Sie sind auf Seite 1von 6

Applied Surface Science 258 (2012) 41294134

Contents lists available at SciVerse ScienceDirect

Applied Surface Science


journal homepage: www.elsevier.com/locate/apsusc

Pit formation on the Ge (1 0 0) surfaces by normal incident Si ion implantation


S.A. Mollick, S. Karmakar, A. Metya, D. Ghose
Saha Institute of Nuclear Physics, Sector-1, Block-AF, Bidhan Nagar, Kolkata 700 064, India

a r t i c l e

i n f o

Article history:
Available online 31 August 2011
Keywords:
Si ion implantation
Ge
Morphology
Pit formation
Simulation

a b s t r a c t
We have observed micron size pit formation on Ge surface due to bombardment of 26 keV Si ion at normal incidence in the uence range 1 1018 and 7 1018 ions/cm2 . Scanning electron microscopy (SEM)
and atomic force microscopy (AFM) are used to follow the evolution of the surface morphology. The pits
are of various shapes, e.g., crescent-shaped, kidney-like or circular structures. The two-eld continuum
model developed for small slope approximations can describe the pit formation and growth at the very
beginning of ion bombardment. The growth of the pits at late times (high uence) can be explained by the
gradient dependent erosion mechanisms due to primary ion beam as well by secondary ux of particles
originating from steep slopes. Energy dispersive X-ray analysis attached to SEM is employed to obtain
the chemical information of the pitted surface. The depletion of Si at the bottom of the pits is explained
due to lower diffusivity of Si in Ge.
2011 Elsevier B.V. All rights reserved.

1. Introduction
Ion implantation in solids is a useful technology for the formation of new alloys irrespective of the solubility and diffusivity
of the doping species. The technique has also other advantages,
such as high precision in the control of ion beam position, controllability in depth prole and concentration, purity and shallow
treatment [13]. Recently, the silicongermanium (SiGe) semiconductor alloys are getting importance for device applications
in semiconductor industry and are competing with direct bandgap compound semiconductors such as GaAs [4]. The SiGe system
has the ability to perform band-gap engineering [5] which has
direct implications in the production of efcient and inexpensive
voltaic cells. SiGe nanostructures synthesized by ion implantation
showed prominent photoluminescence properties in the infrared
and microwave spectral regions [6], which may nd practical
applications in optoelectronic devices and bre-optic information
links. During implantation, the erosion of solids by ion sputtering
can also result in a wide variety of surface structures on length
scales from few nanometers to many micrometers [79]. For example, Hofer et al. [10] studied the morphological modications of
SiGe lms grown epitaxially on Si substrates and sputtered by
normal incident Ar+ ion in the energy range 0.11 keV. They identied two regimes of ion energy for the topographic evolution.
At lower ion energies (<0.75 keV), the pyramidal shape nano-pits
present on the initial lm transformed into shallow troughs and

Corresponding author.
E-mail address: debabrata.ghose@saha.ac.in (D. Ghose).
0169-4332/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.apsusc.2011.08.092

subsequently to smaller craters, while at higher energies


(>0.75 keV) smoothening of the surface occurred. However, to our
knowledge, the change of the substrate topography during synthesis of SiGe by ion implantation is not reported so far. In this
paper, we report the formation of pit structures on Ge surface
due to sputtering during high uence and high energy (26 keV) Si
ion implantation at normal incidence. A model based on gradientdependent sputtering is presented to explain the evolution of the
pits.

2. Experimental
Single-crystal (1 0 0)-oriented one side polished Ge wafers of
thickness 0.5 mm and resistivity > 30  cm were used in the present
study. They were cut into pieces with a size of 10 10 mm2 . The
samples were irradiated at room temperature with mass-analyzed
26 keV Si ions in a low energy negative ion implanter developed in the laboratory. The advantages of using the negative Si ion
are two-fold; rst the voltage built-up on non-conducting samples is minimal and second, the possible contamination of ion
beam by 28 N2 species is eliminated. The beam current density
was 100 A/cm2 and it was incident at a few degree off-normal
to the sample in order to avoid the channelling effect. The
ion uences ranged in between 1 1018 and 7 1018 ions/cm2 .
After irradiation, the surface morphology of the irradiated samples was examined by scanning electron microscope (SEM) as
well as by atomic force microscope (AFM). The AFM images
were taken at contact mode by a glass tip of diameter 20 nm
with opening angle 20 . The cantilever force constant was
3 N/m.

