IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 17, NO. 1, JANUARY/FEBRUARY 2011 79
Ultrasensitive, Room Temperature Detection of THzRadiation Using Nonlinear Parametric Conversion
Mohammad Jalal Khan, Jerry C. Chen, Zong-Long Liau, and Sumanth Kaushik
(Invited Paper)
Abstract—We demonstrate ultrasensitive, room temperature op-tical detection of terahertz by using nonlinear parametric upcon-version. Terahertz radiation is mixed with pump light at 1550 nmin quasi-phase-matched GaAs crystal to generate an optical side-band or idler wave that is coupled into optical fiber and detectedusing a Geiger-mode APD. The resulting terahertz detector has anoise equivalent power of 78 fW/Hz1/2 with a timing resolution of1 ns.
Index Terms—Optical frequency conversion, optical phasematching, submillimeter wave detectors, submillimeter wave fre-quency conversion.
I. INTRODUCTION
T ERAHERTZ radiation has attracted a lot of interest in re-cent years. Its penetrating capability, nonionizing nature,
and the fact that many materials have fingerprint spectra at thisfrequency regime have resulted in increasing use of terahertzradiation for standoff sensing applications like explosive detec-tion [1], penetrative imaging [2], nondestructive evaluation [3],biomedical imaging [4], drug interdiction [5], and vibrationsensing [6]. The capability of these systems depends directly onthe availability of high-power terahertz sources and ultrasensi-tive, fast terahertz detectors. Both these technology areas are be-ing actively researched [5], [7]. However, compared to the nearinfrared, terahertz technology is relatively immature. Commer-cially available, room temperature direct detectors, like Golaycells and pyroelectric, have poor sensitivities. Other commercialterahertz detectors, like bolometers, are much more sensitive butrequire liquid helium cooling, and like their room temperaturecounterparts, have small electrical bandwidths. In contrast, opti-cal detectors are a mature and commercial technology that offersexcellent sensitivity with large bandwidths and room tempera-ture operation. Photon multiplier tubes and Geiger-mode APDs
Manuscript received January 31, 2010; revised March 8, 2010; acceptedMarch 9, 2010. Date of publication April 19, 2010; date of current versionFebruary 4, 2011. This work was supported by the Office of the Secretaryof Defense (OSD)/Director of Defense Research and Engineering (DDR&E)Quick Reaction Fund under Naval Surface Warfare Center Dahlgren Division(NSWCDD) oversight and the U.S. Air Force under Contract FA8721-05-C-0002. Opinions, interpretations, recommendations, and conclusions are those ofthe authors and are not necessarily endorsed by the U.S. Government.
The authors are with the Massachusetts Institute of Technology, LincolnLaboratory, Lexington, MA 02420 USA (e-mail: [emailprotected];[emailprotected]; [emailprotected]; [emailprotected]).
Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JSTQE.2010.2045737
Fig. 1. Terahertz detector concept that leverages optical technologies to per-form ultrasensitive, room temperature detection.
(GM-APDs) enable detection down to the single-photon limitat room temperature, with very fast performance [8], [9].
As a result, researchers have used various optical pumps andnonlinear crystals to frequency upconvert terahertz waves to op-tical signals for detection with optical components. Many groupsused lithium niobate; their pumps were Nd:YAG [10]–[12] andti-sapphire [13]. Nd:YAG pumps were also used with ZnGeP2[14], GaSe [15], and GaP [16] detecting sub-nJ terahertz pulses.At cryogenic temperatures, a group achieved a power conver-sion efficiency of 4.5 × 10−6 with 1.3-μm pump and GaAscrystal [17]. Recently, a 1.3-μm pump was paired with DASTfor terahertz detection [12].
Our scheme to detect terahertz waves leverages telecommu-nications or 1550-nm technologies to enable ultrasensitive roomtemperature detection. We employ GaAs, a χ(2) nonlinear crys-tal pumped with a readily obtainable erbium-doped fiber ampli-fier, to parametrically upconvert a terahertz signal to telecommwavelengths for subsequent detection by an optical detector. Ourinitial work used a low-cost InGaAsP p-i-n [18]. Subsequently,we used a GM-APD achieving 4.5 pW/Hz1/2 noise equivalentpower (NEP) with 1 ns temporal resolution [19]. χ(2) nonlinearinteractions are intrinsically very fast; our temporal bandwidthwas limited by the optical detector.
