inp by mocvd

8/7/2019 InP by MOCVD

1/8

JOURNAL DE PHYSIQUE IVColloque C5, supplkment au Journal de Physique 11, Volume 5 , juin 1995Growth of InP in a Novel Remote-Plasma MOCVD Apparatus: an Approachto Improve Process and Material PropertiesG. Bruno , P. Capezzu to and M . LosurdoCentro di Studio per la Chimica dei Plusmi-CNR, Dipartimento di Chimica-Universita di Bari-via Orabona,4-70126 Bari, Italy

Abstract: Remote plasma metalorganic chemical vapor deposition (RP-MOCVD) technique, thoughrelatively new, is becoming more and more important in the processing of the 111-V semiconductormaterials and, specifically, of indium phosphide. So far different processes have been designed for (a) thecleaning of InP substrates to remove surface native oxide by reduction with H2 plasnla and (b) the InPdeposition under P H 3 plasma preactivation. This paper deals with InP homoepitaxial growth bytrimethylindium (TMI) and plasma preactivated PH3. Optical emission spcctroscopy (OES)measurements evidence the presence of 1'H and pH2 radicals, and of H-atoms in Lhe plasma phase. Massspectrometry (MS) sampling close to the growth surface reveals the presence of alchylphosphinc((CH3)2PH, CH3PH2). indium-phosphorus adduct and biphosphine (F2Hq). whose relative amountsdepend on the growth conditions. Stoichiometric I d ' epilayers having good structure and morphology.and with a very high photolurninescencc intensity are prepared under PH3 plasma preactivation, even atvery low VIIII ratio (=20) and reduced temperalure (-550C).

1. INTRODUCTIONConvent ional MOCVD processes require high deposition temperature, to supply the activation

energy for both gas phase and surface reactions, and high hydride (pH3 or AsH3) f lux to grow goodquality In -V epitaxial layers [I]. Recently, innovative aspects have been introduced in the ITIIV M O C V Dtechnology by using plasmas in remote configurat ion (RP) [2-41, which assures the absence of radiativedamage to epitaxial layers. The plasma, as a secondary source of energy, removes kinetic limitations inthe gro wth mechanism of phosphorus based semiconductors [5], and offers low temperature [6,7.8] andlow VIIII ratio processing [9], mainly by the pre-cracking of the thermally stable PH3. Plasma processesalso offer a valid alternative to the use of the highly toxic PH3, by performing the in situ generation ofPH3 throug h the ablat ion of red-phosphorus in H 2 plasma [lo-121. Another plasma applicat ion o f someimportance for the 111-V epitaxial growth is the in situ substrate cleaning process [13-141. In fact, thepresence of nat ive oxid cs and carbon contaminations can impede the growth of epitaxial layers. Theseox ides can b e r emoved by in sit~rH2 plasma treatment just before the film growth t o yield clean surfaces.

Although there is a great worldwide interest in many research laboratories on the use of plasmasource in the MOCVD technologies, there has hcen little rcal progrcss toward the understanding of theplasma ch emis try and of wh at really determines the surface processes. In reality, the difficulty in theinvestigation of RP-MOCVD processes lies in the large numbers of parameters to examine (pressure, r.f.power, g as flow, tem peratu re) and in the lack of diagnostic techn iques available on the MOCVDapparatus. Thu s, the design of a n ew reactor architecture, al lowing in situ diagnostics for both the processand the material, is essential . With respect ro this. optical diagnostics such as reflectance differenceSpectroscopy ( R D S ) [IS] and spccrl.oscopic ellipsomctiy (SE) [16] (which ar c comp;ltiblc with t l iureactive environment of MOCVD process) a1.e well adapt to investigate thc sul-face chcm~strics.A i o .mass spcctronlccly (MS) is a powerful tcchnii luc to invcsl i~atcM O C V D p r o c c ~ s c s .cvcn though i t has

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1995555
http://www.edpsciences.org/http://dx.doi.org/10.1051/jphyscol:1995555http://dx.doi.org/10.1051/jphyscol:1995555http://www.edpsciences.org/


