3-2 Study for Far-infrared and Faint Light Detection ... · The term “infrared” refers to light at wave-lengths longer than those of visible light. Specifically, infrared rays

1 Introduction

Electromagnetic waves are visible to thehuman eye at wavelengths of 0.3 to 0.75μm.The term “infrared” refers to light at wave-lengths longer than those of visible light.Specifically, infrared rays are electromagneticwaves with wavelengths ranging from 0.75μm to 1 mm. In this paper, far-infrared raysare defined as electromagnetic waves withwavelengths of 30μm to 1 mm.

At temperatures above absolute zero, allmatter emits temperature-dependant electro-magnetic waves due to the motion of surfaceatoms or molecules. In the far-infrared region,rotational spectra of molecules can beobserved. And there are many important sci-entific research fields in this region, such asthe vibration mode of impurities in solids orplasma diagnosis[1]-[3]. In astronomical obser-vation, measurement of far-infrared thermalradiation from interstellar dust has revealed anextraordinary amount of information about thebirth of stars, and important indicators ofgalactic activity. However, there is no estab-

lished method of observation of light in thiswavelength region, due to the technical diffi-culties involved.

In light of the present situation, we initiat-ed a project to develop a far-infrared detectorfor the 50 to 110-μm waveband, to be mount-ed on Japan’s first infrared space observationsatellite, the ASTRO-F, slated for launch bythe Japan Aerospace Exploration Agency(JAXA) in the summer of 2005. A gallium-doped germanium extrinsic semiconductor hasan acceptor level of 10.8 meV[4], and has beenused as a sensitive far-infrared detector with acutoff wavelength of 110 μm. To minimizethermal excitation to the acceptor level and tomaximize sensitivity, the detector is cooled tothe temperature of liquid helium (4.2 K) orlower.

Ge:Ga far-infrared detectors have longbeen known as the most sensitive quantumphotodetectors in this wavelength region,however, necessity of cooling and the fact thatfar-infrared radiation is mostly absorbed bythe atmosphere hinder commercial develop-ment of photodetector in this region. In our

FUJIWARA Mikio et al. 51

3-2 Study for Far-infrared and Faint LightDetection Technology

FUJIWARA Mikio, AKIBA Makoto, and SASAKI Masahide

To realize a sensitive photodetector, cooling down the device is a effective way becauseof reducing thermal noise and dark current. Recent space observation satellites have liquid-Helium cooled detectors. We focus on far-infrared region, in which there are many impor-tant research fields. To our knowledge, we are the first to successfully report a direct hybridtwo-dimensional detector array in the far-infrared region. Moreover, we are trying to developan ultra-sensitive photodetector for the application in quantum information field by usingcryogenic technology.

Keywords Far-infrared detector, Cryogenic temperature, Two-dimensional array, Detection offaint light

52

development efforts, we focused on creating acompact far-infrared detector array that wouldenable efficient and accurate space observa-tion and at the same time be suitable for instal-lation on a satellite. To our knowledge, we arethe first to successfully report a direct hybridstructure of a monolithic two-dimensionalarray and readout circuit with high responsivi-ty and fill factor in the far-infrared region.

In this paper, we describe the performanceof our Ge:Ga direct-hybrid two-dimensionalarray and present the results of research intothe establishment of a 1.5-μm-band photon-number resolving detector based on cryogenicreadout technology for use in today’sadvanced quantum information applications.

2 Ge:Ga far-infrared detectordirect-hybrid two-dimensionalarray

A large format array detector can expandthe available observation region while main-taining the same spatial resolution, forimproved efficiency and accuracy in measure-ment and observation. Conventional Ge:Gafar-infrared detectors of this type were setinside cavities in order to improve quantumefficiency. Such an array structure, however,makes it difficult to achieve a high-densityarray and also increases the overall weight ofthe system. On the other hand, a monolithicarray, in which multiple elements are arrangedon a single wafer, offers a suitable structurefor high-density packaging. Unfortunately, todate the low optical absorption coefficient ofGe:Ga photodetectors has been an obstacle tothe successful development of high-responsiv-ity detectors.

