文档库

最新最全的文档下载
当前位置:文档库 > Ge thin film growth on Si(111) surface using hydrogen surfactant

Ge thin film growth on Si(111) surface using hydrogen surfactant

Ge thin ?lm growth on Si(111)surface using hydrogen surfactant

Toshiaki Fujino *,Takashi Fuse,Jeong-Tak Ryu,Katsuhiko Inudzuka,Toshiaki Nakano,

Koji Goto,Yujin Yamazaki,Mitsuhiro Katayama,Kenjiro Oura

Department of Electronic Engineering,Graduate School of Engineering,Osaka University,2-1Yamadaoka,Suita,Osaka 565-0871,Japan

Abstract

We have investigated the atomic hydrogen (H)-surfactant mediated growth of Ge on Si(111)surface,using coaxial impact-collision ion scattering spectroscopy (CAICISS),time-of-ˉight elastic recoil detection analysis (TOF-ERDA)and scanning electron microscopy (SEM).It has been found that the Ge thin ?lm on the Si(111)1£1-H surface is ˉattened by the H-surfactant,whilst on the Si(111)7£7surface the ˉatness does not change in spite of supplying H.These results indicate that the ˉatness of the Ge thin ?lm is strongly inˉuenced by the structure of the Si(111)substrate surface at the initial stage of Ge thin ?lm growth.q 2000Elsevier Science S.A.All rights reserved.

Keywords :Ion scattering spectroscopy;Elastic recoil detection analysis;Silicon;Germanium;Hydrogen;Surfactant

1.Introduction

In the heteroepitaxial growth of Ge on Si,it is known that covering the Si substrate with a small amount of foreign species changes the growth mode from Stranski±Krastanov (S±K)mode to layer-by-layer growth.This is called surfac-tant-mediated-epitaxy (SME)[1±6].Several elements,such as Ga,As and Bi demonstrate a surfactant effect.However,a certain fraction of these surfactant atoms becomes incorpo-rated in the grown ?lm during SME growth,so that they act as undesirable dopants.It has recently been reported that atomic hydrogen (H)also has a surfactant effect,and it is most likely that H atoms might not cause detrimental effects on the electric properties of the grown layers in contrast to the other surfactants.The successful use of the dynamically supplied atomic hydrogen as a surfactant has been reported for the Ge/Si heteroepitaxy [7±10].With the high ˉux density of H,the island formation has been suppressed and layer-by-layer growth of Ge on Si has been achieved.However,the mechanism of the H-surfactant effect has not yet been fully clari?ed.In particular,the inˉuence of the initial Si(111)surface structure on the morphology of Ge thin ?lm has not been established.

In this study,using coaxial impact-collision ion scattering spectroscopy (CAICISS)[11,12],time-of-ˉight elastic recoil detection analysis (TOF-ERDA)and scanning elec-tron microscopy (SEM),we have investigated the H-surfac-tant mediated growth of Ge on Si(111).The CAICISS and TOF-ERDA have been proved to be useful techniques for in-situ monitoring of surface processes in real time [13±16]by virtue of suf?ciently large ion scattering and recoiling cross-sections due to the low energy of the primary ion http://www.wendangku.net/doc/76b018dd360cba1aa811dacf.htmling these techniques,we can perform in-situ obser-vation of the surface roughness simultaneously with the quantitative analysis of hydrogen atoms on the surface,which is the advantage of CAICISS and TOF-ERDA over microscopies such as STM,AFM and SEM.It has been revealed that the initial structure of the Si(111)substrate surface inˉuences the morphology of Ge thin ?lm much more than the growth conditions such as H ˉux or substrate temperature.This result is different from that found for the Ge growth on Si(001)[17].2.Experimental

Experiments were carried out in an ultrahigh-vacuum (UHV)chamber with a base pressure of 1.0£10210Torr equipped with CAICISS/TOF-ERDA.Since the apparatus has been described elsewhere [11,15],only an outline is given in this paper.A pulsed beam of 2keV He 1impinged on the sample.In CAICISS measurements,the scattered He particles (He 1and He)were detected at a scattering angle of 1808(perfect backscattering)using a time-of-ˉight (TOF)analyzer.The CAICISS spectrum consists of sharp peaks due to the single scattering from individual elements and broad features due to the multiple scattering.In TOF-ERDA measurements,recoiled particles were detected at a recoil

Thin Solid Films 369(2000)

