high reflectivity symmetrically strained znxcdymg1ÀxÀyse-based distributed bragg reflectors for...

4
High reflectivity symmetrically strained Zn x Cd y Mg 1 Àx Ày Se-based distributed Bragg reflectors for current injection devices O. Maksimov, a) S. P. Guo, F. Fernandez, and M. C. Tamargo Department of Chemistry, City College and Graduate Center of CUNY, New York, New York 10031 F. C. Peiris and J. K. Furdyna Department of Physics, University of Notre Dame, Notre Dame, Indiana 46556 ~Received 2 January 2001; accepted 12 March 2001! Distributed Bragg reflectors ~DBRs! with different numbers of periods were grown by molecular beam epitaxy from Zn x Cd y Mg 1 2x 2y Se-based materials on InP substrates. The alternating ZnCdSe/ ZnCdMgSe layers were symmetrically strained to the InP substrate greatly simplifying the growth process and increasing the uniformity. High crystalline quality was also achieved in these structures. A maximum reflectivity of 99% was obtained for a DBR with 24 periods. Chlorine doped ~n-type! DBRs were grown and their electrical and optical properties were investigated. Electrochemical capacitance–voltage profiling indicated that the doping concentrations of the ZnCdSe and ZnCdMgSe layers were 4310 18 and 2 310 18 cm 23 , respectively. The reflectivity of the doped DBR structures was comparable to that of the undoped ones. © 2001 American Vacuum Society. @DOI: 10.1116/1.1374625# I. INTRODUCTION There has recently been considerable effort to obtain highly reflective semiconductor distributed Bragg reflectors ~DBRs!. This is due to the growing interest in a number of semiconductor microstructures, such as vertical cavity sur- face emitting lasers ~VCSELs! that use highly reflective mir- rors, usually in the form of DBRs. These DBRs form the Fabry–Pe ´ rot ~FP! cavities necessary for operation of the de- vices. Although significant success has been achieved in the development of III–V systems ~GaAs/AlAs on GaAs substrates 1 and AlAsSb/GaAsSb on InP substrates 2 !, at- tempts to fabricate similar structures from wide-gap II–VI compounds have been limited. One of the reasons is that the difference in the indices of refraction ~Dn! of many of these materials is relatively small, making it hard to achieve high reflectivity by the growth of an adequate number of periods. Nevertheless, reflectivity in excess of 90% has been reported for ZnSe/ZnTe, ZnMgSe/ZnCdSe, and ZnSe/ZnMgS systems on GaAs substrates. 3–5 However, reflectivity in excess of 99% is necessary for actual device applications. Further- more, at least one electrically conductive semiconductor DBR is required for current injection devices. The growth of Zn x Cd y Mg 1 2x 2y Se alloys has been exten- sively studied due to their applicability for the fabrication of red–green–blue light emitting diodes 6 and high crystalline quality of the materials has been demonstrated. 7 Due to the relatively large difference between the indices of refraction of ZnCdSe ( E g 52.18 eV! and ZnCdMgSe ~E g 52.93 eV! materials, D n / n is on the order of 12% at l5633 nm. 8 The- oretical calculations predict that reflectivity above 99% can be achieved after the growth of 20–26 periods. 9 We have fabricated several DBR structures from Zn x Cd y Mg 1 2x 2y Se materials with different numbers of pe- riods. In order to facilitate the growth process and to increase the reproducibility in the layer composition, alternating sym- metrically strained layers were used. Although a significant amount of strain was present, layers and structures with high crystalline quality were achieved. Since high n-type doping levels, above 10 18 cm 23 , were previously achieved in Zn x Cd y Mg 1 2x 2y Se epilayers doped with chlorine, 10 doped DBRs were also grown. In this article, we report a very high reflectivity ~99%! for undoped II–VI based DBRs as well as promising optical and electrical characteristics for an n-type doped DBR structure. II. EXPERIMENTAL TECHNIQUES Growth was performed in a Riber 2300 molecular beam epitaxy ~MBE! system. This system consists of two growth chambers connected by an ultrahigh vacuum transfer chan- nel. One growth chamber is used for the growth of As-based III–V materials and the other is for wide band gap II–VI materials. All the undoped samples were grown on epi-ready semi-insulating ~Fe-doped! InP ~001! substrates. The sub- strates were de-oxidized in the III–V chamber by heating to 480 °C with As flux impinging on the surface. A lattice matched InGaAs buffer layer ~100 nm! was grown after de- oxidation. Then, the samples were transferred to the II–VI chamber for growth of the Zn x Cd y Mg 1 2x 2y Se layers. Growth was initiated at 170 °C by Zn irradiation 7 followed by growth of a 9 nm thick ZnCdSe interfacial layer. The temperature was then increased to the optimum growth tem- perature of 270 °C, after which an additional 50 nm of ZnCdSe was deposited. Then, alternating symmetrically strained ZnCdMgSe/ZnCdSe layers were grown. The use of symmetrically strained layers allowed us to keep the cell temperatures constant during growth. The Zn to Cd beam equivalent pressure ~BEP! ratio was fixed at 0.6. The Cd, Zn, and Se shutters were kept open while the Mg shutter was alternatively opened and closed, producing ZnCdMgSe and ZnCdSe epilayers. Under these conditions we fabricated a! Author to whom correspondence should be addressed; electronic mail: [email protected] 1479 1479 J. Vac. Sci. Technol. B 194, JulÕAug 2001 1071-1023Õ2001Õ194Õ1479Õ4Õ$18.00 ©2001 American Vacuum Society

