- Email: [email protected]
Interface analysis of Ge ultra thin layers intercalated between GaAs substrates and oxide stacks
Thin Solid Films 518 (2010) S123–S127
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Interface analysis of Ge ultra thin layers intercalated between GaAs substrates and oxide stacks Alessandro Molle a,⁎, Luca Lamagna a, Sabina Spiga a, Marco Fanciulli a,b, Guy Brammertz c, Marc Meuris c a b c
Laboratorio Nazionale MDM, CNR-INFM, Via C. Olivetti 2, 20041 Agrate Brianza (MI), Italy Dipartimento di Scienza dei Materiali, Università di Milano Bicocca, Milano, Italy IMEC, 75 Kapeldreef, B-3001 Leuven, Belgium
a r t i c l e
i n f o
Available online 19 October 2009 Keywords: GaAs III-V High-k High-mobility Atomic H Surface preparation XPS Interface study Logics
a b s t r a c t Capping III–V compound surfaces with Ge ultra-thin layer might be a viable pathway to passivate the electrically active interface traps which usually jeopardize the integration of III–V materials in metal-oxidesemiconductor devices. As the physical nature of such traps is intrinsically related to the chemical details of the interface composition, the structural and compositional features of the Ge/GaAs interface were thoroughly investigated in two different configurations, the atomic layer deposition of La-doped ZrO2 films on Ge-capped GaAs and the ultra-high vacuum based molecular beam deposition of GeO2/Ge double stack on in situ prepared GaAs. In the former case, the intercalation of a Ge interface layer is shown to suppress the concentration of interface Ga–O, As–O and elemental As bonding which were significantly detected in case of the direct oxide deposition on GaAs. In the latter case, the incidence of two different in situ surface preparations, the Ar sputtering and the atomic H cleaning, on the interface composition is elucidated and the beneficial role played by the atomic H exposure in reducing the semiconductor–oxygen bonds at the interface level is demonstrated. © 2009 Elsevier B.V. All rights reserved.
1. Introduction As far as superior transport properties of channel semiconductors are required to face the aggressive shrinking of the gate length in advanced metal-oxide-semiconductor (MOS) devices, III–V compounds are regarded as a viable option to replace the Si-based technology. Indeed, the in-channel effective electron mass of III–V semiconductors, lower than that of Si, leads to a higher electron velocity [1]. Among the III–V compounds, GaAs has so far retained the greater consideration. Nonetheless, it has recently been argued that ternary InxGa1−xAs compound might provide a more forgiving material in terms of density of interface defects [2] and higher mobility in MOS device for high In concentration [3]. An open concern on the integration of GaAs channel in MOS devices relies on the still non-established passivation of semiconductor/gate insulator interface, i.e. how to saturate the electrically active interface defects which cause pinning of the semiconductor Fermi level irrespectively to the externally applied voltage [4]. The physical nature of the defects nearby the GaAs surface region or at the GaAs/oxide interface is currently matter of investigation. A first general distinction can be made by sorting defects with energy level in the midgap region from those with energy level placed in the vicinity of the semiconductor
⁎ Corresponding author. E-mail address: [email protected] (A. Molle). 0040-6090/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2009.10.069
band-edges. The former was originally rationalized by Spacer et al. [5] in the advanced unified model for defects of the GaAs class of semiconductors as bulk bonding disruption and re-organization, e.g. formation of anti-bonding defects (AsGa antisites). However this picture was revisited by Kummel et al. [6], who conversely attributed the midgap defects at the oxide/GaAs interface to oxygen chemisorption to Ga atoms based on local scanning tunneling microscopy/ spectroscopy. On the other hand there is a general consensus in associating band-edge defects with interface bonding states (e.g. semiconductor–oxygen bonding and homopolar As or Ga bonding) induced during the oxide deposition processing [7,8]. The midgap defects are somehow hard to remove because they are related to GaAs bonding re-organization extrinsically induced by subsequent oxide deposition. Only the molecular beam epitaxy of Gd–Ga–O was proved to be successful in passivating As-rich GaAs surfaces by positioning interfacial Ga2O molecules between adjacent As dimers [9]. On the other hand, several solutions have been proposed to improve the interface bonding configuration with ad hoc surface treatments or interface engineering processes. A promising approach to III–V passivation is combining a chemical or in vacuo surface treatments for the native oxide removal along with the deliberate deposition of ultra thin interface passivation layer (IPL) of group V semiconductors [10,11], or nitrides [12]. In particular, the interest for Ge-based IPL has recently increased [13,14] in concomitance with the promising advances in the Ge passivation through various oxidation approaches [15–17]. Indeed, the good crystallographic matching of Ge with III–V
S124
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
compound surface lattice may alleviate bonding disruption in the GaAs surface arrangement and the intercalation of a Ge IPL in between the III–V compound substrate and the gate oxide stack may prevent the formation of interfacial semiconductor–oxygen bonds. In the present work, the structural and compositional details of ultra-thin Ge IPL on GaAs substrates were investigated in two different process configurations, the atomic layer deposition (ALD) of ultra thin Ladoped ZrO2 (L–ZO) films on Ge-capped GaAs (Section 3) and the ultra high vacuum (UHV) based molecular beam deposition (MBD) of GeO2/Ge passivation double stacks on GaAs (Section 4). 2. Experimental details Commercial Si-doped (1 × 1017 cm− 3) GaAs(001) substrates were used either for ALD and MBD-grown samples. In case of the ALD samples the substrates were chemically pre-treated via dipping in NH4 (OH) solution (4%) for 3 min. The deposition of a 2 nm (nominally) thick Ge layer was performed in a high-vacuum chamber by an e-beam evaporator at room temperature (RT). La-doped ZrO2 (L–ZO) films were subsequently grown in an ALD reactor at 300 °C directly onto as-prepared GaAs and after Ge IPL insertion. L–ZO was deposited using O3 as oxidizing precursor in combination with La [(iPrCp)3La] and Zr [(MeCp)2ZrMe(OMe)] metal precursors as described in Ref. [18]. The use of O3 as oxidizing agent addresses the further passivation of the Ge overlayer surface according to the low density of interface states previously achieved after O3 exposure to Ge [16] or O3-based ALD growth of HfO2 [17]. On the other hand, La incorporation in ZrO2 can effectively improve the dielectric features of the oxide stack by changing the polarizability of the oxide molecules and by stabilizing a crystallographic phase endowed with a higher dielectric constant [19]. Interface diagnostic was performed by x-ray photoelectron spectroscopy (XPS) of 3 nm-thick L–ZO films and by spectroscopic ellipsometry (SE) on 10 nm-thick L–ZO films. In parallel, the GaAs(001) substrates were inserted in a multichamber UHV system (base pressure 1 × 10− 11 mbar) to perform MBD growth and in situ characterization of GeO2/Ge double stack layer as a function of the surface treatment. After a preliminary outgas at 450 °C, the substrates were prepared by means of two different in vacuo treatments, 30 min long Ar ion (1 keV) sputtering at 600 °C and 30 min long exposure to an atomic H beam at 400 °C. Ar ions were generated by a standard ion gun working in a differential pumping assembly. The atomic H beam was provided by a radio frequency plasma source at a power of 350 W starting from a forming gas (4% H2, 96 % Ar) supply as
