Growth and doping via gas-source molecular beam epitaxy of SiC and SiCAlN heterostructures and their microstructural and electrical characterization

Growth and doping via gas-source molecular beam epitaxy of SiC and SiCAlN heterostructures and their microstructural and electrical characterization

Diamond ELSEVIER and Related bfaterials 6 (1997) 1282-1288 rowth and doping via gas-source molecular beam epitaxy of heterostructures and their ...

1MB Sizes 0 Downloads 27 Views

Diamond

ELSEVIER

and Related

bfaterials

6 (1997)

1282-1288

rowth and doping via gas-source molecular beam epitaxy of heterostructures and their microstructural SiC an and electrical characterization R.S. Kern I, K. Jtirrendahl, S. Tanaka 2, R.F. Davis * North Carolina State University, Box 7907, Raleigh, NC27695-7907.

USA

Abstract

Gas-source molecular beam epitaxy has been employed to grown thin films of Sic and AlN on vicinal and on-axis 6H-SiC(0001). Growth using the SiH4-C2H4 system resulted in 3C-SiC( 11 I ) epiiayers under all conditions of reactant gas flow and temperatures. Films of SW-SiC(OOO1) were deposited on vicinal 6H-SiC(0001) substrates using the SiH,-C2H4-Hz system at deposition temperatures 2 1350 “C. In situ doping was achieved by intentional introduction of nitrogen and aluminum into the-growing crystal. Monocrystalline AlN was deposited using evaporated Al and ECR plasma derived N or NH3. Films c 50 A grown on the vicinal substrates had higher defect densities compared to those on the on-axis substrates due to the higher density of inversion boundaries forming at most SIC steps in the former material. Metal/AlN/sH-SiC(OOO1) thin film heterostructures which had a density of trapped charges as low as of I x IO” cme2 at room temperature were prepared without post growth treatment. Superior single crystal AIN/SiC hetcrostructures were achieved when very thin AlN was deposited on the on-axis substrates. Single phase monocrystalline solid solutions of (AIN),(SiC), _X were deposited between 0.2 ~~~50.8. A transition from the zincblende to the wurtaite structure was observed at ~~0.25. 0 1997 Elscvier Science S.A.

1. lntroduction

step

Molecular beam cpitaxy (MBE) has attracted interest for cpitaxial SiC [l-S] and AIN [6- 121growth mainly due to the possibilities of a cleaner ambient and lower deposition temperatures. Several SiC [ 13,141 and AIN [ 15-171 deposition studies by gas-source molecular beam epitaxy (GSMBE) have previously been performed in our group. Rowland et al. [I31 reported the deposition of monocrystallinc 3C-Sic on vicinal 6H-SiC(OOOl) by GSMBE using a simultaneous supplyOof CzH4 and S&H, and a growth rate of approx. 100 A h-l. The films contained double positioning boundaries ( DPBs) which formed as a result of 3C-Sic nucleation on terrace sites rather than step sites. Tanaka et al. [I43 achicvcd the first growth of’ 6W-SiCepilayers on GH-SIC substrates by stabilizing * Corrcspoading

author.

’Present address: Hewlett-Packard Optoelectronics CA 95131. USA. ’ Present uddress: (RIKEN ). Saitama 0925.9635/97/$17.00 Pli 8092%9635(97

The Institute 351-01,

of Physical Japan.

0 1997 Elsevier )00066-6

Division,

and Chemical

San Jose, Research

Science S.A. All rights reserved.

flow

growth using low C,H,-to-!&J-I, gus flow

ratios. Rowland et al. [ 15,161 and Tanaka et al. [ 173 used plasma-assisted GSMBE to grow and characterize AIN and the first AIN/SiC heterostructures on vicinal 6H-SiC(OOO1). High-resolution transmission electron microscopy (HRTEM ) images indicated very abrupt interfaces and excellent microstructural quality. Layerby-layer growth was observed on the on-axis 6H-SiC substrates and DPB formation on vicinal substrates due to island coalescence in the vicinity of the steps on the vicinal Sic surface. In the following sections, recent research concerned with the growth of Sic [ IS-21 ] and AIN [l&22-24] thin films via GSMBE in the authors’ laboratory are doscribcd.

2.