Upload: oleg-maksimov

Post on 12-Nov-2014

215 views

Category:

Technology


0 download

DESCRIPTION

J. Vac. Sci. Technol. B 19„4…, Jul/Aug 2001

TRANSCRIPT

Page 1: High reflectivity symmetrically strained ZnxCdyMg1ÀxÀySe-based distributed Bragg reflectors for current injection devices

High reflectivity symmetrically strained Zn xCdyMg1ÀxÀySe-baseddistributed Bragg reflectors for current injection devices

O. Maksimov,a) S. P. Guo, F. Fernandez, and M. C. TamargoDepartment of Chemistry, City College and Graduate Center of CUNY, New York, New York 10031

F. C. Peiris and J. K. FurdynaDepartment of Physics, University of Notre Dame, Notre Dame, Indiana 46556

~Received 2 January 2001; accepted 12 March 2001!

Distributed Bragg reflectors~DBRs! with different numbers of periods were grown by molecularbeam epitaxy from ZnxCdyMg12x2ySe-based materials on InP substrates. The alternating ZnCdSe/ZnCdMgSe layers were symmetrically strained to the InP substrate greatly simplifying the growthprocess and increasing the uniformity. High crystalline quality was also achieved in these structures.A maximum reflectivity of 99% was obtained for a DBR with 24 periods. Chlorine doped~n-type!DBRs were grown and their electrical and optical properties were investigated. Electrochemicalcapacitance–voltage profiling indicated that the doping concentrations of the ZnCdSe andZnCdMgSe layers were 431018 and 231018 cm23, respectively. The reflectivity of the doped DBRstructures was comparable to that of the undoped ones. ©2001 American Vacuum Society.@DOI: 10.1116/1.1374625#

taorou

heet

VIt t

ghdsrtm

feto

-of

io

a

me-as

-antigh

ghs

amthan-

sedI

ady-to

e

VI

eem-ofallye ofellam,wasandted

ma

I. INTRODUCTION

There has recently been considerable effort to obhighly reflective semiconductor distributed Bragg reflect~DBRs!. This is due to the growing interest in a numbersemiconductor microstructures, such as vertical cavity sface emitting lasers~VCSELs! that use highly reflective mir-rors, usually in the form of DBRs. These DBRs form tFabry–Pe´rot ~FP! cavities necessary for operation of the dvices. Although significant success has been achieved indevelopment of III–V systems~GaAs/AlAs on GaAssubstrates1 and AlAsSb/GaAsSb on InP substrates2!, at-tempts to fabricate similar structures from wide-gap II–compounds have been limited. One of the reasons is thadifference in the indices of refraction~Dn! of many of thesematerials is relatively small, making it hard to achieve hireflectivity by the growth of an adequate number of perioNevertheless, reflectivity in excess of 90% has been repofor ZnSe/ZnTe, ZnMgSe/ZnCdSe, and ZnSe/ZnMgS systeon GaAs substrates.3–5 However, reflectivity in excess o99% is necessary for actual device applications. Furthmore, at least one electrically conductive semiconducDBR is required for current injection devices.

