lo~ dit, thermodynamically stable ga203-gaas interfaces: … · 2014. 10. 17. · molecular beam...
TRANSCRIPT
r.
I 214 IEEE TRANSACfIONS ON ELECTRON DEVICES, VOL. 44, NO. 2, FEBRUARY 1997
Lo~ Dit, Thermodynamically StableGa203-GaAs Interfaces: Fabrication,
Characterization, and Modeling,/ I"""---
~.
M. Passlack, Member, IEEE, M. Hong, Member, IEEE, J. P. Mannaerts, R. L. Opila, Member, IEEE,S. N. G. Chu, N. Moriya, F. Ren, Senior Member, IEEE, and J. R. Kwo
('
Abstract- Thermodynamically stable, low Dit amorphousGa203-(I00) GaAs interfaces have been fabricated by extendingmolecular beam epitaxy (MBE) related techniques. We haveinvestigated both in situ and ex situ Ga203 deposition schemesutilizing molecular beams of gallium oxide. The in situ techniqueemploys Ga203 deposition on freshly grown, atomically ordered(100) GaAs epitaxial films in ultrahigh vacuum (UHV); theex situ approach is based on thermal desorption of nativeGaAs oxides in UHV prior to Ga203 deposition. Uniqueelectronic interface properties have been demonstrated for insitu fabricated Ga203-GaAs interfaces including a midgapinterface state density Dit in the low 1010 cm-2 eV-1 rangeand an interface recombination velocity S of 4000 cm/so Theexistence of strong inversion in both n- and p-type GaAs hasbeen clearly established. We will also discuss the excellentthermodynamic and photochemical interface stability. Ex situfabricated Ga203-GBAs interfaces are inferior but still of ahigh quality with S = 9000 cm/s and a corresponding Dit inthe upper 1010 cm-2eV-1 range.We also developed a new numerical heterostructure model for theevaluation of capacitance-voltage (C-V), conductance-voltage(G-V), and photoluminescence (PL) data. The model involvesselfconsistent interface analysis of electrical and optoelectronicmeasurement data and is tailored to the specifics of GaAssuch as band-to-band luminescence and long minority carrierresponse time TR. We will further discuss equivalent circuits instrong inversion considering minority carrier generation usinglow-intensity light illumination.
r-'
".-.
~
r
I. INTRODUCTION
LOW-POWER technologies are key to the widespread useof high-performance portable systems such as portable
computers, cellular telephones, personal communicators,PDA's, medical devices, etc. At the present time, bulk SiCMOS offers significant advantages in terms of integrationlevel and cost; however, reductions in circuit speed of bulkSi CMOS are anticipated as the power supply is scaled
,.-..-
Manuscript received July 2, 1996; revised September 17, 1996. The reviewof this paper was arranged by Editor J. Xu. M. Passlack's work was supportedin part by the Deutsche Forschungsgemeinschaft.
M. Passlack was with AT&T Bell Laboratories, Murray Hill, NJ 07974USA. He is now with Motorola Inc., Phoenix Corporate Research Laborato-ries, Tempe, AZ 85284 USA. .
M. Hong, J. P. Mannaerts, R. L. Opila, S. N. G. Chu, F. Ren, and J. R.Kwo are with Bell Laboratories, Lucent Technologies, Murray Hill, NJ 07974USA.
N. Moriya was with AT&T Bell Laboratories, Murray Hill, NJ 07974 USA.He is now with Netmor Ltd.-Applied Modeling Research, Tel Aviv 61390,Israel.
Publisher Item Identifier S 0018-9383(97)00894-0.".--...
~
to 1 V or below [1], [2]. The decrea~e in circuit speedcan be compensated only to a certain extent by innovativetechnologies (e.g., silicon on insulator [3]) and parallelarchitectures or pipelining (e.g., [4]). Consequently, thenatural choice for low-power, high-speed/frequency devicesremain technologies using high-mo~ility materials such asGaAs and related compounds (e.g., [5], [6]). ComplementaryGaAs technologies exhibit advantages over bulk Si suchas low parasitic capacitance and process simplicity similarto SOl. Beyond the improvements provided by SOl, high-mobility technologies also offer very low "on" voltage andhigh transconductance at low voltage resulting in optimumspeed/power performance at supply voltages of 1V and below.At low supply voltages, complementary GaAs technologieshave demonstrated a 4-9 times lower power consumptionwith the same gate delay when compared to SOl [5]. Further,high transconductance is essential for minimizing interconnectdelay [7]. One of the key remaining issues for GaAs, however,is the lack of insulating layers providing low Dit andstable device operation. The use of MODFET's in presentcomplementary GaAs technologies causes excessive gateleakage currents and results in unacceptable high stand-bypower for portable systems. Gate leakage also affects the gainand limits the maximum output power in power amplifiersusing a single power supply.
This paper presents an approach to overcome the abovedescribed bottleneck of insulating layers on GaAs provid-ing low Dit and stable device operation. We will illustratehow the extension of MBE toward deposition of galliumoxide molecules has provided the required ingredients for theimplementation of thermodynamically stable, low Dit oxide-GaAs interfaces. Pivotal aspects include an extremely lowGaAs surface exposure to impurities «10-100 Langmuirs,1L = 10-6 torr s) and the preservation of GaAs bulk andsurface stoichiometry [8], the complete exclusion of GaAssurface oxidation [9]-[11], and the requirement of a specificatomic structure associated with the interfacial atoms of GaAs
and the deposited molecules [12]. We will emphasize that theserequirements are met when specific molecules are depositedin situ under ultrahigh vacuum conditions. Further, we willdiscuss first results obtained using an ex situ deposition schemewhich is applicable to GaAs surfaces exposed to room airand/or processing environments during fabrication of devices.
0018-9383/97$10.00 @ 1997 IEEE
:
PASSLACK et al.: LOW Dit. THERMODYNAMICALLY STABLE Ga20a-GaAs INTERFACES 215
.~The following sections will cover in situ and ex situ wafertbrication, interface characterization, and' modeling.
H. WAFER FABRICATION
The 2-in wafers were fabricated using a multiple-chamberUHV system [13]. The wafer fabrication was comprised of1.5-J-Lmthick n-type (1.6 x 1016cm-3) or p-type (4.4 x 1016cm-3) GaAs epilayers grown by solid source MBE on heavilySi or Zn-doped (1(0) GaAs substrates, respectively. Aftercompletion of GaAs growth, in situ and ex situ schemes wereinvestigated as described in the following paragraphs.
A. In Situ Deposition
After completion of GaAs growth, the freshly grown film~ith an As-stabilized (2 x 4) surface was transferred under a
vacuum of 6 x 10-11 torr to another chamber (backgroundpressure = 10-10 torr) for oxide deposition. Fig. 1 showsthe pressure (solid line) and the corresponding GaAs surface
r lposure (dashed line) which is typically observed prior toopening the shutter for oxide deposition. The time tc and ta isthe time of completion of GaAs growth and the time at whichthe shutter was opened for oxide deposition, respectively. TheGaAs surface was exposed to a pressure not higher than10-10 torr during the transfer and heating of the substrate
r-) the deposition temperature Ts. For the last 2 min, the e-. beam was turned on to heat up the oxide target. Only then
did the GaAs surface begin to experience the pressure from10-10 to 10-7 torr. The pressure rise (predominantly oxygen)was caused by vaporization and thermal dissociation of theoxide targets during e-beam heating. Note that the typical
~aAs surface exposure prior to opening the shutter for oxide..eposition (tc < t < ts) was:::; 10 Langmuirs. Based ontypical initial sticking coefficients for oxygen (e.g., [14]), the
"'-;~aAssurface impurity coverage is estimated at 10-s-10-3%of a mono-layer or 108_1010 surface impurities/cm2 prior todeposition.