In this paper, we improve our previous NEP by over 50 times.This NEP of 78 fW/Hz1/2 and power conversion efficiency of1.2 × 10−3 are the best reported, to our knowledge. Terahertzdetection is improved by diffusion bonding and AR coatingthe GaAs, and by increasing optical pump powers. This paperpresents both theoretical and experimental results.
II. EXPERIMENTAL LAYOUT
A schematic representation of a terahertz detector using thistechnique is shown in Fig. 1. In our study, we have used GaAsas the nonlinear material to perform upconversion. GaAs waschosen because of its high χ(2) nonlinear coefficient [20] andlow absorption losses at terahertz and optical frequencies [21].Our transmission measurements found losses to be 0.2 and
1077-260X/$26.00 © 2010 IEEE
80 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 17, NO. 1, JANUARY/FEBRUARY 2011
Fig. 2. Experimental layout of terahertz detector.
0.065 cm−1 at terahertz (0.5–0.7 THz) and optical frequencies(near 193 THz), respectively.
The upconversion experimental layout is shown in Fig. 2. Inour experiments, we employed two different terahertz sources.One was a backward wave oscillator (BWO) from MicrotechInstruments. The other was Virginia Diode Inc’s amplified mul-tipler chain (VDI-AMC). Both terahertz sources produced atunable, continuous-wave signal. We operated them around thefrequency that produced maximum power. For the BWO, thiswas around 700 GHz, where it provided about 2.5 mW as mea-sured by using a Thomas–Keating power meter. The VDI-AMCwas typically operated at 820 GHz, where it produced 120 μW.Although the BWO produces higher power, it is multimoded;by comparison, the VDI-AMC produced a near Gaussian mode,which is more suitable for frequency conversion. The terahertzsource radiation was collected and reimaged onto the GaAscrystal by using a pair of 3-in diameter, off-axis parabolic goldmirrors; the crystal position was adjusted such that the beamwaist was centered inside the crystal. In our experiments, weemployed both bulk and quasi-phase-matched (QPM) crystals.With bulk crystals, the length of GaAs was constrained to thecoherent buildup length [22] calculated to be about 4 mm at theoperating terahertz frequency based on the reported indices ofrefraction [21].
Our high-power optical pump source was seeded by an Agi-lent external cavity tunable laser diode operated at 1550 nm. The1550-nm light was routed into a semiconductor optical amplifier(SOA) whose supply current was switched with a 10-ns-widepulse at a 50-kHz repetition rate. The rise and fall time of theSOA was measured to be about 3.5 ns and the resulting pumppulsewidth was about 6.5 ns. The output of the SOA was op-tically filtered, amplified, and passed through a second filterstage to reduce out-of-band amplified spontaneous emission.Finally, this signal was amplified to the operating power levelsby using a high-power, large mode area erbium-doped fiber am-plifier (EDFA). The resulting high-power optical source is veryflexible and allows easy adjustment of power, pulsewidth, andrepetition rate. The typical operating pump power was about1–2 W average, which corresponds to a 2–4 kW peak power.The optical pump radiation was collimated to a spot size of
Fig. 3. Optical spectrum of the high-power, pulsed optical pump.
about 0.9 mm and then passed through two free-space notch orbandpass filters. The filters were each angle tuned for maximumtransmission at 1550 nm and had an insertion of loss of less than1 dB per filter. The filter’s 3-dB bandwidth was about 130 GHzwith an out-of-band rejection of about 30 dB per filter. This fil-tering suppressed the optical noise that was produced by the laststage high-power EDFA, enabling easier detection of the idler.The M2 factor of the last stage power amplifier was about 2. Inour experiments, the typical optical pump intensity inside theGaAs crystal was about 200 kW/cm2 . The two-photon absorp-tion coefficient of GaAs at 1550 nm is about 10 cm/GW [22];consequently, nonlinear absorption does not contribute signifi-cantly to material loss until intensities approach 100 MW/cm2 .Our measurements of two-photon absorption in GaAs were con-sistent with this data. Given the operating pump intensities inour experiments, two-photon absorption effects were negligible.The optical pump was copropagated with the terahertz waves byemploying a 5-mm glass prism centered at the second parabolicgold mirror. The small size of the glass prism ensured that theterahertz beam was minimally occluded.
