- Email: [email protected]
Mn5Ge3 heterostructures induced by interstitial carbon
Thin Solid Films 520 (2012) 3410–3414
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Suppression of Mn segregation in Ge/Mn5Ge3 heterostructures induced by interstitial carbon Minh-Tuan Dau a, Vinh Le Thanh a,⁎, Thi-Giang Le a, Aurélie Spiesser a, Matthieu Petit a, Lisa A. Michez a, Thu-Huong Ngo b, Dinh Lam Vu c, Quang Liem Nguyen c, Pierre Sebban d a
Interdisciplinary Center of Nanoscience of Marseille (CINaM-CNRS), Aix-Marseille University, Campus of Luminy, Case 913, F-13288 Marseille cedex 9, France Faculty of Physics, Hanoi University of Science, 334 Nguyen Trai Road, Hanoi, Viet Nam Institute of Materials Science, Vietnam Academy of Science and Technology (VAST), 18 Hoang Quoc Viet Str., Cau Giay, Hanoi,Viet Nam d University of Science and Technology of Hanoi (USTH), 18 Hoang Quoc Viet Str., Cau Giay, Hanoi,Viet Nam b c
a r t i c l e
i n f o
Available online 6 November 2011 Keywords: Mn5Ge3/Ge heterostructures Intermetallic magnetic compound Segregation length Interstitial diffusion Surfactants
a b s t r a c t Mn5Ge3 compound, with its room-temperature ferromagnetism and possibility to epitaxially grow on Ge, acts as a potential spin injector into group-IV semiconductors. It is shown that the realization of Ge/ Mn5Ge3 heterostructures is highly hampered by Mn segregation toward the Ge growing surface. The Mn segregation length can be estimated in-situ and in real time by means of reflection high-energy electron diffraction. We present here an approach allowing to greatly reduce or even to prevent the Mn segregation, whose principle is based on filling the Mn5Ge3 lattice with interstitial carbon atoms. In addition, we show that interstitial carbon in Mn5Ge3 allows to enhance not only the Curie temperature of Mn5Ge3Cx layers but also in the whole Ge/Mn5Ge3/Ge heterostructures. © 2011 Elsevier B.V. All rights reserved.
1. Introduction In recent years, the epitaxial growth of intermetallic Mn5Ge3 thin films has received growing interest, work in this direction is motivated by the hope to find out an efficient spin injector into group-IV semiconductors [1–9]. Indeed, the use of standard ferromagnetic metals, such as Fe, Co or Ni, directly on Si and Ge is highly hampered by their reactivity with group-IV semiconductors to form interfacial silicides or germanides, which, for most of them, are not ferromagnetic (for example, CoSi2 is metallic [10] while β-FeSi2 is semiconducting [11]). It is also not trivial to obtain epitaxial growth of an oxide in between Ge (or Si) and a ferromagnetic metal, efficiency of spin injection is therefore limited by the interface roughness and the difference in materials conductivity. The Mn5Ge3 compound exhibits numerous features, which may make it as a potential candidate for spin injection into Ge. First, this compound has a Curie temperature near room temperature [12], theoretical calculations predicted a high-spin-injection efficiency along its c axis [3] and a spin polarization up to 42% has been demonstrated [13]. Second, Ge3Mn5 films with thickness up to 165 nm can be epitaxially grown on Ge(111) substrates despite a misfit as high as 3.7% [6]. This feature is of particular importance with respect to the application of Mn5Ge3/Ge(111) heterostructures for the realisation of spintronic devices. Indeed, in a lattice mismatched heteroepitaxial system, a
⁎ Corresponding author. E-mail address: [email protected] (V. Le Thanh). 0040-6090/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2011.10.167
misfit as high as 3.7% often results in highly defected epitaxial layers. For example, Ge has a lattice parameter of ~4% larger than that of Si, pseudomorphic Ge growth is obtained only for a thickness smaller than 4 monolayers, beyond which islanding growth mode occurs [14]. For a Ge thickness of about some tens of nanometers, a threading dislocation density higher than 108 cm− 2 is usually observed and the growth front becomes extremely rough. In addition, in the Ge–Mn phase diagram [15], the Mn5Ge3 compound is not the most stable phase but thanks to its hexagonal structure, which is similar to that of the Ge(111) surface, Mn5Ge3 is the unique epitaxial phase, which has been stabilized on Ge(111) [1,2,5,6]. In this paper, we report on the effect of Mn segregation occurred during Ge overgrowth on Mn5Ge3/Ge heterostructures. During the course of our experiments to produce Ge/Mn5Ge3/Ge multilayers for the realization of spin valves and giant magneto-resistance (GMR) structures, we have observed a long-range Mn segregation upwards the Ge growth front. We show that incorporating carbon atoms into the interstitial sites of the Mn5Ge3 lattice allows to prevent outdiffusion of Mn from Mn5Ge3. In addition, we show that interstitial carbon also allows to enhance the magnetic properties of Ge/ Mn5Ge3/Ge multilayers. 2. Experimental details Materials growth was performed in a molecular beam epitaxial (MBE) system with a base pressure better than 5 × 10 − 10 mbar. The MBE chamber is equipped with a 30-keV reflection high-energy electron diffraction (RHEED) apparatus allowing to monitor in real time
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
