Available online at www.sciencedirect.com

008) 8424–8430www.elsevier.com/locate/tsf

Thin Solid Films 516 (2

Influence of boron surface enrichment on the growth mode of BGaAsepilayers grown on GaAs by metalorganic vapour phase epitaxy

Philippe Rodriguez ⁎, Laurent Auvray, Anthony Favier, Jacques Dazord, Yves Monteil

Laboratoire des Multimatériaux et Interfaces, UMR CNRS 5615, Université Lyon 1, Université de Lyon, 69622 Villeurbanne Cedex, France

Received 16 January 2008; received in revised form 28 February 2008; accepted 7 April 2008Available online 13 April 2008

Abstract

BGaAs epitaxial layers were grown by metalorganic vapour phase epitaxy (MOVPE) on (100) GaAs vicinal substrates using diborane,triethylgallium and arsine as precursors. For growth temperatures of 580 and 610 °C, we studied the boron incorporation in the epilayers, theirboron surface composition and their growth mode as a function of the diborane flow-rate, using respectively X-ray diffraction, X-rayphotoelectron spectroscopy and Atomic Force Microscopy. We observed that increasing the diborane flow-rate strongly favours the developmentof step-bunching. This trend was related to a pronounced boron enrichment of the surface, as a consequence of a surface segregation of boron.These results suggest that boron behaves as a surfactant during the MOVPE growth of BGaAs and particularly increases the surface diffusionlength of gallium adatoms. For excessive diborane flow-rates, a dramatic roughening of the epilayer surface is first observed and then, phaseseparation occurs.© 2008 Elsevier B.V. All rights reserved.

PACS: 81.05.Ea; 82.33.YaKeywords: BGaAs; Growth mode; Atomic Force Microscopy (AFM); Metalorganic vapour phase epitaxy (MOVPE)

1. Introduction

Boron arsenide (BAs) and its related alloys remain some ofthe least explored semiconducting III–V materials. Due to thevery small covalent radius of boron, BInGaAs can be grownlattice-matched to GaAs and has been demonstrated to be apromising material for solar cell applications as an alternative toInGaAsN [1]. This quaternary alloy is also a candidate for theactive layer of devices used for fibre-optical telecommunica-tions networks. Indeed, incorporating boron into InGaAs/GaAsquantum wells may allow extending their emission wavelengthtowards 1.3 µm by reducing the compressive strain. Since thepioneering work of Geisz et al. [2], there were few reports on theepitaxy of BGaAs and BInGaAs alloys on GaAs substrates.BGaAs alloy has been grown by molecular beam epitaxy [3,4]

⁎ Corresponding author. Tel.: +33 472 431 753; fax: +33 472 431 673.E-mail address: poteseso@yahoo.fr (P. Rodriguez).

0040-6090/$ - see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.tsf.2008.04.053

and metalorganic vapour phase epitaxy (MOVPE) [2,5–9]while the quaternary alloy, BInGaAs, has only been grown byMOVPE [2,5–7].

Various factors contribute to the complexity of B(In)GaAsepitaxy. First, a large miscibility gap exists between GaAs andBAs: theoretical calculations predicted a boron solubility limitin bulk GaAs, at thermodynamic equilibrium, ranging between4 and 7% at 600 °C [10,11]. To date, BxGa1− xAs epilayers witha boron concentration x up to about 8% have been grown [4].On the other hand, boron and arsenic can form a rhomboedralB12As2 phase as well as the zinc-blende BAs phase. DuringMOVPE growth, the V/III ratio appears as a key-parameter inorder to stabilize zinc-blende B(In)GaAs alloys. Indeed, high V/III ratios seem to be required to inhibit the formation of B–Bbonds in the solid [5]. Moreover, the mechanism of boronincorporation into B(In)GaAs by MOVPE appears complex andis still not well understood. Low growth temperatures promoteboron incorporation [5–7]. But, whatever the growth tempera-ture and the boron precursor nature (diborane, triethylboron...),

Fig. 1. Detailed XPS spectra of the (a) B1s, (b) Ga3d, and (c) As3d photoelectroncore levels for a BGaAs/GaAs epilayer grown at 580 °C with Xv=62%.