2/8

C5-482 JOURNAL D E P HY S I QU E IV

been utilized, mainly, for ex situ studies of chemistry [17,18] and kinetics [I91 of the precursordecomposition.In this paper, we report on the use of a low pressure RP-MOCVD apparatus, working at 2 ton, togrow InP epilayers from trimethylindium (InMe3) and phosphine (PH3). The PH3 and the carrier gas,H2, are plasma activated, while the organometallic InMe3 is not directly plasma dissociated, but isinjected into the plasma d ownstream flow, where it interacts with plasm a produced re active sp ecies, suchas pHx (x=0-2) radicals and H-atoms. Process diagnostic data, as obtained by mass spectrometry (MS)and optical emission spectroscopy (OES), are used to highlight the chemistry of the PH3 plasma pre-cracking, also during the InP deposition process. From the InP deposition experiments we deriveimportant informations regarding the growth kinetics and the correlations between chemistry at thegrowth surface and the material properties and morphology. The InP epilayers are characterized bystructural (SE M, X-ray diffraction ), and optical (photoluminescence, ellipsometry) techniques.

2. EXPERIMENTALThe I nP growth experiments were carried out in a home-made R P-MOC VD apparatus in which theplasma qua rtz tube is assembled on the stainless-steel MO CV D chamb er, as schematized in fig.1.

Figure 1:Assonometric scheme of the RP-MOCVD experimental apparatus.Th e plasma, fed with H 2 or PH 3-H2 g as mixtures, is generated by supplying r.f. power (13.56 MHz)to two semicircular external electrodes. Trimethylindium (TMI) is directly injected in the plasma

down stream flow throu gh a dispersal ring located in the MOC VD cham ber at the end of the plasma tube.The substrate (Fe-doped semi-insulating (100)-oriented InP) is mounted on a resistively-heatedmolybdenum susceptor whose temperature is controlled by a K-thermocouple. The InP substrates arecleaned in isopropanol and then chemically etched in a H2SO :H2 02: H2 0 (8 : l : l ) solut ion and4imm ediately introduced into a load-lock chamb er evacuated to 10- tom and then transferred into theMOCVD chamber. S om e deposition experiments have been performed on In P substrates which a re in situcleaned by operating H2 plasma before the film growth [20]. InP growths are performed, with andwithout pH3 plasma activation (r.f. power = 20 watt), at the pressure of 2 torr and at the total gas flowrate of 500 sccm. The VnII ratio and the substrate temperature have been changed in the range of 20 -300 and 550 C - 650C , respectively.The study of the PH? plasma preactivation has been performed by investigating thc effect of r.f.power (5-60watt) and of kta l gas flow rate (3-800 sccm).


3/8

Optical emission spectroscop y (OMA EG&G -PAR) has been used to monitor the light em itted, in therange 200 - 800 nm, from the active excited species present in the plasma phase. A quadrupolar massspectrometer (VG-SPX-Elite 600) has sampled gases, through a 100 pm orifice, in the boundary layernear the grow th surface.3. RESULTSAND DISCUSSION

3.1PH3 plasma preactivationFigure 2 sho ws the profiles of the PH3 decomposition and of its partial conversion to biphosphine,P2H4, and phosphorus, Pq, ohtained by MS sampling, as a function of r.f. power and residence time ofthe PH3-H2 mixture in the plasma region. In the same figure, in addition to stable phosphorus products,PH radical and H-atom density profiles, as derived by the actinometric approach of the OES

measurements 1121, are also reported.

Figure 2: Effect of (a) r.f. power and (b) residence time on the gaseous reaction product distribution (left side) and on thePH radical and H atom density (right side) in P H 3 -H 2 ( X P M ~= 0.03) plasma at P = 2 torr. (Other conditions forexperiments (a) gas flow rate = 100 sccm and (b)r.f. power = 20 watt).