A previously described [ 18.25; GSM used to deposit SIC and AIN thin films on

RS, Kern et al. / Diamondand Related Materials6 (1997) 1282-1288

substrates. The substrates were cleaned in a 10% HF solution for 5 min, loaded immediately into the growth chamber and further cleaned in situ using a SiH4 exposure and UHV anneal [IS]. Thin films of 3C- and 6H-SiC were deposited on vicinal 6H-SiC(OOO1) substrates (cut off-axis 3.5 &OS”toward [ 11201).Nominally on-axis 6H-SiC(OOO1)substrates were also used, especially in cases where 3C-SiC( 111) was desired. The SIC deposition experiments were performed at 1000-l 500 “C in either SiH4-C2H., or SiH4-C2H4-Hz environments. Thin films of AlN were deposited on the same type of substrates at temperatures of 900-1300 “C. Two separate sources of nitrogen, namely, molecular N, activated by a ECR plasma source and NH,, were used in conjunction with a standard Al effusion cell operated at 1250 “C. For depositions of (AlN),(SiC), -x solid solutions, C2H4 and S&H, were used as sources for the C and Si, respectively. The films were structurally analyzed with HRTEM, TEM, in-situ reflection high-energy electron diffraction (RHEED), scanning electron microscopy (SEM ), X-ray diffraction (XRD), Auger electron spectroscopy (AES), and secondary ion mass spectrometry (SIMS) [ 1820,22,24]. The electrical properties of the deposited films were characterized by Van der Pauw Hall, currentvoltage (I-V) and capacitance-voltage (C-I’) measurements [18,20,21,23,24].

esuhs and discussion 3. I, Deposition

of3C-SiC(

I I I) in a mm-hydrqcn

iw vironment

Epilayea growth using only SiE-1,and C,lI, always ultcd iu lihns of @C-SiC(1I I) [ 13,14,18 201. The )-labeled curve in Fig. 1 shows a plot of ln(R,) vs. i, where R, is the growth rate (A h-r) and T is the temperature (K), for growth on vicinal 6H-SiC(oOO1) using reactant input flows of 0.75 seem for SiHJ and C2HJ. Analysis of this curve showed that it follows the general Arrhenius equation, R, = R0 exp (- A&/W), where R,,is a pre-exponential factor, dH, is the effective activation energy (kcal mol- ‘) and R is the ideal gas constant ( 1.987cal mol- ‘K-I). From the slope of the curve, the apparent activation energy, Alv,, was determined to be 21.9 kcal mol -I. The rate expression describing this data is given by ln(R,)== 12.3( 11.022/T). From the linear shape of the fitted curve and the size of the activation barrier, the reaction appears to be surface reaction limited. The size of the activation barrier was determined to be independent of nts. No change in the partial pressures of the re flOWrates, &jiH4,Of growth rate was observed for S 0.5, 0.75 and 1.0 seem. These results indicate that the deposition reaction is most likely governed by the

3C-Sic grown with0.75 seem SiH, and variable C-H.

i

10 L.,,l’,l,,‘l,,,‘,,,,‘,,.,I,,,,I 0.55 0.60 0.65 0.70 0.75 0.80 0.85

Fig. 1. Plot of growth rate for 3C-SiC( I1 I) films grown at 1000-1500 “C with 0.75 seem SiH,, and 0.5, 0.75, 1.0 and 1.5 seem CzH,.

decomposition of C2H4 into suitable species to form Sic in the presence of the reaction products fro H,. The CZH4 flow rate, _i&u,, was subsequently varied between 0.5 and 1.5 seem with fsiH4maintained at seem. The identical slopes in, Fig. 1 indicates that the reaction mechanism was unchanged as a ~~~cti~~ of the CZH4 flow rate. Fig. 2 shows a se ln(R,) vs. ln(fc,,,) in the range OS- 1.5 1200 and 1300‘C. The deposition rate a T and j’&, was determined to be In (I 1.000/T)+0.631n(j&rt). Again, the a tion barrier for this process was -22 k the curves wcrc linear over the ra dependence on the C,H4 flow rate, the ~e~o~it~~~process appeared to be controlled by surface ~~ac~i~?~~~~ The 4 appeared to be the rate limiting ste

Fig. 2. Plot of In(R,)vsh(JC2,,,) for 3GSiC( I I1 ) films grown iit I100. 1200 and 1300 “C with 0.75 seem SiHs and OS.- I.5 scctnCtfla.