8425P. Rodriguez et al. / Thin Solid Films 516 (2008) 8424–8430

boron incorporation efficiency remains low [12]. It has alsobeen shown that good-quality BGaAs epilayers could only beobtained on a relatively narrow range of boron gas-phaseconcentrations. Indeed, increasing boron precursor flow-rateleads, beyond a critical point, to a pronounced surfaceroughening and to a structural breakdown [5–9]. For suchepilayers, with a rough surface morphology, we evidenced astrong surface segregation of boron [13].

The aim of this paper is to clarify how the boron gas-phaseconcentration influences the boron surface composition, thesurface morphology and the growth mode of BGaAs epilayersgrown on GaAs byMOVPE. In our growth conditions, we showthat boron surface segregation is a quasi-general phenomenon,not limited to epilayers with degraded surface morphologies.Moreover, we demonstrate that a close correlation existsbetween the boron accumulation on the epilayer surface andthe development of a step-bunching growth mode. This studyproposes an advance in the understanding of the growth modemechanisms of BGaAs epilayers.

2. Experimental details

The growth of BxGa1− xAs layers (x≤2.3%) has beenperformed by atmospheric-pressure MOVPE in a T-shapedhorizontal reactor. The layers were deposited on (100) GaAssubstrates misoriented 1° towards [001] direction. Diborane(B2H6), triethylgallium (TEG) and arsine (AsH3) were used asprecursors and hydrogen as carrier gas. Prior to BGaAs growth,a 70 nm thick GaAs buffer was deposited at 650 °C. BGaAsepilayers were then grown at 580 or 610 °C. AsH3 and TEGflow-rates were kept constant, respectively at 2.7×10−3 and4.2×10−6 mol min−1. Diborane flow-rate was varied from8.9×10−7 to 8.9×10−6 mol min−1, leading to V/III ratiosranging between 125 and 460. We used high V/III ratios in orderto favour BGaAs alloy stabilization. The boron gas-phaseconcentration was quantified by the initial molar flow-rate ratio:Xv=2[B2H6] / (2[B2H6]+ [TEG]). In the range of diborane flow-rates used in this study, Xv varied between 30 and 81%.

Surface morphologies were observed by ex-situ AtomicForce Microscopy (AFM) using a Scientec Molecular Imagingmicroscope in constant-force mode. Surface roughness root-mean square (RMS) was measured on 5×5 µm2 areas. Stepheights were determined using profile cross-sections. The boroncomposition (x) of the epilayers was deduced from X-raydiffraction (XRD) patterns, using the (400) reflection, assumingVegard's law and a coherent strain between the layer and thesubstrate. The Poisson ratio was estimated to 0.31 for ourepilayers, using for BAs the elastic coefficient values calculatedby Meradji et al. [14].

X-ray photoelectron spectroscopy (XPS) experiments wereperformed with a 5950A Hewlett-Packard photoelectronspectrometer equipped with a monochromatic Al Kα X-raysource (hν=1486.6 eV) and a concentric hemisphericalanalyser with a resolution of 0.5 eV. No cleaning procedurewas performed prior to analysing epilayers near-surface. As aconsequence, the analysed samples had always a thin super-ficial naturally-grown oxide layer. In addition to wide binding-

energy range survey scans (0–1000 eV), detailed XPS spectra(20 eV range) were recorded for the strongest photoelectronlines of each element, corresponding to the following corelevels: B 1s, Ga 3d and As 3d. Fig. 1 shows detailed spectraobtained for a BGaAs epilayer grown at 580 °C withXv=62%. The boron surface composition (xs) was estimatedfrom B1s and Ga3d peak areas, using atomic sensitivity factorsfor our instrument which take into account both the X-raycross-section and the transmission function of the spectro-meter. The boron quantification threshold, estimated for our

Fig. 2. Boron bulk composition (x) and boron surface composition (xs)evolutions as a function of boron gas-phase concentration (Xv) for BGaAs layersgrown at 580 °C. The data were respectively derived from XRD and XPSanalyses.