high as 1 × 10− 4 mbar. A 3 nm-thick Ge film was then deposited by an effusion cell with a rate of 1 Å/min (base pressure of 9 × 10− 10 mbar during growth, growth rate calibrated by an in situ quartz balance thickness monitor). Ge growth was monitored by reflection high energy (30 keV) electron diffraction (RHEED). A 1.7 nm-thick GeO2 layer was subsequently formed by 7 min long atomic O exposure at 300 °C to the 3 nm-thick Ge film to subsequently address the Ge surface passivation. Details of the GeO2 formation are reported elsewhere [20,21]. The multistacked GeO2/Ge/GaAs heterostructure can be thus used as interface engineered substrate for the subsequent deposition of high permittivity oxides aiming at the fabrication of GaAs-based MOS [14]. Each step of the surface preparation and double stack deposition on the latter samples was monitored in situ by RHEED and XPS. XPS was provided by a standard Mg Kα source (1253.6 eV) with a pass energy of 20 V as a function of the take-off angle (angle-resolved analysis). Deconvolution of the relevant XPS 3d core-level photoemission lines was performed by using Shirley baseline removal and Lorentzian-Gaussian doublets as fitting components. Doublets accounts for the spin-orbit splitting in d core-level photoemission lines. According to Ref [21,22] the following doublet separations (DS) were used, DSGa3d = 0.44 eV, DSAs3d = 0.67 eV, DSGe3d = 0.60 eV. SE evaluates changes in the polarization state of a reflected light beam in terms of the ellipsometric angles Ψ and Δ thus enabling one to determine the thickness of the GaAs heterostructures through data interpolation based on a multiple layers model by means of the WVASE32 software. SE data are collected by means of a rotating compensator spectroscopic ellipsometer (J. A. Woollam M2000F) in the spectral range of 1.0–5.0 eV and by using a light beam with an incidence angle of 75° [23]. 3. Atomic layer deposition of La-doped ZrO2 A general overview of the XPS compositional analysis accounting for the Ga 3d, In 4d, In 3d5/2, As 3d core-level photoemission lines is reported in Fig. 1(a) and (b) for 3 nm thick L–ZO films on GaAs without and with Ge IPL, respectively. The binding energy (BE) of the spectra was referred to the pre-calibrated position of adventitious carbon in the C 1s line (BE = 284.8 eV). Direct ALD of L–ZO onto GaAs is studied in Fig. 1(a). Here, the Ga 3d line exhibits an interplay of three components, from GaAs bulk (GaB at BE = 18.7 eV), from Ga2O3 interfacial bonding (Ga3+ at BE = 19.9 eV) and from the O 2 s line coming due to the overlying oxide (BE = 22.2 eV). The As 3d line deconvolution involves three
Fig. 1. Ga 3d and As 3d XPS lines recorded for L–ZO/GaAs (a) and L–ZO/Ge/GaAs (b). LZO thickness is 3 nm, Ge IPL thickness is 2 nm. GaB and AsB denote the bulk contributions, the displacement between the bulk peaks in case (a) and (b) is marked. Ga3+ and As3+ components reflect the presence of interfacial Ga2O3 and As2O3, respectively. As0 indicates interfacial elemental As bonding.
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
different contributions which can be assigned to As from GaAs bulk (BE = 40.7 eV), to As2O3 (and no As2O5) species (BE = 43.8 eV) and to elemental As bonding (BE = 41.6 eV). Opposed to the significant interface concentration of semiconductor–oxygen and As homopolar bonding observed after direct ALD processing, the scenario radically changes when Ge IPL is sandwiched between the oxide and GaAs. The related XPS analysis is illustrated in Fig. 1(b). As a result, the formation of As–O bonding can be here completely ruled out from the interface composition as no associated component is observed in the respective As 3d lines, whereas a remarkably lower Ga–O related component can be deduced from the Ga 3d line deconvolution. It should be noted that the position of the Ga 3d and As 3d bulk peaks are mutually displaced by an amount of 0.6 eV which reflects the different alignment of the electronic bands in the two different interface configurations. In particular, the LZO/Ge/GaAs exhibits a XPS line shift of 0.3 eV to higher BE with respect to the line positioning in the clean GaAs surface (BEGa3d = 19.1 eV, BEAs3d = 40.9 eV) as shown in Fig. 3(c) which can be explained by a Ge-induced band bending at the interface band line-up as previously observed [24]. The electrical quality of the GaAs-based heterostructure should be also dictated by the chemical details of the interface that Ge IPLs form with the overlying L–ZO layer. The information from SE and XPS data were mutually combined in order to elucidate this aspect as reported in Fig. 2. From the SE measurements of the NH4(OH) treated GaAs (001), taken immediately before the oxide deposition, the pseudodielectric function (〈ε〉 = 〈ε1〉 + i〈ε2〉) of the substrate [25] can be extracted by means of a Point-to-Point interpolation [26] of the spectra. This enabled us to model the semiconductor substrate as a nearly oxide free. Therefore the calculated 〈ε〉 results in a qualitatively good agreement with the tabulated values expected for GaAs(100) [27] and is then used in all the models to represent the substrate. L–ZO dielectric function was estimated using a Cauchy dispersion relation [28] (oxide is transparent in the 1.0–5.0 eV range) on thicker samples and was then applied to model the thin layers constituting our samples; Ge dielectric function was assumed from tabulated values. The only free parameters used in the two SE fits were oxide thickness (tox) and Ge layer thickness (tge). In Fig. 2(a) a close-up of the experimental Δ spectra acquired is depicted after ALD deposition on GaAs(001) and on Ge/GaAs(001). The pronounced difference between the two curves can be uniquely attributed to the presence of the ultra-thin Ge IPL underneath the oxide. Such a difference can be rationalized in terms of the multi-layers physical model fit to the SE data. In detail, a two layer model (see scheme A in the inset) was used
S125
to represent L–ZO/GaAs therein resulting in an oxide thickness of tox of 8.2 ± 0.2 nm. On the contrary, a three layer model based fitting was necessary to account for the Ge-passivated sample therein indicating the permanence of purely Ge IPL after oxide deposition. Using a three layers model (see the inset) results in tox = 7.4 ± 0.2 nm and tge ~ 0.6 nm. This picture can be corroborated by the complementary XPS analysis in Fig. 2(b) where the Ge 3d line is studied as a function of the sampling depth, i.e. at a low take-off angle measurement (θ = 37°) and under normal photoemission (θ = 90°) to probe near-to-surface and deeper regions of the sample, respectively. Both Ge 3d lines exhibit three components, one at lower BE due to elemental Ge –Ge bonding of the Ge IPL and two (A and B in Fig. 2(b)) at higher BE (BEA = 30.7 eV and BEB = 31.9 eV) being consistent with oxidized Ge bonding. The evidence of an elemental Ge bonding signal indicates the presence of a pure Ge/oxidized Ge interfacial structure. The observed Ge–O bonding signal can be located at the Ge/L–ZO interface as the related XPS signal increases for decreasing take-off angles, i.e. with a lower sampling depth. The physical nature of the Ge–O component is consistent with the formation of a Ge–O bonding region throughout the L–ZO top stack. A more careful deconvolution of the Ge 3d line suggests the interplay of two Ge–O components (A and B in Fig. 2(b)) with different valence states. The A and B components can be assigned to minor GeOx concentration and [20,29] to germanate like region [30], respectively. As the relative A/B intensity ratio slightly increases with increasing θ (from 0.14 at θ = 37° to 0.17 at θ = 90°), the GeOx species presumably places atop the pure Ge layer whereas the germanate is located in between the GeOx and the L–ZO. Such a picture is also supported by the observed position of the relevant Zr 3d line at BE = 182.6 eV(not shown) being consistent with the Ge substitutional bonding inside ZrO2 as described in Ref. [31]. Further investigations of this issue are in progress. 4. Molecular beam deposition of GeO2/Ge double stacks Following a similar approach to that of Section 3, 3 nm-thick Ge layer were deposited in situ on differently treated GaAs(001) surfaces. Aiming at the Ge surface passivation [29], the Ge films have been subsequently capped with ultra-thin GeO2 over-layers by 7 min in situ exposure to atomic O at 300 °C so to have 1.7 nm GeO2/1.5 nm Ge/ GaAs multistacked structures. The bonding configuration at the GeO2Ge/GaAs interface was assessed in situ for two different surface in vacuo preparations, 30 min long Ar+ (700 eV) sputtering at 600 °C
Fig. 2. (a) SE scans registered for L–ZO/GaAs (line) and L–ZO/Ge/GaAs (circles). Inset: sketches of the two multilayer models used to interpolate the SE data in the two different interface configurations, L–ZO/GaAs and L–ZO/Ge/GaAs. (b) Ge 3d XPS lines taken at two take-off angles, θ = 37° (up) and 90° (down). See the text for details of the spectra deconvolution.