The preservation of surface periodicity and atomic orderwas investigated by in situ reflection high-energy electrondiffraction (RHEED) in the time interval tc < t < ts.RHEED pictures acquired from a (1(0) GaAs surface at thetime tc and ts(Ts = 360°C) are identical, showing a (2 x4) reconstructed, As-stabilized surface (Fig. 2). The surface
,-(econstruction (2 x 4) and long streaky RHEED patternsidicate an atomically ordered and contamination free surface.
Consequently, the surface atomic order and periodicity isnot affected by exposure to impurity gases of less than 10Langmuirs in our experiments and the surface stoichiometryis completely preserved prior to opening the shutter for oxidedeposition at time ts.
~
-8. Ex Situ Deposition
Here, the wafers were removed from the MBE chamberafter completion of GaAs growth, exposed to room air for atleast three days, and then loaded into the oxide depositionchamber (background pressure = 10-10 torr). The native
10-6 10
10-7
In-Situ Oxide - (1(0) GaAs
Start deposition (ts) 8 -..~
6 !~8.
4 ~u
~::s2 Cl)
--1;'b 10-8
~Cl)
et 10-9
Turn on e-beam (tJ
Heating to Ts(13 min)
Transfer
\(4 min)Cool down \(8 min)
\10 -10 r
)tc
10 -11-500
02000500 1000
Time (s)15000
Fig. 1. Pressure (solid line) and surface exposure (dashed line) measured be-tween completion of GaAs epitaxial growth (tc) and start of oxide deposition(t.).
oxides which built up on the GaAs surface during air ex-posure were thermally desorbed without As overpressure at asubstrate temperature Ts of 580-600 QC. RHEED was usedto monitor the process of native oxide desorption. No surfacereconstruction was observed and the RHEED streaks were notcontinuous [15]. The latter is indicative of surface roughnessand residual contamination.
Finally, oxide films were deposited at substrate tempera-tures Ta ranging from room temperature to ~ 600 °C usingmolecular beams of gallium oxide. Single crystal Gd3GaS012[16] was used as a source material and evaporated by an e-beam technique. The use of Gd3GaS012 was motivated bythe unavailability of single crystal Ga203 and had led to thefirst successful deposition of gallium oxide molecules formingvery uniform Ga203 films on GaAs [17]. Note that Ga203exhibits a high sublimation temperature of 1725 °C [18]. Thefabrication of MOS capacitors was completed by depositionof TuAu metal dots using a standard shadow mask process.
Reference wafers with identical GaAs epitaxial structure andsubstrate were also fabricated in the same solid-source HI-V
chamber using 1) no oxide deposition (bare samples), and 2)Alo.4sGao.ssAs interfaces.
Ill. STRUCTURAL PROPERTIES
The chemical composition of Ga203-GaAs interfaces hasbeen investigated by X-ray photoelectron spectroscopy (XPS).Compositional profiles and the crystallographic structure ofthe deposited oxide films have been studied by Rutherfordbackscattering spectrometry (RBS), secondary ion mass spec-troscopy (SIMS), transmission electron microscopy (TEM),and RHEED, respectively.
Fig. 3 shows measured (solid line) and simulated (dashedline) RBS spectra of a galli~m oxide film deposited in situon GaAs at Ts = 360°C. The ratio of Ga with oxygen wasdetermined to 43:55.5. Previously published results indicated
216
[110]
/"-
IEEE TRANSACflONS ON ELECTRON DEVICES, VOL. 44, NO. 2, FEBRUARY 1997
[110]
r'-.
Fig. 2. Identical RHEED picture taken at time te and ta. The RHEED picture shows a (2 x 4) reconstructed surface indicating preservation of surfaceatomic order and stoichiometry.
r-
that the nonstoichiometry of the Oa203 film is due to incor-poration of excess elemental Oa [17]. Further, Oa203 filmsusing e-beam evaporation of Od3OaS012 typically exhibit anonuniform Od distribution characterized by a virtually Odfree interfacial region and a peak. in Od concentration atthe film surface (see Fig. 3). Our characterization methods(including XPS and SIMS) can not completely exclude thepresence of Od at the interface and possible effects of Od onelectronic interface properties are not known at this time. Thepresence of Od in the bulk oxide film may affect electricalmeasurements conducted on MOS structures to be discussed
later. In the following, we will refer to the oxide films asGa203 films. Fig. 4 shows the plan view TEM micrograph andthe corresponding electron diffraction pattern of a Oa203 filmdeposited in situ on OaAs at Ts = 360 °C. As observed fromTEM (and RHEED) studies, Oa203 films are amorphous withpartial ordering in a completely disordered state for increaseddeposition temperature [17], [19]. The amorphous phase ofGa203 films eludes problems arising from lattice-mismatchedinterfaces.
Interfacial As 3d core level spectra were acquired using aPerkin Elmer 5600 series XPS spectrometer equipped witha monochromatic Al Ko: X-ray source. The photon energywas 1486.6 eV. Depth profiling was done in situ in a UHVchamber (background pressure = 5 x 10-10 torr) by Arsputtering using an ion gun at 4 keV. The sputtering rateof ~ 0.3 nm/cycle has been significantly smaller than theeffective photoelectron escape depth of ~ 2 nm [20]. Fig. 5shows typical interfacial depth profiles of Oa and As 3d corelevels of a) in situ and b) ex situ fabricated Oa203-GaAsstructures. The 3d binding energies of OaGaAs and ASGaAsare identical to standard XPS lines reported earlier (19.2 and41.2 eV, respectively) [21]. The binding energy of Oaaa203 is21.2 and 20.8 eV for in situ and ex situ fabricated interfaces,respectively. The Oa 3d peak. gradually shifts from the bulk
I""""
I"""'
r"
r-
.r---
In-Situ Oa203 I (1(0) OaAs
Ts =360°Cfox =525A
Od(1.5 at.
\rOa203
Oa (43 at.%)
411e+ '(1.8 MeV)
Detector
OaA~ Experiment.. ... .. Simulation
300Channel
Fig. 3. Measured (solid line) RBS spectrum of 52.5-om thick 08203 filmin situ deposited on OaAs substrate. The simulation (dashed line) assumesconstant Od concentration throughout the film.
100 200 400 500
Oa203 to the bulk OaAs on a length scale consisteqt withthe electron escape depth. The intermediate peak. can easilybe fitted as a sum of two components. Chemical reactionproducts, in particular AS203 (44.6 eV) and As20s (45.7 eV)are not detectable at both in situ and ex situ fabricated
Oa203-GaAs interfaces. Consequently, the chemical reactionAs203 + 2 OaAs -+ Oa203 + 4As (l1G = -62 kcallmol)
[10] resulting in formation of metallic As and degradationof electronic interface properties [11] is excluded. A moredetailed discussion ofXPS spectra acquired from Oa203-0aAsstructures is given in [22].