3.2.

Deppositian qf SC-SIC(

I 11) irr0 h_rdro@Wicm!deat

Additions of Hz (5 seem flow rate) during the growth of SiC produced several advantageous changes in the resulting films [ I S-201. At deposition temperatures c 1350 -‘C. the benefits were increased surface smoothness and a greatly enhanced growth rate of the 3C-Sic films. The activation energy for the films grown with 0.75 seem SiHJ, 0.75 seem C,H, and 5 seem Hz at 1000-1300 “C was 21.6 kcal mol - ‘. Thus, the presence of Hz did not change the magnitude of the barrier over this temperature range. Since the slopes of the curves shown in Fig. I and that from this latter study were nearly identical, the reaction mechanism was both apparently unchanged and independent on the input flow of C#J. The observations arc attributed to the presence of Hz which apparently provided the impetus for at least one of the following processes: (i) adsorption of C2H, onto the surface of the growing film, (ii) sweeping of the unreacted source gas species or unwanted product species from the growth surface and/or (iii) formation of a suitable reactant species in tandem with the decom-

position or reaction of CzH4 in the presence of H,.

which is in exccllcnt :igre~itm~ with values oalculatcd from CVD research ( 13.0 kcal mol -“I using the sitme reactant sources, sub-

strate orientation and crystallographic face [ 26 J. Although the growth rate was still strongly dependent on the deposition temperature, the decrease in the value of this activation barrier was an excellent indicator of the effect of the Hz gas on the gas chemistry, growth kinetics and gas flow dynamics. Limited studies showed that varying the CzH4 input from 0.375 to 1.0 seem did not result in a change in the growth rate. This indicated that the rate controlling factor at these temperatures in the presence in H2 had most likely changed. It is important to note that growth on both vicinal and on-axis substrates proceeded at approximately the same rate (within 5%) under similar growth conditions. However, 3C-Sic was always produced on the on-axis substrates regardless of growth co:lditions, because the smaller atomic dil?‘usion distance relative to the distance between steps.

3.4’. Doping of SC- and 6H-SiCjiltns The presence of the N in Sic grown from the vapor phase is particularly significant, because it is the most shallow donor impurity. Consequently, it was the agent responsible for the n-type character of unintentionally doped Sic films [ l&21,20]. Considerable differences in the background atomic nitrogen and electron concentrations in the SiC films was measured when the CzH3 flow rate WIS modulated in the range 0.375.-0.75 scan. Similar to the “silt-competition epitaxy” arguments of Larkin ct al. [27,28], the N contamination lcvcl was siynilicantlgf &crwxd by increasing rho mout~t or in the gas plms. Fig, 4 shows ;I SIMS C',!!., dclivcrcd prolilc from a (7H-SiC( OOOI ) film grown at 1400 “6‘ using 0.75 SCCI~I SiN,, 5 SC6111 I-I1 arKI :I VUiilblc c‘2t-i4 few. The chgc in N content with C source supply is very apparent from the abrupl cllilngcs that occur in

l:ip. 4. SIMS profile or a OH-SiC(OOO1) lilm grown at 1400 ‘C using 0.75 scan SiH,,. 5 scan Hz and i\ variable C2H, tlow (0.375 swm in Region I. 0.5 seem in Region II. and 0.75 scan in Region 111).

RS. Kern rt al. / Diamond and Related Materials 6 ( 1997) 1282-1288

the N depth profile. A tlow of 0.75 seem C2H4 resulted in the incorporation of N at the detection limit for N (~5 x 10” cm-‘) in the SIMS system. Hall measurements of undoped 6H-SiC films with the lowest atomic nitrogen and electron concentration showed a mobility of 434cm2V-‘s-l. To study n-type doping of the reactant mixture with the lowest total input of source gases, 0.75 seem SiH4, 0.375 seem CzH4, and 5 seem H, (for 6H-Sic), was chosen. Donor doping was performed in situ at 1400 “C on monocrystalline 6H- and 3C-Sic using both NH3 (diluted to 300 ppm in H,) and pure N,. Doping in the range 5 x 101s-8 x 10” cmm3 was achieved using the NH3/H2 mixture and 1 x 1018-4x 10” crne3 with the N2 additions. Figs. 5(a) and (b) show the electron mobility as a function of electron concentration in the close packed plane for SC-SiC( 111) and 6H-SiC( 0001) at room temperature. Epilayers of 3C- and 6H-SiC were also doped p-type by evaporating Al from a standard MRE effusion cell