The fact that PH radical and H-atom densities increase with r.f. power is related to the parallelincrease of the electron density and, hence, to the effectiveness of processes which also govern the PH3and H2 dissociation and , specifically, of the following reactions:

On the contrary, the residence time dependence, shown in fig. 2b, indicates that pH3 decompositionreaches a plateau value (-20%) owing to the competition between PH 3 decomposition and its formationand, hence, to the following partial chem ical equilibrium (PCE):


4/8

C5-484 JOURNAL DE P H Y S I Q U E I V

Among all plasma products, PH radicals and H -atoms are particularly important a s they can promotethe TM I decompo sition [3] and the carbon removal from the growing epilayers [4], through processes:

CH3(ads) + PH, + CH3PHxT ( 6 )

InCHg(ads) + PH + In P + CHqT (8)where the desorptio n of CH 3 radicals is favoured by both H- atoms and pHx radicals, while pHx radicalsdirectly prom ote the InP deposition (see eq. 8). Thus, the control of P H and H amoun ts and, hence, of theplasma parameters is determinant for the ongoing surface process. But, the true values of pHx and Hdensities at the substrate position depend, besides the formation in the plasma phase (eqs. 1-3), on thedisappearance processes in the downstream region, where the following recombination processes areeffective:

The H atom wall recombination process, (H W% 1/2H2 ) has been found to be uneffective underpresent experimental conditions [21]. Nevertheless, the estimation of H-atom and pHx radical densities,at the substrate position in the downstream of, respectively, H2 and PH 3-H 2 plasmas, has been done, forH atoms, on the basis of etching rate measurements of a phosphorus film and, for pHx radicals, throughmeasurements of the phosphorus deposition rate.

3.2. InP deposi t ion a nd m ater ia l character izat ionThe effect of PH3 plasma decomposition on the InP growth kinetics has been evaluated, at fixeddeposition temperature, through the analysis of the growth rate profile as a function of PH3 molar

fraction (XPH~)in the gas feed. The resulting data are included in fig.3, which illustrates the effect of theplasma preactivation.From the figure, it comes out that, at high VlIII ratio (150 - 300), the growth rate is almostindependent on X P H ~and is mainly determinated by TMI partial pressure [22]. Under conditions of verylow VIIII ratio (


5/8

Figure 3; Deposition rate of InP films y~ PH 3 molar fraction in the TMI-PH3-HZ gas mixture. Dashed line is for the effectof the PH3 plasma preactivation. Shadowed area is for the deposition of InP film with metalic indium droplets.(Other experimental conditions: T = 550C, P = 2 tom, gas flow rate = 500 sccm, r.f. power= 20 watt).T h e presence of different chemical environments near the growth surface, depending o n the V nHratio, substrate temperature and on PH3-H2 plasma activation, has been revealed by MS analysis. MSdata for some typical deposition experiments are listed in Tab. I in terms of ion peak intensities of themost significant P-species, norm alized to that of the PHj+ peak in thc fccd.

Table I: Mass spectrometric data for some InP deposition experiments with and without PH3 plasma preactivation, and atdifferent V/III ratio and depositio n temperature.

It can be observed that under plasma activation condition (# InP8) the unreacted PH3 amount is lowerthan that measured for conventional deposition (# InP7), i.e. the PH 3 decomposition is equal to 70% and45% with and without plasma, respectively. Also, a slight increase of the PH 3 decomposition degree ismeasured with a temperature increase (see # InP2, InP6). As for P-decomposition products, the formationof biphosphine, PzHq. and phosphorus. P j , increase with both temperature ( # InP3, InP6) and pIasmaactivation (# InP7, InP 8). Moreove r, the following situations ca n be distinguished:

EXPERIMENTAL RUNVJIII Tg("C)