Fig. 3. AFM images of BGaAs/GaAs epilayers (200 nm thick) grown at 580 °Cwith various diborane flow-rates.

8426 P. Rodriguez et al. / Thin Solid Films 516 (2008) 8424–8430

instrument, corresponded to a boron surface composition ofabout 4%.

3. Results

For 200 nm thick BxGa1−xAs layers grown at 580 °C, wecompared the boron bulk composition (x) with the boron surfacecomposition (xs). Fig. 2 shows x and xs evolutions as a function ofthe boron gas-phase concentration (Xv). The values of x, xs and xs /x, obtained for the various layers, are recapitulated in Table 1.

BGaAs monocrystalline layers could only be obtained forXv≤68%. For the highest Xv value (Xv=81%), we could notobserve anymore an X-ray (400) peak corresponding to the layer,which indicates that a structural breakdown occurred. As aconsequence, the boron bulk composition could not be determinedfor this layer. This structural breakdown can be attributed to aphase separation in the layer due to the limited solubility of boronin GaAs, as reported by Geisz et al. [5]. They showed that phaseseparation in BGaAs alloys leads to the formation of apolycrystalline GaAs phase embedded in a boron-rich amorphousmatrix. xs was determined for all the layers, except for the onegrown with the lowest Xv value. For this layer, boron surfacecomposition was under the XPS detection threshold (e.g. xsb4%).

For epilayers grown with Xv≤62%, the boron bulk composi-tion, x, linearly increases with the boron gas-phase concentration.This trend shows that, in this regime, boron incorporation

Table 1Comparison of boron bulk composition (x) and boron surface composition (xs)for BGaAs epilayers grown at 580 °C

Xv (%) x (%) (DRX) xs (%) (XPS) xs /x

30 1 b4 –52 1.9 10.5 5.662 2.3 24.8 10.968 2.2 55.8 25.681 No X-ray peak 72.5 –

efficiency (R(B)=x/Xv) is independent of diborane flow-rate.Moreover, our results show that boron incorporation efficiency isvery low (R(B)=0.037) at this growth temperature. In this regime,boron surface composition xs also increases with Xv. But, twonoticeable trends are evidenced for Xv≥52%. First, the boroncomposition is much higher on the surface than in the bulk.Second, the boron excess on the surface (quantified by the xs /xratio) is significantly enhanced when Xv increases. Indeed, the xs /xratio is estimated at 5.6 for Xv=52% and reaches 11 for Xv=62%.

For 62%bXv≤68%, the evolution of boron incorporationdeviates from linearity: the boron bulk composition tends tosaturate at about 2.2%. This trend shows that the boron solubilitylimit is reached into the bulk. Nevertheless, phase separation doesnot occur yet for Xv=68% and a metastable epilayer is obtained.While x saturates, a steep increase of the boron surfacecomposition is observed: more than 50% of boron is detected onthe surface of this layer, corresponding to an xs /x ratio of 26. Thisenhanced surface accumulation of boron is accompanied by amorphological breakdown, characterized by a strong increase ofthe epilayer surface roughness, as shown below. It should benoticed that this morphological degradation only occurs beyond acritical BGaAs layer thickness. Indeed, this phenomenon wasevidenced for a 200 nm thick layer but was not observed for a verythin layer (10 nm thick) deposited using the same growthparameters.

For higher Xv values, phase separation occurs and the boronsurface composition further increases: xsN70% for Xv=81%.

Fig. 5. AFM images of (B)GaAs/GaAs epilayers (200 nm thick) grown at 610 °Cwith various diborane flow-rates.

8427P. Rodriguez et al. / Thin Solid Films 516 (2008) 8424–8430

Fig. 3 illustrates the surface morphologies, observed byAFM, for the layers grown at 580 °C. Unexpectedly, the growthmode of BGaAs epilayers is strongly influenced by Xv (e.g. thediborane flow-rate).