Note that virtually identical, AS:r;Oyand metallic As free,interfacial As 3d core level profiles were obtained from insitu fabricated amorphous AI203-(I00) OaAs and amorphous
/
PASSLACK et al.: LOW Dit, THERMODYNAMICALLY STABLE 08203-OaA5 INTERFACES 217
:?'",
r---
.~
Fig. 4. Plan view TEM and the corresponding electton diffraction pattern ofa 26-nm thick Ga203 film in situ deposited on GaAs at 360 0 C.
r--Si02-( 1(0) GaAs structures as well. However, in sharp con-trasl: to Ga203-GaAs interfaces, these interfaces exhibit asurface recombination velocity S comparable to that of a baresurface (~107 cm/s) and a Fermi level pinned at the inter-face as demonstrated by C-V measurements [12]. Evidently,the absence of AszO" and metallic As is a necessary, but
r--- not sufficient, requirement for unpinned interfaces on GaAs.We attribute the fundamentally different electronic interface
r' properties observed at various in situ fabricated oxide-GaAsinterfaces to the specific atomic structure associated with theinterfacial atoms of GaAs and the deposited oxide molecules.Therefore, we have chosen the term intrinsic pinning for ourobservation of Fermi level pinning at in situ fabricated Si02-and A1203-GaAs interfaces.
IV. ELECTRICALPROPEImES
Electrical properties have been investigated by capac-r' itance-voltage (C- V), conductance-voltage (G- V), and
cun'ent-voltage (I-V) measurements. Quasi-static and high-frequency (100 kHz, 1 MHz) C-V measurements have beenperformed on MOS capacitors using a Keithley 590 CVAnalyzer, a Keithley 595 Quasi-static CV Meter, and aKeithley 230 Programmable Voltage Source. Further C- Vcharacterizations have been carried out at frequencies of 100Hz, 1 kHz, and 10kHz using a Hewlett Packard 4284APrecision LCR Meter. All characterizations were done atroom temperature in a shielded probe station. Complementaryoptoelectronic interface characterizations have been performedusing the dependence of the photoluminescence intensity IpLon the incident light power density on the sample surface
In- Situ 0820,- GaA8As - Stabilized C(2x4)T.- 550°CIa= 14.3 nm
Ga-3dAs-3d
A808MG8GaA8
84
Jtu46
48 4t 40 36 28 24
Binding Energy (eV)
(a)
20 18
Ex-situ Ga2O:J
T.=480°C
tu =44.7 om Ga-3d
: °llaaAsAs-3d
oa08203.
50
.!!
40~
3048 40 28 24
Binding Energy (eV)
(b)
162044
Fig. 5. Interfacial depth profile of Ga and As 3d core levels of as deposited(a) in situ and (b) ex situ fabricated Oa203 films. Peak energy positions andfull line width at half-maximum (FWHM) are identical to standard values
reported earlier [2]). The smaJI structure in the oxide film detectable at ~ 45eV is associated with the GdGd20a 5s line.
Po. Thorough studies of the internal quantum efficiency ''1(derived from Ipd over a wide range of Po provide theinterface recombination velocity S [23]
S - rE" Dit(E) dE=VtM'lE 2n' E --E.E" 1+ ---!. cosh ~N kT
which is directly related to Dit obtained from C- V and G- Vmeasurements. Here. Vth, a, Ec, Ev~ E, ni, N, Bi' k, and Tare the average thermal velocity, the capture cross sectionof interface states, the conduction and valence band energy,the energy, the intrinsic carrier density, the doping concentra-tion. the intrinsic energy level, the Boltzmann constant, andthe temperature in Kelvin. respectively. The advantages ofthe above outlined rrPo technique include the capability tocharacterize interfaces with extremely thin insulating films,and the exclusion of effects caused by bulk properties of theinsulating film. The rrPo approach also provides the capturecross section a independent of G- V characterizations.
Based on conventional numerical models for semiconductor
heterostructures (e.g., [24]), we developed a new model which
(I)
,-,"
,
. ..,"r~kl ,lI1,d~"I' n!' L'lcctrical (C-V)iI,.','-di"CrnCIII data and i tailored to
il 'iH' I::,.'" 'i ~I.i/\-- -,\.,,11d'o hand-to-hand luminescence (pho-
illillIl1I1IC""CCIICl'1and long minority carrier response time TR.
The model comprises Poisson' s equation and the steady-statecontinuity equations. Details of the model will be discussed
in 125\.
Excellent electrical interface properties have been obtained
for ~50 qc. :S r~ :S 600°C. however, the interface degradedrapidly for I, < 300 QC. Since the dependence of electrical
interface properties on T~ was discussed in [26] for the entirerange of deposition temperatures (0-600 QC), we focus on
.~50 ')C S 'F- :S 600°C in the following section.
/'
A. In SiTU DeposiTed Ga20:~-GaAs Interfaces
In this section. we discuss measurement results obtained
from in siTufabricated Ga203-GaAs structures including C-V,G V. I-V. and 'I-Po data. Dit and S wilI be derived and wewill thoroughly investigate the strong inversion regime whichhas hecn observed in GaAs for the first time. Figs. 6 and 7show typical C- V and G- V characteristics, respectively, of(a, n-type and (b) p-type Ga203-GaAs MOS structures mea-sured in quasi-static mode and at various frequencies between100 Hz and I MHz. The measured curves have been acquiredusing low-intensity light illumination (10 = 9.4 X 1014 cm-2
s' I). Here. I() is the incident light intensity on the samplesurface. Note that the optical transmissivity l' of our metal dotsis .~ 10-'\ resulting in a significantly lower light intensity Ibentering GaAs. As will be discussed later, low-intensity lightillumination is the only viable approach to observe inversion inC. V measurements on epitaxial GaAs due to the small thermalgeneration/recombination rate of GaAs. We have carefullyinvestigated further effects of low-intensity light illumination,in particular possible implications for the derivation of Dit. InFig. 6. the classical operational modes of ideal MOS structuresarc clearly revealed and we will discuss the regimes of 1)
strong inversion. 2) accumulation. and 3) depletion in thefollowing section. The relative dielectric constant of Ga203tilms obtained by fitting simulation results to the measurementdata is 14.2.