The overlapping terahertz and optical radiation nonlinearlymix in the GaAs to generate the optical idler. The terahertz signalis vertically linearly polarized, whereas the pump is randomlypolarized. We used an in-line polarization rotator prior to thelast EDFA to maximize the generated idler [23]. The opticalidler is then separated from the pump by using three free-spacelong-pass filters in series. Each filter had a transition band ofabout 4 nm and a rejection of about 25 dB. The filters were angletuned to reflect the 1550-nm pump light but to pass wavelengthsexceeding 1554 nm. We verified that pump rejection throughthese three filters exceeded 70 dB; this measurement was limitedby detector noise. The idler suffered an aggregate insertion lossof about 4 dB through the filters. We did not make any specialeffort to separate the terahertz beam from the idler. Since theterahertz beam is a rapidly expanding beam and is not collectedby the 1550-nm optics, this was not a concern; additionally, theoptical detector is not sensitive to the terahertz radiation. Fig. 3shows the spectrum of optical source at a resolution bandwidthof 0.1 nm. The optical spectrum was measured coming out of thelast stage EDFA. The fiber collimator that sends light to the Andooptical spectrum analyzer (OSA) was purposely misaligned toavoid instrument damage. As can be seen, the optical source hasan optical SNR in excess of 25 dB.
KHAN et al.: ULTRASENSITIVE, ROOM TEMPERATURE DETECTION OF THz RADIATION 81
Fig. 4. RF spectrum of New Focus 2153 p-i-n diode output; the optical idlershows up as a clear tone at the chopper frequency.
The idler photons can then be coupled either into fiber orrouted to a p-i-n diode using a flip mirror. We retained theability to route the idler beam into a p-i-n diode, as in [18], toenable independent optimization of the upconversion processand coupling of the idler signal into fiber. Initially, the idlergeneration was maximized by optimally coaligning the terahertzand optical pump beams and then we coupled the idler beaminto fiber by removing the flip mirror and using the six-axisfiber collimator. Once in fiber, the idler is routed via a JDSUniphase switch/attenuator into a Princeton Lightwave GM-APD. Alternately, the light can be coupled into an Ando OSAor to a power meter. The terahertz beam is chopped only whenthe p-i-n diode is employed; the chopper is removed when idlerlight is coupled into the GM-APD via the optical fiber.
III. EXPERIMENTAL RESULTS
A. Bulk Versus QPM GaAs
As part of the upconversion optimization, the idler was ini-tially detected using a New Focus 2153 p-i-n diode with anNEP ≈23 fW/Hz1/2 . The terahertz signal was chopped at49.5 Hz to allow additional discrimination between the gen-erated idler and any remnant pump. The idler power is detectedby looking at the p-i-n diode output on an audio spectrum ana-lyzer; alternately, a lock-in amplifier could have been employed.Since the idler is proportional to the product of the terahertz sig-nal and the optical pump, it is produced at the chopper frequency,unlike the unchopped pump. Therefore, any signal at the chop-per frequency detected by the diode must correspond to idlerpower. Fig. 4 shows the electrical spectrum of the p-i-n diodesignal. We see a clear spectral tone at the chopper frequency,along with higher harmonics, that together correspond to theidler. To confirm that the harmonics are indeed due to idlerpower, we turned OFF the BWO, and the tones at the chopperfrequency disappeared, as is evident in Fig. 4. The peaks at 60,120, and 180 Hz correspond to electrical line noise. By integrat-ing the power spectral density (PSD), measured in volts2 /Hertz,over the spectral peak, we can measure the average idler power,
Fig. 5. Optical idler power versus bulk GaAs length; 2-layer QPM GaAscrystal shows a 5 dB improvement over bulk GaAs.
