I zyxwvutsrqponmlkjih 2548 zyxwv zyxwvutsrqpo zyxwvutsr zyxw IEEF: TRANSACTIONS ON ELECTRON DEVICES, VOL. 38. NO. I I . NOVEMBER 1991 llf Noise in GaAs Filaments Munecazu Tacano and Yoshinobu Sugiyama Abstract-A typical 1 /f noise is excited in a GaAs filament The with the Hooge's noise parameter of about aH = 2 x noise level increases in proportion to the square of the terminal voltage, and decreases approximately in inverse proportion to the total number of carriers within the device. A transition from the typical 1 lfnoise characteristics to the diffusion noise characteristics of MESFET's was observed when the electric field was increased above 1 kV/cm. The noise parameters were also investigated as a function of the device width between 2 and 200 pm. Deep levels within the n-GaAs active layer and the high electric field are the main factors of the nonideal l/fcharacteristics. I. INTRODUCTION HE GaAs metal-semiconductor field-effect transistors (MESFET's) are widely used as low-noise microwave devices, but are reported to have rather large noise levels at low frequencies [I]. Since the low-frequency noise induces phase, FM, or AM noise in microwave circuits, the reduction of the low-frequency noise is enabled through the investigation of its origins and mechanisms. The frequency dependence of the drain noise in a GaAs MESFET has been shown to vary a s f - l below 1 kHz and f-'.5 above 1 kHz [2]-[4]; in ion-implanted resistors of GaAs, it varies as f - ' . 'to f - ' [SI, and in MBE-grown layers of GaAs, as f - ' . 'to f-"' [6]. This frequency dependence could be characterized as the diffusion noise observed in GaAs current limiters [7]. The Hooge's noise parameter, a H ,was determined for a GaAs MESFET to be about 2 X I O p 4 [8], and for a quarter-micrometer GaAs Hall device made by focused ion-beam (FIB) implantation to be 4.5 x [9]. The noise spectrum of both the FIB-implanted device and that of heterostructure 2DEG [IO] suggests that a pure 1 /f noise can be excited in a thin-filament structure of GaAs. Therefore, we have made filament devices of n-GaAs and studied their low-frequency noise. A typical l / f noise, without a large bulge between 1 and 10 Hz usually observed at room temperature in a commercial GaAs MESFET I1 I], is excited in these devices. A transition from the typical 1 lfnoise characteristics to the diffusion noise characteristics of MESFET's together with the exponential increase of a H was observed when the electric field through the filament was increased above 1 kV/cm. T The noise parameters were also investigated as a function of the filament width between 2 and 200 pm. Deep levels within the n-GaAs active layer and the high electric field through the device are the main factors to induce the nonideal 1 lfcharacteristics. 11. DEVICEFABRICATION MEASUREMENTS We made devices from two n-GaAs substrates grown by a Varian MBE-360 system and a commercially available one grown by VPE. The substrates, contact metals, and device dimensions are summarized in Table I. Epitaxial layers of n-GaAs 0.5 pm in thickness were grown on Cr-doped semi-insulating GaAs wafers with Si doping of 5.0 x 10l6 cmP3 in the same conditions as described elsewhere [6]. Using standard integrated circuit processes of photolithography, metal electrode lift-off, and etching, these devices were fabricated with various lengths from 5 to 80 pm or various widths from 2 to 200 pm. The ohmic contacts were made on the epitaxial n-GaAs layer by the deposition of Sn/Ni/Au( 100/50/300 nm) or AuGe/Ni/ Au( 100/50/200 nm) followed by alloying at 450°C for 45 or 30 s. respectively, in flowing H2 gas. The SnAu contacts ha$ specific contact resistances of about 5.0 X l o p 4 Q * cm-. Fig. 1 shows an overview of a prepared device of various lengths with the constant width of 1.4 pm. The current-voltage characteristics of each device were completely linear within the low electric field bias. The noise was measured in a brass-shielded box at room temperature, and a bias dc current was supplied by batteries