At low diborane flow-rate (Fig. 3a), a relatively smoothsurface is obtained, with a RMS roughness close to 5 Å. Thestep/terrace structure of the vicinal surface is indistinct andthe formation of two-dimensional (2D) islands can beobserved. This shows that growth mainly occurs via a 2D-nucleation mechanism rather than by the incorporation ofadatoms at step-edges.When increasing Xv, a growth mode transition is evidenced,characterized by the apparition of a step/terrace structuretypical of a step-bunching mechanism (Fig. 3b and c):terraces are limited by multi-atomic steps and the meanterrace width is much larger than the nominal value(L=16 nm) corresponding to the substrate misorientation.Moreover, when increasing the diborane flow-rate, the meanterrace width and the step height strongly increase, e.g. step-bunching is enhanced. Indeed, while the step height rangesbetween 2 and 7 monolayers (along the observed area) forXv=52%, it reaches 6–17 monolayers for Xv=62%, leadingto an increase of the RMS surface roughness. Thepronounced scattering of step height along the surfaceseems to be a characteristic of BGaAs step-bunching.Another noticeable trend is the fact that step-edges are notstraight and smooth but, on the contrary, jagged.Fig. 3d illustrates the morphological breakdown occurringfor Xv=68%. The metastable epilayer exhibits a muchrougher surface, suggesting a 3D-growth. When phaseseparation occurs (Xv=81%), the roughness becomesmacroscopic and the surface of the layer has a hazy aspectto the naked eye.

In order to investigate the influence of the epilayerthickness on step-bunching development, we performed thegrowth of a very thin BGaAs layer (about 10 nm thick) withan Xv of 62%. Fig. 4 shows the AFM morphology obtainedfor this layer. Comparatively to a thicker layer (see Fig. 3c),

Fig. 4. AFM image of a very thin BGaAs/GaAs epilayer (10 nm thick) grown at580 °C with Xv=62%.

the RMS roughness is significantly reduced. Slight parallelundulations can be noticed, which are not related to surfacesteps but originate from the GaAs buffer layer. Indeed, on thesurface of a buffer-like GaAs layer, undulations with similarorientation, periodicity and amplitude were observed, super-posed on a step/terrace substructure. They were parallel toGaAs step-edges, with a period far larger than the averageterrace width. The surface of a thin BGaAs layer, grown onthe buffer layer, replicates these undulations. The maindifference with a thicker layer is the absence of a distinctstep/terrace structure on the surface. On the other hand, thesurface is covered with 2D island-like features with lateral sizeof about 100 nm. Thus, growth is no longer proceeding bystep-bunching, but via 2D-nucleation. This result shows thatBGaAs step-bunching mechanism only takes place beyond acritical layer thickness. This is an unusual behaviour incomparison with the typical step-bunching observed duringthe growth of classical III–V compounds, like GaAs [15], onvicinal surfaces. In the standard step-bunching mechanism,growth proceeds by step-motion since its early stage; stepheight and terrace width monotonously increase with theepilayer thickness until saturation occurs.

When increasing growth temperature to 610 °C, boronincorporation is significantly reduced (x≤0.7%) as classi-cally observed for BGaAs epitaxy. The corresponding boron

8428 P. Rodriguez et al. / Thin Solid Films 516 (2008) 8424–8430

incorporation efficiency is divided by three (R(B)=0.013).Nevertheless, similarly to layers deposited at 580 °C, a strongboron enrichment of the surface was also evidenced for anepilayer grown with Xv=62%. Indeed, while the boron bulkcomposition was limited to 0.7%, 18% of boron was detected onthe surface of this epilayer.

Fig. 5 recapitulates surface morphologies obtained at 610 °Cas a function of the diborane flow-rate. For comparison, Fig. 5ashows the surface morphology of a GaAs reference epilayer,with similar thickness grown at 610 °C on the buffer layer.This surface exhibits a bunched step/terrace structure. Straightstep-edges are obtained, with step height limited to 1–3monolayers, and the mean terrace width is estimated at40 nm. For BGaAs layers grown with Xv=39% (Fig. 5b) andXv=62% (Fig. 5c), step-bunching is strongly enhanced. Trendsobserved at lower growth temperature are confirmed: stepheight increases with the diborane flow-rate, from 2–10monolayers for Xv=39% to 12–27 monolayers for Xv=62%.The mean terrace width also increases, reaching about 400 nmfor Xv=62%. Comparison with morphologies obtained at580 °C clearly points out that, for a fixed diborane flow-rate,BGaAs step-bunching is strongly favoured when increasingtemperature (compare Figs. 3c and 5c), leading to a significantroughening of the surface. Moreover, one can notice that step-edges appear even more jagged at 610 °C than at 580 °C.