The growth of ZnxCdyMg12x2ySe alloys has been extensively studied due to their applicability for the fabricationred–green–blue light emitting diodes6 and high crystallinequality of the materials has been demonstrated.7 Due to therelatively large difference between the indices of refractof ZnCdSe (Eg52.18 eV! and ZnCdMgSe~Eg52.93 eV!materials,Dn/n is on the order of 12% atl5633 nm.8 The-oretical calculations predict that reflectivity above 99% cbe achieved after the growth of 20–26 periods.9

We have fabricated several DBR structures froZnxCdyMg12x2ySe materials with different numbers of priods. In order to facilitate the growth process and to incre

a!Author to whom correspondence should be addressed; [email protected]

1479 J. Vac. Sci. Technol. B 19 „4…, Jul ÕAug 2001 1071-1023 Õ200

insfr-

-he

he

.eds

r-r

n

n

e

the reproducibility in the layer composition, alternating symmetrically strained layers were used. Although a significamount of strain was present, layers and structures with hcrystalline quality were achieved. Since highn-type dopinglevels, above 1018 cm23, were previously achieved inZnxCdyMg12x2ySe epilayers doped with chlorine,10 dopedDBRs were also grown. In this article, we report a very hireflectivity ~99%! for undoped II–VI based DBRs as well apromising optical and electrical characteristics for ann-typedoped DBR structure.

II. EXPERIMENTAL TECHNIQUES

Growth was performed in a Riber 2300 molecular beepitaxy ~MBE! system. This system consists of two growchambers connected by an ultrahigh vacuum transfer chnel. One growth chamber is used for the growth of As-baIII–V materials and the other is for wide band gap II–Vmaterials. All the undoped samples were grown on epi-resemi-insulating~Fe-doped! InP ~001! substrates. The substrates were de-oxidized in the III–V chamber by heating480 °C with As flux impinging on the surface. A latticmatched InGaAs buffer layer~100 nm! was grown after de-oxidation. Then, the samples were transferred to the II–chamber for growth of the ZnxCdyMg12x2ySe layers.Growth was initiated at 170 °C by Zn irradiation7 followedby growth of a 9 nm thick ZnCdSe interfacial layer. Thtemperature was then increased to the optimum growth tperature of 270 °C, after which an additional 50 nmZnCdSe was deposited. Then, alternating symmetricstrained ZnCdMgSe/ZnCdSe layers were grown. The ussymmetrically strained layers allowed us to keep the ctemperatures constant during growth. The Zn to Cd beequivalent pressure~BEP! ratio was fixed at 0.6. The Cd, Znand Se shutters were kept open while the Mg shutteralternatively opened and closed, producing ZnCdMgSeZnCdSe epilayers. Under these conditions we fabricail:

14791Õ19„4…Õ1479Õ4Õ$18.00 ©2001 American Vacuum Society

Page 2: High reflectivity symmetrically strained ZnxCdyMg1ÀxÀySe-based distributed Bragg reflectors for current injection devices

gn.

1480 Maksimov et al. : ZnxCdyMg1ÀxÀySe DBRs 1480

J. Vac. Sci. Techno

TABLE I. Parameters of the ZnCdSe and ZnCdMgSe calibration layers used for the DBR structure desi

Layera'

~Å!D(a/as)'

~%!ai

~Å!D(a/as) i

~%!a0

~Å!D(a/as)0

~%!

Zn0.58Cd0.42Se 5.828 10.69 5.864 10.08 5.846 10.39Zn0.22Cd0.24Mg0.54Se 5.903 20.58 5.870 2 0.02 5.886 20.29

dsenpep

ron

ng

raSo

poin

omd

lltstic

aan

tiorin

lta

uth00

d

e

ory.of

llyersrreiax-ial

antsn inde-es

lwas

BRingndi-,

ign

edlidnse

llyan

three DBR structures having 12, 18, and 24 periods, allsigned to have nominally the same optical layer thicknes

Chlorine doped DBRs that consisted of 12 periods ahad nominally the same layer thickness as the undoDBRs were grown under the same conditions as the undoDBRs described previously. Semi-insulating andn-type ~S-doped! InP ~001! substrates were used. Chlorine was pvided by a solid ZnCl2 source heated in a Knudsen effusiocell. The ZnCl2 cell temperature was kept constant durigrowth.