through a 100-kQ resistor. The voltage noise across the current terminals was amplified by a low-noise preamplifier. ITHACO 1201, and the output was fed to an ADVANTEST TR9404 spectrum analyzer. The analyzer was controlled by a desktop computer, HP216. The smallest detectable circuit noise of the setup was about - 162 dBV/& between 10 Hz and 100 kHz. The spectrum analyzer formed an average of each 2' noise data, and the computer further averaged the data and subtracted the ac line noise. The noise levels at I Hz and the factorials of frequency were calculated using a least squares approximation of the data between 100 and 0.2 Hz. AND zyxwvut zyxwvutsrqpo zyxwvutsr Manuscript received October 12. 1990: revised April I . 1991. The review of this paper was arranged by Associate Editor M . D. Feuer. The authors are with the Electrotechnical Laboratory. Umerono 1-1-4. Tsukuba, Ibaraki 305. Japan. IEEE Log Nuinber9101415. 111. RESULTSA N D DISCUSSION A. 1 /fNoise Churacteristics The noise spectra of a filament device 10 pm in length are shown for a variety of voltages in Fig. 2. The noise 0018-938319111 100-2548$01 00 'ri 1991 IEEE zyxwvutsrqponmlkjihgfedcbaZYXW zyxwvutsrqponm zyxwvutsrqpo zyxwvutsrqponmlkjih TACANO A N D SUGIYAMA 2549 I ,'f N O I S E IN GdA\ F l L A M t N TS TABLE I Su bat rates MBE, 5 X IO2' m Bufcr ( 1 p m ) VPE. 2 . 7 x IO" in Buffer (4 p m ) Contact Metals ' zyxwvutsrqp zyxwvu Dimensions Sn-Ni-Au 1.3" x 0.5' x ( 5 . 10, 20. 40. 80)' pm AuCe-Ni-Au 1.4" X 0.5' x ( 5 . 10. 20, 40. 80)' p111 (2. 5 . 10, 20. 30. 50. 100. 200)" x 0.5' x 10' p m ' 1c7 I MBE GaAs F l L l FREQUENCY Fig. I. Filamentous device of n-GaAs for lowfrequency noisc mcasurcments with dimensions of 1.3 x 0.5 x (2.5, 5. 10. 20. 30. 40, 5 0 . 60. ( Hr Fig. 3. 1 /,fnoise spectra of n-GaAs filaments at a constant voltage: device length as a parameter. 80) p m . 1o - ~ lo?-: .. . MBE-GaAs FIL 10X1.3x0.5pm - lo-s- MBE-GaAs FIL 1.3x0.5pm 5.0 x 1016 cm-3 a. 10pmL 40pmL 1111111 ' I 1111'11 40.04 0.1 a2 W0.6 1 8OpmL ' ' 2 zyxw I 1 I H 4 6 zyxwvutsrqpo zyxwvutsrq VOLTAGE ( Volt ) Fig. 4. Noise levels of devices with various lengths as a function of terminal voltage. power level S,, decreases with frequency exactly at the rates of f-' throughout the range from 0.1 Hz to 100 kHz. The noise levels increase as the square of the terminal voltage V . Noise spectra are shown for a variety of device lengths in Fig. 3 at a constant voltage of 1 V . The devices were fabricated on the same substrate, and the total effective number of carriers to excite the noise is proportional to the device length. The noise level S,. decreases with increasing device length, that is, with an increase in the total number of carriers. Noise spectra of the devices with various lengths were further measured as a function of the voltage, and noise levels at 1 Hz were plotted as a function of the applied voltage in Fig. 4, with the device length as a parameter. It is evident in the figure that noise levels increase in proportion to the square of the terminal voltage V . These ex- perimental results can thus be fully explained by Hooge's fundamental relationship where V is the voltage across the device, f and N are the frequency and the total number of carriers, respectively, and a H is the Hooge's noise parameter originally determined as aH = 2.0 x [12]. The 1-Hz noise levels normalized at the terminal voltage of 1 V were plotted as a function of the total number of carriers in Fig. 5 . The number of carriers was determined from the carrier concentration and the device volume. The levels decrease in inverse proportion to the total number of carriers. Hooge's parameter derived from the proportionality constant in Fig. 5 was about a H = 2.0 X lop3. 