Ge thin film growth on Si(111) surface using hydrogen surfactant

25±28

0040-6090/00/$-see front matter q 2000Elsevier Science S.A.All rights reserved.PII:S0040-6090(00)00828-2

Ge thin film growth on Si(111) surface using hydrogen surfactant

http://www.wendangku.net/doc/76b018dd360cba1aa811dacf.html/locate/tsf

*Corresponding author.Tel.:181-6-6879-7779;fax:181-6-6879-7780.E-mail address:fujino@ele.eng.osaka-u.ac.jp (T.Fujino).

angle of268using another TOF analyzer.The TOF-ERDA measurements are valid for the quantitative analysis of the surface hydrogen[15±17].We also used SEM(TopoMetrix, TMX-2100)for ex-situ observation of the surface morphol-ogy of Ge thin?lm.

We prepared two kinds of Si substrate surfaces for Ge growth;the

Ge thin film growth on Si(111) surface using hydrogen surfactant

Ge thin film growth on Si(111) surface using hydrogen surfactant

Si(111)7£7and Si(111)1£1-H surfaces.A

Si(111)7£7clean surface was prepared in-situ byˉash

heating at12008C using electron bombardment from

behind.The Si(111)1£1-H surface was prepared by wet-

chemical method as follows.First,organic and metal

contaminants were removed by the RCA method.Then,

the native oxide layer was removed by dipping the sample

into a NH4F solution.It is well known that the surface

etched by HF or NH4F is covered with1£1monohydride

[18,19].

Ge was deposited from a boron nitride Knudsen cell at a

rate of0.9ML/min(1ML 7:8£1014atoms/cm2)onto the

substrate held at3508C.The temperature of the sample was

measured by a digital pyrometer.The thicknesses of Ge

?lms were measured by a quartz microbalance.To supply

atomic hydrogen(H)onto the sample,molecular hydrogen

(H2)was introduced through a molybdenum tube,on the end

of which a tungsten?lament heated at16008C was mounted

3cm away from the sample to dissociate H2.The Hˉux was

estimated from the TOF-ERDA measurement to be about

5.0ML/min when the pressure of H2was1.0£1026Torr

[17].During Ge deposition onto Si(111),the continuousˉux

of H,which was controlled by the pressure of H2in the

chamber,was dynamically supplied.The timing of starting

the exposure of Si(111)surface to H and Ge atoms is simul-

taneous.

3.Results and discussion

Fig.1shows the evolution of the CAICISS spectra with

increasing Ge coverage for the Ge?lm growth on the

Si(111)1£1-H surface.The direction of the primary ion

beam was?xed at incidence angle of908from the surface

(normal incidence).In these spectra there are two peaks at

4810and5220ns,corresponding to the single scattering

from Ge and Si atoms,respectively.The broad structure at

the lower energy side of Ge single scattering peak is due to

multiple scattering in the Ge overlayer and it becomes larger

as the Ge growth proceeds.From the Ge multiple scattering

intensity denoted by the hatched area in Fig.1,one can

estimate theˉatness of Ge overlayer because theˉatter

one shows the larger intensity[13].Fig.2a shows the Ge

multiple scattering intensities as a function of Ge coverage

for the Ge overlayers grown on the Si(111)7£7surface.As

one can see in Fig.2a,the Ge multiple scattering intensities

for the Ge thin?lm grown without H are identical to those

for Ge?lms grown with H even at the H2pressure as high as

2.0£1026Torr,indicating that the Ge overlayer is notˉat-

tened due to H-surfactant.This result is different from the

case of the Ge growth on Si(001)[17].On the other hand,

for the Ge?lms grown on Si(111)1£1-H even without

supplying H,the Ge multiple scattering intensities are larger

than those for the Ge?lms grown on Si(111)7£7(see Fig.

2b).These results suggest that the Ge thin?lm growth on

Si(111)is strongly dependent on the initial structure of the

Si(111)surface.Furthermore,the Ge multiple scattering

intensities for the Ge?lms grown on the Si(111)1£1-H

with supplying H simultaneously at a H2pressure of

1.0£1026Torr is further increased,indicating that the Ge

overlayer becomesˉatter due to H-surfactant effect.