The optical absorption coefficient ofGe:Ga can be increased by increasing the Gadopant density, but high dopant density resultsin an increase in the hopping currents thatflow between impurity levels and degradationof detection limit. Although an increase in thevolume of the detector extends the opticalabsorption length, this method is not suitablefor use in space, where the device is exposed

to enormous amounts of high-energy particles.High-energy particles hitting a Ge:Ga detectorincrease responsivity nonlinearly, resulting insignificant degradation of measurement accu-racy. To minimize the impact of high-energyparticles, miniaturization is essential for anydetector element that is to be mounted on asatellite. Furthermore, since this sort of detec-tor features high impedance, a trans-imped-ance amplifier is necessary for signal extrac-tion. To reduce microphonic noise, trans-impedance must be conducted near the detec-tor. To satisfy these requirements, we madeuse of a monolithic two-dimensional arraywith an ion implanted layer sensitive to far-infrared radiation, a readout circuit operatingat the same temperature as the Ge:Ga detector,and In-bump technology for direct connectionof the array and the circuit. Figure 1 showsthe structure of our direct-hybrid two-dimen-sional array.

Our detector is comprised of 20 elements

Journal of the National Institute of Information and Communications Technology Vol.51 Nos.1/2 2004

(a) Photograph and (b) conceptualcross-sectional view of Ge:Ga directhybrid array.

Fig.1

arranged in 3 rows. A lattice pattern wasformed by the deposition of Cr and Au on thefront surface of the monolithic array, whichhas a role as an electrode and optical separatorfor each pixel. The bottom electrode wasformed by the deposition of Cr and Au overthe entire surface. When far-infrared radiationenters through the surface layer, it generates aphotocurrent, which is then read by the bottomelectrode. The front surface is a transparentelectrode, and is common to all elements.

The elements are separated by grooves onthe bottom electrode measuring 100 μm inwidth and 30μm in depth. The light-receiv-ing area of each element measures 500μm by500μm, and the elements are arranged with acenter-to-center distance of 550 μm and anelectrode-to-electrode distance of 500μm. Acapacitive trans-impedance amplifier (CTIA)constructed with a p-type Si MOSFET isadopted for the readout circuit, which wasdeveloped by a team led by Nagoya Universi-ty. Because the detector is cooled to a temper-ature of 2.5 K, strain of approximately 12μmis generated, due to the different coefficientsof thermal expansion between the Ge:Ga far-infrared detector and the readout circuit withthe base material of Si, when cooling thematerials from room temperature. To absorbthis strain, a direct-hybrid structure was con-structed using indium (In) technology, whichfeatures a low Young’s modulus even at cryo-genic temperatures.

Due to the significant strain, the conven-tional deposition method was not used to formthe In bumps. Instead, indium balls, each witha diameter of 100 μm, were arranged at thedetector intervals. Thermo-sonic was appliedwhen forming the direct-hybrid structure. Fig-ure 2 shows an SEM photograph of the Inbumps and a soft-X-ray photograph of thedirect-hybrid array. These photos indicate thesuccessful formation of smooth bumps.

The entrance surface and the bottom elec-trode of the Ge:Ga semiconductor were dosedwith high-density B having the same acceptorlevel as Ga, thus allowing for easy carrier tun-neling to the electrode. The B injection layer

also plays an important role in improving theresponsivity of the detector. By forming the Binjection layer with a density level below theMott transition, we succeeded in establishingsensitivity to far-infrared radiation. Althougha number of attempts have been made in thepast to realize a BIB structure using the ionimplanting method, there was a notable ten-dency to neglect the activation rate in theinjection of B. In our research, we proved thatthe activation rate of B in the Ge crystals wasan important parameter determining the char-acteristics of the ion injection layer[5].

To verify the effect of the B injection layeron the improvement of Ge:Ga far-infrareddetector performance, we evaluated a longitu-dinal-type Ge:Ga detector using a trans-impedance amplifier (TIA) circuit with feed-back resistance of 6 GΩ, which incorporated aperformance-proven 77-K operating Si JFET.Figure 3 indicates the dependence of sensitivi-ty and noise-equivalent power (NEP) on biasfield strength.

Since the B density is high, the opticalabsorption coefficient takes a large value.Moreover, due to the presence of low-densityGe:Ga between the electrodes, generation of

FUJIWARA Mikio et al. 53

Soft-x-ray photograph of the directhybrid array.

Fig.2

54

dark currents due to hopping currents is sup-pressed. Through this B injection layer, wewere able to increase quantum efficiency to42%, as compared to an efficiency of 19% forthe Ge:Ga bulk region only. As a result, ahigh sensitivity of 15 A/W was achieved,along with a small size (500×500×500μm),and without the need for a cavity. Comparedto the detector[6] installed on the SIRTFinfrared observation satellite launched by theUnited States in August 2003, our detectorarray has more than twice the responsivity at1/6 the size.