the epitaxial growth process and the change of the film surface reconstruction. An Auger electron spectrometer is used to control the cleanliness of the substrate surface prior to growth and the film composition. Standard effusion cells were used for Mn and Ge deposition, the Mn flux was measured by a quartz crystal microbalance and the Ge growth rate was deduced from measurements of RHEED intensity oscillations. The Mn and Ge growth rates used in this work are of 0.9 and 3 nm/min, respectively. Carbon evaporation was carried out using a sublimation source of high purity pyrolytic graphite, the carbon concentration was estimated by combining two approaches: for each sublimation current, it was estimated from δ-doping curves measured on GaAs surfaces and then corrected according to the change of Si(001) surface reconstructions from (2 × 1) to c(4 × 4) upon adsorption of a submonolayer of carbon [16]. The estimated error is about 10%. The substrate temperature was measured by a thermocouple in contact with the backside of the substrate. Mn5Ge3/Ge(111) heterostructures were prepared by a room temperature Mn deposition onto Ge(111) substrates, followed by thermal annealing up to 430–450 °C, the temperature range at which the surface reconstruction characteristic of the Mn5Ge3 phase was generally observed. Structural analyses of post grown films were performed by means of high-resolution transmission electron microscopy (HRTEM) using a JEOL 3010 microscope operating at 300 kV with a spatial resolution of 1.7 Å. HRTEM cross sections were prepared by mechanical polishing, followed by ion polishing (Gatan Precision Ion Polishing System). Magnetic characterizations were carried out using superconducting quantum interference device magnetometer (SQUID) at temperatures ranging from 5 to 350 K. A magnetic field of 0.5 T is applied both in the plane of samples or perpendicular to it. The diamagnetic contributions of Ge substrates were subtracted from the measurements, leaving the magnetic contributions of grown films. More details of sample growth conditions can be found elsewhere [5,6]. 3. Results and discussion It is now well established that epitaxial Mn5Ge3 films on Ge(111) exhibit a (√3x√3)R30° surface reconstruction (denoted hereafter to as √3) with Mn termination [2]. From electron diffraction, such a reconstruction is manifested by the observation of 1 × 1 streaks along [1–10] azimuth and additional 1/3- and 2/3-ordered streaks between 1 × 1 streaks along [11-2] azimuth. The RHEED technique, thanks to its grazing incidence, allows to instantaneously reveal not only the change in the surface morphology but also the surface structure during film growth. To produce Ge/Mn5Ge3/Ge multilayers, we first grow
3411
a thin Mn5Ge3 layer on Ge(111), whose thickness ranges from 10 to 50 nm in order to insure a high crystalline quality and a smooth surface. Starting from those films, Ge overgrowth is carried out at temperatures ranging from 200 to 500 °C. Fig. 1a shows a RHEED pattern taken along [11-2] azimuth of the starting surface of a 40 nm thick Mn5Ge3 layer prior to Ge deposition. Additional 1/3- and 2/3-ordered streaks between the bulk-like 1 × 1 streaks are clearly seen, confirming the Mn-terminated Mn5Ge3 surface. The observed long streaks in the RHEED pattern unambiguously indicate that the corresponding surface is highly smooth. Displayed in Fig. 1b is an example of the RHEED pattern observed after deposition of 5 nm of Ge (corresponding to a deposition time of 100 s) at a substrate temperature of 250 °C. Apart from a certain decrease of the RHEED intensity, which signifies an increase of the surface roughness, the 1/3- and 2/3-ordered streaks still remain visible. This clearly indicates that Mn has floated upwards the Ge growth front and reacted with deposited Ge to form a surface Mn5Ge3-like layer. To quantify the Mn segregation length, we have measured the RHEED intensity evolution as a function of the Ge deposition time of the specular streak (green curve) and of a 2/3-ordered streak. As can be expected, the intensity of both the specular and 2/3 streaks is found to decrease when increasing the Ge deposition time but the specular streak exhibits a decrease more abrupt. The Mn segregation length can be determined from the intersection of the slope of these two curves with the deposition time axis. Taking into account the Ge growth rate of 3 nm/min, the Mn segregation length is found to be of ~10 nm for the specular streak and of ~ 22.5 nm for the 2/3-ordered streak. Since the specular streak is in general more sensitive to the development of the surface roughness than any diffracted streak, we can deduce a value of the Mn segregation length of ~22.5 nm for the Ge overgrowth at 250 °C. It is worth noting that Mn segregation has been reported during Ge deposition on (001)-oriented Ge substrates and a segregation length larger than 20 nm was measured at a substrate temperature as low as 150 K [17]. We also note that in-situ analyses of the RHEED intensity evolution were used to determine the segregation length of In in III–V materials [18] and of Ge during Si growth in Si/Si1-xGex heterostructures [19]. However, in those systems, the segregation length is only of some monolayers and the measurement of the segregation length was based on a RHEED intensity transient at the beginning of growth [18] or the change in the growth rate [19]. In our case, the appearance in RHEED patterns of an additional streaks characteristic of Mn5Ge3 allows a direct estimation of the Mn segregation length. The structural properties of the Ge overgrown layer on Mn5Ge3/ Ge(111) heterostructures are depicted in Fig. 2. Fig. 2a shows a typical cross-sectional TEM image of a 60 nm thick Ge layer grown on a
Fig. 1. RHEED patterns taken along the [2] azimuth: (a) of the √3 reconstruction of the Mn5Ge3 surface prior to the deposition of the Ge capping layers; (b): after deposition of 60 nm of Ge at a substrate temperature of 250 °C; and c) Evolution of the RHEED intensity of the specular streak (green curve) and a 2/3-ordered streak (read curve) with increasing the Ge deposition time. The dotted lines represent the slope of two curves.
3412
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
Fig. 2. (a) Typical cross-sectional TEM image of a 60 nm thick Ge layer grown on a Mn5Ge3/Ge(111) heterostructure at a substrate temperature of 250 °C; and (b) and (c) Crosssectional TEM images taken at interfacial regions of the corresponding sample.