4. Discussion

4.1. Boron incorporation and boron surface enrichment

The linear incorporation of boron as a function of boron gas-phase concentration (Xv), observed at 580 °C for Xv≤62%, issimilar to typical group-III element incorporation in III–III–Vternary alloys like InGaAs [16] or AlGaAs [17]. But, whereasgroup-III element incorporation efficiency is close to unity inthese classical alloys, boron incorporation efficiency in BGaAsepilayers is much lower. As boron vapour pressure is far lowerthan gallium's one in this range of growth temperatures [10],this trend cannot be explained by boron desorption kinetics. Onthe other hand, it has been reported that diborane can react witharsine, even at room temperature, and form very stable arsine–borine compounds like BAsHx (1≤x≤4) [18]. It is also well-known that diborane forms higher boranes (B3H9, B4H10...)when heated [19]. Such parasitic gas-phase reactions, whichlead to the formation of thermally stable compounds much lessvolatile than diborane, considerably reduce the concentrationof boron active species and may explain the low boronincorporation efficiency.

In this linear incorporation regime, we also evidenced apronounced boron surface enrichment (for Xv≥52%) and weobserved that the boron surface excess (e.g. xs /x) significantlyincreases with the diborane flow-rate. This suggests that asurface segregation of boron occurs, consistent with its lowvapour pressure. The precise mechanism for this surfacesegregation is not well clarified. But, the key-factor for boronsurface enrichment may be the adsorption of non-active boroncompounds (above-mentioned) on the surface during growth.

For these layers, boron segregation does not degrade the surfacemorphology and the RMS roughness remains low. It should berecalled that, to date, boron surface segregation had only beenreported for BGaAs layers characterized by a strong surfaceroughening [13].

When the boron solubility limit is reached in the bulk, phaseseparation can still be prevented if the boron excess supplied bythe gas-phase is not too high (Xv=68%). For such a metastablelayer, how can this boron excess be accommodated? Surfaceboron cannot be easily desorbed as discussed previously. Thus,boron excess tends to accumulate on the surface as attested bythe steep increase of the boron surface composition (xsN50%)observed for Xv=68%. The surface is able to accommodatesuch an amount of boron because boron solubility is probablystrongly enhanced near the epilayer surface, as alreadydiscussed for nitrogen incorporation in GaAs [20]. This higherboron surface solubility can be explained as follows. The maincause of the low boron solubility in bulk GaAs is the largedifference between boron and gallium covalent radii. Indeed,the pronounced local compressive strain induced by theintroduction of small boron atoms in the GaAs matrix impedesboron incorporation. Near the epilayer surface, this strain can bemore easily relieved since atoms have more freedom to moveand surface reconstruction may also provide more favourablesites for boron incorporation [20].

The strong surface roughening observed for the metastablelayer grown at Xv=68% is probably a consequence of its veryhigh boron surface content. The fact that this morphologicalbreakdown only occurs beyond a critical epilayer thicknessseems consistent with this analysis. Indeed, boron tends toaccumulate progressively on the surface as growth proceeds.So, boron surface composition is expected to increase with theepilayer thickness. For excessive boron surface coverage, webelieve that the growth mode becomes tri-dimensional anduncontrolled because boron species (adatoms and admolecules)block too many group-III sites, disturbing the incorporation ofgallium adatoms.

At higher diborane flow-rates, the boron excess becomes tooimportant to be accommodated on the layer surface. The alloystability cannot be maintained and phase separation occurs.