2550 zyxwvut zyx zyxwvutsrqponmlkjihg I t E E TRANSACTIONS ON ELECTRON DEVICES. V O L 3X. NO I I . NOVEMBER 1991 I.L2 ( A . p m 2 ) 10-3 Ill 10-2 1 I111111 I I MBE GaAsFI lo-' I Illllll 0.1 1 IO io2 io3 io4 io5 FREQUENCY ( Hz ) io5 106 CARRIER NUMBER 107 ( N ) Fig. 6 . Noise spectra of an n-GaAs filament at high electric fields: voltage across a sample us a parameter. zyxwvut zyxwvutsr zyxwvutsrqp Fig. 5. Noise lerelb at 1 H 7 and 1 V as a lunction of IL' and total carrier number. Hooge's equation ( I ) can be expressed. on the other hand, in terms of the device parameters as [8]. S,, = a,qpV'/(fZL?) (2) where q is the carrier charge, p the mobility, L the device length, and 1the current through the device, respectively. The noise levels at 1 V and 1 Hz are also shown by open circles in Fig. 5 for a variety of ZL', indicating linear decrease in proportion to about (/L2)-o.8.The deviation from the theoretical estimation may be due to a contribution of the finite contact resistance r ( . In the presence of r,, (2) could be modified as s,. = a , q p ~ ' ( ~+ 3 q p r , . ~ / ~ ' ) / ( f ~ ~ ' )(3) and the apparent voltage noise level would increase with the increasing total number of carriers. The contact resistance of the individual device could not be determined in the present pattern, but estimated from the length dependence of resistance as about 5 kQ. Since the carrier concentration was designed rather thin in order to apply the high electric field, large contact resistances were obtained with these devices. This might have caused the deviation from the (ZL') dependence. The electron mobility was derived from the currentvoltage characteristics and the sample dimension as p = 0.29 m2/V s, and the noise parameter was a , = 2.0 x IO-'. Since ( I ) and (2) are mathematically the same. the parameter should agree within experimental error on the sample dimension and the estimation of the mobility. B. High Electric Field Churacteristics The noise spectra of a filament device 40 pm in length are shown for a variety of voltages in Fig. 6. The noise power level S,. decreases with frequency exactly at the rate o f f - ' throughout the range from 0.1 Hz to 100 kHz in low electric fields below about 0.5 kV/cm (2 V), but tends to decrease as f - ' in the electric fields above this value. This tendency was commonly observed in GaAs MESFET's in the current saturation region [4],and explained in terms of the deep levels within the epitaxial layer. " minal voltage in Fig. 7. The levels increase in proportion to V' in low electric fields below 0.5 kV/cm with Hooge's noise parameter of a, = 1.6 X W3, but increase rapidly in the electric field above I kV/cm. Current-voltage characteristics of the filament sample are shown in Fig. 8, where the current saturation is seen to start in the relevant high electric field. Since the carrier concentration would not increase in the electric field around 3 kV/cm, the decrease of the average carrier mobility causes current saturation. The relative mobility of the sample was also plotted as a function of voltage in Fig. 8. Taking into account the mobility saturation in Fig. 8, Hooge's parameters were calculated from (1) and (2) shown by open and closed circles, respectively, in Fig. 7. In low electric fields, (1) gives Hooge's parameter of 1.6 x 10-j, and (2) of 5.7 x IOp3, agreeing quite well with the original value of 2 x lo-'. It must be noted that in the figure the parameter increases quite a bit in the high electric field above 1 kV/cm. The noise parameters of a quarter-micrometer GaAs filament [9] were also plotted in Fig. 7, indicating much smaller values of a , = 4.5 x l o p h increasing at much higher electric fields. Fig. 7 shows explicitly a considerable increase of the noise level and the corresponding noise parameter. A theory of the hot-electron l l ~ f n o i s ein n-GaAs has been studied by Kleinpenning [ 131, who interpreted the relative current noise s//z' [14] in terms of fluctuations in the electron mobility. His theory also introduced the relation S,, = r i S / , r,, being the dynamic resistance in the nonlinear part of the I-Vcharacteristics. The low field and differential resistances of our sample at 7 V (1.8 kV/cm) were Ro = 19 kQ and r,/ = 54 kfl, respectively, and S,. increased 29 dB above the linear relationship S,, = 2 X 