S126
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
Fig. 3. RHEED pattern of an Ar-sputtered GaAs(001) surface along the [1–10] (a1), [110] (a2) and of a 1.5 nm-thick Ge film grown on top at room temperature (a3). Note the (2 × 1) reconstruction occurring at the Ge surface. RHEED pattern of an atomic H-cleaned GaAs(001) surface along the [1–10] (b1), [110] (b2) and of a 1.5 nm-thick Ge film grown on top at room temperature (b3). Note the amorphous structure of the Ge film in this case. (c) Ga 3d and As 3d XPS lines recorded for the as-received GaAs (rich of native oxides), for the atomic H cleaned GaAs, and for the Ar-sputtered GaAs.
and 30 min long atomic H irradiation at 400 °C. Previous reports on the GaAs preparation suggest that both treatments are effective in giving highly ordered and defect-free surfaces [32,33]. However, while the ion sputtering might damage the inner constitutions of the III–V compounds locally, an increasing attention is currently paid to exploit atomic H radicals as non-destructive method to remove native oxides and leave a well-reconstructed III–V surface [34]. A Ga-rich (4 × 6) reconstruction can be recognized in case of the Ar-sputtered GaAs surface from the relevant streak periodicity in the RHEED patterns along the [110] (4× periodicity) and the [1–10] (×6 periodicity) directions shown in Fig. 3(a1)–(a2) as reported in Ref. [32]. Otherwise, based on similar RHEED observation of Fig. 3 (b1)–(b2), the atomic H clean is observed to release an As-rich (2 × 4) reconstructed GaAs surface in agreement with Ref. [33]. The structure of the Ge layer appears to be deeply influenced by the surface preparation. Indeed, the RHEED patterns of the 1.5 nm-thick Ge films grown at room temperature (RT) on the Ar-sputtered surface in Fig. 3 (a3) looses the reconstruction fashion of the underlying substrates, but maintains the primary diffraction streaks thus indicating the epitaxial character of Ge growth and the formation of (2 × 1) reconstructed Ge surface. Conversely, despite the good lattice matching between Ge and GaAs, after atomic H clean the Ge layer exhibits an amorphous arrangement as follows from the absence of diffraction signal in the RHEED image in Fig. 3(b3). The growth of an amorphous Ge film was also documented in case of Ge deposition on atomic H-cleaned In0.15Ga0.85As surfaces [21]. The emergence of an amorphous Ge film might be related to the hypothetical H termination of the GaAs surface. Complementary XPS diagnostic of the differently prepared substrates in Fig. 3(c) rules out residual presence of native oxides and adventitious contaminations in both surface preparations compared to the as-received substrate. The 1.7 nm GeO2/1.3 nm Ge/GaAs heterostructures were fabricated as described in Section 2 by growing Ge films at room temperature. Since the thickness of the GeO2/Ge IPL is still below the mean free paths for bulk photoelectrons, the bonding configuration at the Ge/GaAs interface can be investigated by probing the relevant core level photoemission lines, Ga 3d, Ge 3d, As 3d recorded at a take-off angle of 90°, i.e. normally to the sample surface, as illustrated in Fig. 4(a) and (b) for the atomic H clean sample and the Ar-sputtered sample, respectively. The Ga 3d and As 3d line present a multi-shaped profile arising from the interplay of the contributions from the Ga-As bulk bonds (marked as GaB and AsB) with valence state components at higher BE, Ga3+ (chemical shift Δ = 1.7 eV from GaB) for Ga 3d, As5+ (Δ = 5.2 eV from AsB) for As 3d. The extra valence state components Ga3+ and As5+ denote the presence of interfacial Ga2O3 and As2O5, respectively [7,9,12,14,21]. An additional component Ga* can be observed in shape profile of the Ga
3d line for both surface preparations at lower BEs which can be associated with Ga bonding with more metallic character, e.g. possibly Ga-Ge bonding. From the As 3d line deconvolution no As2O3 species can be detected. Ge oxidation is demonstrated in both cases by the largely dominant GeO2 contribution to the Ge 3d line (see the insets of Fig. 3) as previously reported [16,20,29]. To quantitatively assess the relative amount of interfacial Ga–O and As–O bonds, from the best-fit parameters it is useful to calculate the ratios ζ between the areas of the valence state component(s) (Ga3+ for Ga 3d, As3+ and As5+ for As 3d) and the bulk component, ζGa–O = Ga3+/ GaB and ζAs–O = (As3+ + As5+)/AsB. From the spectra comparison we conclude that the relative amount of oxide bonds at the Ge/GaAs interface strikingly changes with surface preparation. Indeed, the ζGa–O and ζAs–O values are significantly reduced after Ge oxidation from 0.08 and 0.66 in the Ar-sputtered sample down to 0.03 and 0.34 in the atomic H cleaned one. The fraction of interfacial Ga–O bonds is negligible in the latter configuration. In both cases, the nature of the As–O bonding is exclusively identified by the As5+ valence state corresponding to the As2O5 species only. The different chemical quality of the Ge/GaAs interface achieved with the two surface preparations might be also here tentatively attributed to the occurrence of a H-terminated GaAs surface or to the residual presence of contaminants from the atomic gas generation after the atomic H treatment. These two aspects might have
Fig. 4. Ga 3d and As 3d XPS lines recorded for GeO2/Ge/GaAs heterostructure after 30 min atomic H cleaning at 400 °C (a) and after 30 min Ar sputtering at 600 °C (b). Insets: Ge 3d XPS lines for the two cases in (a) and (b).