I j STrong Inversion: We have investigated different mech-anisms of inversion carrier generation/recombination includ-ing thermal generation/recombination and low-intensity lightgeneration. Thermal generation/recombination of minority (in-version) carriers is dominated by bulk trap levels Et nearmidgap located within the depletion layer where the Fermilevel f.;r crosses Et (crossover point [27]). In our structures,the crossover point is located in the lightly doped epitaxiallayers which typically exhibit bulk minority carrier lifetimesTJ}11in excess of 100 ns [23]. The minority carrier responsetime TH for thermal generation/recombination is given by [27]
r-----
r"
"...-....
r
r---
1
(N
)Tn = J2 n; T
/""'"'
where T == JTnTp' With Tp = Tn = T > 100 ns, n.i =2.7 x lOb cm-:3. and N in the low 1016 cm-3 range, (2) givesTn 2 400 s. Consequently, we did not observe inversion due
IEEE TRANSACTIONS ON ELECTRON DEVICES. VOL. 44. NO. 2. FEBRUARY 1997
In-Situ Ga203 / n-type (100) GaAs
ND = 1.6 X 1016 cm-3Area =2.1 x 10-3 cm2
Sweep Rate =120 mV/s0
-8 -6 -4 -2 0 2
Voltage (V)
(a)
4 6 8
700
In-Situ Ga203 / p-type (100) GaAs600
100 Hz500 ~1kH
10kHz1 MHi
100 kHz
quasi-static
G:'0-~ 400Co)c5.~ 3000-~U
1 kHz
100 Hz
200
100NA =4.4 x 1016 cm-3Area =2.1 x 10-3 cm2
Sweep Rate =120 mV/s
0-8 -6 -4 -2 0 2
Voltage (V)
(b)
4 6 8
Fig. 6. Typical C-V characteristics of in situ fabricated Ga201-GaAs (a)n-type and (b) p-type samples. Oxide thickness tox and flatband capacitanceCJ are 46.2 nm and 310 pF, and 59.4 nm and 304 pF, respectively. Thesubstrate deposition temperature Ts is ~ 600 QC, the light intensity duringthe measurement 10 = 9.4 X 10]4 cm-2 s-].
(2)
to thermalgeneration/recombinationin ourC-V measurementseven at very low sweep rates. Instead, the samples operate inthe deep depletion regime with identical high-frequency andquasi-static capacitance [25].
In contrast to thermal generation/recombination, high gen-eration rates can be readily achieved with low-intensity lightillumination. Using our standard microscope illumination andobjectives with magnification 2, 10, and 20, optical generationrates for electronsand holes Gn = Gp = IO(Ao~ 870 nm)ranging from 4.6 x 1013_2.6 x 1015cm-2 S-1 (12.7 ~ Po ::;715 J-LW/cm2)are obtained. Here, Ao is the wavelength of theincident light. Note that a light pulse is also often used for C-Vmeasurements on state-of-the-art Si02-Si (ni = 1.4 x 1010
cm-3) with Si bulk lifetimes of the order of milliseconds in
800
700 r quasi-static
I600
5000Co)
! 400.
g.300u
200
100II kHz
\
PASSLACK et al.: LOW Dit. THERMODYNAMICALLY STABLE Ga203-GaAs INTERFACES
~ order to generate minority carriers when the structure is biasedin inversion. In our experiments, leakage currents of up to200 pA [28] corresponding to 6 x 1011carriers/cm2 s requiredsteady state low-intensity light illumination to maintain stronginversion.
We have further evaluated the measured conductance G in
strong inversion (Fig. 7). G is related to the inversion carriergeneration/recombination conductance Ggr by [27]
G - G (wCox)2- grG;r + w2(Cox + CD)2
where w, Cox, and CD are the angular frequency, the oxidecapacitance, and the depletion layer capacitance, respectively.In Fig. 7, G is described by (3) for frequencies f ~ 10 kHz.Series resistance loss, which will be discussed later in con-
f' junction with the accumulation regime, significantly increasesfor f > 10 kHz and dominates at high frequencies. Ggris a central component of the MOS equivalent circuit instrong inversion and can be derived from G or from the
~ measured capacitance C in strong inversion [27]. For example,Fig. 8(a) shows the conductance G as a function of frequencywith 10 as a parameter for a Ga203/p-type GaAs structurein strong inversion. Ggr is inferred from the plateaus inthe measurement curves using (3) and Ggr is plotted as afunction of 10 in Fig. 8(b). Linear Ggr-Io relations have been
r' obtained with a regression coefficient of 6.7 x 10-16 and8.3x 10-16/-LScm2s forp- andn-typestructures,respectively.Identical values of Ggr have been inferred using the frequencyWmwhich marks the transition from low-frequency to high-frequency behavior in the measured C-V characteristics [27].Fig. 8(b) also shows the minority carrier response time TR =Ggr/CD. No attempt has been made to theoretically derive
~ an expression for Ggr as a function of 10' Note that Ggrdoes not only depend on the light intensity 10 entering the
r semiconductor through the metal dot but is also influenced byminority carriers generated within a diffusion length of themetal dot edge.
The capacitance overshoot at the onset of inversion (Fig. 6)has been consistently observed on both n- and p-type samplesand may be attributed to filling of trapping centers in the oxide[28].
2) Accumulation: The frequency dispersion observed inaccumulation is caused by oxide layers of relatively lowresistivity adjacent to the interface. This reduces the effective
r oxide thickness and increases the observed capacitance withdecreasing measurement frequency (Maxwell-Wagner effect[29]). It also causes an additional component in the G- Vmeasurements which can be described as a series resistance(Rs) loss [27]. The Maxwell-Wagner effect can be approxi-mated by an equivalent circuit comprising multiple RC circuits
/"-. connectedin series. The equivalentcircuit parametershavebeen inferred from C- V and G- V measurements in accu-mulation. For the C- V and G- V results shown in Figs. 6and 7, respectively, the nonuniforrnity extends over a depthof ~ 10 nm from the interface and may be related to thedepth profile of the Gd concentration (see Section Ill). Theincorporation of Gd gives also rise to trapping effects and
~_._--
In-Situ Ga203/ n-type (lOO) GaAs10-3 I I I I I I
-6 -4 -2 0 2 4Voltage (V)
(a)
(3)
en~8 101~C,)
.g 100c0U
10-1
en~4.) 10 1
~C,)
.g 100c0U
10-1
10-2
10-2
219
104 II I I
ND =1.6 x 1016 cm-3
103 ~Area =2.1 x 10-3 cm2
Sweep Rate =Z I MRz
-Y;- 100kHz~IOkHZ~ -=:::
102
6 8
104 -, I J I
NA = 4.4 x 1016 cm-3Area =2.1 x 10-3cm2 ':Sweep Rate =120 mV/s
103
102
:1~~I MHz ':-
lOOkHz110 kHz
====<1 kHz\100 Hz
In-Situ Ga203 lp-type (100) GaAs10-3 I I I . I I
-6 -4 -2 0 2 4
Voltage (V)
(b)
6 8
Fig. 7. Typical G-V characteristics of in situ fabricated Ga203-GaAs (a)n-type and (b) p-type samples. The oxide thickness tOT is 46.2 and 59.4 nm,respectively. The substrate deposition temperature Ts ~ 600°C. the lightintensity during the measurement 10 = 9.4 x 1014 cm - 2 S- I .
charge injection into the oxide which cause long term drift ofthe capacitance and of device parameters in accumulation [28].
Another interesting feature disclosed in Fig. 6(a) is thequasi-static capacitance being typically smaller in accumula-tion than in inversion. This is attributed to the low effectivedensity of states in the GaAs conduction band comparedto that of the valence band. This characteristic feature has
been experimentally observed for the first time and confirmedby calculations based on our selfconsistent device model forsemiconductor heterostructures.