Prior to growth of the DBR structures, a series of calibtion layers of ternary ZnCdSe and quaternary ZnCdMgalloys were grown to determine the composition, indicesrefraction, and growth rates of the alloys used. The comsition of the ternary ZnCdSe alloys was calculated assuma linear dependence of the lattice constant on the alloy cposition ~Vegard’s law!. The lattice constant was estimateusing double crystal x-ray diffraction~DCXRD! measure-ments. Since the calibration layers could be partiastrained,~115! a and b asymmetrical XRD measuremenwere made to obtain the perpendicular and parallel latconstants:a' andai. The bulk lattice constant (a0) was thencalculated from the following equation:

a05a'$12@2n/~11n!#@~a' /ai!/a'#%. ~1!

A value of 0.28 was used forn, which is the Poisson ratio.The composition of the quaternary ZnCdMgSe alloys w

determined by combining the bulk lattice constant and bgap energy data as described elsewhere.9 The refractive in-dices of the epilayers grown were estimated by extrapolafrom the dispersion curve8 that was previously obtained foZnCdMgSe layers with different band gap energies usthe prism coupler technique.11

Electrical characteristics ofn-type doped DBRs werestudied by Hall effect measurements, current–voltage~I –V!measurements, and electrochemical capacitance–vo~ECV! profiling. For the Hall effect measurements, 0.530.5cm squares~van der Pauw geometry! were cut from the DBRstructure grown on a semi-insulating substrate. ForI –Vmeasurements, gold dots~0.3 mm2! were deposited on thetop surface of the DBR structure grown on an1-InP sub-strate. Gold wires were attached to the back of the InP sstrate, which was covered with In and to the gold dots ontop. ECV profiling was performed using a BioRad Pn 43ECV profiler. A solution of 1 M NaOH and 1.25 M Na2SO3

was used as an electrolyte. This solution was previouslyveloped for electrochemical etching of ZnxCdyMg12x2ySealloys.12

The room-temperature reflectance of the DBRs was msured using a Cary 500 ultraviolet~UV!-visible spectropho-

l. B, Vol. 19, No. 4, Jul ÕAug 2001

e-s.dd

ed

-

-ef-g-

y

e

sd

n

g

ge

b-e

e-

a-

tometer with a variable angle specular reflectance accessThe data were calibrated using an Ag-coated mirrorknown reflectivity as reference.

III. RESULTS AND DISCUSSIONS

The calibration layers were designed to be symmetricastrained to the InP substrate. Although the calibration laywere relatively thick~640 nm for ZnCdSe and 980 nm foZnCdMgSe!, the x-ray analysis indicates that the layers wenearly pseudomorphic. The ZnCdSe layers were under bial tension and the ZnCdMgSe layers were under biaxcompression. The ZnCdSe and ZnCdMgSe lattice constand their lattice mismatch to the InP substrates are giveTable I. The composition and the lattice mismatch weresigned to provide a relatively large difference in the indicof refraction between the constituent layers~Dn/n512%!and to make the mean~perpendicular! lattice constant equato that of the InP substrate. The mean lattice constantcalculated from the perpendicular lattice constants (a') ofthe calibration layers. The mean lattice mismatch of the Dto InP, (Da/a)' , was designed to be less than 0.01%. Usthe dispersion relations, the desired thicknesses of the ividual layers~i.e.,l/4n! were calculated to be 59 and 67 nmrespectively, for ZnCdSe and ZnCdMgSe at the deswavelength of 633 nm.

The DCXRD rocking curve obtained from an undopDBR structure with 12 periods is shown in Fig. 1. The soline represents the experimental spectrum. The most inte

FIG. 1. Double crystal x-ray rocking curve obtained from a symmetricastrained ZnCdSe/ZnCdMgSe DBR structure with 12 periods grown onInP substrate.