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
deep implications in the different structure of the Ge layer observed as a function of the surface treatment. However, further theoretical and experimental investigations are needed to elucidate the interaction between atomic H and III–V compound surfaces. 5. Conclusions Given the potential application of Ge as passivating material for III–V compound surfaces, Ge ultra-thin films were grown on differently prepared GaAs(001) substrates and the relevant interface bonding was elucidated by XPS. Intercalation of Ge IPL between NH4 (OH) treated GaAs and La-doped ZrO2 films grown by O3-based ALD confers a remarkably improved chemical quality to the interface with GaAs compared to the case of the direct oxide deposition. The Ge IPL leads to an efficient removal of potentially defective Ga2O3, As2O3 and As elemental bonding from the interface. The use of O3 as oxidizing agent is concomitantly intended to gain superior dielectric properties of the oxide [18] and to pursue the electrical passivation of the Ge IPL surface [17] thereby providing an ALD suited approach to fabrication of Ge IPL/GaAs-based MOS capacitors. On the other hand, the chemical details of the Ge/GaAs interface were also studied through the molecular beam deposition of GeO2/Ge double stacks on GaAs surfaces prepared either by Ar sputtering up to 600 °C and by atomic H cleaning at 400 °C in an UHV environment. The GaAs surface arrangement and the structure of the subsequently grown Ge films appear to be strictly dependent on the surface preparation, i.e. resulting in a Ga-rich (4× 6) reconstruction and an epitaxial Ge layer in the former case, and in an As-rich (2× 4) reconstruction and an amorphous Ge layer in the latter one. Given the good crystallographic matching between Ge and GaAs, the amorphous Ge layer grown on atomic H cleaned GaAs can be tentatively associated with the presence of surface termination or contaminants artificially induced during the atomic H exposure and not sensed by XPS, e.g. a H-termination. Despite the high structural order of the as-prepared GaAs surface, oxygenrelated bonding can be observed at the Ge/GaAs interface once that the GeO2/Ge double stack is grown upon atomic O exposure to the Ge layer. The entity of the Ga–O and As–O bonds is significantly lowered when using atomic H as surface cleaner. We acknowledge REALISE European Project NMP4-CT-2006016172 and ARAMIS Project for partially funding this activity. References [1] S. Takagi, T. Tezuka, T. Numata, S. Nakazari, N. Hirashita, Y. Moriyama, K. Usuda, E. Toyoda, S. Dissanayake, M. Shichijo, R. Nakane, S. Sugahara, M. Takenaka, N. Sugiyama, IEEE Trans. Electron. Dev. 55 (2008) 21.
S127
[2] G. Brammertz, M. Heyns, M. Meuris, M. Caymax, D. Jiang, Y. Mols, S. Degrote, M. Leys, G. Borghs, Appl. Phys. Lett. 90 (2007) 134102. [3] P.D. Ye, J. Vac. Sci. Technol. A 26 (2008) 697. [4] G. Brammertz, H.-C. Lin, K. Martens, D. Mercier, S. Sioncke, A. Delabie, W.E. Wang, M. Caymax, M. Meuris, M. Heyns, Appl. Phys. Lett. 93 (2008) 183504. [5] W.E. Spicer, Z. Liliental-Weber, E. Weber, N. Newman, T. Kendelewicz, R. Cao, C. McCants, P. Mahowald, K. Miyano, I. Lindau, J. Vac. Sci. Technol. B 6 (1988) 1245. [6] M.J. Hale, S.I. Yi, J.Z. Sexton, A.C. Kummel, M. Passlack, J. Chem. Phys. 119 (2003) 6719. [7] C.L. Hinkle, M. Milojevic, B. Brennan, A.M. Sonnet, F.S. Aguirre-Tostado, G.J. Hughes, E.M. Vogel, R.M. Wallace, Appl. Phys. Lett. 94 (2009) 162101. [8] J. Robertson, Appl. Phys. Lett. 94 (2009) 152104. [9] R. Droopad, K. Rajagopalan, J. Abrokwah, L. Adams, N. England, D. Uebelhoer, P. Fejes, P. Zurcher, M. Passlack, J. Cryst. Growth 301–302 (2007) 139. [10] D. Shahrjerdi, M.M. Oye, A.L. Holmes, S.K. Banerjee, Appl. Phys. Lett. 89 (2006) 043501. [11] C.L. Hinkle, A.M. Sonnet, E.M. Vogel, S. McDonnell, G.J. Hughes, M. Milojevic, B. Lee, F.S. Aguirre-Tostado, K.J. Choi, J. Kim, R.M. Wallace, Appl. Phys. Lett. 91 (2007) 163512. [12] W.P. Li, X.W. Wang, Y.X. Liu, S.I. Shim, T.P. Ma, Appl. Phys. Lett. 90 (2007) 193503. [13] H.-S. Kim, I. Ok, M. Zhang, F. Zhu, S. Park, J. Yum, H. Zhao, J.C. Lee, J. Oh, P. Majhi, Appl. Phys. Lett. 92 (2008) 102904. [14] A. Molle, G. Brammertz, L. Lamagna, M. Fanciulli, M. Meuris, S. Spiga, Appl. Phys. Lett. 95 (2009) 023507. [15] F. Bellenger, M. Houssa, A. Delabie, V. Afanasiev, T. Conard, M. Caymax, M. Meuris, K. De Meyer, M.M. Heyns, J. Electrochem. Soc. 155 (2008) G33. [16] D. Kuzum, T. Krishnamohan, A.J. Pethe, A.K. Okyay, Y. Oshima, Y. Sun, J.P. McVittie, P.A. Pianetta, P.C. McIntyre, K.C. Saraswat, IEEE Electron Device Lett. 29 (4) (2008) 328. [17] S. Spiga, C. Wiemer, G. Tallarida, G. Scarel, S. Ferrari, G. Seguini, M. Fanciulli, Appl. Phys. Lett. 87 (2005) 112904. [18] D. Tsoutsou, L. Lamagna, S.N. Volkos, A. Molle, S. Baldovino, S. Schamm, P.E. Coulon, M. Fanciulli, Appl. Phys. Lett. 94 (2004) 053504. [19] D. Fischer, A. Kersch, Appl. Phys. Lett. 92 (2008) 012908. [20] A. Molle, S. Spiga, M. Fanciulli, J. Chem. Phys. 129 (2008) 011104. [21] A. Molle, S. Spiga, A. Andreozzi, M. Fanciulli, G. Brammertz, M. Meuris, Appl. Phys. Lett. 93 (2008) 133504. [22] NIST Database, see http://srdata.nist.gov/xps/. [23] Guide to using WVASE 32, J. A. Woollam Co., Inc., (www.jawoollam.com). [24] S.A. Chambers, T.J. Irwin, PRB 38 (1988) 7858. [25] H. G. Tompkins and E. A. Irene, Handbook of Ellipsometry, William Andrew Publishing and Springer (2005). [26] E.D. Palik, Handbook of Optical Constants of Solids, Academic Press, 1985. [27] S. Zollner, D. Zarr, Compound Semiconductors, 2000 IEEE Int. Symp, 2000, p. 13. [28] H.G. Tompkins, A user's guide to Ellipsometry, Dover Publications, Inc., Mineola, New York, 1993. [29] A. Molle, M.N.K. Bhuiyan, G. Tallarida, M. Fanciulli, Appl. Phys. Lett. 89 (2006) 083504. [30] O. Renault, L. Fourdrinier, E. Martinez, L. Clavelier, C. Leroyer, N. Barrett, C. Crotti, Appl. Phys. Lett. 90 (2007) 052112. [31] P. Tsipas, S.N. Volkos, A. Sotiropoulos, S.F. Galata, G. Mavrou, D. Tsoutsou, Y. Panayiotatos, A. Dimoulas, C. Marchiori, J. Fompeyrine, Appl. Phys. Lett. 93 (2008) 082904. [32] J.J. Kolodzieij, B. Such, M. Szymonski, Phys. Rev. B 71 (2005) 165419. [33] A. Khatiri, J.M. Ripalda, T.J. Krzyzewski, G.R. Bell, C.F. McConville, and T.S. Jones, Surf. Sci. 548, L1 (2004) and references therein. [34] F.S. Aguirre-Tostado, M. Milojevic, C.L. Hinkle, E.M. Vogel, R.M. Wallace, S. McDonnell, G.J. Hughes, Appl. Phys. Lett. 92 (2008) 171906.