given by
PIdler =1
Rv
[∫ +f
−f
PSD df
]1/2
(1)
where Rv is the responsivity of the p-i-n diode in volts/watts.We measured the idler power produced by bulk GaAs crystals
of varying lengths using this technique. Fig. 5 shows a plot ofmeasured idler power versus length. As expected, the conversionefficiency only increases up to the coherent buildup length andthen decreases. This is because the optical and terahertz waveswalk out of phase such that when the phase difference is π,the idler and the terahertz waves recombine to form the pump.Theoretically, we can estimate the detected power by using thefollowing expression [23]:
PIdler =8π2(deff )2L2IPumpTFiltersTFresnel
cεonT nI nP λ2I
× sin c
(ΔkL
2
)2
PTHz (2)
where deff is the effective nonlinearity, L is the length of thecrystal, IPump is the pump intensity, and nT , nI , and nP are therefractive indexes at the terahertz, idler, and pump frequencies,respectively. Δk = kP − kI − kT is the wave vector mismatch.TFilters is the combined insertion loss of the filters and TFresnelis the transmission coefficient of the idler at an uncoated GaAscrystal interface. There was excellent agreement between thetheoretically calculated idler power and the measured result.For these measurements, the terahertz source was operated at700 GHz; at this frequency the coherent buildup length is nearly4 mm based on reported refracted indexes of GaAs [21]. Fromtheory, we expect no upconversion to occur at twice the coherentbuildup length; this agrees well with the observed results (seeFig. 5).
The noise performance of the terahertz receiver improveslinearly with the terahertz-to-optical conversion efficiency. Theconversion efficiency is quadratically related to the length of
82 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 17, NO. 1, JANUARY/FEBRUARY 2011
Fig. 6. Diffusion bonding process is used to create QPM GaAs crystals byfusing together two and three orthogonal pieces of bulk GaAs.
nonlinear medium up to the coherent buildup length. To in-crease the efficiency further, we must overcome the walk offby using specially designed QPM GaAs crystals. A QPM GaAscrystal reverses the polarity of the nonlinearity just as the op-tical and terahertz beams have slipped out of phase by π ra-dians [23], by stacking alternating orthogonal orientations ofGaAs crystal. We fabricated QPM crystals by a diffusion bond-ing process [24], [25] depicted in Fig. 6. Bulk 4-mm-thick GaAscrystals of orthogonal orientation were put inside a graphitecontainer that was shimmed to tightly fit into a quartz cylinderand then heated to 700 ◦C. Differential thermal expansion ofthe graphite and quartz compresses the wafer stack into atomiccontact and fuses the GaAs pieces together. The fabricationprocess was refined to create multilayer GaAs QPM crystalswith clean bond interfaces. The exposed surfaces were repol-ished after the bonding process. Using a two-layer QPM GaAscrystal, we demonstrated a near 5 dB improvement in the con-version efficiency; the theoretical limit is 6 dB (see Fig. 5). Wealso developed an antireflection (AR) coating at 1550 nm usingaluminum oxide and tantalum pentoxide. These materials werespecifically chosen because they have low losses at terahertzfrequencies. These AR coatings allow more optical pump lightinto and idler light out of the nonlinear crystal, without absorb-ing virtually any terahertz. The result is an additional 2.5 dB ofimprovement in the terahertz-to-optical conversion efficiency.
B. Geiger-Mode APD
After optimizing the conversion efficiency using the p-i-ndiode, we coupled the light into the output optical fiber usinga matched collimator. Fig. 7 shows the optical spectrum of theoutput light measured using an OSA at a resolution bandwidthof 0.1 nm. Here, we used a two-layer QPM GaAs crystal ARcoated for the 1550-nm optical pump; the terahertz source wasthe VDI-AMC operated near 820 GHz with an output power of120 μW. When the terahertz source was turned on, an idler signalappeared at 1556.6 nm, which comes from the nonlinear mixingof the terahertz signal and the 1550-nm pump. Energy conser-vation dictates that the idler frequency is the difference betweenthe optical pump and the terahertz frequency. For an 820-GHzterahertz signal, we expect an idler exactly at 1556.6 nm as wasobserved. Conversely, our detection scheme enables the precisedetermination of the input terahertz wavelength, since the opticalpump is well known and the idler wavelength can be accurately
Fig. 7. Optical spectrum of light coupled in the fiber clearly shows the gener-ated optical idler when the terahertz source is on.