The experimental results mentioned above are consistent

with our SEM observations.Figs.3a,b show the surface

after50ML Ge deposition on Si(111)7£7without H and

that on Si(111)1£1-H conducted at a H2pressure of

1.0£1026Torr,respectively.In Fig.3b,the size of Ge

islands is found to be smaller and their number density to

T.Fujino et al./Thin Solid Films369(2000)25±28

26

Fig.1.The evolution of the CAICISS spectra for Ge grown on a

Si(111)1£1-H surface at a growth temperature of3508C without supplying

H.The hatched area indicates the energy window for multiple scattering

intensity due to Ge atoms.

Fig.2.Intensities of multiple scattering as a function of Ge coverage for Ge

overlayers grown on(a)Si(111)7£7and(b)Si(111)1£1-H at a growth

temperature of3508C with various Hˉuxes(expressed by a H2pressure).

be higher than those in Fig.3a,indicating that the ˉatness of Ge thin ?lm on the Si(111)1£1-H with H is improved.In order to clarify the behavior of H during Ge growth,we have measured the amount of H on the growth front.Fig.4a shows TOF-ERDA spectra for the Ge overlayer grown on Si(111)1£1-H with H.For comparison,the spectrum for the Si(111)1£1-H surface is also shown,which is well known to contain 1ML of hydrogen [19].These TOF-ERDA measurements were carried out after intermitting the supply of H and Ge and cooling down the samples to room temperature.There are two peaks at 4350and 5000ns,corresponding to the recoiled H and forward scattering of He particles from Ge and Si atoms.Fig.4b shows hydrogen coverages measured by TOF-ERDA as a function of Ge coverage for the Ge grown on Si(111)1£1-H,with and without H.As one can see,at Ge coverages exceeding 10ML the negligible amount of hydrogen is left on the surface.However,the H-surfactant effect is still observed in Fig.2b.This suggests that similar to the case of the H-assisted Ge growth on Si(001),continuously supplied H atoms inˉuence dynamically the growth of Ge;H atoms adsorb on the surface for a short residence time,act as surfactant and then desorb from the surface.

Sakai et al.[7]reported the results on the Ge growth on Si(001)and Si(111)using H-surfactant.In their work,the layered structure of Ge with a much ˉatter surface morphol-ogy was achieved not only on Si(001)but also on Si(111)using H-surfactant at a H 2pressure of 1.0£1024Torr.The reason why their experimental result is different from the present one is thought to be connected with the difference in the density of H ˉux;the H ˉux in their experiments was about two order higher than that in our experiments.On the other hand,in our previous study of the H-assisted Ge growth on Si(001),the H-surfactant effect is observed at a H 2pressure of 1.0£1026Torr [17].This is explained by the fact that the sticking coef?cient of H on Si(111)surface is lower than that on Si(001)surface,which is found from TOF-ERDA measurements (not shown).

The results of our present study demonstrates that the Ge thin ?lms grown on the Si(111)1£1-H surface are ˉatter than those on the Si(111)7£7surface.It is well known that Si(111)1£1-H surface prepared by NH 4F treatment is atomically ˉat over a large domain.However,the improve-ment of the ˉatness of Ge ?lm on this surface should not result from the initial ˉatness of the substrate,since the surface migration of Ge should be enhanced in this case,leading to the formation of large islands.Therefore,it is most likely that the promotion of layer-by-layer growth on the Si(111)1£1-H surface may stem from the presence of H-surfactant and the initial 1£1surface structure.It is inferred that the 7£7dimer adatom stacking-fault (DAS)structure [20]needs to turn into the bulk like 1£1structure

T.Fujino et al./Thin Solid Films 369(2000)25±28

Ge thin film growth on Si(111) surface using hydrogen surfactant

27

Fig.4.(a)TOF-ERDA for the Si(111)1£1-H surface (corresponds to 1ML of H)and for Ge layers deposited on Si(111)1£1-H at 3508C with H at a H 2pressure of 1.0£1026Torr.(b)Hydrogen coverages measured by TOF-ERDA as a function of Ge coverage for Ge grown on Si(111)1£1-H at 3508C without and with H at a H 2pressure of 1.0£1026

Ge thin film growth on Si(111) surface using hydrogen surfactant

Torr.

