Atomically Controlled Epitaxial Growth of Single-Crystalline Germanium Films on a Metallic Silicide
aa r X i v : . [ c ond - m a t . m t r l - s c i ] O c t Atomically controlled epitaxial growth ofsingle-crystalline germanium films on a metallicsilicide
Shinya Yamada, Kohei Tanikawa, Masanobu Miyao, and Kohei Hamaya ∗ Department of Electronics, Kyushu University, 744 Motooka, Fukuoka 819-0395, Japan
E-mail: [email protected]
Abstract
We demonstrate high-quality epitaxial germanium (Ge) films on a metallic silicide, Fe Si,grown directly on a Ge(111) substrate. Using molecular beam epitaxy techniques, we can ob-tain an artificially controlled arrangement of silicon (Si) or iron (Fe) atoms at the surface onFe Si(111). The Si-terminated Fe Si(111) surface enables us to grow two-dimensional epitax-ial Ge films whereas the Fe-terminated one causes the three-dimensional epitaxial growth ofGe films. The high-quality Ge grown on the Si-terminated surface has almost no strain, mean-ing that the Ge films are not grown on the low-temperature-grown Si buffer layer but on thelattice matched metallic Fe Si. This study will open a new way for vertical-type Ge-channeltransistors with metallic source/drain contacts. ∗ To whom correspondence should be addressed significant progresses with gate-stacking, source-drain, and thin-film chan-nel technologies have so far been reported for Ge-based metal-oxide-semiconductor filed-effect transistors (MOSFETs). Whereas the use of the Ge channel can be good solution in or-der to keep the development of complementary metal-oxide-semiconductor (CMOS) technologies,critical limitations still remain from the viewpoint of scalability because of their lateral devicestructures fabricated by the lithography processes.On the other hand, vertical-type device structures have also been proposed. To achieve ultrahigh-density nanoscale devices for three-dimensional integrated circuits, oriented epitaxial Ge nanowireswere explored.
In particular, growth and impurity doping technologies compatible with Si-based CMOS technologies have been well investigated.
However, there are still many tech-nological issues such as precise controls of crystal orientation and of the doping profile for theGe nanowires. That is, it is difficult for vertical-type Ge nanodevice structures to achieve low-resistance contacts by using an impurity doping technique. Consequently, the artificially fabricatedmetal/Ge/metal structures without impurity doping will be desirable for next-generation devices.In general, there is a large difference in crystallization energy between element semiconduc-tors such as Si or Ge and metals because of the difference in bonding energy between covalent andmetallic bonds. Also, Si or Ge essentially requires relatively high thermal energy to crystallize,whereas metals can crystallize with a relatively low one. As a result, if one tries to form high-quality Si or Ge on a metal in conventional growth conditions, the formation of other compoundssuch as silicide or germanide cannot be prevented. Up to now, to realize Si-based three-dimensionalintegrated devices, the crystal growth of Si/metal/Si vertical double heterostructures has alreadybeen studied.
To overcome the above issues in the crystallization of Si, complicated meth-ods including the combination of molecular beam epitaxy (MBE), solid phase epitaxy (SPE), andseveral annealing techniques were explored intensively.
Considering an achievement of metal/Ge/metal structures, we should first form high-qualitysingle-crystalline Ge films on a metal. In this paper, we demonstrate a high-quality Ge film on a2etallic silicide grown on Ge(111). Actually, this structure is Ge/metal/Ge as well as Si/metal/Sireported so far. However, this structure can be achieved only by using an MBE technique with anartificially controlled arrangement of the surface atoms on the metallic silicide. This study willopen a new way for vertical-type Ge-channel transistors with metallic source/drain contacts.To realize epitaxial growth of Ge films on a metal, we focus on one of the silicide compounds,Fe Si, as a metallic material in this study. The positive reasons are as follows. First, wehave so far developed single-crystalline Fe Si on Si or Ge, indicating the low electrical re-sistivity and good compatibility for the SiGe technologies. Second, we have well understood themechanism of the crystal growth of high-quality Fe Si films on Ge. Since the formed Fe Si/Geheterointerface is atomically flat, the surface of the grown Fe Si is also atomically smooth, leadingto a good condition for the growth of Ge films. Finally, the atomically flat Fe Si/Ge junctions havepositive possibilities to solve the Fermi-level pinning problems at metal/Ge interfaces.