Page 3: High reflectivity symmetrically strained ZnxCdyMg1ÀxÀySe-based distributed Bragg reflectors for current injection devices

rls

tich

os

thttic

iceTho

is

Bh

deaaeflec-s i-

threthinm

t

.1

lktra-r-to

enotatetiontedet

h-and

ri-res.

r ad

1481 Maksimov et al. : ZnxCdyMg1ÀxÀySe DBRs 1481

narrow peak in the center represents the~004! reflection fromthe InP substrate and the zero order peak from the supetice is seen as a shoulder on the InP peak. More thansatellite peaks on each side of the zero order peak~whichcorresponds to the mean lattice constant of the superlat!are visible in the DCXRD rocking curve, indicative of a higquality periodic structure. The observed satellite peak ptions are plotted in the form of sinu in Fig. 2. The linearrelation to the order of the peaks gives evidence thatobserved peaks are diffraction peaks from the superlastructure. From the slope, the thickness of the period~L! isestimated to be 124.5 nm, very close to the designed thness of 126 nm. The dotted line in Fig. 1 represents a thretical simulation based on the period calculated above.position and the intensity of the satellite peaks are in goagreement with the experimental data.

The reflectivity spectrum for the sample of Fig. 1shown in Fig. 3~a!. A maximum reflectivity of 88% wasobtained at around 617 nm. The peak reflectance of the Dis somewhat blueshifted relative to the intended value. Tshift is in agreement with the difference between thesigned and measured thickness. The reflectivity spectrDBR structures that consisted of 18 and 24 periodsshown in Figs. 3~b! and 3~c!, respectively. Increasing thnumber of periods to 18 and 24 periods increases the retivity to 98% and 99%, respectively. The reflectivity spetrum of a chlorine-doped DBR that consisted of 12 periodshown in Fig. 3~d!. The maximum reflectivity of that structure is 89% at around 597 nm.

Using the thickness for each layer calculated frompositions of the x-ray satellite peaks and the indices offraction at the specific stop-band centers obtained fromdispersion curves, we calculated the optical thickness ofdividual layers in the four DBR structures. These are sumarized in Table II. The optical thicknesses are very closel/4 for the four DBR structures grown.

The electrical characteristics of then-type DBR structurewere studied. Room-temperature Hall effect measuremen

FIG. 2. Position~sinu! of satellite peaks plotted vs satellite peak order fosymmetrically strained ZnCdSe/ZnCdMgSe DBR structure with 12 perio

JVST B - Microelectronics and Nanometer Structures

at-ix

e

i-

ee

k-o-ed

Ris-ofre

c-

s

e-e--

to

of

the DBR structure gave a free electron density of 331018 cm23 and a parallel mobility of 72 cm2/V s. Thiselectron mobility was lower than that measured for buZnCdSe and ZnCdMgSe epilayers at this doping concention ~'200 cm2/V s),9 possibly due to the interface scatteing in the DBR structure. ECV profiling was performeddetermine the net carrier concentration (nd2na) in the indi-vidual layers. The ECV profile plot is shown in Fig. 4. Sincthe electrochemical etching rate in these materials wasknown precisely, etching was continued until the substrwas reached. At this point the measured carrier concentrachanged abruptly. The etching depth was then estimafrom the known overall thickness of the DBR structure. Nelectron concentrations of 4.431018 cm23 for ZnCdSe and1.931018 cm23 for ZnCdMgSe were measured by this tecnique, in general agreement with the Hall measurementsthe expected values.

Electron transport perpendicular to the layers is of pmary importance for device applications of these structu

s.

FIG. 3. Reflectivity spectra of four DBR structures with~a! 12, ~b! 18, and~c! 24 periods, and~d! n-type doped with 12 periods.

Page 4: High reflectivity symmetrically strained ZnxCdyMg1ÀxÀySe-based distributed Bragg reflectors for current injection devices

1482 Maksimov et al. : ZnxCdyMg1ÀxÀySe DBRs 1482

J. Vac. Sci. Techno

TABLE II. Parameters of the DBR structures grown.

DBR Compositiondi

~nm! n Nlmax

~nm!R

~%!Optical thickness

~l!

Nominal Zn0.58Cd0.42Se 59 2.68 633 0.25Zn0.22Cd0.24Mg0.54Se 67 2.36 0.25

a Zn0.58Cd0.42Se 58.3 2.72 12 617 88 0.26Zn0.22Cd0.24Mg0.54Se 66.2 2.37 0.25

b Zn0.58Cd0.42Se 63 2.67 18 646 98 0.26Zn0.22Cd0.24Mg0.54Se 71.6 2.35 0.26

c Zn0.58Cd0.42Se 65.9 2.65 24 661 99 0.26Zn0.22Cd0.24Mg0.54Se 74.9 2.33 0.25

d Zn0.58Cd0.42Se:Cl 53.7 2.75 12 597 89 0.25Zn0.22Cd0.24Mg0.54Se:Cl 61 2.38 0.24

uc

.i

vi

g

mb

man

ec-estheitu-thaten

l-S-o.

h.ewstIn-c-

nce

ch-

st.

