-T. We employed this microstrip mount because the geometry of the fabricated test device was not suitable for direct measurement with a ground-signal-ground probe.
Figure 5 shows the photocurrent versus the optical power under different values of applied bias. For 50 mW input optical power and a reverse bias of 2 V the photodiode exhibits a responsivity of 0.1 A/W at 1.55 µm. This value matches very well the theoretical responsivity calculated for our 120 nm thick absorption layer, assuming an absorption constant for In0.53Ga0.47As of approximately 7,000 cm−1 at 1.55 μm.
For RF power and relative frequency response measurements a Mach-Zehnder modulator was used for the external modulation of the optical signal. The device relative frequency response, measured with the Lightwave component analyser calibrated to an open, a short and a 50 Ω load, is shown in Fig. 6 ; the photodiode exhibits a 3dB bandwidth of approximately 12.5 GHz at a bias of −2.4 V.
The output RF power delivered over the frequency range is shown in Fig. 7 . Because of the mount employed for these measurements, the effective load connected to the photodiode is 25 Ω, that is the parallel connection of two 50 Ω loads (the termination and the measuring instrument). The RF power plotted in Fig. 7 is the power delivered to this effective 25 Ω load. The local minima of power at 10 GHz, 15 GHz and 20 GHz visible in Fig. 7 and consistently also present in Fig. 6, are due to discontinuities at the connections between the mount and the loads. The RC limited bandwidth, given the photodiode 200 fF capacitance and 24 Ω resistance, and the 25 Ω load, is 1/2πRC ≈16 GHz.
The maximum output RF power delivered by the photodiode to the effective 25 Ω load at 10 GHz, for a −4 V bias and a 13.7 dBm input optical power, was −5.8 dBm; in these conditions the device exhibited a photocurrent of 4.8 mA. This measurement was performed at 10 GHz, falling in the frequency response local minimum observable at the same frequency. This RF power is comparable to the output power generated by other large area (from 25 µm to 40 µm diameter) normal incidence UTCs for similar values of photocurrent [30–32]; these referenced UTCs show very high photocurrent and responsivity, having been fabricated from epitaxial layers optimised for vertical coupling with absorption layer thicknesses ranging from 450 nm to 1.5 µm. The maximum DC photocurrent generated by our normal incidence test devices was 7.5 mA for a −6 V bias and a 17 dBm input optical power; in these conditions the output RF power was −10 dBm (at 10 GHz). The main factor limiting the photocurrent of our devices is due to the fact that the layer structure we grew by SSMBE is optimised for edge coupled waveguide devices and is not suited to the fabrication of high performance vertically illuminated devices; the absorption layer is thin (120 nm) and provides limited responsivity under vertical illumination while is optimised for the edge coupled waveguide devices that we aim to fabricate and study in our future research. The second factor is related to the fact that we did not perform any optimisation of the laser spot size on the optical window; as explained in the above referenced papers [28,29], this optimisation plays an important role when coupling light into large area vertically illuminated devices. Other secondary factors may have played a role, such as the high series resistance and the significant undercut of sulphuric acid on InGaAs which may have shrunk the absorption layer in the mesa.
We have reported the first uni-travelling carrier photodiodes fabricated on InGaAsP-based material grown by solid source MBE. Large area vertically illuminated devices have been fabricated to test the material.
This research aims to combine the merits of InGaAsP materials for the fabrication of uni-travelling carrier photodiodes with the advantages of Solid Source Molecular Beam Epitaxy. InGaAsP has been extensively employed in the literature to produce record breaking performance UTCs and is also preferred to quaternary materials containing aluminium because high optical quality aluminium compounds are difficult to grow and devices constructed with it can suffer reliability problems caused by aluminium oxidation. By using Solid Source Molecular Beam Epitaxy it is possible to solve the major problem associated with the diffusion of zinc in MOVPE and exploit the superior control provided by MBE growth techniques without the costs and the risks of handling toxic gases in Gas Source MBE.
In our future research we will test new versions of the same layer structure by fabricating coplanar waveguide (CPW) integrated edge coupled waveguide UTC-PDs. The integration of the CPW on the chip will allow the microstrip mount to be removed from the measurement system, while the edge coupled waveguide design is expected to improve significantly the device performance in terms of bandwidth and responsivity [2,8–10].