Theseare based on the controlled molecular beam epitaxy (MBE) techniques for a metal/Ge interface.The above features arise from a special condition between Fe Si and Ge. Figure 1a illustratescrystal structures of the ideal ( D -ordered) Fe Si and Ge, which are bcc and diamond structures,respectively, where the lattice mismatch between Fe Si (0.565 nm) and Ge (0.565 nm) is almostzero. When we look at the atomic arrangement at the (111) plane, the ideal Fe Si has a periodicalstacking structure consisting of three Fe layers and one Si layer as shown in Figure 1b. Fortunately,the atomic arrangements between Fe Si(111) and Ge(111) are well matched. By utilizing thespecial conditions, we have demonstrated the two-dimensional epitaxial growth of Fe Si films onGe(111) with extremely high-quality heterointerfaces.
Using the surface of the Fe Si/Ge(111) structure, we explore epitaxial growth of Ge films. Thefollowing is the detailed procedure for the fabrication of the Ge( ∼
100 nm)/Fe Si( ∼
25 nm)/Ge(111)heterostructures by using MBE techniques. Prior to the growth, we chemically cleaned nondopedGe(111) substrates ( r ∼ W cm, sample size: 2 × ) using 1 % HF solution to removecontamination and native oxide from the surface. The cleaned substrates were loaded immediatelyinto an ultra high vacuum chamber with a base pressure of ∼ − Pa. After the heat treatment at3 (a)(b)
Ge(111) substrate Ge D -Fe Si Fe Si disordered Si-terminated Fe Si layer ...
GeFeFeFeSiGe
DisorderDisorder DisorderDisorder
Fe-terminated ...
Figure 1: (a) The crystal structures of D -ordered Fe Si and Ge. (b) Schematic diagrams of theatomic arrangements of our as-grown Fe Si films on Ge(111) (lower) and the concept with anartificially controlled arrangement of the surface atoms on Fe Si (upper). o C400 o CGe layer disordered Si-terminated o CFe Si layer
Fe-terminated
Figure 2: RHEED patterns for the grown Ge films (upper) and the various surfaces of the Fe Sifilms (lower), observed along [211] azimuth. 450 ◦ C for 20 min, the substrate temperature was reduced down to 130 ◦ C. After a reflectionhigh energy electron diffraction (RHEED) pattern of the surface of the Ge(111) substrate showedit to be atomically smooth, we grew Fe Si films directly on the Ge(111) substrate by coevaporatingFe and Si using Knudsen cells. The deposition rate of Fe and Si is 2.1 nm/min and 1.2 nm/min,respectively. In-situ RHEED patterns of the Fe Si layers clearly exhibited symmetrical streaks,indicating good two-dimensional epitaxial growth (see the lower left RHEED pattern in Figure 2).Although the D -ordered structure should have an Fe or a Si atomic layer on the top of Fe Si, theactual top layer of the grown Fe Si consists of the mixed layer with Fe and Si atoms because ofsome structural disorder (see upper left in Figure 1b). However, an atomically smooth surfaceis guaranteed even for the mixed layer as shown in the lower left in Figure 2 (disordered surface).As a preliminary experiment, we try to grow Ge films (deposition rate: 0.3 nm/min) on the top ofthe disordered Fe Si with increasing the growth temperature from 200 to 400 ◦ C. The RHEEDpattern during the growth is weekly spotty [see arrows in Figure 2 (upper left)], indicating that theactual top layer of the grown Fe Si, consisting of the mixed layer with Fe and Si, cannot defendthe three-dimensional epitaxial growth of Ge