Mn5Ge3/Ge(111) heterostructure at a substrate temperature of 250 °C. Two general features regarding the Ge overlayer can be noticed. First, the Ge overlayer is found to be relatively defected and contains some clusters. This implies that during Ge deposition, a part of Mn floats upwards the growth frond and acts as a surfactant while another part of Mn is incorporated inside the Ge layers, resulting in the formation of Mn-rich clusters. The incorporation of surfactant elements inside the upper grown layers is a common effect and has been observed for most of surfactant atoms during epitaxial growth [20,21]. Second, the interface between Ge overlayer and Mn5Ge3 is extremely rough as compared to the initial Mn5Ge3 surface (bi-dimensional RHEED pattern in Fig. 1a). Displayed in Fig. 2b and c are two TEM images taken around the interface regions of the corresponding sample. The interface roughness, the presence of defects as well as of Mn-rich clusters are clearly visible. These features are unambiguously a direct consequence of Mn segregation during Ge overgrowth. The phenomenon of Mn segregation in the Ge–Mn system has been previously reported [17,22,23]. Based on total energy calculations, Zhu et al. [22] predicted that along the [001] orientation of Ge, Mn can act as a surfactant while on Ge(111) it can easily diffuse into the bulk. Pursuing their idea, the authors have experimentally demonstrated that during subsequent Ge deposition on Mn/Ge(001) [17], a great part of Mn floats upwards the growth front and a small fraction of Mn atoms is incorporated in Ge capping layer. Our present results, together with a previous work [23], clearly indicates that along the (111) orientation of Ge, Mn also floats at the growth front and act as a surfactant. To prevent or to retard the out-diffusion of an element, it is common to use a diffusion barrier and materials used to make diffusion barriers must be not only non-reactive but also are able to strongly adhere to adjacent materials. In electronic or memory devices, multilayers of metals or insulators, such as tungsten nitride, RuTiN or RuTiO, are generally used to prevent the out-diffusion of dopants (B and P) or the oxidation of devices [24–26]. Such materials are,
however, difficult to be used in a heterostructure where epitaxial growth is needed. Commonly, the segregation phenomenon of an element involves a rapid and also long-range diffusion process, which, in general, occurs via interstitial mechanism. In Ge, par example, the diffusivity of metallic elements via interstitial sites can be larger by 8 to 12 orders of magnitude compared to the diffusion processes via vacancy or substitutional mechanisms, respectively [27]. In order to reduce or to suppress Mn segregation, our approach consists of filling the interstitial sites of the Mn5Ge3 lattice with carbon atoms. Indeed, since the atomic radius of carbon is almost twice smaller than that of Mn and Ge, it follows that in an adequate growth process where carbon atoms can diffuse, it becomes possible to incorporate them in interstitial sites. In a recent work, we have implemented the solid-phase epitaxy (SPE) technique to enhance incorporation of carbon into the interstitial sites of Mn5Ge3[28]. The SPE growth technique consists of co-deposition at room temperature of Mn and carbon on a Ge substrate followed by a thermal annealing. There are two interstitial sites per Mn5Ge3 unit cell and we have shown that it is possible to fill with carbon up to a saturation content x ~ 0.6, beyond which Mn–C clusters are formed. Prior to Ge over-deposition, we have then grown Mn5Ge3 layers doped with carbon. To test this approach, the carbon content was chosen to be ~ 0.4 to avoid the formation of Mn–C clusters. The Ge growth was carried out at two substrate temperatures, at 250 and 450 °C. From RHEED measurements, we clearly observe that the Mn segregation length on carbon-doped Mn5Ge3 layers is greatly reduced. For a Ge deposition at 250 °C, it is only of a few nm compared to 22.5 nm measured on carbon-free Mn5Ge3 layers. Fig.3a and b represents cross-sectional TEM images of a ~ 60 nm thick Ge layer grown on a Mn5Ge3C0.4 layer at substrate temperatures of 250 and 450 °C, respectively. Despite of the fact that the Ge overlayers are not of perfect crystalline quality due to the presence of some defects, the interface becomes much smoother and almost no Mn-rich clusters are found in the layers, as compared to the case of Ge deposition on carbon-free Mn5Ge3 layers. This result indicates that filling
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
3413
Fig. 3. Cross-sectional TEM images of a ~ 60 nm thick Ge layer grown on a Mn5Ge3C0.4 layer at a substrate temperature of 250 °C (a) and at 450 °C (b).
Mn5Ge3 interstitial sites with carbon effectively allows to greatly reduce the Mn segregation and that a complete suppression of Mn segregation probably requires a higher carbon filling level. The above results are in agreement with our recent work, which shows that adsorption of some monolayers of carbon on top of Mn5Ge3 prior to Ge deposition can prevent out-diffusion of Mn from Mn5Ge3 due to carbon occupation of the interstitial sites of the underneath Mn5Ge3 layers [29]. Fig. 4 shows the evolution of normalized magnetization as a function of the temperature of a Mn5Ge3C0.4 layer, which has been capped with ~60 nm thick Ge grown at 250 and 450 °C. For comparison, we also display the temperature evolution of magnetization of two reference samples (without capping): a carbon-free Mn5Ge3 layer and a Mn5Ge3C0.4 layer having the same thickness. The Curie temperature of the carbon-free Mn5Ge3 layer, measured at the inflection point of the curve M versus T, is of about 296 K, which is in agreement with previous works [1,2,5]. Extrapolation of the M(T) curve of the Mn5Ge3C0.4 sample gives a TC of about 380 K. The TC enhancement in carbon-doped Mn5Ge3Cx has been observed in polycrystalline films [30] and also recently confirmed in epitaxial films [28]. Such TC enhancement in carbon-doped Mn5Ge3Cx films has been attributed
to Mn–Mn interactions mediated by carbon atoms incorporated in octahedral voids of the hexagonal Mn5Ge3 cell [7]. Interestingly, the figure reveals that a further enhancement in TC is observed for Mn5Ge3C0.4 films capped with Ge at both temperatures of Ge deposition. This behaviour at present is not clearly understood. A possible explanation is that under the growth conditions performed in these experiments, all carbon atoms have not been inserted into the interstitial sites of Mn5Ge3C0.4, some of them may be accumulated near the surface region due to strain present at a free surface. Upon Ge deposition on top of the Mn5Ge3C0.4 layer, accumulated carbon clusters may be destabilized and dissolved. The released carbon can diffuse down to the underneath Mn5Ge3Cx layer, resulting in a further TC enhancement. Further characterizations, in particular by depth-profile surface sensitive techniques before and after Ge overgrowth are needed in order to obtain a better understanding of this TC enhancement.