4.2. BGaAs growth mode

Our results demonstrate that step-bunching is favoured whenincreasing the diborane flow-rate and the epilayer thickness.Moreover, we have shown that an increase of the diborane flow-rate leaded to an enhanced boron surface segregation. Similarly,boron surface composition is expected to increase with the layerthickness as previously discussed. Step-bunched layers arethus characterized by a relatively high boron surface content (upto about 25%). This suggests that step-bunching developmentand boron surface enrichment are tightly related phenomena.According to the usualmodels [21,22], step-bunching of epilayersgrown by MOVPE is mainly governed by group-III adatomsurface diffusion and incorporation kinetics. As the boron bulkcomposition is low for our BGaAs epilayers (xb2.5%), onlygallium adatom surface kinetics may significantly affect this

8429P. Rodriguez et al. / Thin Solid Films 516 (2008) 8424–8430

growth mode. Thus, our results suggest that boron surfaceenrichment strongly influences these kinetics.

At 580 °C, a growth mode transition from 2D-nucleation tostep-bunching was evidenced when increasing the diboraneflow-rate. 2D-nucleation occurs on vicinal surfaces when thesurface diffusion length of group-III adatoms is smaller than thenominal terrace width. On the contrary, step-bunching mechan-ism is associated to diffusion lengths significantly higher thanthe nominal terrace width, so that wider terraces may be formed.The observed growth mode transition thereby points out anincrease of gallium adatom surface mobility with the diboraneflow-rate, e.g. with the boron surface composition. This trendcan also contribute to the increase of step heights and terracewidths with the diborane flow-rate evidenced at 580 and 610 °C.

We believe that the influence of boron surface species(adatoms or admolecules) on Ga surface mobility and, moregenerally, on BGaAs growth mode can be interpreted as asurfactant effect. Surfactants used for the epitaxy of semicon-ducting materials are typically elements characterised by a lowvapour pressure, at the growth temperature of interest, and a lowsolubility within the solid. As a consequence, they tend tosegregate to and accumulate on the growth front, giving rise tochanges in kinetically and/or thermodynamically based surfaceprocesses. Particularly, they may affect surface reconstruction,step structure, adatom attachment at step-edges and surfacediffusion coefficients. Even though, unlike usual surfactants,boron incorporates significantly in our BGaAs layers, itstendency to accumulate onto the surface is similar as a surfactantone. Moreover, cases where the surfactant species is one of theatoms of the growing film, such as indium in the growth of InAson GaAs, have already been reported [23]. It should also benoticed that boron has already been used as a surfactant for thegrowth of Ge quantum dots on Si [24,25] and for the growth ofthin Si films on CaF2/Si substrates [26]. Even though surfactantsare commonly used to modify the growth mode of epitaxiallayers, corresponding mechanisms are complex and stillcontroversial. A typical example is the heteroepitaxial growthof strained layers, like Ge or SiGe on Si substrates. In this case,suitable surfactants may suppress 3D islanding, allowing thestabilization of a layer-by-layer 2D growth mode [27,28]. It wasalso shown that the introduction of antimony during the growthof GaN-rich GaNAs epilayers could induce a transition from arough surface covered with hillocks to a step-bunched surfacewith atomically flat terraces [29].

The influence of surfactants on adatom surface diffusiondepends on the nature of both the surfactant and the epitaxialmaterial. For Ge/Si growth, the suppression of 3D islanding withthe addition of arsenic surfactant was related to a large decreasein adatom mobility [28]. Similarly, for the epitaxy of InAs/GaAsquantum dots, it has been shown that the addition of antimonyincreased the dot density and suppressed coalescence, suggestinga reduction of indium adatom mobility [30]. On the contrary forGaAs homoepitaxy on (001) patterned substrates, Wixom et al.showed that Sb or Bi surfactants strongly increased the [110]lateral growth rate [31]. This effect was attributed to an enhancedGa surface diffusion. For GaN epitaxy on sapphire substrates, thesame conclusion was drawn concerning the effect of antimony

addition on Ga surface diffusion [32]. Surface boron seemsto have a similar influence during BGaAs epitaxy on GaAs.Nevertheless, the mechanism by which boron surface speciesenhance Ga adatom surface mobility is still unknown. A possiblecause could be a change of surface reconstruction on terraces.