measured. The noise floor of the spectrum was limited by thesensitivity of the OSA. We confirmed that at the pump wave-length (1550 nm), no signal was visible above the noise floor. Byintegrating the power under the spectral peak, we calculated theidler power to be about 85 pW. The terahertz-to-optical photonconversion efficiency is given by
ηTHz→optical =P
(p)Idler
PTHz× νTHz
νIdler
where the peak idler power P(p)Idler is related to the average idler
power by the duty factor (20 μs/10 ns). From the OSA measuredidler power, we calculate a photon conversion efficiency of 5.9×10−6 .
The idler was then routed via an attenuator into the PrincetonLightwave GM-APD [8], which is triggered by a delayed copyof the electrical pulse that drives the SOA. The trigger initiatesdetection of idler photons in a 1-ns gate by the GM-APD. TheGM-APD is equipped with a counter, which records the totalnumber of 1-ns gates with photons over a 1 s time interval. Thesensitivity or NEP of the GM-APD is limited by the dark countrate (DCR) or the number of spurious counts per second in theabsence of any signal; its sensitivity is given by
NEP =hν
η
√DCR (3)
where η is the quantum efficiency of the GM-APD, and ν isthe frequency of the incident radiation. The DCR of the GM-APD receiver is 20 kHz, which corresponds to an NEP = 9.3 ×10−17 W/Hz1/2 . To experimentally measure the sensitivity of theterahertz detector, one can attenuate the terahertz power until thecounts registered by the GM-APD are equal to the fluctuationsin the dark count over the 1 s interval. This attenuated terahertzpower would then correspond to the minimum detectable power.Since calibrated terahertz attenuators with a large dynamic rangeare not readily available, we chose to attenuate the optical idlersignal instead. The linear relationship between the idler andterahertz power justifies this approach. Fig. 8 shows a plot of theGM counts as the attenuation of the idler beam was varied. Thex-axis values correspond to the equivalent input terahertz power,
KHAN et al.: ULTRASENSITIVE, ROOM TEMPERATURE DETECTION OF THz RADIATION 83
Fig. 8. Plot of GM-APD counts versus input terahertz power normalized bygating signal bandwidth. A linear fit of GM-APD counts (dashed line) intersectsnoise level (solid line) at 78 fW/Hz1/2 .
for a given optical attenuation, in units of NEP (W/Hz1/2). Thex-axis values are given by scaling the input terahertz power,120 μW, by the optical attenuation of the idler, and normalizingit to the square root of the equivalent bandwidth of the gatingsignal. The equivalent bandwidth of the 1-ns pulse train witha pulse separation of 20 μs and integrated over 1 s is 20 kHz[(20 μs/1 ns)× 1 Hz]. The mean number of dark counts in 1 s wasmeasured to be 1.6 and was slightly higher than the theoreticallyexpected value of 1 [(1 ns/20 μs) × DCR]. To calculate thesensitivity of the resulting terahertz detector, we extrapolate alinear fit to the data to the noise level. The fluctuation in thedark counts or noise level is governed by Poisson statistics andis given by the square root of the mean dark counts, 1.6. Usingthis method, we obtained a terahertz NEP ≈ 78 fW/Hz1/2 , asis shown in Fig. 8. This corresponds to a minimum detectablepulse energy of 1.1 × 10−20 J in the 1-ns GM-APD gate; ofcourse, we integrate over 5 × 104 pulses in 1 s.
The relationship between the NEP of the GM-APD and theresulting terahertz detector is given by
NEPGM -APD
νIdler= ηTHz→optical
NEPTHz
νTHz(4)
where ηTHz→optical = (Output idler photons)/(Input terahertzphotons) is the system photon conversion efficiency. νTHz andνIdler are the terahertz and idler frequencies, respectively. Us-ing the aforementioned equation, we calculate the end-to-endreceiver photon conversion efficiency of the system to be 5 ×10−6 ; the equivalent system power conversion efficiency is 1.2×10−3 . These numbers agree very well with the conversion ef-ficiency calculated from the OSA data. The intrinsic photonconversion efficiency is significantly higher and can be calcu-lated by taking into account the 2 dB Fresnel loss at the crystalinterface for the terahertz (the crystal is AR coated for opticalwavelengths); the 4 dB of insertion losses the idler experiencesdue to the three long-pass dielectric filters and 4 dB of fibercoupling losses.