Our cryogenic readout circuit is a CTIAthat uses a p-type Si MOSFET and offers anopen loop gain of 1,000 times with a feedbackcapacity of 7 pF. The noise of the detectorand readout circuit exhibited a 1/f-dependentspectrum of 20μV/Hz1/2 at 1 Hz. It is knownthat this noise is generated by the readout cir-cuit. The time signal is integrated in 0.14 sec-onds in the survey mode of the ASTRO-F.Estimation of the NEP using correlation dou-ble sampling (CDS) yielded a value of 1.8×10-17 W, indicating that performance is suffi-cient for use in observation equipment. Withthese results we have developed the world’sfirst far-infrared detector direct-hybridarray[7]. Figure 4 shows the output waveform

from the direct hybrid array.

3 Faint-light detection technology

Cryogenic circuit technology can beapplied not only to far-infrared detection butalso to the detection of faint light at any wave-length. Devices such as a high-sensitivityphotodetector with high impedance and a low-noise readout circuit that performs trans-impedance and amplification can be powerfuldevice for applications of detecting faint lightsources such as spectroscopy. These devicesare expected to play key roles in the quantuminformation field in the future, particularly inthe realization of high-volume, and uncondi-tionally secure. A photon number resolvingdetector (which accurately counts the photonsin an optical pulse) can improve quantumcryptography, and will be an essential devicein quantum computation in the optical region.For example, by combining non-classical lightsuch a single-photon state or squeezed statewith a photon-number resolving detector and afeedback system, it will be possible to con-struct a general-purpose quantum computingmachine. This technology will prove critical

Journal of the National Institute of Information and Communications Technology Vol.51 Nos.1/2 2004

Responsivities and NEPS as functionsof the bias field. Squares, step changein the photon influx (dc); triangles, 7.5-Hz chopped light; circles, 15-Hzchopped light. Solid curves, responsiv-ities; dashed curves, NEPS.

Fig.3

Output waveform of the Ge:Ga directhybrid array at 2.15K. Integration timewas 1 sec, and the bias field was1.8V/cm. Solid curve is for the BBsource on; dashed curve, for the BBsource off.

Fig.4

not only in quantum coding technology forfuture large-capacity communications, but willalso serve as a basis for the construction ofhigh-security quantum information networks.Therefore, the impact of this technology haspotential influence.

Stanford University in the United Stateshas developed a visible-light photon counter(VLPC) operating at 10 K [8]. Further, a pho-ton number resolving detector was developedin NIST and was made from a superconductor,with an operation temperature of 100 mK [9].It can count photon numbers in the 1.5-μminfrared light band, in which the attenuationrate of an optical fiber is minimum. Thesedetectors have a number of shortcomings:high dark count, susceptibility to backgroundlight, and low quantum efficiency, to cite afew. Furthermore, these detectors require spe-cial fabrication techniques, a major obstacle towidespread use.

Our research is aimed at the developmentof a photon-number resolving detector for 1.5-μm-band infrared light through a combinationof commercially available devices operating atcryogenic temperatures. The method we haveadopted counts photons in the incident lightthrough accurate determination of the electriccharge generated in the InGaAs pin photodi-ode. The readout circuit system uses a chargeintegration amplifier (CIA), which is suitablefor faint-light detection and requires a mini-mum number of components. The circuit dia-gram is shown in Fig.5. The section surround-ed by the red line is cooled to a temperature of4.2 K. For the InGaAs pin photodiode, weused aφ30-μm type, from Kyosemi.

Low-capacity circuit packaging and reduc-

tion of noise in the readout circuit determinesthe feasibility of photon number resolvingdetection. In our research we adopted a GaAsJFET for the primary amp. A number of FETsare available that operate at cryogenic temper-atures, such as MOS FETs, MES FETs,HEMTs, and compound JFETs. In n-typeMOS FETs, the kink phenomenon—a suddenincrease in current—is observed among thevarious current and voltage characteristicsseen at cryogenic temperatures. To avoid thiskink phenomenon, it is necessary to use a p-type MOS FET [10] or to increase the amountof dopant [6]. Even with such measures, noiseremains at approximately 10μV/Hz1/2 at 1 Hz,insufficiently low.