Apart from their influence on Ga adatom mobility, boronsurface species may enhance BGaAs step-bunching by otherways. According to the classical model developed bySchwoebel, step-bunching occurs when surface adatoms,migrating on terraces, are preferentially incorporated to down-ward step sites rather than to upward step sites [21]. Thisimplies that the energy barriers for adatom incorporation to astep-edge from the two adjacent terraces are asymmetrical. Theaccumulation of boron on the surface may affect the stepstructure and modify these energy barriers. To promote step-bunching, the energy barrier for downward incorporationshould be lowered and/or the energy barrier for upwardincorporation should be increased. On the other hand, boronspecies adsorbed at step-edges may impede the step-motion byblocking some of the group-III step sites, e.g. by reducing Gaincorporation probability. By analogy with the kinematicalmodel of impurity-induced step-bunching proposed by Frank[22], this could also contribute to the development of BGaAsstep-bunching.

We have noticed that step-bunched BGaAs surfaces werecharacterized by rough and jagged step-edges. The reason forsuch step morphology is still uncertain. But, some argumentsmay be discussed. XPS analyses do not provide information onthe spatial distribution of boron surface species. If we assumethat boron species are not uniformly distributed along terracesand step-edges, this could result in lateral fluctuations of thediffusion rate of Ga adatoms and, possibly, of their probabilityto incorporate to a given step-edge. Both effects couldcontribute to lateral variations of the step velocity, and thusgenerate the observed step-edge roughening.

Finally, the enhancement of step-bunching when increasingBGaAs growth temperature is a usual trend, well-known forhomoepitaxial growth of GaAs [33] and InP [34] by MOVPE.The thermal activation of adatom surface diffusion is a key-parameter to explain such an evolution. But, for BGaAsepitaxy, growth temperature has probably a much morecomplex influence. Particularly, it may affect boron surfac-tant behaviour by modifying surface reconstruction and Gaadatoms incorporation kinetics.

5. Conclusion

We have studied the influence of the diborane flow-rate onthe growth mode of BGaAs epilayers grown at 580 or 610 °C.We showed that increasing the diborane flow-rate stronglyfavours step-bunching. On the basis of X-ray photoelectronspectroscopy analyses, we demonstrated that the developmentof BGaAs step-bunching could be correlated to a pronouncedboron enrichment of the surface due to a boron surfacesegregation. These results suggest that boron behaves as asurfactant during the MOVPE growth of BGaAs and couldparticularly enhance the surface mobility of gallium adatoms.

8430 P. Rodriguez et al. / Thin Solid Films 516 (2008) 8424–8430

Beyond a critical diborane flow-rate, the boron solubility limitis reached in the bulk layer but a metastable alloy can still begrown if the boron excess supplied by the gas-phase is not toohigh. For such a metastable epilayer, the surface is able toaccommodate the boron excess but the strong accumulation ofboron on the surface leads to a morphological breakdown. Forhigher diborane flow-rates, phase separation occurs.

References

[1] G. Leibiger, C. Krahmer, J. Bauer, H. Herrnberger, V. Gottschalch, J.Cryst. Growth 272 (2004) 732.

[2] J.F. Geisz, D.J. Friedman, J.M. Olson, S.R. Kurtz, R.C. Reedy, A.B.Swartzlander, B.M. Keyes, A.G. Norman, Appl. Phys. Lett. 76 (2000) 1443.

[3] V.K. Gupta, M.W. Koch, N.J. Watkins, Y. Gao, G.W. Wicks, J. Electron.Mater. 29 (2000) 1387.

[4] M.E. Groenert, R. Averbeck, W. Hösler, M. Schuster, H. Riechert, J. Cryst.Growth 264 (2004) 123.

[5] J.F. Geisz, D.J. Friedman, S. Kurtz, J.M. Olson, A.B. Swartzlander, R.C.Reedy, A.G. Norman, J. Cryst. Growth 225 (2001) 372.

[6] V. Gottschalch, G. Leibiger, G. Benndorf, J. Cryst. Growth 248 (2003) 468.[7] P. Rodriguez, L. Auvray, H. Dumont, J. Dazord, Y. Monteil, J. Cryst.