Alternately, at a fixed attenuation setting, we can delay theGM-APD’s 1-ns gate with respect to the upconverted pulse andthereby measure the idler’s temporal characteristics, and thus,
Fig. 9. Idler’s pulse shape versus time measured by GM-APD.
infer the terahertz temporal shape. Fig. 9 plots the GM-APDcounts versus time. We can see that the upconverted idler sig-nal has a temporal width of about 6 ns. The terahertz sourceis continuous wave signal, therefore, the temporal shape of theupconverted light mirrors that of the optical pump; the measuredpulsewidth agrees well with that of the optical pump, given a1-ns timing resolution. However, if we had used a pulsed tera-hertz source, our terahertz detector would provide a 1-ns time-resolved measurement of the terahertz pulse. As is clear fromthe figure, with the terahertz source OFF, there is no upconvertedidler signal; the noise floor of the terahertz detector is only lim-ited by the noise characteristics of the GM-APD, resulting inexcellent room temperature detection sensitivity.
IV. CONCLUSION
We have used optical frequency upconversion at 1550 nm tocreate a fast, ultrasensitive room temperature terahertz receiver.Nonlinear conversion was enhanced by quasi-phase matching,where two GaAs crystals were diffusion bonded. Increased opti-cal pump powers on these AR-coated crystals resulted in record-breaking power conversion efficiency of 1.2 × 10−3 , whichagrees with theory. We show that terahertz can be detected atan unprecedented 78 fW/Hz1/2 NEP at room temperature. Thissensitivity is nearly 40 dB better than commercial room temper-ature terahertz detectors and over 15 dB better than commercial4.2-K bolometers. The detector’s temporal resolution is 1 ns.
ACKNOWLEDGMENT
The authors would like thank P. O’Brien of the MassachusettsInstitute of Technology Lincoln Laboratory for AR-coated GaAscrystals and for help with imaging the bond interfaces.
REFERENCES
[1] Y. C. Shen, T. Lo, P. F. Taday, B. E. Cole, W. R. Tribe, and M. C. Kemp,“Detection and identification of explosives using terahertz pulsed spectro-scopic imaging,” Appl. Phys. Lett., vol. 86, pp. 241116-1–241116-8, Jun.2005.
[2] A. M. Sinyukov, A. Bandyopadhyay, A. Sengupta, R. B. Barat, D. E.Gary, Z. Michalopoulou, D. Zimdars, and J. F. Federici, “Terahertzinterferometric imaging of concealed object,” Proc. SPIE, vol. 6373,pp. 63730K-1–63730K-8, Oct. 2006.
[3] J. White and D. Zimdars, “Time domain terahertz non destructive eval-uation of water intrusion in composites and corrosion under insulation,”
84 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 17, NO. 1, JANUARY/FEBRUARY 2011
presented at the Quantum Electron. Laser Sci. Conf., Baltimore, MD, May6–11, 2007, paper JWA112.
[4] E. Pickwell and V. P. Wallace, “Biomedical applications of terahertz tech-nology,” J. Phys. D, Appl. Phys., vol. 39, pp. R301–R310, Sep. 2006.
[5] M. Tonouchi, “Cutting-edge terahertz technology,” Nature Photon., vol. 1,pp. 97–105, Feb. 2007.
[6] J. C. Chen and S. Kaushik, “Terahertz interferometer that senses vibrationsbehind barriers,” IEEE Photon. Tech. Lett., vol. 19, no. 7, pp. 486–488,Apr. 2007.
[7] P. Siegel, “Terahertz technology,” IEEE Trans. Microw. Theory Tech.,vol. 50, no. 3, pp. 910–928, Mar. 2002.