This work has been supported by the United Kingdom Engineering and Physical Sciences Research Council (EPSRC), grant number: EP/J017671/1.
References and links
1. T. Nagatsuma, H. Ito, and T. Ishibashi, “High-power photodiodes and their applications,” Laser Photonics Rev. 3, 123–137 (2009). [CrossRef]
2. E. Rouvalis, C. Renaud, D. Moodie, M. Robertson, and A. Seeds, “Continuous wave terahertz generation from ultra-fast InP-based photodiodes,” IEEE Trans. Microw. Theory Tech. 60(3), 509–517 (2012). [CrossRef]
3. A. Beling, Z. Li, Y. Fu, H. Pan, and J. C. Campbell, “High-power and high-linearity photodiodes,” IEEE Photonic Society 24th Annual Meeting 1, 19–20 (2011).
4. X. Wang, N. Duan, H. Chen, and J. C. Campbell, “InGaAs – InP photodiodes with high responsivity and high saturation power,” IEEE Photon. Technol. Lett. 19(16), 1272–1274 (2007). [CrossRef]
5. H. Ito, T. Furuta, S. Kodama, and T. Ishibashi, “InP/InGaAs uni-travelling-carrier photodiode with 310 GHz bandwidth,” Electron. Lett. 36(21), 1809–1810 (2000). [CrossRef]
6. H. Ito, T. Nagatsuma, A. Hirata, T. Minotani, A. Sasaki, Y. Hirota, and T. Ishibashi, “High-power photonic millimetre wave generation at 100 GHz using matching-circuit-integrated uni-travelling-carrier photodiodes,” in IEE Proc.-Optoelectron. - (IET, 2003), 150, 138–142.
7. H. Ito, F. Nakajima, T. Furuta, and T. Ishibashi, “Continuous THz-wave generation using antenna-integrated uni-travelling-carrier photodiodes,” Semicond. Sci. Technol. 20(7), S191–S198 (2005). [CrossRef]
8. C. Renaud, D. Moodie, M. Robertson, and A. Seeds, “High output power at 110 GHz with a waveguide uni-travelling carrier photodiode,” Lasers and Electro-Optics Society 2007. LEOS 2007. The 20th Annual Meeting of the IEEE 782–783 (2007).
9. C. Renaud, M. Robertson, D. Rogers, R. Firth, P. Cannard, R. Moore, and A. Seeds, “A high responsivity, broadband waveguide uni-travelling carrier photodiode,” Proc. SPIE 6194, 61940C (2006).
10. E. Rouvalis, C. C. Renaud, D. G. Moodie, M. J. Robertson, and A. J. Seeds, “Traveling-wave uni-traveling carrier photodiodes for continuous wave THz generation,” Opt. Express 18(11), 11105–11110 (2010). [CrossRef] [PubMed]
11. C. Doerr, “InP-based high-speed photonic devices,” in Proceedings of Optical Fiber Communications (OFC)/National Fiber Optic Engineers Conf. (2008).
12. J. Baillargeon, A. Cho, F. Thiel, R. J. Fischer, P. J. Pearah, and K. Y. Cheng, “Reproducibility studies of lattice matched InGaAsP on (100) InP grown by molecular beam epitaxy using solid phosphorus,” Appl. Phys. Lett. 65(2), 207–209 (1994). [CrossRef]
13. C. C. Wamsley, M. W. Koch, and G. W. Wicks, “Solid source molecular beam epitaxy of InGaAsP/InP: growth mechanisms and machine operation,” J. Vac. Sci. Technol. B 14(3), 2322–2324 (1996). [CrossRef]
14. M. Pessa, M. Toivonen, M. Jalonen, P. Savolainen, and A. Salokatve, “All-solid-source molecular beam epitaxy for growth of III–V compound semiconductors,” Thin Solid Films 306(2), 237–243 (1997). [CrossRef]
15. P. A. Claxton, J. S. Roberts, J. P. R. David, C. M. Sotomayor-Torres, M. S. Skolnick, P. R. Tapster, and K. J. Nash, “Growth and characterisation of quantum wells and selectively doped heterostructures of InP/Ga0.47In0.53As grown by solid source MBE,” J. Cryst. Growth 81(1-4), 288–295 (1987). [CrossRef]
16. E. Schubert, S. Downey, C. Pinzone, and A. Emerson, “Evidence of very strong inter-epitaxial-layer diffusion in Zn-doped GaInPAs/InP structures,” Appl. Phys. Mater. Sci. 60, 525–527 (1995).