films.Considering the above preliminary data, for the growth of high-quality Ge films on a metal,we suggest a new concept with an artificially controlled arrangement of the surface atoms onFe Si. Since there is a Si atomic layer in the ideal ( D -ordered) Fe Si at the (111) plane, wecan artificially form a Si(111) atomic layer on the disordered surface of the grown Fe Si(111) byprecisely controlling the evaporation of Si atoms. Fortunately, even if the surface of the Fe Silayers is terminated with a few Si atomic layers, the RHEED pattern can still show streaks asdisplayed in the lower center of Figure 2. We note that there is almost no difference in the RHEEDpatterns between disordered and Si-terminated surfaces. We hereafter define the atomically smoothsurface shown as the Si-terminated surface. When we formed a 2-nm-thick Si layer (six Si atomiclayers) on Fe Si(111) at several temperatures from 130 to 350 ◦ C, the RHEED pattern darkenedand disappeared, indicating that we could not crystallize the Si layers on Fe Si(111). Thus, we havealready confirmed that there is a large difference in the surface quality between the Si-terminated5urface and 2-nm-thick Si layer on Fe Si(111). Using the Si-terminated surface, we grow Ge layerson Fe Si. Interestingly, we can demonstrate two-dimensional epitaxial growth of the Ge layer(see the upper center in Figure 2) even with the same conditions shown in the above experiment.During the growth, we hardly observed the change in the RHEED pattern. As a result, a 100-nm-thick Ge epilayer can be grown even on a metallic silicide. In order to confirm the effect ofthe Si termination, we also investigate the growth for the Fe-terminated surface, where the surfaceof the Fe Si layers is terminated with a few Fe atomic layers as shown in the upper right panelin Figure 1b. Although the Fe-terminated surface is also atomically flat, we cannot demonstratetwo-dimensional epitaxially grown Ge layer (see the right RHEED patterns in Figure 2). Notethat these features were well reproduced. Therefore, we conclude that the Si termination of thedisordered Fe Si layers is very important to obtain the high-quality Ge films on Fe Si(111).To examine the structural characteristics of the grown Ge films on Fe Si, we observed cross-sectional transmission electron micrographs (TEM) and nanobeam electron diffraction (NED) pat-terns. In Figure 3b, we can see a high-quality Ge layer grown uniformly on Fe Si. The latticeimages of the grown Ge layer are very clear. Although the Ge layer was grown directly on Fe Si,Fe was not detected surprisingly in the Ge layer from the energy dispersive X-ray spectroscopy(EDX) measurement, indicating no atomic interdiffusion between Ge and Fe Si. Also, the NEDpattern of the grown Ge layer (point Si junc-tion clearly shows an atomically smooth heterointerface (Figure 3c). It is just as the artificiallycontrolled heteroepitaxy of the next-generation semiconductor, Ge. We also show the surface mor-phology of the grown Ge layer by dynamic force mode atomic force microscopy (AFM) in Figure3d. Since a very smooth surface with rms roughness of ∼ Si, easilyenabling us to achieve metallic silicide/Ge/metallic silicide vertical structures.On the other hand, there are lots of stacking faults in the Ge layer on the Fe-terminated surface6 c)(e) (f)(b) Ge layer (semiconductor)Ge(111) sub.