[8] X. Jiang, M. A. Itzler, R. Ben-Michael, and K. Slomkowski, “InGaAsP-InP avalanche photodiodes for single photon detection,” IEEE J. Sel. Top.Quantum Mech., vol. 13, no. 4, pp. 895–905, Jul./Aug. 2007.
[9] S. Verghese, J. P. Donnelly, E. K. Deurr, K. A. McIntosh, D. C. Chapman,C. J. Vineis, G. M. Smith, J. E. Funk, K. E. Jensen, P. I. Hopman, D. C.Shaver, B. F. Aull, J. C. Aversa, J. P. Frechette, J. B. Glettler, Z. L. Liau,D. C. Oakley, E. J. Ouellette, M. J. Renzi, and B. M. Tyrrell, “Arrays ofInP-based avalanche photodiodes for photon counting,” IEEE J. Sel. Top.Quantum Electron., vol. 13, no. 4, pp. 870–884, Jul./Aug. 2007.
[10] A. A. Babin, V. N. Petryakov, and G. I. Freidman, “Use of stimulatedscattering by polaritons in detection of submillimeter radiation,” Sov. J.Quantum Electron., vol. 13, pp. 958–960, Jul. 1983.
[11] R. Guo, T. Ikari, H. Minamide, and H. Ito, “High sensitive detection ofTHz-wave at room temperature using stimulated polariton scattering inMgO:LiNbO3,” Appl. Phys. Lett., vol. 93, pp. 21106-1–21106-8, Jul.2008.
[12] H. Minamide and H. Ito, “Coherent terahertz-wave generation and de-tection over a wide frequency range using DAST crystals,” Proc. SPIE,vol. 7179, pp. 71790C1-1–71790C1-8, 2009.
[13] G. H. Ma, Q. B. Zhu, G. Kh. Kitaeva, and I. I. Naumova, “Narrow-bandterahertz wave generation and detection in one periodically poled lithiumniobate crystal,” Opt. Commun., vol. 273, pp. 549–553, May 15, 2007.
[14] Y. J. Ding and W. Shi, “Observation of THz to near-Infrared parametericconversion in ZnGeP2 ,” Opt. Exp., vol. 14, pp. 8311–8316, Sep. 2006.
[15] W. Shi, Y. J. Ding, N. Fernelius, and F. K. Hopkins, “Observation ofdifference-frequency generation by mixing of terahertz and near-infraredlaser beams in a GaSe crystal,” Appl. Phys. Lett., vol. 88, pp. 101101-1–101101-3, Mar. 2006.
[16] Y. J. Ding and W. Shi, “Efficient THz generation and frequency upcon-version in GaP crystals,” Solid State Electron., vol. 50, pp. 1128–1136,Jun. 2006.
[17] S. S. Dhillon, C. Sirtori, S. Barbieri, A. de Rossi, M. Calligaro, H. E. Beere,and D. A. Ritchie, “THz sideband generation at telecom wavelengths in aGaAs-based quantum cascade laser,” Appl. Phys. Lett., vol. 87, pp. 71101-1–71101-3, Aug. 2005.
[18] M. J. Khan, J. C. Chen, and S. Kaushik, “Optical detection of terahertzradiation by using nonlinear parametric upconversion,” Opt. Lett., vol. 32,pp. 3248–3250, Nov. 15, 2007.
[19] M. J. Khan, J. C. Chen, and S. Kaushik, “Optical detection of terahertzradiation using nonlinear parametric upconversion,” Opt. Lett., vol. 33,pp. 2725–2727, Dec. 2008.
[20] V. G. Dmitriev, G. G. Gurzadyan, and D. N. Nikogosyan, Handbook ofNonlinear Optical Crystals. ser. Springer Series in Optical Sciences 64.New York: Springer-Verlag, 1997.
[21] E. D. Pallik, Handbook of Optical Constants of Solids. San Francisco,CA: Academic, 1998.
[22] R. W. Boyd, Nonlinear Optics, 2nd ed. San Francisco, CA: Academic,2003.
[23] W. C. Hurlbut, Y.-S. Lee, K. L. Vodopyanov, P. S Kuo, and M. M. Fejer,“Multiphoton absorption and nonlinear refraction of GaAs in the mid-infrared,” Opt. Lett., vol. 32, pp. 668–670, Mar. 2007.