3) Depletion: We have inferred Dit using results of C-Vand G- V measurements in depletion. Fig. 9 shows typicalresults obtained for Dit as a function of bandgap energy. Wewill first discuss Dit obtained from C- V measurements usingthe quasi-static/high-frequency technique [27]. Dit indicated
'\
220
101
10°
In-Situ Ga203 / p-type (100) GaAs
- V =2 V 10(cm-2s-1) =V =3 V ~ 3.9x1015
.:?'- en.:;11)Co)
~ 10-1Co).gc:::0U
9.4xlO14
10-2 4.6xl013
Area =2.1 x 10-3 cm2
10-3101 102 103
Frequency (Hz)
(a)
104 105
102 10-1'I
In-Situ Ga203 / (1(0) GaAs0 n-typeD p-type
5'<~
-I 10-2 S'=(j
3.~
10-3~"8=
(I)n~
10-4 S'n~
~
r;;.:;O~ 101
r--8 =5Co)
.g§ 10°u....11)>.Cl:!
~
~.§ 10-1Cl:)....11);>
.s
IO...
"'..(: -.. .-::~
0 .l:i::::::.. ::::;.::::::::::..
. .":.':"'::'"0..........e',.'" '0..., """
-).".' '..:0'..""',,,:8'
10-21013
10-5
1016
.. ..
~
1014 1015
Light Intensity 10(cm-2s-l)
(b)
Fig. 8. (a) Measured conductance G as a function of frequency with Ioas a parameter for the in situ fabricated Ga203/p-type GaAs structure of
r Fig. 7(b) in strong inversion, (b) G 9' and TR as a function of Io for the insitu fabricated Ga203-GaAs structures of Fig. 7 in strong inversion,
by open circles and triangles has been derived from p- andn-type samples, respectively. The inferred midgap interfacestate density Dit is in the low 1010 cm-2 ey-1 range.Note that the quasi-static/high-frequency technique requiresconclusive evidence that interface traps do not respond to thehigh-frequency signal. This has been clearly demonstrated inFig. 6 which shows virtually identical measurement results in
/'iepletion above measurement frequencies of 1kHz.Further, we have thoroughly studied the effect of low-
intensity light illumination and its implications for the deriva-tion of Dit. The application of low-intensity light illumi-nation during the measurement causes the electron quasi-Fermi level EFn to depart from the hole quasi-Fermi level
r~E Fp in the semiconductor and modifies the quasi-static andligh-frequency C- V curves. This is demonstrated in Fig. 10
IEEE TRANSACTIONS ON ELECTRON DEVICES. VOL. 44. NO. 2. FEBRUARY 1997
Fig, 9. Typical Dd as a function of bandgap energy for in situ fabricatedGa20:j-GaAs interfaces. The open circles and triangles have been obtainedfrom the quasi-static and I MHz C-V measurements shown in Fig. 6 (cor-rected for the Maxwell-Wagner effect [29]), the filled triangles have beeninferred from G-V measurements of another in situ fabricated Ga~03/n-typeGaAs structure.
..-..~
for a Ga203/n-type GaAs structure. Two consequences ofillumination on the derivation of Dit have been revealed.Since EFn f:; EFp, 1) weak inversion already occurs forEFn > Ei (n-type, see Fig. 10) or Epp < Ei (p-type), and2) the occupancy of all interface states located at Et betweenEFn and EFp is changed when (EFn + EFp)/2 crosses themidgap level Ei. Whereas the first implication only restrictsthe validity range of the quasi-staticlhigh-frequency techniqueto E > Ei + 0.17 eV (n-type) and E < Ei - 0.17 eV (p-type)in our experiments, the second effect can lead to an artificiallyhigh result for Dit. With respect to the latter finding, theinferred midgap Dit should be considered as an upper limit.More details on the effects of low-intensity light illuminationon C-V characteristics will be reported in [25].
For G- V analysis, our measurement results need to becorrected for Rs [27]. Fig. I I shows the Rs corrected high-frequency G- V curves for a Ga203/n-type GaAs structure.Note that the corrected curves shown in Fig. I I were notobtained from the G- Y results shown in Fig. 7 since interfacetrap loss is entirely masked by oxide loss (Rs) for the latterG-Y curves. Using the corrected G-Y curves, Dit is obtainedusing
, , Gp
Dit = 2 { / }2Aq iDUJ (Cox - Cp) Cox
(4)
where A. iD. Gp. Cp, and q are the sample area, a universalfunction (iD ~ 0.12 - 0.4), the measured conductance andcapacitance at the conductance peak [27], and the electroniccharge. Using (4), Dit = (1.2::1: 0.7) x 1011 and (1.5 ::I:0.8) x 1011 cm-2 ey-1 have been obtained in the vicinityof the conduction band edge for E = Ec - 0.29 eV andE = Ec - 0.17 eY, respectively (filled triangles in Fig. 9).The relatively large error range for Dit stems from the
1014r
In-Situ Ga203 / (100) GaAsr..-..
1013 p-type I n-typee
. "v.I I l15 1012 I ICl 0 I11)
0I0
I0I00ci5 0 I
11) 0 I
;11Co) D
<t 10 II0 I0 I£ 0
0 I.s I l1l1
IJ#>.
t101°
Ev E. Ec1
Energy (eV)
-~----~---
PASSLACK et al.: LOW Dd. THERMODYNAMICALLY STABLE Ga:z03-GaAs INTERFACES 221
(a)
~~(.)
~ 400.~g.
U
(b)
>" 0.5~cif
I
LIJ 0.0
800
In-Situ Ga203 / n-type (lOO) GaAsSimulation
600
200 -10'=1013 cm-2s-
. . . ... 10'= 0.n :r' .
0
1.5
1.0EFn
-0.5
EFn+EF2
-10'=1013 cm-2s.1
.. . ... 10'=0-1.0
-5 -4 -3 -2 -I 0 1Voltage (V)
2 3 4 5
Fig. 10. Calculated (a) quasi-static and high-frequency C-V and (b) Fermilevel with E as a parameter as a function of voltage for no light ut) = 0) and
low-intensity light entering the semiconductor Ub= 101:1 cm-2 s-I).
assumption 0.12 ~ ID ~ 0.4 [27]. To accurately determineID would require a conductance measurement as a function offrequency which was not performed. Note that the divergenceof Dit results obtained from G- V and C- V measurementsrapidly increases when EF moves toward Ec. For Ec -0.17 eV, Dit is about an order of magnitude higher whendetermined from C- V curves as compared to G- V results.This is indicative of a slow trapping process in the oxidewhich is partially reflected by the quasi-static/high-frequencyC-V method but is completely invisible in high-frequencyG- V measurements. The time constants of charge trappingin the oxide are > 1 s [28] and depend only indirectly on theposition of E F at the semiconductor surface via the carrierconcentrations in the semiconductor bands [30].
The interface state density Dit is directly related to the in-terface recombination velocity S (1). Therefore, we performedcomplementary studies of the internal quantum efficiency TJ(derived from Ipd from low injection (p < N) to veryhigh injection (p > N) over a wide range of incident lightpower densities Po(l ~ Po ~ 104 W/cm2). Here, p isthe injected carrier density. This technique is based on the
102 --./-.
In-Situ Ga203/ ootype (lOO) GaAs
101
r;;6~(.)
§ 100(.)='
~c::0U
I MHz --J
100kHz10-1
ND = 1.6 x 1016 cm-3Area =2.1 x 10-3 cm2
10,2-4
I
4 60 2
Voltage (V) ---J
Fig. 11. R.. corrected high-frequency G- V curves for an in situ fabricatedGa20:I/n-type GaAs structure (T.. ~ 360°C).