4130

S.A. Mollick et al. / Applied Surface Science 258 (2012) 41294134

Fig. 1. Some representative SEM images of etch pits formed on Ge (1 0 0) surface by 26 keV Si bombardment at different uences. (a and b) Fluence = 1.5 1018 ions/cm2 . Note
the pits are crescent-circled with stepped rims, one of them is zoomed in (b); (c and d) uence = 3.5 1018 ions/cm2 , where (c) shows kidney-shaped pits and (d) terrace-like
structures with overlapping pits; (e and f) uence = 4.5 1018 ions/cm2 showing secondary structures at the bottom of the pits; (g and h) uence = 5 1018 ions/cm2 showing
almost circular-shaped pits.

S.A. Mollick et al. / Applied Surface Science 258 (2012) 41294134

4131

Fig. 2. (a) A typical AFM image showing kidney-shaped pits; (b) the AFM cross-section prole of a pit corresponding to the line scan in (a); (c and d) show the variation of
pit slopes and depth with the ion uence.

3. Results and discussion


Fig. 1 shows some selected SEM images of the sputtered Ge surfaces at different ion uences. Etched pits are found to develop and
distributed randomly over the whole bombarded area. The AFM
images also show similar features. For quantitative analysis the
AFM images were used. The average density of pits is found to be of
the order of 108 cm2 and the lateral size of the pits lies in the range
0.35 m. In some regions, where the number densities and sizes
of pits are large, the pits overlap to form quasi-repetitive terrace
structures [cf., Fig. 1(d)]. The shape of the pits is, however, dependent on the ion uence. Within the range of uence studied, the
pits are mainly crescent-shaped [cf., Fig. 1(a)] at the lower uence

region, kidney-like [cf., Fig. 1(c)] at the intermediate region and


almost full circular structure [cf., Fig. 1(g)] at the higher uence.
The slopes of the pit walls can be determined from the AFM crosssectional proles as illustrated in Fig. 2(ab). Fig. 2(c) depicts the
slopes of the two opposite side walls as a function of uence. Clearly
the slopes at the two faces are different because of the asymmetric structure of the pit. Similar results are obtained if we measure
the slopes for other sides of the pit walls. The depth of pits is also
measured utilizing AFM images, which appears to increase linearly
with increasing uence [see Fig. 2(d)].
Basically, the ion-induced surface morphology is the result of
two competitive effects, the roughening process and the smoothing process [11]. When a beam of energetic ions encounters a solid

Fig. 3. The hole morphology calculated by Eq. (2) for r = 5 at times: (a) t = 10 and (b) t = 100; the system size is L = 128 (256 nodes).

4132

S.A. Mollick et al. / Applied Surface Science 258 (2012) 41294134

target, the ions penetrate and lose energy in the solid by elastic
and inelastic collision processes until the ions are trapped by the
cohesive forces of the solid. In the cascade of collisions, the recoiled
lattice atoms which lie very close to the surface and energies greater
than the binding energy of the solid will leave the surface (sputtering). This results in the roughening of the surface. A fraction of
the sputtered atoms can redeposit and diffuse along the target surface. There is also diffusion of the implanted species as well as other
defects (vacancies and interstitials) created in the collision cascade.
In addition, an amorphous layer can be generated on the semiconductor surface due to large number of ion impacts and thus the
viscous ow in the layer comes into play for structure formation.
All these mechanisms are incorporated successfully into a coupled

et al. [12].
two-eld model developed recently by Munoz-Garca
For isotropic sputtering, i.e., at normal ion incidence, the non-linear
two-eld theory gives the temporal evolution of surface height h
as:
h
2
2
= v 2 h K 4 h + 1 ( h) + 2 2 ( h) ,
t