4. Conclusion In conclusion, by investigating the epitaxial overgrowth of Ge on Mn5Ge3/Ge(111) heterostructures we were able to determine in-situ and in real time the Mn segregation length by means of RHEED. We provide evidence that along the (111) orientation of Ge, Mn also floats upwards the Ge growth front and acts as a surfactant. This feature has not been predicted in previous energy calculations reported in ref. 22. Of particular interest, we have presented an approach to prevent the segregation phenomenon, whose principle is based on filling the interstitial sites of the lattice to freeze long-range diffusion. Example with interstitial carbon has proven to be efficient but other elements, such as boron or nitrogen, can produce the same effect. Probably, a high level of interstitial filling may conduct to a complete suppression of Mn segregation. In addition, we have shown that interstitial carbon in Mn5Ge3 allows to enhance not only the Curie temperature of Mn5Ge3Cx layers but also in whole Ge/Mn5Ge3/ Ge(111) heterostructures.
Acknowledgments Fig. 4. Evolution of normalized magnetization as a function of the temperature of a Mn5Ge3C0.4 layer capped with ~ 60 nm thick Ge grown at 250 and 450 °C. For comparison, reference samples of a carbon-free Mn5Ge3 layer and of a Mn5Ge3C0.4 layer having the same thickness are shown.
A part of this work was supported by the French National Research Agency (ANR PNANO MnGe-SPIN). One of the authors, T.H Ngo, would like to thank the support of the project TRIG A funded by the Vietnamese government.
3414
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
References [1] C. Zeng, S.C. Erwin, L.C. Feldman, A.P. Li, R. Jin, Y. Song, J.R. Thompson, H.H. Weitering, Appl. Phys. Lett. 83 (2003) 5002. [2] C. Zeng, W. Zhu, S.C. Erwin, Z. Zhang, H.H. Weitering, Phys. Rev. B 70 (2004) 205340. [3] S. Picozzi, A. Continenza, A.J. Freeman, Phys. Rev. B 70 (2004) 235205. [4] A. Stroppa, M. Peressi, Phys. Status Solidi A 204 (2007) 44. [5] S. Olive-mendez, A. Spiesser, L.A. Michez, V. Le Thanh, A. Glachant, J. Derrien, T. Devillers, A. Barski, M. Jamet, Thin Solid Films 517 (2008) 191. [6] A. Spiesser, S.F. Olive-Mendez, M.-T. Dau, L.A. Michez, A. Watanabe, V. Le Thanh, A. Glachant, J. Derrien, A. Barski, M. Jamet, Thin Solid Films 518 (2010) S113. [7] I. Slipukhina, E. Arras, Ph. Mavropoulos, P. Pochet, Appl. Phys. Lett. 94 (2009) 192505. [8] P. De Padova, J.-M. Mariot, L. Favre, I. Berbezier, B. Olivieri, P. Perfetti, C. Quaresima, C. Ottaviani, A. Taleb-Ibrahimi, P. Le Fee`vre, F. Bertran, O. Heckmann, M.C. Richter, W. Ndiaye, F. D'Orazio, F. Lucari, C.M. Cacho, K. Hricovini, Surf. Sci. 605 (2011) 638. [9] E. Arras, D. Caliste, T. Deutsch, F. Lançon, P. Pochet, Phys. Rev. B 83 (2011) 174103. [10] J. Derrien, J. Chevrier, V. Le Thanh, J.E. Mahan, Appl. Surf. Sci. 56–58 (1992) 382. [11] V. Le Thanh, J. Chevrier, J. Derrien, Phys. Rev. B 46 (1992) 15946. [12] Y. Tawara, K. Sato, J. Phys. Soc. Jpn. 18 (1963) 773. [13] R.P. Panguluri, C. Zeng, H.H. Weitering, J.M. Sullivan, S.C. Erwin, B. Nadgorny, Phys. Status Solidi B 242 (2005) R67. [14] See, for example ad references therein,. V. Le Thanh, Surf. Sci. 492 (2001) 255; V. Le Thanh, Physica E 23 (2004) 401. [15] T.B. Massalki, 2nd Edition, Binary alloy phase diagrams, Vol 1, 2, ASM International, 1992.
[16] M. Stoffel, L. Simon, J.L. Bischoff, D. Aubel, L. Kubler, G. Castelein, Thin Solid Films 380 (2000) 32. [17] Ch. Zeng, Z. Zhang, K. van Benthem, M.F. Chisholm, H.H. Weitering, Phys. Rev. Lett. 100 (2008) 066101. [18] C. Renard, X. Marcadet, J. Massies, I. Prévot, R. Bisaro, P. Galtier, J. Crystal Growth 259 (2003) 69. [19] N. Ohtani, S.M. Molder, M.H. Xie, J. Zhang, B.A. Joyce, Surf. Sci. 284 (1993) 305. [20] M. Copel, M.C. Reuter, E. Kaxiras, R.M. Tromp, Phys. Rev. Lett. 63 (1989) 632. [21] N. Grandjean, J. Massies, V.H. Etgens, Phys. Rev. Lett. 69 (1992) 796. [22] W. Zhu, H.H. Weitering, E.G. Wang, E. Kaxiras, Z. Zhang, Phys. Rev. Lett. 93 (2004) 126102. [23] M.-T. Dau, A. Spiesser, T.G. Le, L.A. Michez, S.F. Olive-Mendez, V. Le Thanh, M. Petit, J.-M. Raimundo, A. Glachant, J. Derrien, Thin Solid Film 518 (2010) S266. [24] Yoon Dong-Soo, Roh Jae Sung, Lee Sung-Man, Baik Hong Koo, Prog. in Mats. Sci. 48 (2003) 275. [25] J.S. Becker, R.G. Gordon, Appl. Phys. Lett. 82 (2003) 2239. [26] C.Y. Ting, Thin Solid Films 119 (1984) 11. [27] H. Mehrer, Diffusion in Solids, Springer, Berlin, 2007. [28] A. Spiesser, I. Slipukhina, M.-T. Dau, E. Arras, V. Le Thanh, L. Michez, P. Pochet, H. Saito, S. Yuasa, M. Jamet, J. Derrien, Phys. Rev. B 84 (2011) 165203. [29] M.-T. Dau, V. Le Thanh, T.-G. Le, A. Spiesser, M. Petit, L.A. Michez, R. Daineche, Appl. Phys. Lett. 99 (2011) 151908. [30] M. Gajdzik, C. Sürgers, M.T. Kelemen, H.v. Löhneysen, J. of Magn. Magn. Mat. 221 (2000) 248.