n-type

XXiC(

I%.5

during the growth of Sic epilayers. All growth e ments were performed at 1450 “C using 0.75 seem 0.75 seem C2H4, and 5 seem Hz (for 6H-Sic). Thk higher C2H, flow rate was used to take advantage of the site-competition process which resulted in a decrease in the concentration of background N, a compensating impurity in p-type Sic. The higher temperature was used in an attempt to aid in dopant activation. Acceptor doping of 2 x lo”-8 x 1Ol8cme3 was achieved. Figs. 6(a) and (b) show the hole mobility as a function of hole concentration in the close packed plane for 3C-SiC( 111) and 6H-SiC( 0001) at room temperature. 3.5. Deposition of AIN thin$filnts and AI/AIN,fSiC MIS-structures Thin films of wurtzite structure (2H) AlN were deposited on 6H-SiC substrates [ 181. The primary problem encountered using an ECR source as a nitrogen source was the unintentional incorporation of impurity atoms

p-type 3cxK( I 11)

1I I) epilayer

epilayer

-r-vrmm~+rr”Tl~-_,-rmn”

n-type

61-LSiC(OOOI)

@layer

1

fi0I i

Fig. 5.Roo’n temperature rier

conccntrntion

for

6H-SiC(OOO1 ) epilayers.

measurements n-type

(a)

of ckclrol~ 3C-(

mobility

11I ) epilayers

vs. ci’rand

(b)

Fig. 6. kmm concentration

temperclturs measurements of hole mohlsty VT cxricr for p-type (a) 3c’-( I I i ) eplla&Xs md lb)

6%SiC(OOO1)

epilayers.

from the matcriat that mitdc up ihe liner crucible. Substitution to N ti, improved litm qualily and grow ~1~ rate. Growth rates of about X25 ii h ’ ww uchicvod for films grown al 1050 (1 using Lhe ECR, whereas. growth rates of about 1000 A h ’ were obtained for films deposited using NI-I., at I I00 C. With the addition of !=I?. higher growth rates and smoolher surfaces were achieved. A comparison of ihin (0001 ) oriented AIN fihns t-z 50 A) grown simultaneously on vicinat and

on-axis

wrc made. Films of surface exhibited ~1 lough sin-face. whereas. the tihns grown on the on-axis substrate possessed a very smooth surface and excelten~ indicative of two-dimensional rhickness uniformity, growth. The achievement of highcr quality AIN films in the taller case were due to the low density of surface irregularities (i.e. steps and kinks) on the on-axis subslrates. Because of rhe very tow critical rhickness of the

AIN

OH-SiC(

grwvn

011

0001 ) the

suhstr;itcs

viciilitl

AlN epilayers grown on the Sic substrates, the quality of the AlN films thicker than approx. 150 A did not depend on the substrate. Above this thickness. the misfit/threading dislocations becpme the dominate defect, and the effect of substrate orientation on crystal quality became of decreasing importance. The wide band gap and low dielectric constant of AlN permits its use as the gate material in AI/AIN/SiC MIS structures [ 18,231. The low lattice mismatch and excellent thermal stability of these materials relative to the SiOz/SiC system make this a very attractive system for insulated-gate device technology using Sic and, potentially, GaN semiconductor microelectronic devices. The AI/AIN/SiC MIS-structures grown in this research had low interface charge trap densities ( 1 x IO” cm ‘) and very small hystereses when swept from deep depletion into accumulation and back. The interface states were negatively charged (acceptor-like) on the Si-face 6H-SiC substrates. Positively charged traps that were attributed to deep level states became the dominant source of interface states and hysteresis when the MIS structures were heated or illuminated under a halogen lamp.