Fig.3.The SEM images for 50ML of Ge grown (a)on Si(111)7£7without H and (b)on Si(111)1£1-H with H at a H 2pressure of 1.0£1026Torr.Both of the growth temperatures are 3508C.

for the layer-by-layer growth of Ge on Si(111)surface[21], but the high energy barrier for this rearrangement leads to the disruption of layer-by-layer growth of Ge,even when H-surfactant is used.This is nothing but the reason for the dif?culty of forming theˉat Ge thin?lm on the Si(111)7£7surface as mentioned above.Consequently, preparing the Si(111)1£1-H surface is of great importance for promoting the layer-by-layer growth of Ge during subse-quent Ge deposition.

4.Conclusion

We have investigated the Ge thin?lm growth on Si(111) surface with H-surfactant.It has been found from the CAICISS measurements that the Ge growth on Si(111)is strongly dependent on the initial structure of the substrate surface.Similar to the Ge growth on Si(001),the H-surfac-tant effect is observed in the case of the Ge growth on the Si(111)1£1-H surface;theˉatness of the Ge overlayer is improved.From the TOF-ERDA measurements,it has been revealed that the continuously supplied H atoms inˉuence the growth of Ge dynamically.From these results,we can conclude that the Si(111)1£1-H surface is more suitable for the H-surfactant mediated layer-by-layer growth of Ge than the Si(111)7£7surface.

Acknowledgements

This work was supported by a Grant-in-Aid for Scienti?c Research from the Ministry of Education,Science,Sports and Culture.References

[1]M.Copel,M.C.Reuter,E.Kaxiras,R.M.Tromp,Phys.Rev.Lett.63

(1989)632.

[2]M.Copel,M.C.Reuter,M.Horn-von Hoegen,R.M.Tromp,Phys.

Rev.B42(1990)11682.

[3]R.M.Tromp,M.C.Reuter,Phys.Rev.Lett.68(1992)954.

[4]M.Horn-von Hoegen,M.Copel,J.C.Tsang,M.C.Reuter,R.M.

Tromp,Phys.Rev.B50(1994)10811.

[5]M.Horn-von Hoegen,B.H.Mullar,A.Al-Falou,Phys.Rev.B50

(1994)11640.

[6]B.Voigtlaènder,A.Zinner,Surf.Sci.351(1996)L233.

[7]A.Sakai,T.Tatsumi,Appl.Phys.Lett.64(1994)52.

[8]S.-J.Kahng,Y.H.Ha,J.-Y.Park,S.Kim,D.W.Moon,Y.Kuk,Phys.

Rev.Lett.80(1998)4931.

[9]S.Zaima,Y.Yasuda,J.Cryst.Growth163(1996)105.

[10]G.Ohta,S.Fukatsu,Y.Ebuchi,T.Hattori,http://www.wendangku.net/doc/76b018dd360cba1aa811dacf.htmlami,Y.Shiraki,

Appl.Phys.Lett.65(1994)2975.

[11]M.Katayama,E.Nomura,N.Kanekama,H.Soejima,M.Aono,Nucl.

Instrum.Methods B33(1988)857.

[12]K.Sumitomo,K.Oura,I.Katayama,F.Shoji,T.Hanawa,Nucl.

Instrum.Methods B33(1988)871.

[13]M.Katayama,T.Nakayama,M.Aono,C.F.McConville,Phys.Rev.

B54(1996)8600.

[14]M.Katayama,R.S.Williams,M.Kato,E.Nomura,M.Aono,Phys.

Rev.Lett.66(1991)2762.

[15]T.Fuse,T.Fujino,J.-T.Ryu,M.Katayama,K.Oura,Surf.Sci.420

(1999)81.

[16]T.Fuse,J.-T.Ryu,T.Fujino,K.Inudzuka,M.Katayama,K.Oura,

Jpn.J.Appl.Phys.38(1999)1359.

[17]T.Fujino,T.Fuse,E.Tazou,T.Nakano,K.Inudzuka,K.Goto,Y.

Yamazaki,M.Katayama,K.Oura,Nucl.Instrum.Methods B,(in press).

[18]T.Takahagi,I.Nagai,A.Ishitani,H.Kuroda,Y.Nagasawa,J.Appl.

Phys.64(1988)3516.

[19]G.S.Higashi,Y.J.Chabal,G.W.Trucks,K.Raghavachari,Appl.

Phys.Lett.56(1990)656.

[20]K.Takayanagi,Y.Tanishiro,S.Takahashi,M.Takahashi,Surf.Sci.

164(1985)367.

[21]Y.Shigeta,Surf.Rev.Lett.5(1998)865.

T.Fujino et al./Thin Solid Films369(2000)25±28 28