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Suppression of Mn segregation in Ge/Mn5Ge3 heterostructures induced by interstitial carbon Minh-Tuan Dau a, Vinh Le Thanh a,⁎, Thi-Giang Le a, Aurélie Spiesser a, Matthieu Petit a, Lisa A. Michez a, Thu-Huong Ngo b, Dinh Lam Vu c, Quang Liem Nguyen c, Pierre Sebban d a
Interdisciplinary Center of Nanoscience of Marseille (CINaM-CNRS), Aix-Marseille University, Campus of Luminy, Case 913, F-13288 Marseille cedex 9, France Faculty of Physics, Hanoi University of Science, 334 Nguyen Trai Road, Hanoi, Viet Nam Institute of Materials Science, Vietnam Academy of Science and Technology (VAST), 18 Hoang Quoc Viet Str., Cau Giay, Hanoi,Viet Nam d University of Science and Technology of Hanoi (USTH), 18 Hoang Quoc Viet Str., Cau Giay, Hanoi,Viet Nam b c
a r t i c l e
i n f o
Available online 6 November 2011 Keywords: Mn5Ge3/Ge heterostructures Intermetallic magnetic compound Segregation length Interstitial diffusion Surfactants
a b s t r a c t Mn5Ge3 compound, with its room-temperature ferromagnetism and possibility to epitaxially grow on Ge, acts as a potential spin injector into group-IV semiconductors. It is shown that the realization of Ge/ Mn5Ge3 heterostructures is highly hampered by Mn segregation toward the Ge growing surface. The Mn segregation length can be estimated in-situ and in real time by means of reflection high-energy electron diffraction. We present here an approach allowing to greatly reduce or even to prevent the Mn segregation, whose principle is based on filling the Mn5Ge3 lattice with interstitial carbon atoms. In addition, we show that interstitial carbon in Mn5Ge3 allows to enhance not only the Curie temperature of Mn5Ge3Cx layers but also in the whole Ge/Mn5Ge3/Ge heterostructures. © 2011 Elsevier B.V. All rights reserved.
1. Introduction In recent years, the epitaxial growth of intermetallic Mn5Ge3 thin films has received growing interest, work in this direction is motivated by the hope to find out an efficient spin injector into group-IV semiconductors [1–9]. Indeed, the use of standard ferromagnetic metals, such as Fe, Co or Ni, directly on Si and Ge is highly hampered by their reactivity with group-IV semiconductors to form interfacial silicides or germanides, which, for most of them, are not ferromagnetic (for example, CoSi2 is metallic [10] while β-FeSi2 is semiconducting [11]). It is also not trivial to obtain epitaxial growth of an oxide in between Ge (or Si) and a ferromagnetic metal, efficiency of spin injection is therefore limited by the interface roughness and the difference in materials conductivity. The Mn5Ge3 compound exhibits numerous features, which may make it as a potential candidate for spin injection into Ge. First, this compound has a Curie temperature near room temperature [12], theoretical calculations predicted a high-spin-injection efficiency along its c axis [3] and a spin polarization up to 42% has been demonstrated [13]. Second, Ge3Mn5 films with thickness up to 165 nm can be epitaxially grown on Ge(111) substrates despite a misfit as high as 3.7% [6]. This feature is of particular importance with respect to the application of Mn5Ge3/Ge(111) heterostructures for the realisation of spintronic devices. Indeed, in a lattice mismatched heteroepitaxial system, a
⁎ Corresponding author. E-mail address: [email protected] (V. Le Thanh). 0040-6090/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2011.10.167
misfit as high as 3.7% often results in highly defected epitaxial layers. For example, Ge has a lattice parameter of ~4% larger than that of Si, pseudomorphic Ge growth is obtained only for a thickness smaller than 4 monolayers, beyond which islanding growth mode occurs [14]. For a Ge thickness of about some tens of nanometers, a threading dislocation density higher than 108 cm− 2 is usually observed and the growth front becomes extremely rough. In addition, in the Ge–Mn phase diagram [15], the Mn5Ge3 compound is not the most stable phase but thanks to its hexagonal structure, which is similar to that of the Ge(111) surface, Mn5Ge3 is the unique epitaxial phase, which has been stabilized on Ge(111) [1,2,5,6]. In this paper, we report on the effect of Mn segregation occurred during Ge overgrowth on Mn5Ge3/Ge heterostructures. During the course of our experiments to produce Ge/Mn5Ge3/Ge multilayers for the realization of spin valves and giant magneto-resistance (GMR) structures, we have observed a long-range Mn segregation upwards the Ge growth front. We show that incorporating carbon atoms into the interstitial sites of the Mn5Ge3 lattice allows to prevent outdiffusion of Mn from Mn5Ge3. In addition, we show that interstitial carbon also allows to enhance the magnetic properties of Ge/ Mn5Ge3/Ge multilayers. 2. Experimental details Materials growth was performed in a molecular beam epitaxial (MBE) system with a base pressure better than 5 × 10 − 10 mbar. The MBE chamber is equipped with a 30-keV reflection high-energy electron diffraction (RHEED) apparatus allowing to monitor in real time
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
