hypocriticalCHIN.PHYS.LETT.Vol.25,No.1(2008)242 Evolution of Ge and SiGe Quantum Dots under Excimer Lar Annealing∗HAN Gen-Quan(韩根全)1∗∗,ZENG Yu-Gang(曾玉刚)1,YU Jin-Zhong(余金中)1,CHENG Bu-Wen(成步文)1,
trillionYANG Hai-Tao(杨海涛)2
1State Key Laboratory on Integrated Optoelectronics,Institute of Semiconductors,Chine Academy of Sciences,
Beijing100083
2Tsinghua-Foxconn Nanotechnology Rearch Center,Department of Physics,Tsinghua University,Beijing100084
(Received15September2007)
We prent different relaxation mechanisms of Ge and SiGe quantum dots under excimer lar annealing.Inves-tigation of the coarning and relaxation of the dots shows that the strain in Ge dots on Gefilms is relaxed by dislocation since there is no interface between the Ge dots and the Ge layer,while the SiGe dots on Si0.77Ge0.23film relax by lattice distortion to coherent dots,which results f
rom the obvious interface between the SiGe dots and the Si0.77Ge0.23film.The results are suggested and sustained by Vanderbilt and Wickham’s theory,and also demonstrate that no bulk diffusion occurs during the excimer lar annealing.
PACS:68.65.Hb,68.35.Fx,68.35.Md,68.37.Ps
Ge and SiGe lf-asmbled quantum dots (SAQDs)are widely studied for their promis-ing application in optoelectronics due to three-dimensional(3D)quantum confinement.[1]Many works have focud on the growth mechanism,[2,3] shape transition,[4,5]and the coarning process under thermal annealing[6]of the SAQDs in S-K mode.Re-cently,we obtained SiGe quantum dots with small size and high density by excimer lar annealing(ELA).[7] The nanocond pul duration of the excimer,which induces rapid heating and cooling of the sample sur-face,ensuring that the lar induced quantum dots (LIQDs)are formed only by surface atoms diffusion.[8] We obtained Ge and SiGe lar induced quantum dots by ELA of the Ge and SiGefilms,respectively.
In this Letter,we report that the lar-induced Ge and SiGe quantum dots undergo different relax-ation mechanisms.Atomic-force-microscopy(AFM) measurements indicate that the Ge LIQDs on the Ge film relax by formation of dislocation,while the SiGe LIQDs on the Si0.77Ge0.23film relea the s
train by the lattice tetragonal distortion and then form coher-ent dots.The theory developed by Vanderbilt and Wickham has shown[9]that the interface between the dots and the wetting layer plays a pivotal role in the relaxation process of the strained dots.For the SiGe LIQDs on the Si0.77Ge0.23film,our calculation shows that SiGe quantum dots with the Ge composition of about83%are formed on the Si0.77Ge0.23film,which indicates an obvious interface between the dots and the Si0.77Ge0.23film.The interface leads to the for-mation of the coherent SiGe relaxed dots.However, for the Ge LIQDs on the Gefilm,no interface between the dots and the wetting layer results in the formation of the dislocated dots.The are suggested and sus-tained by Vanderbilt and Wickham’s theory,and also demonstrates that no bulk diffusion occurs during the excimer lar annealing.
The Ge and SiGefilms were grown by an ultra-high-vacuum chemical vapour deposition(UHV-CVD) system on(001)-oriented Si substrates at500◦C and 550◦C,respectively.The Gefilm is in thickness of about1nm(8monolayers),and the SiGefilm is about 20nm.The sources of Si and Ge are disilane and ger-mane,respectively.The Si substrates were cleaned in an ex-situ chemical etch process and loaded into an UHV growth chamber with basic pressure lower than 10−7Pa,and then heated up to950◦C to deoxidize. The thickness and Ge composition of the Si0.77Ge0.23film are determined by double-crystal x-ray diffraction (XRD).
A193nm ArF excimer lar operating frequency in 40Hz,was ud to ex-situ anneal the samples,which were annealed in argon ambient.A top-flat beam profile of10×10mm2with the energy density of about180mJ/cm2was obtained by using a homog-enizer.This was carried out to ensure uniform an-nealing of samples’surface.The surface morphology of the samples was measured by an SPA-300HV AFM, performed in tapping mode.