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Interface analysis of Ge ultra thin layers intercalated between GaAs substrates and oxide stacks Alessandro Molle a,⁎, Luca Lamagna a, Sabina Spiga a, Marco Fanciulli a,b, Guy Brammertz c, Marc Meuris c a b c
Laboratorio Nazionale MDM, CNR-INFM, Via C. Olivetti 2, 20041 Agrate Brianza (MI), Italy Dipartimento di Scienza dei Materiali, Università di Milano Bicocca, Milano, Italy IMEC, 75 Kapeldreef, B-3001 Leuven, Belgium
a r t i c l e
i n f o
Available online 19 October 2009 Keywords: GaAs III-V High-k High-mobility Atomic H Surface preparation XPS Interface study Logics
a b s t r a c t Capping III–V compound surfaces with Ge ultra-thin layer might be a viable pathway to passivate the electrically active interface traps which usually jeopardize the integration of III–V materials in metal-oxidesemiconductor devices. As the physical nature of such traps is intrinsically related to the chemical details of the interface composition, the structural and compositional features of the Ge/GaAs interface were thoroughly investigated in two different configurations, the atomic layer deposition of La-doped ZrO2 films on Ge-capped GaAs and the ultra-high vacuum based molecular beam deposition of GeO2/Ge double stack on in situ prepared GaAs. In the former case, the intercalation of a Ge interface layer is shown to suppress the concentration of interface Ga–O, As–O and elemental As bonding which were significantly detected in case of the direct oxide deposition on GaAs. In the latter case, the incidence of two different in situ surface preparations, the Ar sputtering and the atomic H cleaning, on the interface composition is elucidated and the beneficial role played by the atomic H exposure in reducing the semiconductor–oxygen bonds at the interface level is demonstrated. © 2009 Elsevier B.V. All rights reserved.
1. Introduction As far as superior transport properties of channel semiconductors are required to face the aggressive shrinking of the gate length in advanced metal-oxide-semiconductor (MOS) devices, III–V compounds are regarded as a viable option to replace the Si-based technology. Indeed, the in-channel effective electron mass of III–V semiconductors, lower than that of Si, leads to a higher electron velocity [1]. Among the III–V compounds, GaAs has so far retained the greater consideration. Nonetheless, it has recently been argued that ternary InxGa1−xAs compound might provide a more forgiving material in terms of density of interface defects [2] and higher mobility in MOS device for high In concentration [3]. An open concern on the integration of GaAs channel in MOS devices relies on the still non-established passivation of semiconductor/gate insulator interface, i.e. how to saturate the electrically active interface defects which cause pinning of the semiconductor Fermi level irrespectively to the externally applied voltage [4]. The physical nature of the defects nearby the GaAs surface region or at the GaAs/oxide interface is currently matter of investigation. A first general distinction can be made by sorting defects with energy level in the midgap region from those with energy level placed in the vicinity of the semiconductor
⁎ Corresponding author. E-mail address: [email protected] (A. Molle). 0040-6090/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2009.10.069
band-edges. The former was originally rationalized by Spacer et al. [5] in the advanced unified model for defects of the GaAs class of semiconductors as bulk bonding disruption and re-organization, e.g. formation of anti-bonding defects (AsGa antisites). However this picture was revisited by Kummel et al. [6], who conversely attributed the midgap defects at the oxide/GaAs interface to oxygen chemisorption to Ga atoms based on local scanning tunneling microscopy/ spectroscopy. On the other hand there is a general consensus in associating band-edge defects with interface bonding states (e.g. semiconductor–oxygen bonding and homopolar As or Ga bonding) induced during the oxide deposition processing [7,8]. The midgap defects are somehow hard to remove because they are related to GaAs bonding re-organization extrinsically induced by subsequent oxide deposition. Only the molecular beam epitaxy of Gd–Ga–O was proved to be successful in passivating As-rich GaAs surfaces by positioning interfacial Ga2O molecules between adjacent As dimers [9]. On the other hand, several solutions have been proposed to improve the interface bonding configuration with ad hoc surface treatments or interface engineering processes. A promising approach to III–V passivation is combining a chemical or in vacuo surface treatments for the native oxide removal along with the deliberate deposition of ultra thin interface passivation layer (IPL) of group V semiconductors [10,11], or nitrides [12]. In particular, the interest for Ge-based IPL has recently increased [13,14] in concomitance with the promising advances in the Ge passivation through various oxidation approaches [15–17]. Indeed, the good crystallographic matching of Ge with III–V
S124
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
compound surface lattice may alleviate bonding disruption in the GaAs surface arrangement and the intercalation of a Ge IPL in between the III–V compound substrate and the gate oxide stack may prevent the formation of interfacial semiconductor–oxygen bonds. In the present work, the structural and compositional details of ultra-thin Ge IPL on GaAs substrates were investigated in two different process configurations, the atomic layer deposition (ALD) of ultra thin Ladoped ZrO2 (L–ZO) films on Ge-capped GaAs (Section 3) and the ultra high vacuum (UHV) based molecular beam deposition (MBD) of GeO2/Ge passivation double stacks on GaAs (Section 4). 2. Experimental details Commercial Si-doped (1 × 1017 cm− 3) GaAs(001) substrates were used either for ALD and MBD-grown samples. In case of the ALD samples the substrates were chemically pre-treated via dipping in NH4 (OH) solution (4%) for 3 min. The deposition of a 2 nm (nominally) thick Ge layer was performed in a high-vacuum chamber by an e-beam evaporator at room temperature (RT). La-doped ZrO2 (L–ZO) films were subsequently grown in an ALD reactor at 300 °C directly onto as-prepared GaAs and after Ge IPL insertion. L–ZO was deposited using O3 as oxidizing precursor in combination with La [(iPrCp)3La] and Zr [(MeCp)2ZrMe(OMe)] metal precursors as described in Ref. [18]. The use of O3 as oxidizing agent addresses the further passivation of the Ge overlayer surface according to the low density of interface states previously achieved after O3 exposure to Ge [16] or O3-based ALD growth of HfO2 [17]. On the other hand, La incorporation in ZrO2 can effectively improve the dielectric features of the oxide stack by changing the polarizability of the oxide molecules and by stabilizing a crystallographic phase endowed with a higher dielectric constant [19]. Interface diagnostic was performed by x-ray photoelectron spectroscopy (XPS) of 3 nm-thick L–ZO films and by spectroscopic ellipsometry (SE) on 10 nm-thick L–ZO films. In parallel, the GaAs(001) substrates were inserted in a multichamber UHV system (base pressure 1 × 10− 11 mbar) to perform MBD growth and in situ characterization of GeO2/Ge double stack layer as a function of the surface treatment. After a preliminary outgas at 450 °C, the substrates were prepared by means of two different in vacuo treatments, 30 min long Ar ion (1 keV) sputtering at 600 °C and 30 min long exposure to an atomic H beam at 400 °C. Ar ions were generated by a standard ion gun working in a differential pumping assembly. The atomic H beam was provided by a radio frequency plasma source at a power of 350 W starting from a forming gas (4% H2, 96 % Ar) supply as