the epitaxial growth process and the change of the film surface reconstruction. An Auger electron spectrometer is used to control the cleanliness of the substrate surface prior to growth and the film composition. Standard effusion cells were used for Mn and Ge deposition, the Mn flux was measured by a quartz crystal microbalance and the Ge growth rate was deduced from measurements of RHEED intensity oscillations. The Mn and Ge growth rates used in this work are of 0.9 and 3 nm/min, respectively. Carbon evaporation was carried out using a sublimation source of high purity pyrolytic graphite, the carbon concentration was estimated by combining two approaches: for each sublimation current, it was estimated from δ-doping curves measured on GaAs surfaces and then corrected according to the change of Si(001) surface reconstructions from (2 × 1) to c(4 × 4) upon adsorption of a submonolayer of carbon [16]. The estimated error is about 10%. The substrate temperature was measured by a thermocouple in contact with the backside of the substrate. Mn5Ge3/Ge(111) heterostructures were prepared by a room temperature Mn deposition onto Ge(111) substrates, followed by thermal annealing up to 430–450 °C, the temperature range at which the surface reconstruction characteristic of the Mn5Ge3 phase was generally observed. Structural analyses of post grown films were performed by means of high-resolution transmission electron microscopy (HRTEM) using a JEOL 3010 microscope operating at 300 kV with a spatial resolution of 1.7 Å. HRTEM cross sections were prepared by mechanical polishing, followed by ion polishing (Gatan Precision Ion Polishing System). Magnetic characterizations were carried out using superconducting quantum interference device magnetometer (SQUID) at temperatures ranging from 5 to 350 K. A magnetic field of 0.5 T is applied both in the plane of samples or perpendicular to it. The diamagnetic contributions of Ge substrates were subtracted from the measurements, leaving the magnetic contributions of grown films. More details of sample growth conditions can be found elsewhere [5,6]. 3. Results and discussion It is now well established that epitaxial Mn5Ge3 films on Ge(111) exhibit a (√3x√3)R30° surface reconstruction (denoted hereafter to as √3) with Mn termination [2]. From electron diffraction, such a reconstruction is manifested by the observation of 1 × 1 streaks along [1–10] azimuth and additional 1/3- and 2/3-ordered streaks between 1 × 1 streaks along [11-2] azimuth. The RHEED technique, thanks to its grazing incidence, allows to instantaneously reveal not only the change in the surface morphology but also the surface structure during film growth. To produce Ge/Mn5Ge3/Ge multilayers, we first grow
3411
a thin Mn5Ge3 layer on Ge(111), whose thickness ranges from 10 to 50 nm in order to insure a high crystalline quality and a smooth surface. Starting from those films, Ge overgrowth is carried out at temperatures ranging from 200 to 500 °C. Fig. 1a shows a RHEED pattern taken along [11-2] azimuth of the starting surface of a 40 nm thick Mn5Ge3 layer prior to Ge deposition. Additional 1/3- and 2/3-ordered streaks between the bulk-like 1 × 1 streaks are clearly seen, confirming the Mn-terminated Mn5Ge3 surface. The observed long streaks in the RHEED pattern unambiguously indicate that the corresponding surface is highly smooth. Displayed in Fig. 1b is an example of the RHEED pattern observed after deposition of 5 nm of Ge (corresponding to a deposition time of 100 s) at a substrate temperature of 250 °C. Apart from a certain decrease of the RHEED intensity, which signifies an increase of the surface roughness, the 1/3- and 2/3-ordered streaks still remain visible. This clearly indicates that Mn has floated upwards the Ge growth front and reacted with deposited Ge to form a surface Mn5Ge3-like layer. To quantify the Mn segregation length, we have measured the RHEED intensity evolution as a function of the Ge deposition time of the specular streak (green curve) and of a 2/3-ordered streak. As can be expected, the intensity of both the specular and 2/3 streaks is found to decrease when increasing the Ge deposition time but the specular streak exhibits a decrease more abrupt. The Mn segregation length can be determined from the intersection of the slope of these two curves with the deposition time axis. Taking into account the Ge growth rate of 3 nm/min, the Mn segregation length is found to be of ~10 nm for the specular streak and of ~ 22.5 nm for the 2/3-ordered streak. Since the specular streak is in general more sensitive to the development of the surface roughness than any diffracted streak, we can deduce a value of the Mn segregation length of ~22.5 nm for the Ge overgrowth at 250 °C. It is worth noting that Mn segregation has been reported during Ge deposition on (001)-oriented Ge substrates and a segregation length larger than 20 nm was measured at a substrate temperature as low as 150 K [17]. We also note that in-situ analyses of the RHEED intensity evolution were used to determine the segregation length of In in III–V materials [18] and of Ge during Si growth in Si/Si1-xGex heterostructures [19]. However, in those systems, the segregation length is only of some monolayers and the measurement of the segregation length was based on a RHEED intensity transient at the beginning of growth [18] or the change in the growth rate [19]. In our case, the appearance in RHEED patterns of an additional streaks characteristic of Mn5Ge3 allows a direct estimation of the Mn segregation length. The structural properties of the Ge overgrown layer on Mn5Ge3/ Ge(111) heterostructures are depicted in Fig. 2. Fig. 2a shows a typical cross-sectional TEM image of a 60 nm thick Ge layer grown on a
Fig. 1. RHEED patterns taken along the [2] azimuth: (a) of the √3 reconstruction of the Mn5Ge3 surface prior to the deposition of the Ge capping layers; (b): after deposition of 60 nm of Ge at a substrate temperature of 250 °C; and c) Evolution of the RHEED intensity of the specular streak (green curve) and a 2/3-ordered streak (read curve) with increasing the Ge deposition time. The dotted lines represent the slope of two curves.
3412
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
Fig. 2. (a) Typical cross-sectional TEM image of a 60 nm thick Ge layer grown on a Mn5Ge3/Ge(111) heterostructure at a substrate temperature of 250 °C; and (b) and (c) Crosssectional TEM images taken at interfacial regions of the corresponding sample.