Figure1shows the AFM images of Ge and SiGe LIQDs obtained by ELA of Ge and Si0.77Ge0.23films, respectively.The height profiles of the dots are also in Fig.1.The diameters of the Ge and SiGe LIQDs are20–25nm and15–20nm,respectively.The ther-mal process induced by the excimer lar pul is only veral tens nanoconds,so during the ELA,only surface diffusion occurs.The dot energy can be ex-presd by E=4ΓV2/3tan1/3θ−6AV tanθ,[2]where Γ=γd cscθ−γs cotθis the increa of surface energy,
∗Supported by the National Natural Science Foundation of China under Grant No60576001.∗∗Email:hgquan@red.mi.ac
c 2008Chine Physical Society an
d IOP Publishing Ltd
No.1HAN
Gen-Quan et al.243
γs and γd are the surface energy per unit area of the wetting layer and dot facet,respectively,θis the facet angle with respect to the surface of the wetting layer,V is the volume of the dot,A =σ2(1−ν)/(2πG )where σis the in-plane misfit strain,and νand G are Poisson’s ratio and shear modulus,respectively.For the LIQDs,only surface energy should be stud-ied,and the cond term on the right can be con-sidered as the effect of strain on the surface energy.From the formula,we can e that the slightly strained dots are not stable during the ELA.We speculate that the heavily strained LIQDs will grow,relax the strainpocket pc
in them with longer annealing time.To investigate the relaxation of the LIQDs,we prolong the anneal-ing time with the lar energy density of 180mJ/cm 2.As the ELA continues,We obrve the relaxation and the shrinking of the LIQDs,while it is surprisingly found that Ge quantum dots on the Ge wetting layer and SiGe dots on the Si 0.77Ge 0.23layer underwent the different relaxation modes:the Ge dots relax through the formation of the dislocation,while the strain in the SiGe quantum dots on the Si 0.77Ge 0.23wetting layer is relead by lattice tetragonal distortion.
Fig.1.AFM images (500nm ×500nm)of LIQDs:(a)Ge LIQDs on the Ge film and the height profiles along the line marked,(b)SiGe LIQDs on the Si 0.77Ge 0.23film and the height profiles along the line marked.
成都培训班
Figure 2shows a ries of AFM images of the mor-phology of Ge LIQDs on the Ge film at different a
mion-nealing times.When the annealing time is prolonged to 3.5hours,coarning of the quantum dot,as shown in Fig.2(a),occurs.The contacting of the small and large dots in Fig.2(a)and 2(b)can be interpreted to be the losing materials of small dots to the near large dots,which is analogous to the anomalous coarning in the SAQDs.[10]As the ELA proceeds,the density of the dots further decreas,and when the annealing time is up to 5hours,almost all the LIQDs disappear (shown in Fig.2(c)).After 7-h ELA,no new LIQDs are obrved.We speculate that the relaxation of the lar induced Ge dots is by the dislocations and the strained film is also relaxed by the dislocations.Fig-ure 3shows the schematic of the relaxation process of
bulgarithe Ge quantum dots on the Ge film.
Figure 4(a)shows the coarning and the growth of the SiGe dots on the Si 0.77Ge 0.23film.After 4-h an-nealing,the SiGe dots become larger and the density decreas.As the annealing continues (5h),some new LIQDs appear.This indicates that the growth and disappearing of the SiGe dots give ri to the restora-tion of the strain in the Si 0.77Ge 0.23film.This will decrea the surface energy and increa the strain en-ergy.The recovered stress in the film drives the new LIQDs under ELA.This reveals that the SiGe dots grow and relax to be the coherent ,the strain in the SiGe dots is relaxed by the lattice distortion.Figure 5shows the schematic of the relaxation process of the
SiGe quantum dots on the Si 0.77Ge 0.23film.
244HAN Gen-Quan et al.Vol.25
Fig.2.AFM images (1µm ×1µm)of the Ge LIQDs on the Ge film with different annealing times:(a)annealed for 3.5h,(b)annealed for 4h,(c)annealed for 5h,(d)annealed for 7h.
Fig.3.Schematic diagram of the relaxation mode of the Ge quantum dots on the Ge film.
商品英语Fig.4.AFM images (1µm ×1µm)of the SiGe LIQDs on the Si 0.77Ge 0.23film for different annealing times:(a)annealed for 4h,(b)annealed for 5h.despicable
The results reveal the existence of two different relaxation mechanisms:generation dislocation in the dots and formation coherent relaxed dots.When the quantum dots grow,the relaxation of quantum dots is the competing of the lattice distortion (coherent re-laxed dots)with the formation of the dislocation (dis-located relaxed dots).The theory developed by Van-derbilt and Wickham [9]compares the two mechanisms of elastic relaxation and yields a pha diagram of a lattice misma
tched system in which all possible mor-phologies are ,uniform films,dislocated dots,and coherent dots.