-2
/
relative weight of nonradiative (RSRH)and radiative (Rrad)recombination as a function of Po resulting in a unique curveshape of TJversus Po for a specific 5 (see, e.g., [31], [32]).This is demonstrated in Fig. 12(a) which shows the measured(symbols) and simulated (lines) efficiency TJversus P~ of irsitu and ex situ (to be discussed later) fabricated Ga203/n--.,ttype GaAs structures and, for comparison purposes, 1/ versusP~ of a reference Alo.45Gao.55As-GaAs interface (n-type).Here, p~ = T Po where T is the optical transmissivity ofthe samples and Aa= 514.5 nm is the excitation wavelength.The simulation results have been obtained from calculated PIdepth profiles using our self-consistent, numerical interfa~analysis device model [25], [32]. Note that the analysis hasbeen performed from low to very high injection levels sincethe calculated carrier densities pare 6.5 x 1014and 7.8 x 1017cm-3 at the semiconductor surface (5 = 5000 cm/s) fo-,/the lowest and highest excitation densities Po of 1 and 104W/cm2, respectively. Since the PL intensity is not measuredin absolute units, the measured curves are rigidly shifted tothe calculated ones [31]. The best fit of the simulations tothe measurement data has been attained for 5 = 4000-5000and 1000 cm/s for Ga203- and Alo.45Gao.55As-GaAsstruc-tures, respectively. The above obtained 5 at in situ fabricatedGa203-GaAs interfaces corresponds to a capture-cross sectiona of ~ 10-15 cm2. Further, Fig. 12(b) shows TJversus P~of in situ fabricated Ga203/p-type GaAs structures. Here, theinferred recombination velocity S is ~ 104 cm/so In contrastto the results on n-type samples, TJdoes not saturate for veryhigh injection levels and exhibits a weaker dependence on P~.The observed differences in the TJ-P~ characteristics for n- arv-l
p-type samples are discussed in [25], [32]. '.,J
B. Ex Situ Deposited Ga203-GaAs Interfaces
The in situ process requires complex equipment and more-over, it is not applicable to electronic passivation of GaAssurfaces exposed to processing environments during device
....-.
" ,
1r
/"""-
~
--
l
IPASSLACK et al.: LOW Dit. THERMODYNAMICALLY STABLE Ga203-GaAs INTERFACES 223
102
~'S::s
~ 101..J
...e-~ 0'(i; 10c::8.s85 10-1(,)~Cl)c::
's.a lO-Z9]~
In-Situ Ga203/ (1 (0) GaAs
Ts=560°C Tstress' 30s(OC)=- 1000".." 700
as deposited
800" 700
as deposited
10-3800 850 900
Luminescence Wavelength (nm)950
1 Fig. 14. Typical PL spectra of in situ fabricated n- and p-type as deposited(Ts = 560°C) and temperature stressed Ga203-GaAs structures as well asof corresponding bare samples.
properties to be discussed later. Note that a 36-nm thick Si02cap layer has been deposited prior to rapid thennal annealing inorder to avoid interface degradation caused by penetrating gasspecies. This relatively thick Si02 layer introduces additionalsurface roughness in the course of sputtering and in turn,decreases the height and increases the spatial extension (notshown) of the acquired GaGa203 spectra.
Electronic interface properties of temperature stressed, insitu fabricated Ga203-GaAs interfaces have been investigatedby standard steady-state PL measurements. Fig. 14 showstypical PL spectra of n- and p-type as-deposited and tem-perature stressed Ga203-GaAs structures as well as of thecorresponding bare samples. The incident light power densityPo and the excitation wavelength Ao are 1060 W/cm2 and514.5 nm, respectively. As indicated in Fig. 14, the PL spectraafter RTAin fonning gas for 30 s at 700 or 800 °C are virtuallyidentical to those obtained from corresponding as-depositedsamples. These results imply preservation of excellent elec-tronic interface properties. IpL decreases by only 21% aftertemperature stress of 1000 °C for 30 s giving a slightlyincreasedrecombinationvelocity S ~ 8500 cm/soNote thatidentical results have been obtained after temperature stress ininert gases such as N 2.
B. Photochemical Stability
Fig. 15shows the relative peak photoluminescence intensityof in situ fabricated n-type Ga203-GaAs samples and of ann-type Alo.45Gao.55As-GaAsreference structure as a functionof time t measured for a high-incident light power densityPo = 1060 W/cm2. Here, the unit relative photoluminescenceis defined as the PL intensity of the structures under testnormalized to the PL of a corresponding bare surface at zerotime determined at Po = 1060 W/cm2. Furthennore, theas-measured PL signals have been corrected for alterationsof optical transmissivity T. The slope of the measured PL
-----
--~
104
~'r;;c::8 103.sCl)(,)
5(,)~Cl)c::
.s 10z
.a9.8~
Cl)
.~ 101COd
~
n-type (100) GaAs
1.0= 514.5 nm
Po = 1060 W/cmzAIGaAs-GaAs
In-situ GaZ03-GaAs, as deposited
In-situ Ga203-GaAs (1000 QC, 30 s,85% Nz, 15% H2)
(NH4}zS -Treatment
1000 10 15
Time (Hours)
20 255
Fig. 15. Measured PL as a function of time normalized to the PL of acorresponding bare surface (at zero time) for in situ fabricated Ga203-GaAsinterfaces, Ala .45Gaa .55 As-GaAs structures, and Si3 N4 -GaAs samples us-ing (NH4 h S treatment and ECR plasma deposition. All samples compriseidentical GaAs epitaxial layers and substrates.
versus time characteristics is +0.028, +0.041, and -0.079%/hfor as-deposited and temperature stressed Ga203-GaAs inter-faces, and for the Alo.45Gao.55As-GaAsreference structure,respectively. Thus, in situ fabricated Ga203-GaAs interfacesshow excellent photochemical stability with no degradationof PL intensity to be observed after 22 h of high-intensitylaser excitation. These results imply preservation of excellentinterface properties in agreement with the conclusions obtainedfrom the above described temperature stress experiments.
For comparison purposes, an optimized (NH4hS treatmentof a GaAs surface combined with electron cyclotron resonance(ECR) plasma deposition of Si3N4 and subsequent anneal-ing has been perfonned on n-type bare reference samples.Fig. 15 also shows the best result obtained using the lattertechnique. In sharp contrast to in situ fabricated Ga203-GaAsinterfaces, the relative PL intensity of (NH4hS treated samplesdecays rapidly from 20 to 2.7 during the first hour of high-intensity light exposure. Note that the relatively low PL of the(NH4hS treatedsamplemeasuredat zerotimeis indicativeofS ~ 105cm/s for an As-depositedstructure.
VI. SUMMARY
The extension of MBE toward deposition of gallium oxidemolecules has provided required ingredients for the imple-mentation of thennodynamically stable, low Dit oxide-GaAsinterfaces for low-power, GaAs-based applications. The fol-lowingresults have been obtained. "
1) Excellent electronic interface properties includingDit in the low 1010 cm-2 eV-1 range andS = 4000-5000 cm/s for in situ fabricated Ga203-GaAs
interfaces, and Dit in the upper 1010 cm-2 ey-1 rangeand S = 9000 cm/s for ex situ fabricated Ga203-GaAsinterfaces.
2) Strong inversion in both n- and p-type GaAs.
r . ~24/
3) Thermodynamic interface stability up to at least 800 DCand photochemical interface stability.