Mn5Ge3/Ge(111) heterostructure at a substrate temperature of 250 °C. Two general features regarding the Ge overlayer can be noticed. First, the Ge overlayer is found to be relatively defected and contains some clusters. This implies that during Ge deposition, a part of Mn floats upwards the growth frond and acts as a surfactant while another part of Mn is incorporated inside the Ge layers, resulting in the formation of Mn-rich clusters. The incorporation of surfactant elements inside the upper grown layers is a common effect and has been observed for most of surfactant atoms during epitaxial growth [20,21]. Second, the interface between Ge overlayer and Mn5Ge3 is extremely rough as compared to the initial Mn5Ge3 surface (bi-dimensional RHEED pattern in Fig. 1a). Displayed in Fig. 2b and c are two TEM images taken around the interface regions of the corresponding sample. The interface roughness, the presence of defects as well as of Mn-rich clusters are clearly visible. These features are unambiguously a direct consequence of Mn segregation during Ge overgrowth. The phenomenon of Mn segregation in the Ge–Mn system has been previously reported [17,22,23]. Based on total energy calculations, Zhu et al. [22] predicted that along the [001] orientation of Ge, Mn can act as a surfactant while on Ge(111) it can easily diffuse into the bulk. Pursuing their idea, the authors have experimentally demonstrated that during subsequent Ge deposition on Mn/Ge(001) [17], a great part of Mn floats upwards the growth front and a small fraction of Mn atoms is incorporated in Ge capping layer. Our present results, together with a previous work [23], clearly indicates that along the (111) orientation of Ge, Mn also floats at the growth front and act as a surfactant. To prevent or to retard the out-diffusion of an element, it is common to use a diffusion barrier and materials used to make diffusion barriers must be not only non-reactive but also are able to strongly adhere to adjacent materials. In electronic or memory devices, multilayers of metals or insulators, such as tungsten nitride, RuTiN or RuTiO, are generally used to prevent the out-diffusion of dopants (B and P) or the oxidation of devices [24–26]. Such materials are,
however, difficult to be used in a heterostructure where epitaxial growth is needed. Commonly, the segregation phenomenon of an element involves a rapid and also long-range diffusion process, which, in general, occurs via interstitial mechanism. In Ge, par example, the diffusivity of metallic elements via interstitial sites can be larger by 8 to 12 orders of magnitude compared to the diffusion processes via vacancy or substitutional mechanisms, respectively [27]. In order to reduce or to suppress Mn segregation, our approach consists of filling the interstitial sites of the Mn5Ge3 lattice with carbon atoms. Indeed, since the atomic radius of carbon is almost twice smaller than that of Mn and Ge, it follows that in an adequate growth process where carbon atoms can diffuse, it becomes possible to incorporate them in interstitial sites. In a recent work, we have implemented the solid-phase epitaxy (SPE) technique to enhance incorporation of carbon into the interstitial sites of Mn5Ge3[28]. The SPE growth technique consists of co-deposition at room temperature of Mn and carbon on a Ge substrate followed by a thermal annealing. There are two interstitial sites per Mn5Ge3 unit cell and we have shown that it is possible to fill with carbon up to a saturation content x ~ 0.6, beyond which Mn–C clusters are formed. Prior to Ge over-deposition, we have then grown Mn5Ge3 layers doped with carbon. To test this approach, the carbon content was chosen to be ~ 0.4 to avoid the formation of Mn–C clusters. The Ge growth was carried out at two substrate temperatures, at 250 and 450 °C. From RHEED measurements, we clearly observe that the Mn segregation length on carbon-doped Mn5Ge3 layers is greatly reduced. For a Ge deposition at 250 °C, it is only of a few nm compared to 22.5 nm measured on carbon-free Mn5Ge3 layers. Fig.3a and b represents cross-sectional TEM images of a ~ 60 nm thick Ge layer grown on a Mn5Ge3C0.4 layer at substrate temperatures of 250 and 450 °C, respectively. Despite of the fact that the Ge overlayers are not of perfect crystalline quality due to the presence of some defects, the interface becomes much smoother and almost no Mn-rich clusters are found in the layers, as compared to the case of Ge deposition on carbon-free Mn5Ge3 layers. This result indicates that filling
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
3413
Fig. 3. Cross-sectional TEM images of a ~ 60 nm thick Ge layer grown on a Mn5Ge3C0.4 layer at a substrate temperature of 250 °C (a) and at 450 °C (b).
Mn5Ge3 interstitial sites with carbon effectively allows to greatly reduce the Mn segregation and that a complete suppression of Mn segregation probably requires a higher carbon filling level. The above results are in agreement with our recent work, which shows that adsorption of some monolayers of carbon on top of Mn5Ge3 prior to Ge deposition can prevent out-diffusion of Mn from Mn5Ge3 due to carbon occupation of the interstitial sites of the underneath Mn5Ge3 layers [29]. Fig. 4 shows the evolution of normalized magnetization as a function of the temperature of a Mn5Ge3C0.4 layer, which has been capped with ~60 nm thick Ge grown at 250 and 450 °C. For comparison, we also display the temperature evolution of magnetization of two reference samples (without capping): a carbon-free Mn5Ge3 layer and a Mn5Ge3C0.4 layer having the same thickness. The Curie temperature of the carbon-free Mn5Ge3 layer, measured at the inflection point of the curve M versus T, is of about 296 K, which is in agreement with previous works [1,2,5]. Extrapolation of the M(T) curve of the Mn5Ge3C0.4 sample gives a TC of about 380 K. The TC enhancement in carbon-doped Mn5Ge3Cx has been observed in polycrystalline films [30] and also recently confirmed in epitaxial films [28]. Such TC enhancement in carbon-doped Mn5Ge3Cx films has been attributed
to Mn–Mn interactions mediated by carbon atoms incorporated in octahedral voids of the hexagonal Mn5Ge3 cell [7]. Interestingly, the figure reveals that a further enhancement in TC is observed for Mn5Ge3C0.4 films capped with Ge at both temperatures of Ge deposition. This behaviour at present is not clearly understood. A possible explanation is that under the growth conditions performed in these experiments, all carbon atoms have not been inserted into the interstitial sites of Mn5Ge3C0.4, some of them may be accumulated near the surface region due to strain present at a free surface. Upon Ge deposition on top of the Mn5Ge3C0.4 layer, accumulated carbon clusters may be destabilized and dissolved. The released carbon can diffuse down to the underneath Mn5Ge3Cx layer, resulting in a further TC enhancement. Further characterizations, in particular by depth-profile surface sensitive techniques before and after Ge overgrowth are needed in order to obtain a better understanding of this TC enhancement.
4. Conclusion In conclusion, by investigating the epitaxial overgrowth of Ge on Mn5Ge3/Ge(111) heterostructures we were able to determine in-situ and in real time the Mn segregation length by means of RHEED. We provide evidence that along the (111) orientation of Ge, Mn also floats upwards the Ge growth front and acts as a surfactant. This feature has not been predicted in previous energy calculations reported in ref. 22. Of particular interest, we have presented an approach to prevent the segregation phenomenon, whose principle is based on filling the interstitial sites of the lattice to freeze long-range diffusion. Example with interstitial carbon has proven to be efficient but other elements, such as boron or nitrogen, can produce the same effect. Probably, a high level of interstitial filling may conduct to a complete suppression of Mn segregation. In addition, we have shown that interstitial carbon in Mn5Ge3 allows to enhance not only the Curie temperature of Mn5Ge3Cx layers but also in whole Ge/Mn5Ge3/ Ge(111) heterostructures.