No.1
HAN Gen-Quan et al.245
It was shown by Vanderbilt and Wickham that morphology of the mismatched system is determined by the ratio of the energy of interface between dots and the wetting layer (E interface )to the change of the sur-face energy (∆E surf ).[9]The deposited material wets the substrate firstly,and then the 2D strained film transforms to the 3D quantum dots.If ∆E surf is posi-tive and large,or if the energy of the interface between the dots and the wetting layer is relatively small,the formation of coherently strained dots is not favoured.With an increa in the amount of deposited material,a transition occurs from uniform film to dislocated dots,and the coherently strained dots are not formed.If ∆E surf is positive and small,or if the energy of the dislocated interface is relatively large,with an increa in the amount of deposited material,a transition oc-curs from a uniform film to coherent dots.Further de-position may cau the ont of dislocations.The de-tailed calculation and the pha diagram can be found in Ref.[9].
初中升旗仪式演讲稿
Fig.5.Schematic diagram of the relaxation process of the SiGe quantum dots on the Si 0.77Ge 0.23film.
This theory can be ud to interpret the differ-ent relaxation modes of the Ge and SiGe dots.It is sure that the pyramidal lar induced Ge dots,with the diameter of about 20–25nm and density of about 6×1010cm −2,do not exhaust the Ge film with the thickness more than 1nm (8monolayers).Becau no bulk diffusion occurs during the annealing,atoms intermixing between the dots and the wetting layer need not be considered.We think that the pure Ge LIQDs are formed on the Ge fi,there is no in-terface between the dots and the wetting layer.For the SiGe LIQDs on the Si 0.77Ge 0.23film,bad on the surface chemical potential calculation,we show that the heavily strained SiGe quantum dots must have a misfit above 0.035corresponding to a Ge composi-tion of about 83%,to promi E surf <0(the dots stable under ELA).[7]This indicates the SiGe dots are Ge ri
cher than the Si 0.77Ge 0.23film,which also results from that the surface diffusion coefficient of Ge is 102–103times greater than that of Si.[11]If the atoms interdiffusion is neglected,there should be an obvious interface between the SiGe quantum dots and the Si 0.77Ge 0.23wetting layer.It is suggested theoret-ically by Vanderbilt and Wickham and supported by our experiments that the interface between the quan-tum dots and the wetting layer plays a pivotal role in the competition between the lattice distortion and the formation of dislocation.
Vanderbilt and Wickham’s theory is proven by our results and also confirms and enforces our previous conclusion that the pure Ge dots and an abrupt inter-face between the dots and wetting layer are available
which is attributed to no bulk atoms diffusion under ELA.
In conclusion,we have studied the different relax-ation mechanisms of the Ge and SiGe quantum dots on Ge and Si 0.77Ge 0.23films,respectively,under ELA.We recover the pivotal role of the interface between the dots and the wetting layer.The relaxation of Ge dots by dislocation is attributed to no interface between Ge dots and the Ge layer,and that of SiGe dots by lattice tetragonal distortion results from the obvious interface between SiGe dots and the Si 0.77Ge 0.23film.This is sustained by Vanderbilt and Wickham’s theory.
References
[1]Baribeau J M,Wu X,Rowell N L and Lockwood D J 2006
J.Phys.:Condens.Matter 18R139
[2]TersoffJ and LeGoues F K 1994Phys.Rev.Lett.723570[3]Sutter P,Schick I,Ernst W and Sutter E 2003Phys.Rev.
Lett.91176102
[4]Rastelli A,Stoffel M,TersoffJ,Kar G S and Schimidt O G
2005Phys.Rev.Lett.95026103
[5]Montalenti F,Raiteri P,Migas D B,von K¨a nel H,Rastelli
A,Manzano C,Costantini G,Denker U,Schimidt O G,Kern K and Miglio L 2004Phys.Rev.Lett.93216102[6]Kamins T I,Medeiros-Ribeiro G,Ohlberg D A A and
Williams R S 1999J.Appl.Phys.851159
[7]Han G Q,Zeng Y G,Yu J Z,Cheng B W and Yang H T
2007J.Cryst.Growth (submitted)
[8]Misra N,Xu L,Pan Y L,Cheung N and Grigoropoulos C
P 2007Appl.Phys.Lett.90111111
[9]Vanderbilt D and Wickham L K 1991Mater.Res.Soc.
Symp.Proc.202555
[10]Rastelli A,Stoffel M,TersoffJ,Kar G S and Schmidt O G
2005Phys.Rev.Lett.95026103
[11]Huang L,Liu F,Lu G-H and Gong X G 2000Phys.Rev.
schoolbagLett.96016103