It has been reported that noise may bereduced to low levels (below 1μV/Hz1/2 at 1Hz) in MES FETs and HEMTs with a draincurrent of approximately 1 mA [11], but theseFETs consume a large amount of power,which is a fatal defect for cryogenic electron-ics, and also feature significant gate leak cur-rent; these devices are therefore unsuitable foruse in the readout circuits of high-impedancedetectors. On the other hand, a compound(GaAs) JFET uses a p-n junction in its gatestructure and can provide higher gate imped-ance than HEMTs and other devices. Further-more, the carrier-traveling channel and gateelectrode distances in a compound (GaAs)JFET are longer than in other FETs, resultingin a lower input capacity given the same gatesize. This means that a compound JFET hasthe major advantage of providing a higher S/Nratio in an integration-type readout circuit.

We evaluated the performance of theSONY GaAs JFET at cryogenic temperatures.Figure 6 (a) shows the I-V characteristic of aGaAs JFET with a gate width of 5μm and agate length of 50μn at 4.2 K, and Figure 6 (b)indicates the dependence of mutual conduc-tance on gate voltage. As shown in the graph,the I-V characteristic was favorable at 4.2 K,thus confirming the possibility of achievingtrans-conductance of approximately 10μS.

Figure 7 shows the dependence of gatecapacity on the gate voltage at room tempera-

FUJIWARA Mikio et al. 55

Faint light detection system at 1.5µmwith charge integration amplifier.Parts in the frame are cooled to 4.2K.

Fig.5

56

ture, 77 K, and 4.2 K. The lower the tempera-ture, the lower the capacity. At 4.2 K, opera-tion below 0.1 pF becomes possible. At thiscryogenic temperature, a type of noise [12]referred to as “random telegraph signal”(RTS) is generated in the GaAs JFET (Fig.8).This switching phenomenon results in signifi-cantly disruptive measurement deviation.

We have developed a method of substan-tially reducing the probability of RTS genera-tion, in which the elements are heated above35 K with a drain current, followed by re-cooling to 4.2 K. We have designated thisprocess as “thermal cure” (TC). Figure 9shows noise levels with and without TC.

Without TC, noise was approximately 3μV/Hz1/2 at 1 Hz, while the application of TCreduced this noise to 0.5μV/Hz1/2 at 1 Hz. Fordetails of the mechanism involved, pleaserefer to Document [13].

Further, measurement of gate input capaci-ty of the GaAs JFET returned a value of 0.06pF, while the InGaAs pin photodiode dis-played a capacity of 0.026 pF at a cryogenictemperature. When this was combined withthe input capacity of the GaAs JFET, totalcapacity was 0.086 pF (Cs), lower than 0.1 pF.We also attempted to detect faint light using aCIA-type readout circuit. If the signal isexpressed by QGM/Cs (V), noise is Vn,CDS, Q is

Journal of the National Institute of Information and Communications Technology Vol.51 Nos.1/2 2004

Gate capacitance as a function ofgate-source voltage with operationtemperature as a parameter.

Fig.7

Drain current vs drain voltage curveswith gate bias as a parameter. (b)Transimpedance as a function of gatevoltage with drain voltage as aparameter.

Fig.6

Typical fluctuation in drain current IDwith an RTS amplitude of ~0.1%,VD=0.75V, VG=0.32V, at 4.2K.

Fig.8

Noise spectra of the GaAs JFET bothbefore and after TC. (a) VD=0.75V,VG=0.32V, (b) VD=0.5V, VG=0.21V.

Fig.9

the elementary electric charge (1.6×10-19 C),and GM is the source follower gain (0.8 to0.9), then noise (Vn,CDS) at CDS can beexpressed by the following equation.

Here, T is the integration time (0.5 or 1 s),and fc (100 Hz) is the circuit’s cut-off frequen-cy. Vn ( f ) refers to the noise spectrum. Figure10 shows the output waveform when the inci-dent light was attenuated to approximately 40photons per second. Quantum efficiency wasapproximately 60%. Although there was anacquisition failure due to RTS, we were ableto achieve detection with a deviation of 2 pho-tons, corresponding to the estimated accuracybased on the applicable noise level. The leakcurrent of this circuit was 500 electrons/hourand thus had no practical effect on measure-ment. In short, this can be regarded as themost sensitive detector in the world, with min-imal dark count in the 1.5-μm band. Our nextgoals consist of reducing noise further in thereadout circuit and developing a detector fea-turing single-photon accuracy.