Growth 298 (2007) 81.[8] H. Dumont, J. Dazord, Y. Monteil, F. Alexandre, L. Goldstein, J. Cryst.

Growth 248 (2003) 463.[9] D.A. Pryakhin, V.M. Danil'tsev, Y.N. Drozdov, M.N. Drozdov, D.M.

Gaponova, A.V. Murel, V.I. Shashkin, S. Rushworth, Semiconductors 39(2005) 11.

[10] H. Dumont, Y. Monteil, J. Cryst. Growth 290 (2006) 419.[11] R. Asomoza, V.A. Elyukhin, R. Pena-Sierra, Appl. Phys. Lett. 78 (2001) 2494.[12] J.F. Geisz, D.J. Friedman, S. Kurtz, R.C. Reedy, G. Barber, J. Electron.

Mater. 30 (2001) 1387.

[13] H. Dumont, D. Rutzinger, C. Vincent, J. Dazord, Y. Monteil, F. Alexandre,J.L. Gentner, Appl. Phys. Lett. 82 (2003) 1830.

[14] H. Meradji, S. Drablia, S. Ghemid, H. Belkhir, B. Bouhafs, A. Tadjer,Phys. Status Solidi, B 241 (2004) 2881.

[15] M. Kasu, T. Fukui, Jpn. J. Appl. Phys. 31 (1992) L864.[16] M.J. Ludowise, C.B. Cooper, R.R. Saxena, J. Electron. Mater. 10 (1981) 1051.[17] Y. Mori, N. Watanabe, J. Appl. Phys. 52 (1981) 2972.[18] F.G.A. Stone, A.B. Burg, J. Am. Chem. Soc. 76 (1954) 386.[19] H. Fernandez, J. Grotewold, C.M. Previtali, J. Chem. Soc. Dalton Trans.

20 (1973) 2090.[20] S.B. Zhang, A. Zunger, Appl. Phys. Lett. 71 (1997) 677.[21] R.L. Schwoebel, J. Appl. Phys. 40 (1969) 614.[22] F.C. Frank, in: R.H. Doremus, B.W. Roberts, D. Turnbill (Eds.), Growth

and Perfection of Crystals, Wiley, New-York, 1958, p. 411.[23] J. Behrend, M.Wassermeier, K.H. Ploog, J. Cryst. Growth 167 (1996) 440.[24] P.S. Chen, Z. Pei, Y.H. Peng, S.W. Lee, M.J. Tsai, Mater. Sci. Eng., B 108

(2004) 213.[25] H. Takamiya, M. Miura, N. Usami, T. Hattori, Y. Shiraki, Thin Solid Films

369 (2000) 84.[26] C.R. Wang, B.H. Muller, E. Bugiel, T. Wietler, M. Bierkandt, K.R.

Hofmann, P. Zaumseil, J. Vac. Sci. Technol., A 22 (2004) 2246.[27] T. Schmidt, J. Falta, G. Materlik, J. Zeysing, G. Falkenberg, R.L. Johnson,

Appl. Phys. Lett. 74 (1999) 1391.[28] M. Copel, M.C. Reuter, E. Kaxiras, R.M. Tromp, Phys. Rev. Lett. 63

(1989) 632.[29] A. Kimura, Z. Liu, T.F. Kuech, J. Cryst. Growth 272 (2004) 432.[30] D. Guimard, M. Nishioka, S. Tsukamoto, Y. Arakawa, Appl. Phys. Lett. 89

(2006) 183124.[31] R.R. Wixom, L.W. Rieth, G.B. Stringfellow, J. Cryst. Growth 265 (2004) 367.[32] L. Zhang, H.F. Tang, J. Schieke, M. Mavrikakis, T.F. Kuech, J. Appl. Phys.

92 (2002) 2304.[33] M. Shinohara, N. Inoue, Appl. Phys. Lett. 66 (1995) 1936.[34] V. Thevenot, V. Souliere, H. Dumont, Y. Monteil, J. Bouix, P. Regreny,

Tran-Minh Duc, J. Cryst. Growth 170 (1997) 251.