Acknowledgments Fig. 4. Evolution of normalized magnetization as a function of the temperature of a Mn5Ge3C0.4 layer capped with ~ 60 nm thick Ge grown at 250 and 450 °C. For comparison, reference samples of a carbon-free Mn5Ge3 layer and of a Mn5Ge3C0.4 layer having the same thickness are shown.
A part of this work was supported by the French National Research Agency (ANR PNANO MnGe-SPIN). One of the authors, T.H Ngo, would like to thank the support of the project TRIG A funded by the Vietnamese government.
3414
M.-T. Dau et al. / Thin Solid Films 520 (2012) 3410–3414
References [1] C. Zeng, S.C. Erwin, L.C. Feldman, A.P. Li, R. Jin, Y. Song, J.R. Thompson, H.H. Weitering, Appl. Phys. Lett. 83 (2003) 5002. [2] C. Zeng, W. Zhu, S.C. Erwin, Z. Zhang, H.H. Weitering, Phys. Rev. B 70 (2004) 205340. [3] S. Picozzi, A. Continenza, A.J. Freeman, Phys. Rev. B 70 (2004) 235205. [4] A. Stroppa, M. Peressi, Phys. Status Solidi A 204 (2007) 44. [5] S. Olive-mendez, A. Spiesser, L.A. Michez, V. Le Thanh, A. Glachant, J. Derrien, T. Devillers, A. Barski, M. Jamet, Thin Solid Films 517 (2008) 191. [6] A. Spiesser, S.F. Olive-Mendez, M.-T. Dau, L.A. Michez, A. Watanabe, V. Le Thanh, A. Glachant, J. Derrien, A. Barski, M. Jamet, Thin Solid Films 518 (2010) S113. [7] I. Slipukhina, E. Arras, Ph. Mavropoulos, P. Pochet, Appl. Phys. Lett. 94 (2009) 192505. [8] P. De Padova, J.-M. Mariot, L. Favre, I. Berbezier, B. Olivieri, P. Perfetti, C. Quaresima, C. Ottaviani, A. Taleb-Ibrahimi, P. Le Fee`vre, F. Bertran, O. Heckmann, M.C. Richter, W. Ndiaye, F. D'Orazio, F. Lucari, C.M. Cacho, K. Hricovini, Surf. Sci. 605 (2011) 638. [9] E. Arras, D. Caliste, T. Deutsch, F. Lançon, P. Pochet, Phys. Rev. B 83 (2011) 174103. [10] J. Derrien, J. Chevrier, V. Le Thanh, J.E. Mahan, Appl. Surf. Sci. 56–58 (1992) 382. [11] V. Le Thanh, J. Chevrier, J. Derrien, Phys. Rev. B 46 (1992) 15946. [12] Y. Tawara, K. Sato, J. Phys. Soc. Jpn. 18 (1963) 773. [13] R.P. Panguluri, C. Zeng, H.H. Weitering, J.M. Sullivan, S.C. Erwin, B. Nadgorny, Phys. Status Solidi B 242 (2005) R67. [14] See, for example ad references therein,. V. Le Thanh, Surf. Sci. 492 (2001) 255; V. Le Thanh, Physica E 23 (2004) 401. [15] T.B. Massalki, 2nd Edition, Binary alloy phase diagrams, Vol 1, 2, ASM International, 1992.
[16] M. Stoffel, L. Simon, J.L. Bischoff, D. Aubel, L. Kubler, G. Castelein, Thin Solid Films 380 (2000) 32. [17] Ch. Zeng, Z. Zhang, K. van Benthem, M.F. Chisholm, H.H. Weitering, Phys. Rev. Lett. 100 (2008) 066101. [18] C. Renard, X. Marcadet, J. Massies, I. Prévot, R. Bisaro, P. Galtier, J. Crystal Growth 259 (2003) 69. [19] N. Ohtani, S.M. Molder, M.H. Xie, J. Zhang, B.A. Joyce, Surf. Sci. 284 (1993) 305. [20] M. Copel, M.C. Reuter, E. Kaxiras, R.M. Tromp, Phys. Rev. Lett. 63 (1989) 632. [21] N. Grandjean, J. Massies, V.H. Etgens, Phys. Rev. Lett. 69 (1992) 796. [22] W. Zhu, H.H. Weitering, E.G. Wang, E. Kaxiras, Z. Zhang, Phys. Rev. Lett. 93 (2004) 126102. [23] M.-T. Dau, A. Spiesser, T.G. Le, L.A. Michez, S.F. Olive-Mendez, V. Le Thanh, M. Petit, J.-M. Raimundo, A. Glachant, J. Derrien, Thin Solid Film 518 (2010) S266. [24] Yoon Dong-Soo, Roh Jae Sung, Lee Sung-Man, Baik Hong Koo, Prog. in Mats. Sci. 48 (2003) 275. [25] J.S. Becker, R.G. Gordon, Appl. Phys. Lett. 82 (2003) 2239. [26] C.Y. Ting, Thin Solid Films 119 (1984) 11. [27] H. Mehrer, Diffusion in Solids, Springer, Berlin, 2007. [28] A. Spiesser, I. Slipukhina, M.-T. Dau, E. Arras, V. Le Thanh, L. Michez, P. Pochet, H. Saito, S. Yuasa, M. Jamet, J. Derrien, Phys. Rev. B 84 (2011) 165203. [29] M.-T. Dau, V. Le Thanh, T.-G. Le, A. Spiesser, M. Petit, L.A. Michez, R. Daineche, Appl. Phys. Lett. 99 (2011) 151908. [30] M. Gajdzik, C. Sürgers, M.T. Kelemen, H.v. Löhneysen, J. of Magn. Magn. Mat. 221 (2000) 248.