4 Summary

Working to develop highly sensitive low-

noise photodetectors through experimentationat cryogenic temperatures, we have designedthe world’s first direct-hybrid structure for aGe:Ga far-infrared detector, and have succeed-ed in the detection of far-infrared light. Fur-thermore, we have constructed a 1.5-μm-bandfaint-light detector using an InGaAs pin pho-todiode and a GaAs JFET, successfully detect-ing light at about 40 photons per second with aquantum efficiency of 60% and a deviation of2 photons. We plan to reduce noise further inthe future, as part of our efforts to furtherimprove photon-number resolution.

FUJIWARA Mikio et al. 57

References1 C. A. Beichman, “The IRAS View of the Galaxy and the Solar System”, Ann. Rev. Astron. Astrophysics, 35,

521-563, 1987.

2 D. M. Mittleman, R. H. Jacobsen, and M. C. Nuss, “T-ray Imaging”, IEEE J. Sel Topics Quantum Electron, 2,

679-692, 1996.

3 M. Hermann, M. Tani, K. Sakai, and R. Fukasawa, “Terahertz imaging of silicon wafers”, J. Appl. Phys. 91

(3), 1247-1250, 2002.

4 P. R. Bratt, “Impurity germanium and silicon infrared detector” in Semiconductors and Semimetals, R. K.

Willardson and A. C. Beer, eds. (Academic Press, New York, 1997), pp. 33-142.

5 M. Fujiwara, “Development of far-infrared Ge:Ga photoconductor having a longitudinal configuration,” Appl.

Phys. Lett. 77, 3099-3101, 2000.

6 E. T. Young, J.T. Davis, C.L. Thompson, G.H. Rieke, G. Rivlis, R.Shchnurr, J. Cadien, L. Davidson, G. S.

Winters, and K. A. Kormos, “Far-infrared imaging array for SIRTF,” in Infrared Astronomical Instrumentation,

A.M. Fowler eds,. Proc. SPIE 3354, 57-65, 1998.

(1)

Photo-counting waveform of theInGaAs pin photodiode with CIA at4.2K.

Fig.10

58 Journal of the National Institute of Information and Communications Technology Vol.51 Nos.1/2 2004

7 M. Fujiwara, T. Hirao, M. Kawada, H. Shibai, S. Matsuura, H. Kanneda, M. Patrashin, and T. Nakagawa,

“Development of a gallium-doped germanium far-infrared photoconductor direct hybrid two-dimensional

array,” Appl. Opt. Vol. 42, No. 12, 2166-2173, 2003.

8 E. Waks, K. Inoue, W. D. Oliver, E. Diamanti, and Y. Yamamoto, “High efficiency photon number detection

for quantum information processing,” quant-ph/0308054v1, 2003.

9 A. J. Miller, S. W. Nam, J. M. Martinis, and A. V. Sergienko, “Demonstration of a low-noise near-infrared

photon counter with multiphoton discrimination,” Appl. Phys. Lett. Vol. 83, No. 4, 791-793, 2003.

10 T. Hirao, Y. Hibi, M. Kawada, H. Nagata, H. Shibai, T. Watabe, M. Noda, and T. Nakagawa, “Cryogenic

readout electronics with silicon P-MOSFETs for the infrared astronomical satellite, ASTRO-F,” Advances in

Space Research, 2002.

11 R. K. Kirschman and J. A. Lipa, “Further evaluation of GaAs FETs for cryogenic readout,” in Infrared Detec-

tors and Instrumentation, W.H. Makky eds,. Proc. SPIE 1946, 350-364, 1993.

12 K. Kandiah, “Random telegraph signal currents and low-frequency noise in junction field effect transistors,”

IEEE trans. Elec. Dev. Vol. ED-41, No. 11, 2006-2015, 1994.

13 M. Fujiwara, M. Sasaki, and M. Akiba, “Reduction method for low-frequency noise of GaAs junction field

effect transistor at a cryogenic temperature,” Appl. Phys. Lett. Vol. 80, No. 10, 1844-1846, 2002.

FUJIWARA Mikio, Ph. D.

Senior Researcher, Quantum Informa-tion Technology Group, Basic andAdvanced Research Department

Photodetection Technology

SASAKI Masahide, Ph. D.

Leader, Quantum Information Technol-ogy Group, Basic and AdvancedResearch Department

Quantum Information Theory

AKIBA Makoto, Ph. D.

Senior Researcher, OptoelectronicsGroup, Basic and Advanced ResearchDepartment

Optical Sensing