high as 1 × 10− 4 mbar. A 3 nm-thick Ge film was then deposited by an effusion cell with a rate of 1 Å/min (base pressure of 9 × 10− 10 mbar during growth, growth rate calibrated by an in situ quartz balance thickness monitor). Ge growth was monitored by reflection high energy (30 keV) electron diffraction (RHEED). A 1.7 nm-thick GeO2 layer was subsequently formed by 7 min long atomic O exposure at 300 °C to the 3 nm-thick Ge film to subsequently address the Ge surface passivation. Details of the GeO2 formation are reported elsewhere [20,21]. The multistacked GeO2/Ge/GaAs heterostructure can be thus used as interface engineered substrate for the subsequent deposition of high permittivity oxides aiming at the fabrication of GaAs-based MOS [14]. Each step of the surface preparation and double stack deposition on the latter samples was monitored in situ by RHEED and XPS. XPS was provided by a standard Mg Kα source (1253.6 eV) with a pass energy of 20 V as a function of the take-off angle (angle-resolved analysis). Deconvolution of the relevant XPS 3d core-level photoemission lines was performed by using Shirley baseline removal and Lorentzian-Gaussian doublets as fitting components. Doublets accounts for the spin-orbit splitting in d core-level photoemission lines. According to Ref [21,22] the following doublet separations (DS) were used, DSGa3d = 0.44 eV, DSAs3d = 0.67 eV, DSGe3d = 0.60 eV. SE evaluates changes in the polarization state of a reflected light beam in terms of the ellipsometric angles Ψ and Δ thus enabling one to determine the thickness of the GaAs heterostructures through data interpolation based on a multiple layers model by means of the WVASE32 software. SE data are collected by means of a rotating compensator spectroscopic ellipsometer (J. A. Woollam M2000F) in the spectral range of 1.0–5.0 eV and by using a light beam with an incidence angle of 75° [23]. 3. Atomic layer deposition of La-doped ZrO2 A general overview of the XPS compositional analysis accounting for the Ga 3d, In 4d, In 3d5/2, As 3d core-level photoemission lines is reported in Fig. 1(a) and (b) for 3 nm thick L–ZO films on GaAs without and with Ge IPL, respectively. The binding energy (BE) of the spectra was referred to the pre-calibrated position of adventitious carbon in the C 1s line (BE = 284.8 eV). Direct ALD of L–ZO onto GaAs is studied in Fig. 1(a). Here, the Ga 3d line exhibits an interplay of three components, from GaAs bulk (GaB at BE = 18.7 eV), from Ga2O3 interfacial bonding (Ga3+ at BE = 19.9 eV) and from the O 2 s line coming due to the overlying oxide (BE = 22.2 eV). The As 3d line deconvolution involves three
Fig. 1. Ga 3d and As 3d XPS lines recorded for L–ZO/GaAs (a) and L–ZO/Ge/GaAs (b). LZO thickness is 3 nm, Ge IPL thickness is 2 nm. GaB and AsB denote the bulk contributions, the displacement between the bulk peaks in case (a) and (b) is marked. Ga3+ and As3+ components reflect the presence of interfacial Ga2O3 and As2O3, respectively. As0 indicates interfacial elemental As bonding.
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
different contributions which can be assigned to As from GaAs bulk (BE = 40.7 eV), to As2O3 (and no As2O5) species (BE = 43.8 eV) and to elemental As bonding (BE = 41.6 eV). Opposed to the significant interface concentration of semiconductor–oxygen and As homopolar bonding observed after direct ALD processing, the scenario radically changes when Ge IPL is sandwiched between the oxide and GaAs. The related XPS analysis is illustrated in Fig. 1(b). As a result, the formation of As–O bonding can be here completely ruled out from the interface composition as no associated component is observed in the respective As 3d lines, whereas a remarkably lower Ga–O related component can be deduced from the Ga 3d line deconvolution. It should be noted that the position of the Ga 3d and As 3d bulk peaks are mutually displaced by an amount of 0.6 eV which reflects the different alignment of the electronic bands in the two different interface configurations. In particular, the LZO/Ge/GaAs exhibits a XPS line shift of 0.3 eV to higher BE with respect to the line positioning in the clean GaAs surface (BEGa3d = 19.1 eV, BEAs3d = 40.9 eV) as shown in Fig. 3(c) which can be explained by a Ge-induced band bending at the interface band line-up as previously observed [24]. The electrical quality of the GaAs-based heterostructure should be also dictated by the chemical details of the interface that Ge IPLs form with the overlying L–ZO layer. The information from SE and XPS data were mutually combined in order to elucidate this aspect as reported in Fig. 2. From the SE measurements of the NH4(OH) treated GaAs (001), taken immediately before the oxide deposition, the pseudodielectric function (〈ε〉 = 〈ε1〉 + i〈ε2〉) of the substrate [25] can be extracted by means of a Point-to-Point interpolation [26] of the spectra. This enabled us to model the semiconductor substrate as a nearly oxide free. Therefore the calculated 〈ε〉 results in a qualitatively good agreement with the tabulated values expected for GaAs(100) [27] and is then used in all the models to represent the substrate. L–ZO dielectric function was estimated using a Cauchy dispersion relation [28] (oxide is transparent in the 1.0–5.0 eV range) on thicker samples and was then applied to model the thin layers constituting our samples; Ge dielectric function was assumed from tabulated values. The only free parameters used in the two SE fits were oxide thickness (tox) and Ge layer thickness (tge). In Fig. 2(a) a close-up of the experimental Δ spectra acquired is depicted after ALD deposition on GaAs(001) and on Ge/GaAs(001). The pronounced difference between the two curves can be uniquely attributed to the presence of the ultra-thin Ge IPL underneath the oxide. Such a difference can be rationalized in terms of the multi-layers physical model fit to the SE data. In detail, a two layer model (see scheme A in the inset) was used
S125
to represent L–ZO/GaAs therein resulting in an oxide thickness of tox of 8.2 ± 0.2 nm. On the contrary, a three layer model based fitting was necessary to account for the Ge-passivated sample therein indicating the permanence of purely Ge IPL after oxide deposition. Using a three layers model (see the inset) results in tox = 7.4 ± 0.2 nm and tge ~ 0.6 nm. This picture can be corroborated by the complementary XPS analysis in Fig. 2(b) where the Ge 3d line is studied as a function of the sampling depth, i.e. at a low take-off angle measurement (θ = 37°) and under normal photoemission (θ = 90°) to probe near-to-surface and deeper regions of the sample, respectively. Both Ge 3d lines exhibit three components, one at lower BE due to elemental Ge –Ge bonding of the Ge IPL and two (A and B in Fig. 2(b)) at higher BE (BEA = 30.7 eV and BEB = 31.9 eV) being consistent with oxidized Ge bonding. The evidence of an elemental Ge bonding signal indicates the presence of a pure Ge/oxidized Ge interfacial structure. The observed Ge–O bonding signal can be located at the Ge/L–ZO interface as the related XPS signal increases for decreasing take-off angles, i.e. with a lower sampling depth. The physical nature of the Ge–O component is consistent with the formation of a Ge–O bonding region throughout the L–ZO top stack. A more careful deconvolution of the Ge 3d line suggests the interplay of two Ge–O components (A and B in Fig. 2(b)) with different valence states. The A and B components can be assigned to minor GeOx concentration and [20,29] to germanate like region [30], respectively. As the relative A/B intensity ratio slightly increases with increasing θ (from 0.14 at θ = 37° to 0.17 at θ = 90°), the GeOx species presumably places atop the pure Ge layer whereas the germanate is located in between the GeOx and the L–ZO. Such a picture is also supported by the observed position of the relevant Zr 3d line at BE = 182.6 eV(not shown) being consistent with the Ge substitutional bonding inside ZrO2 as described in Ref. [31]. Further investigations of this issue are in progress. 4. Molecular beam deposition of GeO2/Ge double stacks Following a similar approach to that of Section 3, 3 nm-thick Ge layer were deposited in situ on differently treated GaAs(001) surfaces. Aiming at the Ge surface passivation [29], the Ge films have been subsequently capped with ultra-thin GeO2 over-layers by 7 min in situ exposure to atomic O at 300 °C so to have 1.7 nm GeO2/1.5 nm Ge/ GaAs multistacked structures. The bonding configuration at the GeO2Ge/GaAs interface was assessed in situ for two different surface in vacuo preparations, 30 min long Ar+ (700 eV) sputtering at 600 °C
Fig. 2. (a) SE scans registered for L–ZO/GaAs (line) and L–ZO/Ge/GaAs (circles). Inset: sketches of the two multilayer models used to interpolate the SE data in the two different interface configurations, L–ZO/GaAs and L–ZO/Ge/GaAs. (b) Ge 3d XPS lines taken at two take-off angles, θ = 37° (up) and 90° (down). See the text for details of the spectra deconvolution.