.

na,

.une

Therefore,I –V measurements were performed on the strture. TheI –V characteristics of then-type DBR stack mea-sured at room temperature is shown in the inset of Fig. 4rectifying behavior, most probably due to the band offsetsthe conduction band, was observed. The rectifying behais expected to be minimized by the use of step gradeddigitally graded layers13 as well as by the modulation dopinof the graded interfaces.14

IV. CONCLUSION

We have fabricated distributed Bragg reflectors frosymmetrically strained ZnCdSe/ZnCdMgSe layers grownmolecular beam epitaxy. High crystalline quality was deonstrated by DCXRD measurements. A 99% reflectivity wobtained from a 24 period DBR structure. The optical aelectrical properties of chlorine-dopedn-type ZnCdSe/

FIG. 4. ElecrochemicalC–V profile for a ZnCdSe/ZnCdMgSen-type DBRstructure with 12 periods, showing the net electron concentration as a ftion of depth. The inset shows theI –V characteristic of the same structurmeasured at room temperature.

l. B, Vol. 19, No. 4, Jul ÕAug 2001

-

Anoror

y-sd

ZnCdMgSe DBR structures were also investigated. Refltivity spectra, similar to those of undoped DBR structurwith the same number of periods, were observed fromdoped structures. High carrier concentration in the constent layers was achieved. These results demonstrateZnxCdyMg12x2ySe is a promising material system for thdesign of highly reflective, conductive DBRs for applicatioin VCSELs.

ACKNOWLEDGMENTSThe group from the City College of New York acknow

edge the National Science Foundation, Grant No. EC9707213, and the Army Research Laboratory, Grant NDAAD17-99-C-0072, for support provided for this researcThis work was performed under the auspices of the NYork State Center for Advanced Technology on UltrafaPhotonics and the Center for Analysis of Structures andterfaces. The group from the University of Notre Dame aknowledges the support provided by the National ScieFoundation through Grant No. DMR-0072897.

1V. Bardinal, R. Legros, and C. Fountaine, Appl. Phys. Lett.67, 3390~1995!.

2O. Blum, M. J. Hafich, J. F. Klem, and K. L. Lear, Appl. Phys. Lett.67,3233 ~1995!.

3F. C. Peiris, S. Lee, U. Bindley, and J. K. Furdyna, Semicond. Sci. Tenol. 14, 878 ~1999!.

4F. C. Peiris, S. Lee, U. Bindley, and J. K. Furdyna, J. Appl. Phys.86, 719~1999!.

5I. Tawara, I. Suemune, and S. Tanaka, J. Cryst. Growth214Õ215, 1019~2000!.

6M. C. Tamargo, W. Lin, S. P. Guo, Y. Luo, and Y. C. Chen, J. CryGrowth 214Õ215, 1058~2000!.

7L. Zeng, S. P. Guo, Y. Y. Luo, W. Lin, M. C. Tamargo, H. Xing, and GS. Cargill III, J. Vac. Sci. Technol. B17, 1255~1999!.

8S. P. Guo, O. Maksimov, M. C. Tamargo, F. C. Peiris, and J. K. FurdyAppl. Phys. Lett.77, 4107~2000!.

9O. Maksimov, S. P. Guo, L. Zeng, and M. C. Tamargo, J. Appl. Phys.89,2202 ~2001!.

10W. Lin, A. Cavus, L. Zeng, and M. C. Tamargo, J. Appl. Phys.84, 1472~1998!.

11F. C. Peiris, S. Lee, U. Bindley, and J. K. Furdyna, J. Appl. Phys.84,5194 ~1998!.

12W. Lin, S. P. Guo, and M. C. Tamargo, J. Vac. Sci. Technol. B18, 1534~2000!.

13K. Tai, L. Yang, J. D. Wynn, and A. Y. Cho, Appl. Phys. Lett.56, 2496~1990!.

14E. F. Schubert, L. W. Tu, G. J. Zydzik, R. F. Kopf, A. Benvenuti, and MR. Pinto, Appl. Phys. Lett.60, 466 ~1992!.

c-