InP2300 55 0

InP6150 600

InP720 550

InP8-plasma20 550

pH3+

0.5

0.47

0.55

0.3

I ~ P H ~ +

l . l ~ - ~

P(CH3)+

&lo-4

3.10-3

6.10-3

5-10-3

P(CH3)2+

2.10-3

P2H4+

4.10-4

2.10-3

8 . 1 0 ~ ~

1 . 10 -~

P4+

9.10-4

2.10-3

2 . 10 . ~

2.10-3


6/8

C5-486 JOURNALDE P H Y S I Q U E IVi) the ion fragment InPH3 + appears only for the experiment # InP2, i.e. at very high VAII ratio (-300).This ion fragmen t can be related to the presence of Me3InPH3 adduc t in gas phase.ii) methylphosphines formation increases with a VDII ratio decrease and, in particular, thedimethylphosphine (P(CH3)2+) is present at very low VDII ratio (# InW). The methylphosphinesformation can be correlated to the low availability of H-atoms at the growth surface, so as the CH3desorption process of equation (6) prevales on that of equation (5). Th e presence of d ialchylphosphine,subtracting active phosphorus from the growing InP surface, causes the growth of non stoichiometricmaterial, as confirmed by SEM -ED S joined m easurements (# InW) revealing the presence of metallicindium droplets. On the contrary, when in presence of large amount of H atoms a t the growth surface, asin the case of plasma activated process (see # InP8), the dimethylphosphine disappears, the availability ofactive phosphorus at the growth surface increases, and stoichiometric material is obtained even at lowV/JII ratio.As regards the material characterization, X-ray diffraction (XRD), spectroscopic ellipsorneuy (SE)and photoluminescence measurements have been performed on some typical samples deposited with(#InP8) and without (# InP6) plasma preactivation. Samples produced without plasma and undercondition of very low VDII ratio (# InP7) are not interesting as they are not stoichiometric, In-richepilayers.Figure 4 shows the XRD spectra of #InP6, #InP8 samples and, for a comparison, of the (100) InPsubstrate. The X-ray diffraction peaks are centered at the same Bragg angle ( O =31.57"), so indicatingthe epitaxial (100) oriented nature of the deposited InP epilayers. In addition, the narrow half widthevidences the high cristallinity of the InP6 film, whereas the broader features, characterizing the XRDpeak of the sample #InP8, indicates some structural disorder. In the latter case, the presence of InPmicrocluster with a slight different lattice constant is supposed. The origin of these microclusters has notbeen established, but, in the opinion of the present authors, it could be related to the presence of hydrogenin the m aterial bulk.

-2000 -TOW 0 1000 2000OrnegaIPThetc (seconds)

Figure 4: X-ray diffraction peaks for InP film deposited (a) without and (b)with PH3 plasma preactivation. The peak of the(100) InP substrate is shown for comparison.Figure 5 shows the S E spectra of the imaginary part, Ei , of the pseudod ielectric function for the InP

epilayers deposited under conditions of # InP2, # InP6, and # InP8 experiments. The inset shows thethickness and composition values of each layer as obtained by the multilayer BEMA model [23] used forthe f i t of the experime ntal &i curves.


7/8

Figure 5: Ei ellipsometric spectra vs energy for InP epilayers deposited from experiments (a) #InP8-plasma, (b) #In N a nd(c) #InP2 (see Tab.1).

The most important feature is the lower oxide layer thickness (241)for the InP8-plasma sample thanthat for sam ples deposited under conventional conditions (# InP2, # InP6). The formation kinetics of thenative oxide layer can be used to speculate on the effectiveness of H2-PH3 plasma in passivating andstabilizing the InP surface [24]. In addition, the sample (# InP2) deposited at high VIIII ratio (=150) andlow temperature (550C) shows a complex multilayer structure on a disordered InP epilayer film whichincludes 2% of voids.The goodness of the InP epilayers can also be evaluated by optical properties such as refractive inde x, n,and absorption coefficient, a, which can be derived by SE measurements. Figure 6 shows the n and avalues energ y for the sam e samples of figure 5 and for a clean and well-ordered c-InP sample.

11L ~ ~ ~ 1 ~ ~ ' ~ 1 ~ ~ ' ~ 1 ~ 4 ~ ~ 1 ' 1 ' a " ' " 1 " ' " " " 1

1 1,5 2 2,5 3 3,5 4 4.5 5Energy, (eV) 1 1.5 2 2,5 3 3.5 4 4.5 5Energy, (eV)

Figure 6: Refractive index, n. and optical absorption coefficien~a, y~ energy for the same InP films of fig.5. The literaturedata of a clean and well-ordered c-In P film are also shown.