S126
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
Fig. 3. RHEED pattern of an Ar-sputtered GaAs(001) surface along the [1–10] (a1), [110] (a2) and of a 1.5 nm-thick Ge film grown on top at room temperature (a3). Note the (2 × 1) reconstruction occurring at the Ge surface. RHEED pattern of an atomic H-cleaned GaAs(001) surface along the [1–10] (b1), [110] (b2) and of a 1.5 nm-thick Ge film grown on top at room temperature (b3). Note the amorphous structure of the Ge film in this case. (c) Ga 3d and As 3d XPS lines recorded for the as-received GaAs (rich of native oxides), for the atomic H cleaned GaAs, and for the Ar-sputtered GaAs.
and 30 min long atomic H irradiation at 400 °C. Previous reports on the GaAs preparation suggest that both treatments are effective in giving highly ordered and defect-free surfaces [32,33]. However, while the ion sputtering might damage the inner constitutions of the III–V compounds locally, an increasing attention is currently paid to exploit atomic H radicals as non-destructive method to remove native oxides and leave a well-reconstructed III–V surface [34]. A Ga-rich (4 × 6) reconstruction can be recognized in case of the Ar-sputtered GaAs surface from the relevant streak periodicity in the RHEED patterns along the [110] (4× periodicity) and the [1–10] (×6 periodicity) directions shown in Fig. 3(a1)–(a2) as reported in Ref. [32]. Otherwise, based on similar RHEED observation of Fig. 3 (b1)–(b2), the atomic H clean is observed to release an As-rich (2 × 4) reconstructed GaAs surface in agreement with Ref. [33]. The structure of the Ge layer appears to be deeply influenced by the surface preparation. Indeed, the RHEED patterns of the 1.5 nm-thick Ge films grown at room temperature (RT) on the Ar-sputtered surface in Fig. 3 (a3) looses the reconstruction fashion of the underlying substrates, but maintains the primary diffraction streaks thus indicating the epitaxial character of Ge growth and the formation of (2 × 1) reconstructed Ge surface. Conversely, despite the good lattice matching between Ge and GaAs, after atomic H clean the Ge layer exhibits an amorphous arrangement as follows from the absence of diffraction signal in the RHEED image in Fig. 3(b3). The growth of an amorphous Ge film was also documented in case of Ge deposition on atomic H-cleaned In0.15Ga0.85As surfaces [21]. The emergence of an amorphous Ge film might be related to the hypothetical H termination of the GaAs surface. Complementary XPS diagnostic of the differently prepared substrates in Fig. 3(c) rules out residual presence of native oxides and adventitious contaminations in both surface preparations compared to the as-received substrate. The 1.7 nm GeO2/1.3 nm Ge/GaAs heterostructures were fabricated as described in Section 2 by growing Ge films at room temperature. Since the thickness of the GeO2/Ge IPL is still below the mean free paths for bulk photoelectrons, the bonding configuration at the Ge/GaAs interface can be investigated by probing the relevant core level photoemission lines, Ga 3d, Ge 3d, As 3d recorded at a take-off angle of 90°, i.e. normally to the sample surface, as illustrated in Fig. 4(a) and (b) for the atomic H clean sample and the Ar-sputtered sample, respectively. The Ga 3d and As 3d line present a multi-shaped profile arising from the interplay of the contributions from the Ga-As bulk bonds (marked as GaB and AsB) with valence state components at higher BE, Ga3+ (chemical shift Δ = 1.7 eV from GaB) for Ga 3d, As5+ (Δ = 5.2 eV from AsB) for As 3d. The extra valence state components Ga3+ and As5+ denote the presence of interfacial Ga2O3 and As2O5, respectively [7,9,12,14,21]. An additional component Ga* can be observed in shape profile of the Ga
3d line for both surface preparations at lower BEs which can be associated with Ga bonding with more metallic character, e.g. possibly Ga-Ge bonding. From the As 3d line deconvolution no As2O3 species can be detected. Ge oxidation is demonstrated in both cases by the largely dominant GeO2 contribution to the Ge 3d line (see the insets of Fig. 3) as previously reported [16,20,29]. To quantitatively assess the relative amount of interfacial Ga–O and As–O bonds, from the best-fit parameters it is useful to calculate the ratios ζ between the areas of the valence state component(s) (Ga3+ for Ga 3d, As3+ and As5+ for As 3d) and the bulk component, ζGa–O = Ga3+/ GaB and ζAs–O = (As3+ + As5+)/AsB. From the spectra comparison we conclude that the relative amount of oxide bonds at the Ge/GaAs interface strikingly changes with surface preparation. Indeed, the ζGa–O and ζAs–O values are significantly reduced after Ge oxidation from 0.08 and 0.66 in the Ar-sputtered sample down to 0.03 and 0.34 in the atomic H cleaned one. The fraction of interfacial Ga–O bonds is negligible in the latter configuration. In both cases, the nature of the As–O bonding is exclusively identified by the As5+ valence state corresponding to the As2O5 species only. The different chemical quality of the Ge/GaAs interface achieved with the two surface preparations might be also here tentatively attributed to the occurrence of a H-terminated GaAs surface or to the residual presence of contaminants from the atomic gas generation after the atomic H treatment. These two aspects might have
Fig. 4. Ga 3d and As 3d XPS lines recorded for GeO2/Ge/GaAs heterostructure after 30 min atomic H cleaning at 400 °C (a) and after 30 min Ar sputtering at 600 °C (b). Insets: Ge 3d XPS lines for the two cases in (a) and (b).