(1)

where the coefcients , K, 1 , and 2 can be determined by the


parameters of the ion sputtering process and diffusion processes.
Eq. (1) can be simplied through an appropriate rescaling of the
coefcients [7]. This reads:
h
2
2
= 2 h 4 h + ( h) + r 2 ( h) ,
t

(2)

where r = (2 )/(K1 ). The value of r is positive for dot formation, whereas a negative value of r describes the hole
formation. It should be mentioned that the damped version of the
KuramotoSivashinsky (dKS) equation as proposed by Facsko et al.
[13] can predict the hole formations as well [14], although the
physical meaning of the damping term is not yet clear.
After a short time of bombardment, the top surface of Ge
becomes amorphous and a layer of Si and Ge mixture is formed.
Since the amorphous layer has a lower density than the substrate,
it tends to expand and compressive stresses build up on the surface
[15]. The implantation of foreign species can induce a new chemical phase in the near surface region. It is a known fact that GeSi
bonds are stronger than GeGe bonds [16,17]. The implanted Si
ions could replace GeGe bonds by the energetically more favorable GeSi bonds. This can change signicantly the stress state of
the surface [15]. In a separate experiment with Si (1 0 0)-self sputtering under the same experimental conditions, we found no etch
pit formation on the surface of Si. This might suggest that the precursors of the pits in the present study are the local inhomogeneity
in surface stress due to the incorporation of Si in Ge. The mixed layer
can extend up to 2Rp (50 nm as calculated by SRIM [18]), where
Rp is the projected range of the incident ion. The stress-induced
roughening can be expressed by a third order derivative term as
C3 h, where C is a parameter that depends on the elastic constants
of the substrate [19].
A qualitative simulation study of Ge surface evolution during
normal incidence 26 keV Si sputtering has been performed by
numerical integration of Eq. (2) using a fourth order RungeKutta
method and the improved spatial discretization introduced by
Lam and Shin [20] for the nonlinear terms. We have used periodic boundary conditions, lattice constant x = 0.5, and time step
t = 0.01. The initial height values were chosen uniformly random
between 0 and 1. The standard system size of our simulation was
L = 128 (256 nodes). At short-time scale shallow depressions are
developed, which in the long-time limit evolve to pits of nanoscale
dimensions (Fig. 3). The experimental results, however, show
macroscopically large structures different from that predicted by
the theory, because of the high uence used in the experiment.

Fig. 4. (a) Schematic illustration of the expansion of a pit during erosion with ion ux
J ions/cm2 s1 . For simplicity, one half of the vertical section of the pit is shown. In the
plane (A) the beam is incident normally, while in plane (B) the incident angle is ; d
and d are respectively, the horizontal and vertical displacements of the intersection
O at time t1 . (b) The temporal evolution of an arbitrary two-dimensional surface
prole bombarded by ion beam. (c) The displacement velocity of the intersection
between the two planes at  and at 0 , as calculated by SRIM [18], is shown as a
function of the incident angle.

Nevertheless, the theoretical formalisms indicate how the sputter


pits are initiated and grow at the early stages of bombardment.
Once the structures are nucleated and coarsen, the small slope
approximations as assumed in the derivation of the continuum
equations (1) and (2) may no longer be valid. For large slope structures, the local incidence angle or gradient dependent sputtering is
more important for nal topography evolution [21,22]. As shown
schematically in Fig. 4(a), we consider the movement of two intersecting planes where in one plane (A) the beam is incident in normal

S.A. Mollick et al. / Applied Surface Science 258 (2012) 41294134

4133

Fig. 5. The SEM image of pits (a) and the corresponding EDX elemental map (bd) (the green colour for Si-K line and the red colour for Ge-L line) showing the deciency
of Si at the bottom of the pits. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

direction and in the other plane (B) the beam is incident at an angle
 with respect to the surface normal. The structure is equivalent
to one half of the vertical section of the two-dimensional pit morphology. Here we assume a smooth S()  function with one peak
at   , i.e., the angular dependence of the sputtering yield S() curve
for amorphous solids. For a two-dimensional surface (xOy) erosion
[Fig. 4(b)], the vertical velocity and the horizontal velocity of a reference point on the bombarded surface are, respectively, given by
[23], (y/t)x = (J/N) S() and (x/t)y = (J/N) S() cot , where J is
the ion ux and N is the density of target atoms. Using simple geometrical arguments it can be shown [24,25] that the intersection
(O) moves sideways with a velocity.
V () =