Pseudomorphic hetcrostructures of wurtzitic AIN and 3C-Sic wcrc grown on vicinal and on-axis 6H-SiC( 0001 ) substrates under a variety of conditions of reactant input, substrate orientation iltld tcmpcraturc [ 181. TIC substrate temperature and orientation wcrc dctcrmincd to all’ect the growth of both AIN i\ntl Sic’. ‘Ib pr<)d~cc SiC lihns with i\ h bd 01’ ddkcls. hycrs d’ AlN having a subcriticill thickness wcrc used. E~~11iIycrs01’ 3C-SiC( I I I ) 011thcsc AIN l~ycrs had cxcellc~~t clcctrical propertics with room tcmpcrnture clcctron mobilitics ilS high as 72 I cm1 V ’ s ’ for unil~tetltlot~~~llydoped fihns grown with a high Cdl-I, flow in order to take alvantago of “site-competitiol1i epitaxy” [ 27,281. Single-phase, monocryst‘lllinc ( AIN ),( SIC ), ., solid solutions were deposited [ 18,241 at 900 I300 using C2H, and Si&, together with the Al clTusion cell and the ECR source. By ranging the composition bclwccn 0.2 5s ~0.8 both zincblendc (3C ) ilIId wurtzitc ( 2H) single-phase lihns were grown with the transition from cubic to hexagotlal structure at about .S= 0.25. Figs. 7(a) and (b) show HRTEM micrographs and RlIEED patterns from lilms with the coml>osition I AIN ),,.gSiC I,,,, and (AIN ),j.3(SiC)o,-,. rcspcctivcly.

ury onocrystalline thin lih-ns of 3C-SiC( I I I ) wcrc grown by GSMBE on both vicinal and on-axis GH-SiC(0001) substrates using SiHJ and C2M.,. Growth

of homoepitaxial -Sic ~~~0~ 1 and an increase in growth rate were ieved between 1350 and 850~ C on vicinal 6H-SiC( 0001 ) substrates with the add~ti~~~~ of Hz to the reactants. The cpilaysrs were doped n- or p-type in situ by adding N or Al. res processing environment. Films of AIN were deposited on 6~-SiC(~Q~ ) substrates. The high quality of the films allowed the fabrication of metal!A1N!SiC MIS structm-es and AIN/SiC heterostructures. Single-phase. monocrystalline (AIN),(SiC), _s solid solutions of both zincblende (0.21~5025) and wurtzite (0.25 1~10.8) structures were also deposited.

[ 161 LB. Rowland, R.S. Kern. S. Tanuka, R.F. Davis. Appl. Pbys. J.&e:162 (1993) 3333. [ 131S.Tanaka. R.S. Kern. R.F. Dnvis. Appl. Phys. Lett. 66 ( 1995) 37. [IS] R.S. Kern, Ph.D. Thesis. North Carolina State University, Raleigh, NC, 1996. [ 191 R.S. Kern, S. Tanaka, L.B. Rowland, R.F. Davis, J. Cryst. Growth.. submitted. [20] R.S. Kern. K. Jtirrendnhl, S. Tanaka, R.F. Davis, Phys. Stat. Sol. A., accepted. [21] R.S. Kern. R.F. Davis, Appl. Phys. Lett., accepted. [X2] U.S. Kern. L.B. Rowland. S. Tanaka, R.F. Davis, J. Mater. Rrs. 8 (1993) 1477.

[23] M.O. Aboelfotoh. R.S. Kern, S. Tanaka. R.F. Davis, C.I. Harris, Appl. Phys. Lett. 69 (1996) 2873. [ 241 R.S. Kern, L.B. Rowland, S. Tanaka, R.F. Davis, J. Mater. Res., submitted. [25] L.B. Rowland. S. Tanaka, R.S. Kern. R.F. Davis. in Amorphous and Crystalline Silicon Carbide IV. C.Y. Yang. M.M. Rahman, G.L. Harris (Eds.). Springer, Berlin, 1992. p. 84. [26] Y.C. Yang, R.F. Davis, J. Electron. Mater. 20 ( 1991) 869. [27] D.J. Larkin, PG. Neudeck, J.A. Powell, L.G. Matus. Appl. Phys. Lett. 65 (1994) 1659. [28] D.J. Larkin, S.G. Sridhara, R.P. Devaty, W.J. Choyke, J. Electron. Mater. 24 (1995) 289.