[24] Z. L. Liau and D. E. Mull, “Wafer fusion: A novel technique for optoelec-tronic device fabrication and monolithic integration,” Appl. Phys. Lett.,vol. 56, pp. 737–739, Feb. 1990.
[25] L. Gordon, G. L. Woods, R. C. Eckardt, R.R. Route, R. S. Feigelson,M.M. Fejer, and R. L. Byer, “Diffusion-bonded stacked GaAs for quasi-phase-matched second-harmonic generation of a carbon dioxide laser,”Electron. Lett., vol. 29, pp. 1942–1944, Oct. 1993.
Mohammad Jalal Khan was born in Karachi, Pakistan. He received the S.B.,S.M., and Ph.D. degrees in electrical engineering and computer science fromthe Massachusetts Institute of Technology (MIT), Cambridge, in 1994, 1996,and 2002, respectively. His Ph.D. study has been concerned with integratedphotonics for telecommunication optical networks.
He was at Clarendon Photonics as an R&D Engineer, where he was engagedin developing integrated optical devices for the telecommunications industry.He is currently a Technical Staff Member at the MIT Lincoln Laboratory,Lexington, MA, where he is involved in developing active optical systems forsensing and imaging applications. His research interests include using novelnonlinear techniques to detect and generate terahertz radiation.
Dr. Khan is a member of Sigma Xi, Tau Beta Pi, and Eta Kappa Nu. He isa recipient of the Lemelson-MIT Award for innovation in Telecommunicationsand Networking.
Jerry C. Chen received the B.S.E. degree in electrical engineering and theCertificate in engineering physics from Princeton University, Princeton, NJ in1989, and the S.M., E.E., and Ph.D. degrees in electrical engineering fromthe Massachusetts Institute of Technology (MIT), Cambridge, in 1991, 1995,and 1996, respectively.
From 1996 to 1997, he was Postdoctoral Research Fellow at Bell Laborato-ries, Lucent Technologies, Holmdel, NJ, where he modeled and improved densewavelength-division multiplexing demultiplexers. Afterward, he performed re-search on novel optical communication links at the MIT Lincoln Laboratory,Lexington, until 2000. From 2000 to 2002, he was at the Tellabs ResearchCenter, where he was involved in designing and implementing metro fiber net-works. In 2002, he rejoined the MIT Lincoln Laboratory, where he was engagedin research on designing and testing laser radar systems, and for the last fewyears, he has been engaged in research on terahertz receivers and sensor appli-cations. He has coauthored 23 journal articles and holds 9 patents.
Dr. Chen is a member of Optical Society of America, Phi Beta Kappa, SigmaXi, and Tau Beta Pi and also involved with the Terahertz Technical ProgramCommittee for Conference on Lasers and ElectroOptics of IEEE.
Zong-Long Liau received the B.S. degree in physics from the National TaiwanUniversity, Taipei, Taiwan, in 1972, and the Ph.D. degree in applied physicsfrom the California Institute of Technology, Pasadena, in 1979.
Since December 1978, he has been a Staff Member in the Electro-OpticalMaterials and Devices Group, Lincoln Laboratory, Massachusetts Institute ofTechnology, Lexington. He pioneered a mass transport process for semiconduc-tor microstructures, with which he demonstrated a first surface-emitting diodelaser array, and a wafer fusion process (also known as direct wafer bonding ordiffusion bonding), which has stimulated a wide range of new device applica-tions. Since 2000, he has been engaged in highly powerful microoptics, and hascontributed to several laser- and detector-array applications.
Dr. Liau was an Associate Editor for the IEEE JOURNAL OF QUANTUM
ELECTRONICS from 1993 to 2002.
Sumanth Kaushik received the B.S., M.S., and Ph.D. degrees in electrical en-gineering and computer science from the Massachusetts Institute of Technology(MIT), Cambridge, in 1987, 1989, and 1994, respectively.
He is currently the Group Leader of the Space Control Systems Group, MITLincoln Laboratory, Lexington. His research interests include advanced tech-nologies in the areas of remote sensing, metrology, and imaging.