4) Charge trapping in the oxide has been revealed as the
dominant trapping mechanism. Alternative Ga203 depo-sition techniques need to be explored to further reducethe total density of trapping centers in the structure andto eliminate the Maxwell-Wagner effect observed inaccumulation.
r
ACKNOWLEDGMENT
M. Hong and J. P. Mannaerts gratefully acknowledge A. Y.Cho for his support and technical discussions.
REFERENCES
[I) A. Chandrashekaran and R. Broderson, "Low-power CMOS digitaldesign." IEEE J. Solid-State Circuits, vo\. 27. pp. 473-484, April1992.
(2) G. Schrom, C. Pichler, T. Simlinger, and S. Selberherr, "On the lowerbounds of CM OS supply voltage:' Solid-State Electron., vo\. 39, pp.425-430, 1996.
[3] W. M. Huang, K. Papworth, M. Racanelli, J. P. John, J. Foerstner, H. C.0 Shin, H. Park, B. Y. Hwang, T. Wetteroth, S. Hong, H. Shin, S. Wilson,
and S. Cheng, "TFSOI CM OS technology for sub I V microcontrollercircuits," in IEDM Tech. Dig., 1995, pp. 59-62.
[4J P. K. Vasudev, "CMOS device and interconnect technology enhance-ments for low-power/low-voltage applications," Solid-State Electron.., vo\. 39, pp. 481-488, 1996.
. , D. E. Fulkerson, S. Baier, J. Nohava, and R. Hochhalter, "Complemen-tary heterostructure FET technology for low-power, high-speed digitalapplications," Solid-State Electron., vo\. 39, pp. 461-469, 1996.
[6] B. Bernhardt, M. LaMacchia, J. Abrokwah, J. Hallmark, R. Lucero,B. Mathes, B. Crawforth, D. Foster, K. Clauss, S. Emmert, T.Lien, E. Lopez, V. Mazzotta, and W. Ooms, "Complementary GaAs(CGaAs TM): A high-performance BiCMOS alternative," in IEEE GaAs" Symp. Tech. Dig.. 1995, pp. 18-21.M. Passlack, M. Uhle, and H. Elschner, "Analysis of propagation delaysin high-speed VLSI circuits using a distributed line model," IEEE Trans.Computer-Aided Design, vo\. 9, pp. 821-826, Aug. 1990.
(8) W. E. Spicer, I. Lindau, P. Skeath, C. Y. Su, and P. Chye, "Unifieddefect model and beyond," J. Vac. Sci. Technol.. vo\. 17, pp. 1019-1027,1980.
(9) G. P. Schwartz, "Analysis of native oxide films and oxide-substrate reac-tions on Ill-V semiconductors using thermochemical phase diagrams,"
I"'" Thin Solid Films, vo\. 103, pp. 3-16, 1983.I J O. Kubaschewski and C. B. Alcock, Metallurgical Thermochemistry.
Oxford U.K.: Pergamon, 1979.
r.1. M. Wood all and J. L. Freeouf, "GaAs metallization: Some problemsand trends," J. Vac. Sci. Technol., vo\. 19. pp. 794-798, 1981.
[ 12) M. Passlack and M. Hong, "Low interface state density oxide-GaAsstructures fabricated by in situ molecular beam epitaxy," in Proc. MRSSpring Mtg., Materials Research Society, San Francisco, CA, 1996, vo\.421 (Compound Semiconductor Electronics and Photonics).
[13J M. Hong, "New frontiers of molecular beam epitaxy with in situprocessing," J. Cryst. Growth, vo\. 150, pp. 277-284, 1995.
(14) R. Ludeke and A. Koma, "Surface studies on clean and oxygen exposedGaAs and Ge surfaces by low-energy electron loss spectroscopy," CRCCrit. Rev. Solid-State Sci., vo\. 5, p. 259, 1975.
[15] M. Hong, M. Passlack, J. P. Mannaerts, T. D. Harris, M. L. Schnoes, R.L. Opila, and H. W. Krauttner, "A Ga203 passivation technique com-
" patible with GaAs device processing," Solid-State Electron.. acceptedfor publication.
[loJ W. Piekarczyk and A. Pajaczkowska, "Dissociation process and crystalgrowth of gadolinium gallium garnet," J. Cryst. Growth, vo\. 46, pp.483-486, 1979.
[17] M. Passlack, E. F. Schubert, W. S. Hobson, M. Hong, N. Moriya, S.N. G. Chu, K. Konstadinidis, 1. P. Mannaerts, M. L. Schnoes, and G. J.Zydzik, "Ga203 films for electronic and optoelectronic applications,"J. Appl. Phys., vo\. 77, pp. 68(r-693, 1995.
[~ CRC Handbook of Chemistry and Physics. 76th ed. Boca Raton, FL:CRC, 1995, pp. 4-58.
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 44. NO. 2, FEBRUARY 1997
[19J M. Passlack, M. Hong, E. F. Schubert. J. R. Kwo, J. P. Mannaerts, S. N.G. Chu, N. Moriya, and F. A. Thiel, "In .fitu fabricated Ga',!O:I-GaAsstructures with low interface recombination velocity," Appl. Phy.f. Lett..vo\. 66, pp. 625-627, 1995. .
[20J S. Tanuma, C. J. Powell, and D. R. Pen, "Calculations of electron
inelastic mean free paths," Surf. Interface Anal., vo\. 17, pp. 927-939,1991.
[21] Y. Mizokawa, H. Iwasaki, R. Nishitani, and S. Nakamura, "Quantitativechemical depth profiles of anodic oxide on GaAs obtained by X-ray photoelectron spectroscopy," J. Electrochem. Soc.: Solid-State Sci.Technol., vo\. 126, pp. 1370-1374, 1979.
[22] M. Passlack, M. Hong, R. L. Opila, J. P. Mannaerts, and 1. R. Kwo,"GaAs surface passivation using in situ oxide deposition," Appl. SurfSci.. vols. 104/105, pp. 441-447, 1996.
[23] R. K. Ahrenkiel and M. S. Lundstrom, Minority Carrier.\' in Ill-VSemiconductors: Physics and Applications. Boston MA: Academic,1993, pp. 48, 54, 58, 119-121, and 129-134.
[24] M. S. Lundstrom and R. J. Schuelke, "Numerical analysis of heterostruc-ture semiconductor devices," IEEE Trans. Electron Devices, vo\. ED-30,pp. 1151-1159, Sept. 1983.
[25J M. Passlack, M. Hong, J. P. Mannaerts, E. F. Schubert, G. J. Zydzik,W. S. Hobson. and T. D. Hams, "Advancing metal-oxide-semiconductortheory: Steady-state nonequilibrium conditions," submitted for publica-tion.
(26) M. Hong, M. Passlack. J. P. Mannaerts, J. R. Kwo, S. G. N. Chu, N.
Moriya. S. Y. Hou, and V. J. Fratello, "Low interface state densityoxide-GaAs structures fabricated by in situ molecular beam epitaxy," J.Vac. Sci. Technol. 8, vo\. 14, pp. 2297-2300. 1996.
[27J E. H. Nicollian and J. R. Brews, MOS (Metal Oxide Semiconduc-
tor) Physics and Technology. New York: Wiley, 1982, pp. 104-156,17(r-222, 222-226, and 331-333.