10 nm Fe Si (metal) Si
Si-terminated sampleFe-terminated sample [ μ m] [nm] (d)(h) 2 nm stacking faults Ge layerGe(111) sub.Fe Si
10 nm
Fe-terminated Fe Si Figure 3: Cross-sectional TEM images of the epitaxial Ge/metallic Fe Si/Ge(111) heterostructurefor the growth on the Si-terminated Fe Si (b,c) and those on the Fe-terminated Fe Si (f-h). NEDpatterns of the epitaxial Ge film and the Ge substrate for (a) Si- and (e) Fe-terminated samples.The zone axis of the incident electron beam is parallel to [110] direction. (d) AFM image of thesurface of the Ge film grown on the Si-terminated Fe Si.7Figure 3f). NED pattern measured at point Si(111), as shown in Figure 3g (also see the areas marked by reddashed lines in Figure 3f). An enlarged TEM image of the stacking faults in the Ge layer is shownin Figure 3h. We can confirm that the stacking faults in the Ge layer occur at the position nearthe amorphous phases. Interestingly, Fe was also not detected in the Ge layer from the EDXmeasurement, indicating no interdiffusion between Ge and Fe Si even on the Fe-terminated Fe Si.From the results obtained, we found that a formation of some germanide compounds (Fe-Ge)was not induced when we grew Ge films on the Fe Si(111) surface at a low growth temperature( ∼ ◦ C). On the contrary, compared to the Si-terminated surface, some parts of Ge films werenot sufficiently crystallized on the Fe-terminated surface. We can speculate that, compared to theFe-terminated surface, the Si-terminated one reduces the crystallization temperature of Ge and/orenhances the surface migration of Ge on the Fe Si(111), which is caused by the difference in thebonding energy between Ge-Si bond and Ge-Fe bond. Although further study is required to clearlyunderstand the growth mechanism of Ge on Fe Si, we find that the Si-terminated surface on theFe Si(111) is quite effective to realize the low-temperature crystallization of Ge.A wide-area crystal orientation of the grown Ge layer was evaluated by electron backscatteringdiffraction (EBSD). Figure 4a,b shows a scanning electron micrograph (SEM) (sample area: 250 × m m ) and an EBSD (sample area: 50 × m m ) image, respectively, of the Ge layer grownon the Si-terminated Fe Si. The surface structure of the grown Ge layer is almost uniform and thecrystal orientation parallel to the growth direction is kept in the <111> direction over the entiremeasured area, indicating that a wide-area single-crystalline Ge(111) film was achieved even on ametal.To further evaluate the lattice strain in the Ge layer grown at the (111) plane, we measuredmicroprobe Raman spectra (spot size: ∼ m m f , excitation laser wavelength: 532 nm, effectiveresolution: ∼ − ) of the 100-nm-thick Ge layer grown on Fe Si. Figure 5 displays a room-temperature Raman spectrum of the Ge layer grown on the Si-terminated Fe Si, together with8 μ m (a) (b) μ m Figure 4: (a) SEM and (b) EBSD images of the Ge film grown on the Si-terminated Fe Si.that of the homoepitaxial Ge layer grown on Ge(111) at 400 ◦ C. For the Ge layer grown on theSi-terminated surface, a sharp peak originating from Ge-Ge bonding is clearly observed at ∼ − . The peak position of the homoepitaxial Ge layer is also ∼