A. Molle et al. / Thin Solid Films 518 (2010) S123–S127
deep implications in the different structure of the Ge layer observed as a function of the surface treatment. However, further theoretical and experimental investigations are needed to elucidate the interaction between atomic H and III–V compound surfaces. 5. Conclusions Given the potential application of Ge as passivating material for III–V compound surfaces, Ge ultra-thin films were grown on differently prepared GaAs(001) substrates and the relevant interface bonding was elucidated by XPS. Intercalation of Ge IPL between NH4 (OH) treated GaAs and La-doped ZrO2 films grown by O3-based ALD confers a remarkably improved chemical quality to the interface with GaAs compared to the case of the direct oxide deposition. The Ge IPL leads to an efficient removal of potentially defective Ga2O3, As2O3 and As elemental bonding from the interface. The use of O3 as oxidizing agent is concomitantly intended to gain superior dielectric properties of the oxide [18] and to pursue the electrical passivation of the Ge IPL surface [17] thereby providing an ALD suited approach to fabrication of Ge IPL/GaAs-based MOS capacitors. On the other hand, the chemical details of the Ge/GaAs interface were also studied through the molecular beam deposition of GeO2/Ge double stacks on GaAs surfaces prepared either by Ar sputtering up to 600 °C and by atomic H cleaning at 400 °C in an UHV environment. The GaAs surface arrangement and the structure of the subsequently grown Ge films appear to be strictly dependent on the surface preparation, i.e. resulting in a Ga-rich (4× 6) reconstruction and an epitaxial Ge layer in the former case, and in an As-rich (2× 4) reconstruction and an amorphous Ge layer in the latter one. Given the good crystallographic matching between Ge and GaAs, the amorphous Ge layer grown on atomic H cleaned GaAs can be tentatively associated with the presence of surface termination or contaminants artificially induced during the atomic H exposure and not sensed by XPS, e.g. a H-termination. Despite the high structural order of the as-prepared GaAs surface, oxygenrelated bonding can be observed at the Ge/GaAs interface once that the GeO2/Ge double stack is grown upon atomic O exposure to the Ge layer. The entity of the Ga–O and As–O bonds is significantly lowered when using atomic H as surface cleaner. We acknowledge REALISE European Project NMP4-CT-2006016172 and ARAMIS Project for partially funding this activity. References [1] S. Takagi, T. Tezuka, T. Numata, S. Nakazari, N. Hirashita, Y. Moriyama, K. Usuda, E. Toyoda, S. Dissanayake, M. Shichijo, R. Nakane, S. Sugahara, M. Takenaka, N. Sugiyama, IEEE Trans. Electron. Dev. 55 (2008) 21.
S127
[2] G. Brammertz, M. Heyns, M. Meuris, M. Caymax, D. Jiang, Y. Mols, S. Degrote, M. Leys, G. Borghs, Appl. Phys. Lett. 90 (2007) 134102. [3] P.D. Ye, J. Vac. Sci. Technol. A 26 (2008) 697. [4] G. Brammertz, H.-C. Lin, K. Martens, D. Mercier, S. Sioncke, A. Delabie, W.E. Wang, M. Caymax, M. Meuris, M. Heyns, Appl. Phys. Lett. 93 (2008) 183504. [5] W.E. Spicer, Z. Liliental-Weber, E. Weber, N. Newman, T. Kendelewicz, R. Cao, C. McCants, P. Mahowald, K. Miyano, I. Lindau, J. Vac. Sci. Technol. B 6 (1988) 1245. [6] M.J. Hale, S.I. Yi, J.Z. Sexton, A.C. Kummel, M. Passlack, J. Chem. Phys. 119 (2003) 6719. [7] C.L. Hinkle, M. Milojevic, B. Brennan, A.M. Sonnet, F.S. Aguirre-Tostado, G.J. Hughes, E.M. Vogel, R.M. Wallace, Appl. Phys. Lett. 94 (2009) 162101. [8] J. Robertson, Appl. Phys. Lett. 94 (2009) 152104. [9] R. Droopad, K. Rajagopalan, J. Abrokwah, L. Adams, N. England, D. Uebelhoer, P. Fejes, P. Zurcher, M. Passlack, J. Cryst. Growth 301–302 (2007) 139. [10] D. Shahrjerdi, M.M. Oye, A.L. Holmes, S.K. Banerjee, Appl. Phys. Lett. 89 (2006) 043501. [11] C.L. Hinkle, A.M. Sonnet, E.M. Vogel, S. McDonnell, G.J. Hughes, M. Milojevic, B. Lee, F.S. Aguirre-Tostado, K.J. Choi, J. Kim, R.M. Wallace, Appl. Phys. Lett. 91 (2007) 163512. [12] W.P. Li, X.W. Wang, Y.X. Liu, S.I. Shim, T.P. Ma, Appl. Phys. Lett. 90 (2007) 193503. [13] H.-S. Kim, I. Ok, M. Zhang, F. Zhu, S. Park, J. Yum, H. Zhao, J.C. Lee, J. Oh, P. Majhi, Appl. Phys. Lett. 92 (2008) 102904. [14] A. Molle, G. Brammertz, L. Lamagna, M. Fanciulli, M. Meuris, S. Spiga, Appl. Phys. Lett. 95 (2009) 023507. [15] F. Bellenger, M. Houssa, A. Delabie, V. Afanasiev, T. Conard, M. Caymax, M. Meuris, K. De Meyer, M.M. Heyns, J. Electrochem. Soc. 155 (2008) G33. [16] D. Kuzum, T. Krishnamohan, A.J. Pethe, A.K. Okyay, Y. Oshima, Y. Sun, J.P. McVittie, P.A. Pianetta, P.C. McIntyre, K.C. Saraswat, IEEE Electron Device Lett. 29 (4) (2008) 328. [17] S. Spiga, C. Wiemer, G. Tallarida, G. Scarel, S. Ferrari, G. Seguini, M. Fanciulli, Appl. Phys. Lett. 87 (2005) 112904. [18] D. Tsoutsou, L. Lamagna, S.N. Volkos, A. Molle, S. Baldovino, S. Schamm, P.E. Coulon, M. Fanciulli, Appl. Phys. Lett. 94 (2004) 053504. [19] D. Fischer, A. Kersch, Appl. Phys. Lett. 92 (2008) 012908. [20] A. Molle, S. Spiga, M. Fanciulli, J. Chem. Phys. 129 (2008) 011104. [21] A. Molle, S. Spiga, A. Andreozzi, M. Fanciulli, G. Brammertz, M. Meuris, Appl. Phys. Lett. 93 (2008) 133504. [22] NIST Database, see http://srdata.nist.gov/xps/. [23] Guide to using WVASE 32, J. A. Woollam Co., Inc., (www.jawoollam.com). [24] S.A. Chambers, T.J. Irwin, PRB 38 (1988) 7858. [25] H. G. Tompkins and E. A. Irene, Handbook of Ellipsometry, William Andrew Publishing and Springer (2005). [26] E.D. Palik, Handbook of Optical Constants of Solids, Academic Press, 1985. [27] S. Zollner, D. Zarr, Compound Semiconductors, 2000 IEEE Int. Symp, 2000, p. 13. [28] H.G. Tompkins, A user's guide to Ellipsometry, Dover Publications, Inc., Mineola, New York, 1993. [29] A. Molle, M.N.K. Bhuiyan, G. Tallarida, M. Fanciulli, Appl. Phys. Lett. 89 (2006) 083504. [30] O. Renault, L. Fourdrinier, E. Martinez, L. Clavelier, C. Leroyer, N. Barrett, C. Crotti, Appl. Phys. Lett. 90 (2007) 052112. [31] P. Tsipas, S.N. Volkos, A. Sotiropoulos, S.F. Galata, G. Mavrou, D. Tsoutsou, Y. Panayiotatos, A. Dimoulas, C. Marchiori, J. Fompeyrine, Appl. Phys. Lett. 93 (2008) 082904. [32] J.J. Kolodzieij, B. Such, M. Szymonski, Phys. Rev. B 71 (2005) 165419. [33] A. Khatiri, J.M. Ripalda, T.J. Krzyzewski, G.R. Bell, C.F. McConville, and T.S. Jones, Surf. Sci. 548, L1 (2004) and references therein. [34] F.S. Aguirre-Tostado, M. Milojevic, C.L. Hinkle, E.M. Vogel, R.M. Wallace, S. McDonnell, G.J. Hughes, Appl. Phys. Lett. 92 (2008) 171906.