10-'Vv'/(fN). The relative increase of the dynamic resistance from its low-field value was about 2.8 times, resulting in the increase of SI about 20 dB. The normalized noise levels, S,,/ V' and S///2,together with the current through the sample, were compared at the same voltage in Fig. 9. We used a commercially available zyxwvutsr zyxwvuts zy zyxwvut -' 9 The noise levels at 1 Hz, where every spectrum has the 1 lfcharacteristics, were plotted as a function of the ter- zyxwvutsrqponmlkjih zyxwvutsrqponm zyxwvutsrqponm zyxwvutsrqponm zyxwvutsrqpon TACANO A N D SUGIYAMA: I / f NOISE IN GaAs FILAMENTS ELECTRIC FIELD ( KV/cm) 0.1 0.2 0.4 1 2 4 \ N I” VPE-GaAs I zyxwvutsrqponmlk 1 2 4 10 20 VOLTAGE (volt ) 1 P A1 Fig. 7 . Noise levels at I H z , 1 V and Hooge’s noise parameter as a function of voltage or electric field across a sample. 0.8 P t I-v MBE-GaAs FIL 40~1.3~0.5 pm 5.0~10 ~ ~m - ~ 0.1 12 , 0 VOLTAGE (volt) Fig. 8. Current-voltage characteristics and mobility change due to the high-field effect. The kink at 8 V is due to the Gunn effect. 7 1- zyxwvu 0.2 0.4 >a 255 1 -8 - 0 40 x 1.4 x 05 p n 4 8 VOLTAGE 12 16 (Volt) Fig. 9. Relative voltage and current noises in comparison with currentvoltage characteristics. substrate with concentration of 2.8 X 10’’ ~ m - 6~times , greater than that used in the previous experiments. S , , / V 2 showed a rapid increase above 12 V (3 kV/cm). Shorter samples of the length between 5 and 30 pm were also studied, and all showed rapid increase of S , , / V 2 above 3 kV/cm, corresponding to the threshold voltage for the r - L transition [15]. In contrast to the rapid increase of S t , / V 2 ,the current noise S,/Z2 increased slowly with bias voltage. The relative increase of the dynamic resistance was about 1 . 8 times at 17 V, introducing the expected increase of S,/Z2 by an amount of 1 1 dB, whereas the experiment gave the value of 21 dB. It must be noted that the low-field Hooge noise parameter aH estimated from I 1 I I I 10 102 103 FREQUENCY ( Hr ) I lb Y lo” Fig. 10. 1 lfnoise spectra of n-GaAs devices at a constant voltage; device width as a parameter. S,, was 1 x lop4 and that from SI was 2.3 X lop6, respectively. The voltage noise S,, will represent the carrier number fluctuation associated with the r - L transition, and the current noise SI the mobility fluctuation. C. Sample Width Dependence Noise spectra are shown for a variety of device widths in Fig. 10 at a constant voltage of 1 V . The noise levels decrease approximately in inverse proportion to the frequency, although there are several bulges of the GR noise caused by the deep levels in n-GaAs. Most of the GR noise appeared at around 10 kHz, as shown by the plots for 100 and 200 pm in the figure. The factorials over f are about 1.2 for samples with deep levels. An ideal 1 /f characteristic was always observed, on the other hand, by samples without deep levels. The device length was kept constant as 10 pm, and the electric field through the device was 1 kV/cm in each case. The noise levels decrease with increasing sample width, i.e., with increasing total number of carriers. Hooge’s noise parameters aH of various widths were derived from ( 1 ) and plotted in Fig. l l . Hooge’s noise pato 3 x l o p 3in VPE samrameters ranged from 4 x ples. zyxwvu D. a H of Various Devices Hooge’s noise parameters for various epitaxial devices are plotted in Fig. 1 1 as a function of the total number of carriers. These data were obtained recently in our group. Data sets #1 and #2 correspond to the result of the present experiments. The filamental devices made from MBEgrown substrates gave a Hooge’s noise parameter close to 2 X and those from VPE-grown substrates about 1 x l o p 3on average. Data sets #3 and #4 correspond to the TLM patterned devices of GaAlAs /GaAs two-dimensional electron gases [ 101 and ion-implanted InP layers [16], respectively, #5 and #7 those of GaAs FET’s [SI, [17], #6 those of HEMT [ l l ] . Hooge’s noise parameters are indicated as #8 for GaAs quarter-micrometer filaments and #9 