17. N. Otsuka, M. Kito, M. Ishino, Y. Matsui, and F. Toujou, “Control of double diffusion front unintentionally penetrated from a Zn doped InP layer during metalorganic vapor phase epitaxy,” J. Appl. Phys. 84(8), 4239 (1998). [CrossRef]
18. M. Henini, “Recent developments in InP and related compounds,” III-Vs Rev. 13(2), 34–43 (2000). [CrossRef]
19. M. Panish, “Molecular beam epitaxy of GaAs and InP with gas sources for As and P,” J. Electrochem. Soc. 127(12), 2729–2733 (1980). [CrossRef]
20. D. Miller and S. Bose, “Design and operation of a valved solid-source As2 oven for molecular beam epitaxy,” J. Vac. Sci. Technol. B 8(2), 311–315 (1990). [CrossRef]
21. G. Wicks, M. Koch, J. Varriano, F. G. Johnson, C. R. Wie, H. M. Kim, and P. Colombo, “Use of a valved, solid phosphorus source for the growth of Ga0.5In0.5P and Al0.5In0.5P by molecular beam epitaxy,” Appl. Phys. Lett. 59(3), 342–344 (1991). [CrossRef]
22. M. Toivonen, A. Salokatve, M. Jalonen, J. Näppi, H. Asonen, M. Pessa, and R. Murison, “All solid source molecular beam epitaxy growth of 1.35 μm wavelength strained-layer InGaAsP quantum well laser,” Electron. Lett. 31(10), 797–799 (1995). [CrossRef]
23. C. P. Liu, A. Seeds, J. S. Chadha, P. N. Stavrinou, G. Parry, M. Whitehead, A. B. Krysa, and J. S. Roberts, “Design fabrication and characterisation of normal-incidence 1.56-mum multiple-quantum-well asymmetric Fabry-Perot modulators for passive picocells,” IEICE Trans. Electron. E Series C 86, 1281–1289 (2003).
24. F. Fiedler, A. Schlachetzki, and G. Klein, “Material-selective etching of InP and an InGaAsP alloy,” J. Mater. Sci. 17(10), 2911–2918 (1982). [CrossRef]
25. D. Pasquariello, E. Bjorlin, D. Lasaosa, Y.-J. Chiu, J. Piprek, and J. E. Bowers, “Selective undercut etching of InGaAs and InGaAsP quantum wells for improved performance of long-wavelength optoelectronic devices,” J. Lightwave Technol. 24(3), 1470–1477 (2006). [CrossRef]
26. S. Phatak and G. Kelner, “Material-selective chemical etching in the system InGaAsP/ InP,” J. Electrochem. Soc. 126(2), 287 (1979). [CrossRef]
27. S. Adachi, “Chemical etching of InP and InGaAsP/InP,” J. Electrochem. Soc. 129(3), 609 (1982). [CrossRef]
28. K. Williams and R. D. Esman, “Design considerations for high-current photodetectors,” J. Lightwave Technol. 17(8), 1443–1454 (1999). [CrossRef]
29. G. Davis, R. Weiss, R. LaRue, K. J. Williams, and R. D. Esman, “A 920-1650-nm high-current photodetector,” IEEE Photon. Technol. Lett. 8(10), 1373–1375 (1996). [CrossRef]
30. N. Duan, X. Wang, N. Li, H. Liu, and J. C. Campbell, “Thermal analysis of high-power InGaAs–InP photodiodes,” IEEE J. Quantum Electron. 42(12), 1255–1258 (2006). [CrossRef]
31. M. Chtioui, D. Carpentier, S. Bernard, B. Rousseau, F. Lelarge, F. Pommereau, C. Jany, A. Enard, and M. Achouche, “Thick absorption layer uni-traveling-carrier photodiodes with high responsivity, high speed, and high saturation power,” IEEE Photon. Technol. Lett. 21(7), 429–431 (2009). [CrossRef]
32. M. Chtioui, F. Lelarge, A. Enard, F. Pommereau, D. Carpentier, A. Marceaux, F. van Dijk, and M. Achouche, “High responsivity and high power UTC and MUTC GaInAs-InP photodiodes,” IEEE Photon. Technol. Lett. 24(4), 318–320 (2012). [CrossRef]