302 cm − . Comparing these twodata, we can recognize that there is almost no lattice strain even for the Ge layer grown on Fe Si.By estimating the lattice spacing from the NED patterns in Figure 3a, we also confirmed that thelattice constant of the Ge layer grown on the Si-terminated Fe Si is almost equal to that of bulkGe. These features exactly support that there is no lattice mismatch between Fe Si and Ge, and theSi-terminated surface on Fe Si is not a low-temperature-grown Si buffer layer on Fe Si(111) but apart of a Si atomic layer in Fe Si(111). Hence, the well-matched atomic arrangement between theSi atomic layer in Fe Si and Ge contributes largely to the quality of the grown Ge layer.We finally comment on the full-width at half maximum (FWHM) value of the Raman spectra.For the Ge layer grown on Si-terminated Fe Si, the FWHM value was ∼ − . We plot theFWHM values for the Ge layer grown on the Si- or Fe-terminated Fe Si and those of various Gecrystallines grown by the different methods in the inset of Figure 5. Although the FWHMvalue for the Ge layer grown on the Si-terminated Fe Si is larger than those for the Ge films grownon an insulator by a liquid-phase epitaxy ( ≤ − ), it is at least smaller than that for the Ge9 n t e n s it y ( a r b . un it s ) Raman shift (cm ) -1 Si-terminatedGe/Ge(111) F W H M ( c m - ) o C) c-Ge Ref. 30Ref. 29Ref. 31Ge/Ge(111)Si-terminated RT Fe-terminated
Figure 5: Raman spectra of the 100-nm-thick Ge film grown on the Si-terminated Fe Si and thehomoepitaxial Ge film grown on Ge(111) at 400 ◦ C, measured at room temperature. The insetshows the FWHM values of various Ge crystallines grown by the different methods as afunction of growth temperature.layer grown on the Fe-terminated Fe Si ( ∼ − ) and those for the crystalline Ge fabricated byother low-temperature growth techniques such as metal-induced lateral crystallization (MILC) and solid-phase crystallization. Considering these facts, we can also say that high-quality Gelayers were obtained even on a metallic silicide. Since we can easily fabricate Fe Si layers onthe grown Ge layer, metallic silicide/Ge/metallic silicide vertical structures can also be eas-ily achieved. From these results, this study will open a new way for vertical-type Ge-channeltransistors with metallic source/drain contacts.In summary, we demonstrated high-quality single-crystalline Ge films on a metallic silicide,Fe Si, by individually developing a novel growth technique. Using molecular beam epitaxy tech-niques, we obtained an artificially controlled surface structure terminated with a few Si or Featomic layers at the Fe Si(111) plane. Only the Si-terminated Fe Si surface enabled us to growtwo-dimensional epitaxial Ge films. The high-quality Ge films grown have almost no strain, mean-ing that the Ge films are not grown on the low-temperature-grown Si buffer layer but on the wellmatched Fe Si. This technique can be applied for vertical-type Ge-channel transistors with metal-lic source/drain contacts. 10 cknowledgement
The authors would like to thank Prof. T. Asano and Prof. T. Kimura of Kyushu University for theirexperimental support. This work was partly supported by Grant-in-Aid for Young Scientists (A)from The Japan Society for the Promotion of Science (JSPS) and Industrial Technology ResearchGrant Program from NEDO. S.Y. acknowledges JSPS Research Fellowships for Young Scientists.11 eferences (1) Tezuka, T.; Nakaharai, S.; Moriyama, Y.; Sugiyama, N.; Takagi, S.;