8/8

C 5 - 4 8 8 JOUR NAL DE P H Y S I Q U E IVThe higher the epilayer quality, the higher the refractive index and the lower the absorption coefficient.Again, the optical quality of the InP8-plasma epilayer is comparable to that of the best film depositedwithout plasma (# InP6).Photoluminescence measurements, performed at 2K, have revealed for the InP8- plasma film a PLefficiency five times higher than that of InP6, which is, in turn, fifty times higher than that of the c-InPsubstrate. The PL efficiency for plasma deposited samples remains higher than that for samples depositedwithout plasma, even at room temperature.

AcknowledgementsThe authors wish to thank dr. L. Tapfer (PASTISICNRSM) for the XRD measurements and fruitfuldiscussion.

References[ l ] G.B. S tringfellow , Org anom etallic Vapor Phase Epitaxy:Theory and Practice (Acad emic Press Inc.,1989).[2] S.W. C hoi, K.J. Bachma nn, G . Lucovsky, J. Vac. Sci. Techno l. A, l l (3 ) (1993) 626.[3] S.W. Choi, G. Lucovsky, K.J. Bachmann, J. Vac. Sci. Techn ol. B , lO(3) (1992) 1070.[4] T. Sug ino, K. Kawarai, M. Maeda, J. Shirafuji, J. Crys. Grow th, 139 (1994) 15.[5] M Behe t, A. Brau ers, P. Balk, J. Cryst. Growth, 107 (1991) 209.[6] S. Sakai, S. Yamamoto, M. Umeno, Jap. J. Appl. Phys., 25(8) (1986) 1156.[7] U. Suda rsan, N.W. C od y, T. Dos luogolu , R. Solanki, J. Crys. Growth, 9 4 (1989) 978.[8] H. Takei, T. Hamada, T . Hariu, J. Crys. Growth, 115 (1991) 309.[91 M. Behe t, A. Brauers, H. Luth, P. Balk, J. C ~ y s .Growth, 107 (1991) 209.[l o] S.F. Fang, T. Hariu, J. Vuc. Sci. Tech nol. A, 9(4) (1981) 2253.[ l l ] R. Schu tz, H.L. Hartnagel, Int. J. Electronics, 67(2) (1989) 233.1121G. Bruno, M. Losurdo, P. Capezzuto, J. Vuc. Sci. Technol. A, 13(2) (1995) 1.[13] A.J. N elson, S. Frigo, D. Mancini, R. Rosenberg, J. Appl. Phys., 70(10) (1991) 5619.[14 1P.G. Hofstra, D.A. Thompson, B.J. Robinson, R.W. Streater, J. Vac. Sci. Technol. B, 11(3) (1993)985.[15] D.E. Aspnes, R. B hat, C. Caneau, E. Colas, L.T. Florez, S. Gregory, J.P. Harbison, I. Kamiya, V.G.

Keramidas, M.A. Koza, M.A.A. Pudensi, W.E. Quinn, S.A. Schwarz, M.C. Tamargo, H. Tanaka,. Crys. Growth, 120 (1992) 71.[16] B. Drevillon, E. B ertran, P. A lnot, J. Oliver, M. Razeghi, J.Appl. Phys, 60(10) (1986) 3512.[17] C.A. L arsen, G.B. String fellow, J. Crys. Grow th, 7 5 (1986) 247.[18]N. I. Buchan, C.A. L arsen, G.B. Stringfellow, J. C~ys .Growth, 92 (1988) 605.[19] J.A. M cCau lley, R.J. Shu l, V.M. D onnelly, J. Vac. Sci. Tec hnol. A, 9(6) (1991) 2872.[20] G. Brun o, M. Losurdo, P. Capezzu to, this conference.[21]G. Bruno, M. Losurdo, P. Capezzuto, Appl . Phys. Lett., to bc published.[22] H. Heine cke, A. Bra uers, H. L uth, P. Balk, J. Crys. Grow th, 77 (1986) 241.[23] D. A. G. B ruggemann, Ann. Phys., 24 (1985) 636.[24] R.W. Glew , A.R. Adarns, C.G. Crookes, P.D. Grecn c, S.N. Holmes, S.A. Kitching, P.C. Klipstein,

D. Lanccficld, R.A. Stradling, R.A. Woolley, Semicond. Sci. Technol. 6 (1991) 1088.

inp by mocvd

Documents