[28] M. Passlack, M. Hong, and J. P. Mannaerts, "Quasi-static and high-frequency capacitance voltage characterization of Ga203-GaAs struc-
tures fabricated by in situ molecular beam epitaxy," Appl. Phys. Lett.,vo\. 68, pp. 1099-1101, 1996.
[29] A. R. von Hippell, Dielectrics and Waves. New York: Wiley, 1954,p.412.
[30] F. P. Heiman and G. Warfield, "The effects of oxide traps on the MOScapacitance," IEEE Tram. Electron Devices. vo\. ED-12, pp. 167- 178,Apr, 1965.
[31] O. Brandt, K. Kanamoto, M. Gotoda, T. Isu, and N.
Tsukada, "Comparative analysis of the optical quality of singleIno.1 Gao.9As-Alo.3:IGao.6i As quantum wells grown by molecularbeam epitaxy on (I (0) and (311) GaAs substrates," Appl. Phys. Lett.,vo\. 67, pp. 1885-1887, 1995.
[32] M. Passlack, M. Hong, J. P. Mannaerts, J. R. Kwo, and L. W. Tu.
"Recombination velocity at oxide-GaAs interfaces fabricated by in situmolecular beam epitaxy," Appl. Phys. Lett., vo\. 68, pp. 3605-3607.1996.
M. PaS$lack (M'93) received the Dip\.-Ing. de-gree in 1984, the Dr.-Ing. degree in 1988, andthe Dr.-Ing. habi\. degree in 1990, all in electricalengineering from Technische Universitat Dresden,Germany.
From 1984 to 1988, he worked on modeling,design, simulation, and characterization of GaAsdevices and integrated circuits. He was an AssistantProfessor and the Principal Investigator of a researchproject on multiple-state quantum functional devicesand integrated circuits from 1989 to 1991 at Tech-
nische Universitat Dresden. In 1992, he was with the Daimler-Benz ResearchCenter, Ulm, Germany, where he worked on thermal modeling of GaAspower HBT's. From 1993 to 1995, he was engaged in research activitieson insulator-GaAs interfaces and 980-nm high-power pump lasers at AT&TBell Laboratories, Murray Hill, NJ. In 1995, he joined the Motorola PhoenixCorporate Research Laboratories, Tempe, AZ, where he is currently workingon insulator-GaAs interfaces and GaAs heterostructure devices.
/
I'""
PASSLACK et al.: LOW D". THERMODYNAMICALlY S IAHU' (ia!() < .(jaA, I:--JTERI-'ACES
M. Hong <M'83) rCl:clved the H.S. degree In phY-'Il:-'from National Taiwan Univer-.It), Taipei, in 1973.After a two-year military -.erVll.:ein Taiwan', navy,he attended the Universlly 01 California, Berkeley,where he received the MS and Ph.D. degrees inmaterials science and engll1eering in 1978 and 1980.respectively.From 1980 to 198 I. he wa, a Staff Sl.:ientist 1\
at Lawrence Berkeley Laboratory, where he was incharge of a superconducting mall:rial program forapplications in high-energy physll.:' accelerator and
fusion reactor magnets, In 1981, he joined the Material-. Research Laboratory,Bell Laboratories, Murray Hill, NJ, where he wa-. engaged In rescarchactivities of A 15 superconductors. magnetic rare earth ,uperlattlcc-" magnetKthin films for magneto-optical recording. and high-temperature superl.:on-ductivity. In 1988. he joined the Semiconductor Research Laboratory. Bell
Laboratories. where he has developed an in SItu proce'-' combining MBE ofIll-V compound semiconductors, metals. and oxide, wIth dectron cyclotronresonance (ECR) plasma etching.
r---
J. P. Mannaerts, photograph and biography not <lHl1lahh: at the tllnc 01publication.
..---
r'.
R. L. Opila <M'H3) received the B.S. degreein chemistry from the L;OIversity of Illinois.Champaign-Urbana. in 1l)75, and the Ph.D. degreein physical chemistry from the Univcrsity ofChicago. Chicago, IL. 111IqH~.
He joined AT&T Bell Laboratories. Murray Hill.NJ, in 1982 as a Memher of Technical Staff inthe Materials Science DivIsIOn. In 1l)93, he was
promoted to Distinguished Member of TechnicalStaff. Currently. he " a TechOlcal Manager atBell Laboratories. Lucent Technologies. (formerly
AT&T Bell Laboratories), and manages the Surface Preparation and InterfaceReliability Group,
r.
.----"S. N. G. Chu received the B.S. degree In physicsfrom National Taiwan University. Taipei. andthe M.S. and Ph.D. degrees in materials sciencefrom the University of Rochester. Rochester,NY.
He joined AT&T Bell Laboratories. MurrayHill. NJ, in 1980 and has been associated with
fiber-optic communications technology workingon various compound semiconductor materials.device structures. and reliability. Currently, he is aDistinguished Member of the Technical Staff in the
Quantum Phenomena and Device Research Department, Bell Laboratories.Lucent Technologies, Murray Hill, NJ, working on materials for high-speedelectronic devices for wireless applications. He has published over 230 papersin major journals, and holds three patents in these areas.
Or, Chu is a Fellow of The Electrochemical Society and has been activein the Electrochemical Society affairs. He is also a member of the AmericanPhysical Society. ASM International. the Mineral, Metal. and MaterialsSociety. and the Materials Research Society.
1'"--,
".-
-~--.- ~- ~~._------
---" ~ - ---~
simulation for complex
225
N. Moriya received the M.S. and Ph.D, degreesin physics from the Technion-Israel Institute ofTechnology, Haifa, in 1988 and 1991. respectively.
From 1992 to 1995, he was with AT&T BellLaboratories, Murray Hill, NJ, where he inves-tigated diffusion and bandgap effects in strainedsemiconductor layers. Currently, he is with NetmorLtd.-Applied Modeling Research, Tel Aviv. Israel,where he is involved with research and developmentin the area of virtual reality. His research and inter-est areas include detection methods and numerical
systems.
F. Ren (SM'93). photograph and biography not available at the time ofpublication.
J. R. Kwo received the B. S. degree in physicsfrom National Taiwan University, Taipei. in 1975,and the M.S. and Ph.D degrees in applied physicsfrom Stanford University. Stanford. CA. in 1977 and1981. respectively.
In 198 I. she joined AT&T Bell Laboratories,Murray Hill, NJ, as a Member of Technical Staffin the basic research area.' Since joining Bell Lab-oratories, her work has been primarily on thin-filmsynthesis of a variety of novel electronic materials.She was a pioneer in developing metal molecular
beam epitaxy <MBE) methods for epitaxial growth of metallic superlatticesduring 1982-1986. Her work on synthetic rare earth magnetic superlatticesdemonstrated the modulation effect in magnetic properties through artificiallayering for the first time. A variety of new, coherent magnetic order wasdiscovered in the artificial magnetic heterostructures. From 1987 to 1992,she worked on high-temperature superconducting (HTSC) oxide films byboth MBE and sputtering methods for studying superconducting and transportproperties as well as for HTSC devices. In recent years. her interest has shiftedtoward investigating new transparent oxide films for electrophoto applications.and high dielectric constant films for charge.