for InP ones [ 181. The solid line indicates the noise zyxwvutsr zyxwv zyx zyxwv IEEE TRANSACTIONS ON ELECTRON DEVICES. VOL. 38. NO. II. NOVEMBER 1991 2552 3 ) Hooge’s noise parameter increased exponentially with the electric field above 1 kV/cm, and a transition from the typical ,f - I noise characteristics to the diffusion noise characteristics of MESFET’s was observed. The kink frequency from f - I to f - 2 characteristics was about 1 kHz. 4) The GR noise caused by the deep levels within the n-GaAs active layer was observed mostly at about 25 Hz and to a lesser degree at about 1 and 20-50 kHz in devices with MBE-grown substrates, while mostly at 3-5 Hz and to a lesser extent about 1 and 20 kHz with VPE substrates at room temperature. Deep levels within the active layers and the high electric field are the main factors of the nonideal f characteristics, and they enhance the equivalent Hooge’s noise parameter. It must also be noted that the transition from f to f characteristics is due to the presence of GR noise. zyxwvutsrqponm zyxwvutsrq zyxwvutsrqponmlkjihgfedcba zyxwvutsrqp zyxwvutsrqponml lo4 lo5 lo6 10’ lo8 loQ lolo CARRIER NUMBER, N Fig. 1 I . a,, versus carrier number of various compound semiconductor devices: #Iand #2 present work. #3 [ I O ] . #4 1161. #S 181. #6 [ I ll. #7 1161. #8 [SI. and #9 [IS]. parameter originally proposed by Hooge, which seems well confirmed in the filamental form of the present experiment. Comparing the data set #I with #7 of n-GaAs or #2 with #5 of GaAlAs/GaAs 2DEG, structures and surfaces as well as contacts of devices still play an important role in determining the noise parameter. It must also be pointed out that there is no correlation between the total number of carriers and Hooge’s parameter. As indicated in Fig. 1 1 , some of the recent GaAs FET’s and HEMT have far smaller noise parameters; on the order of l o P 5and below. We may now need a new relationship to explain all of these 1 / f noise levels consistently [ 191. A N D CONCLUSIONS IV. DISCUSSION We have studied low-frequency noise excited in n-GaAs epitaxial layers for various device dimensions or contact metals, and a typical 1 lfnoise is observed with Hooge’s noise parameter of about aH= 2 X lop3. The noise level increases in proportion to the square of the terminal voltage, and decreases in inverse proportion to the total number of carriers within the device. A transition from the typical 1/ f noise characteristics to the diffusion noise characteristics typical in MESFET’s was observed when the electric field applied across the device was above 1 kV/cm. Deep levels within the n-GaAs active layer and the high electric field through the device are main factors in inducing the nonideal 1 /f characteristics with factorials other than - 1. Experimental results are summarized as follows. 1) The device made from the MBE-grown substrate to a filament shape of 1.3 x 0.5 x ( 5 , 10, 20, 30, 40, 80) pm had typicalf - characteristics with Hooge’s noise parameter of aH = 2 x l o P 3 . This must be compared to f - 1 , 5 characteristics of TLM-shaped devices in the earlier experiment. 2) The device made from the VPE-grown substrate noise with n- buffer thickness of 4 pm also had an f characteristic with a Hooge’s noise parameter of 4 x I O P 4 < aH < 3 X l o P 3 .These values are between those obtained in the device with MBE-grown substrates and those in commercial FET’s or FIB-implanted filaments. ’ -’ ACKNOWLEDGMENT The authors would like to thank the referee who suggested the theoretical analysis of hot-electron 1 / f noise by Kleinpenning. REFERENCES [ I ] K. Takagi and A . van der Ziel. “High frequency excess noise and flicker noise in GaAs FETs,” Solid-Stute Electron., vol. 22, p. 285. 1979. 121 C . H. Suh, A . van der Ziel. and R. P . Jindal, “ I / f n o i s e in GaAs MESFETs,” Solid-Stutc Electron., vol. 24. p. 717, 1981. 131 K . H. Duh, X . C. Zhu, and A . van der Ziel, “Low-frequency noise in gallium arsenide MESFETs,“ Solid-Stare Electron., vol. 27. p. 1003. 1984. 