IEEE Electron DeviceLett. , , 243-245.(2) Miyao, M.; Toko, K.; Tanaka, T.; Sadoh, T.; Appl. Phys. Lett. , , 022115.(3) Ritenour, A.; Khakifirooz, A.; Antoniadis, D. A.; Lei, R. Z.; Tsai, W.; Dimoulas, A.; Mavrou,G.; Panayiotatos, Y.; Appl. Phys. Lett. , , 132107.(4) (a) Lee, C. H.; Nishimura, T.; Tabata, T.; Wang, S. K.; Nagashio, K.; Kita, K.; Toriumi, A.; IEDM Tech. Dig. , 416-419. (b) Nishimura, T.; Lee, C. H.; Tabata, T.; Wang, S. K.;Nagashio, K; Kita, K.; Toriumi, A.;
Appl. Phys. Express , , 064201.(5) Hirayama, K.; Ueno, R.; Iwamura, Y.; Yoshino, K.; Wang, D.; Yang, H.; Nakashima, H.; Jpn.J. Appl. Phys. , , 04DA10.(6) Nakaharai, S.; Tezuka, T.; Sugiyama, N.; Moriyama, Y.; Takagi, S.; Appl. Phys. Lett. , , 3516-3518.(7) Dimoulas, A.; Tsipas, P.; Sotiropoulos, A;. Evangelou, E. K.; Appl. Phys. Lett. , ,252110.(8) Takeuchi, S.; Shimura, Y.; Nakatsuka, O.; Zaima, S.; Ogawa, M.; Sakai, A.; Appl. Phys. Lett. , , 231916.(9) Nishimura, T.; Kita, K.; Toriumi, A.; Appl. Phys. Express , 051406.(10) Yamane, K.; Hamaya, K.; Ando, Y.; Enomoto, Y.; Yamamoto, K.; Sadoh, T.; Miyao, M.; Appl. Phys. lett. , , 162104.(11) Kasahara, K.; Yamada, S.; Sawano, K,; Miyao, M.; Hamaya, K.; Phys. Rev. B , ,205301.(12) Ikeda, K.; Maeda, T.; Takagi, S.: Thin Solid Films , , 359-362.1213) Myronov, M.; Sawano, K. Shiraki. Y.; Mouri, T.; Itoh, K. M. Appl. Phys. Lett. , ,082108.(14) Nakamura, Y.; Murayama, A.; Ichikawa, M.; Crystal Growth & Design , , 3301-3305.(15) Hoshi, Y.; Sawano, K.; Hamaya, K.; Miyao, M.; Shiraki, Y.; Appl. Phys. Express , ,015701.(16) (a) Adhikari, H.; Marshall, A. F.; Chidsey, C. E. D.; Mclntyre, P. C.; Nano Lett. , ,318-323. (b) Woodruff, J. H.; Ratchford, J. B.; Goldthorpe, I. A.; Mclntyre, P. C.; Chidsey,C. E. D. Nano Lett. , , 1637-1642.(17) Kamins, T. I. and Li, X. and Williams, R. Stanley and Liu, X.; Nano Lett. , , 503-506.(18) Fukata, N.;Sato, K.; Mitome, M.; Bando, Y.; Sekiguchi, T.; Kirkham, M.; Hong, J.; Wang, Z.L.; Snyder, R. L.; ACS Nano , , , 3807-3816.(19) Saitoh, S.; Ishiwara, H.; Furukawa, S.; Appl. Phys. Lett. , , 203-205.(20) Bean, C. J.; Poate, M. J.; Appl. Phys. Lett. , , 643-646.(21) (a) Ohsima, T.; Nakamura, N.; Nakagawa, K.; Miyao, M.; Thin Solid Films , , 275-282. (b) Miyao, M.; Nakagawa, K.; Nakamura, N.; Ohshima, T.; J. Crystal Growth , , 957-960.(22) Hines, W. A.; Menotti, A. H. ; Budnick, J. I. ; Burch, T. J.; Litrenta, T.; Niculescu, V.; Raj,K.; Phys. Rev. B , , 4060-4068.(23) Niculescu, V. A., Burch, T. J.; Budnick, J. I.; J. Magn. Magn. Mater. , , 223-267.(24) Bansil, A.; Kaprzyk, S.; Mijnarends, P. E.; Tobola, J.; Phys. Rev. B , , 13396-13412.(25) Ionescu, A.; Vaz, C. A. F.; Trypiniotis, T.; Gürtler, C. M.; García-Miquel, H.; Bland, J. A. C.;Vickers, M. E.; Dalgliesh, R. M.; Langridge, S.; Bugoslavsky, Y.; Miyoshi, Y.; Cohen, L. F. ;Ziebeck, K. R. A.; Phys. Rev. B , , 094401.1326) (a) Sadoh, T.; Kumano, M.; Kizuka, R.; Ueda, K.; Kenjo, A.; Miyao, M.; Appl. Phys. Lett. , , 182511. (b) Hamaya, K.; Ueda, K.; Kishi, Y.; Ando, Y.; Sadoh, T.; Miyao, M.; Appl. Phys. Lett. , , 132117.(27) (a) Hamaya, K.; Murakami, T.; Yamada, S.; Mibu K.; Miyao, M. Phys. Rev. B , ,144411. (b) Hamaya, K.; Ando, Y.; Sadoh, T.; Miyao, M.; Jpn. J. Appl. Phys. , ,010101.(28) Ueda, K.; Ando, Y.; Kumano, M., Sadoh, T., Maeda, Y.; Miyao, M.; Appl. Surf. Sci. , , 6215-6217.(29) Kanno, H.; Toko, K.; Sadoh, T.; Miyao, M.; Appl. Phys. Lett. , , 182120.(30) Toko, K.; Nakao, I.; Sadoh, T.; Noguchi, T.; Miyao, M.; Solid-State Electron , , 1159-1164.(31) Toko, K.; Kurosawa, M.; Yokoyama, H.; Kawabata, N.; Sakane, T.; Ohta, Y.; Tanaka, T.;Sadoh, T.; Miyao, M.; Appl. Phys. Express3