141 L. M. Rucker and J . R . Hellums, “Low-frequency noise characteristics of gallium arsenide MESFETa,” Solid-State Elec.fron.,vol. 27, p. 947. 1984. 151 J . R . Hellums and L. M. Rucker, “An anomalous behavior in lowfrequency GaAs resistor noise.“ Solicl-State Electron.. vol. 2 8 . p. 549, 1985. 161 M. Tacano, Y . Sugiyama. T . Taguchi. and H. Soga. “Low-frequency noise i n GaAs layers grown by molecular beam epitaxy.” S(~lid-St~tc, E/ecIroti., vol. 31, p. 1215. 1988. 171 A . Peczalski. A . van der Ziel. and R . Zuleeg. “Low-frequency noise in GaAs current limitters.” Solicl-Stufc, Electron.. vol. 26. p. 861. 1983. [8] K . H. Duh and A. van der Ziel. ”Hooge parameters for various FET structures.“ lEEE Trans. Electron Devices, vol. ED-32, p. 662, 1985. 191 M . Tacano. T . Kanayama. H. Hiroshima. M. Komuro, and Y. Sugiyama, “ I /f noise in a quarter-micron GaAs Hall device made by focused ion-beam implantation.” J . A / ~ p l ,Phys., vol. 58, p. 4301. 1987. [ I O ] M. Tacano. Y. Sugiyama, and H. Soga, “Device process dependence of low-frequency noise in GaAIAs/GaAs heterostructure.” SoliclSrutc, Electron.. vol. 32, p. 49, 1989. [ I I ] M. Tacano and Y . Sugiyama, I /f noise of GaAIAs/GaAs HEMT in comparison with that o f GaAs MESFET,” Solid-Stcrtc Electron.. vol. 34, to be published. [ 121 F. N . Hoogc, ’‘Il f n o i s e is no surface ctfect,” Phys. Let/. , vol. 29, p. 139, 1969. [ 131 T. G . M. Kleinpenning, “On hot-electron 1 lfnoise in GaAs and Si,” Physicti. vol. 142B, pp. 229-232. 1986. [ 141 M. E. Levinshtein and S . L. Rumyantsev. “Hot-electron I /f noise in GaAs,“ Sor. Phys.~Ser,7icorld.,vol. 19, pp. 1015-1018. 1985. [ I S ] S . M. Sze, Physics ofScw~iconducrorDevices. New York: Wiley, 1981. p. 638. 1161 M . Tacano. K. Oigawa, and Y. Sugiyama. ” l / f noise in ion-implanted indium phosphide layers.” Solid-Srute Electron.. vol. 31, pp. 1243-1249. 1988. zyxwvutsrqp zyxwvutsr -’ “ zyxwvutsrqponmlkjih zyxwvutsrqponmlkjih zyxwvutsrqpo TACANO AND SUGIYAMA: I / j NOISE I N GaAs FILAMENTS 1171 C. Y. Su, H. Rohdin, and C . Stolte, " l / f noise in GaAs MESFETs," in IEDM83 Dig. Tech. Pupers, 1983. pp. 601-604. [I81 M. Tacano, T . Kanayama, and Y. Sugiyama, " I /,fnoise in quartermicron filaments of GaAs and InP made by focused ion-beam implantation." Solid-Stutp Electron., vol. 34. pp. 193-196, 1991. [ 191 M. Tacano, "A new approach to the Hooge parameter for I /fnoise in semiconductors," Solicl-Sture Elecvon.. vol. 34, to be published. Munecazu Tacano was born in Tokyo. Japan, in 1940. He received the B.Sc. and M.Sc. degrees, both from Yokohama National University i n 1964 and 1966, respectively. and the Ph.D. degree in applied physics from the University of Tokyo in 1982. In 1966. he joined the Electrotechnical Laboratory, Ibaraki, Japan. where he has been working on compound semiconductor physics and devices. He was a Research Associate at the University of Washington during 1972- 1973 working on semi- 2553 conductor plasma instabilities, and a Manager, Research & Development Association for Future Electron Devices during 1988- 1990. Dr. Tacano is a member of the Applied Physics Society of Japan, and the IEE of Japan zyxwvutsrq * Yoshinobu Sugiyama was born in Gifu, Japan, in 1945. He received the B.Sc. and M.Sc. degrees in fine measurement engineering from the Nagoya Institute of Technology in 1968 and 1970, respectively, and the Ph.D. degree in physical electronics from the Tokyo Institute of Technology in 1984. In 1970, he joined the Electrotechnical Laboratory, Agency of Industrial Science and Technology, Ministry of International Trade and Industry. Tokyo. Japan. He has been working on semiconductor magnetic sensing devices and superconductor microwave sensing devices. and was a Visiting Researcher at the University of California. Berkeley, during 198 I - 1982. His current research interests include quantum devices with fine heterostructure semiconductors for intelligent sensing. Dr. Sugiyama is a member of the Applied Physics Society of Japan, the IEE of Japan, and the IEICE of Japan.