J
[S() S(0)]cot .
N

(3)

Similarly, for the other half of the pit-section, the intersection


shifts in the opposite direction. In other words, the pit expands
outwards with a velocity that depends primarily on the local surface
slopes. Fig. 4(c) shows the variation of the displacement velocity of
the intersection as a function of angle of ion incidence as calculated
by SRIM [18]. When the different sides of the pit vary in slope, the
differential displacement velocities can lead to the expansion of
pits into various shapes, e.g. formations of crescent-, kidney-, or
circular-shaped as observed in the present experiment.
The appearance of ne structures in some of the pits, i.e., the
stepped rims [Fig. 1(a and b)] can be explained by considering the
rotation of local slope under ion bombardment. The time evolution
of the surface slope due to sputtering is given by [23].

J
S() 
= cos2 
.
N
t
 x

(4)

The S()  curve for an amorphous solid reveals that the equilibrium slopes are those at 0 ,   or 90 for which S/ = 0. Under
these boundary conditions, Bayly [25] argued that if the sides of
the pits have subsidiary irregularities with local slopes exceeding

  , then steps with 90 walls and 0 bases can be formed, which,


however, disappears as the pit expands with sputter time.
The increase of pit depth with increasing uence [Fig. 2(d)] cannot be explained by the primary ion beam erosion alone since at the
pit bottom the ion incidence angle,  0 . Therefore, with respect to
the outer reference at surface, the pit depth is conserved. When
considering the erosion at the foot of steep slopes, the effects of secondary beam should be taken into account. The secondary particle
beam consists of scattered ions, reected ions and energetic sputtered atoms from the inclined pit walls. This will add to the erosion
by primary beam causing faster recession of the bottom surface. The
potentiality of such effects was also recognized in the formation of
grooves at the bases of large conical structures [26]. The repetitive formation of secondary pits at the bottom of large primary
pits at higher uences [Fig. 1(f)] might also be the consequences of
secondary particle uxes.
Finally, we have made the compositional study of the pit morphology by energy dispersive X-ray (EDX) analysis. A typical SEM
image of the pitted area along with the X-ray elemental mapping is
depicted in Fig. 5. The brighter areas of Si X-ray map in Fig. 5(b) represent higher concentration of Si which is obviously present outside
of the pits. Similarly, for Ge X-ray map [Fig. 5(c)], the bottoms of
the pits are brighter than outside thus indicating higher concentration of Ge there. These ndings are clearer when Si and Ge X-ray
mappings are superimposed as shown in Fig. 5(d). A representative X-ray spot analysis inside and outside of the pit shows that the
Si/Ge intensity ratio inside the pit is about 0.07, while at outside it
is nearly 1 [Fig. 6(a)]. We also observed that the relative percentage
of the implanted Si increases with increasing uence at the outside of the pit, while inside it is largely insensitive to the uence
[Fig. 6(b)]. All these results can be understood by the difference in
mobility of the Si and Ge atoms due to their atomic size differences.
The diffusion of dopant atoms and self-diffusion in Ge, both are considered to be vacancy-mediated [27] which in turn depends on the
size of the diffusing atoms interacting with the vacancies. Recent

4134

S.A. Mollick et al. / Applied Surface Science 258 (2012) 41294134

The number density of pits is about 108 cm2 , the sizes vary in the
range 0.35 m and the depth lies in the range 50550 m. For the
uence range studied the slope of the pit walls is initially 3 and
nally becomes 3545 . The continuum models developed for small
slope approximations can describe the pit formation and growth at
the very beginning of bombardment. The later growth of the pits
with a large slope at the side walls can be explained by the local
gradient dependent erosion instability caused by the primary ion
beam and the secondary beam of particles.
Acknowledgement
The authors thank Souvik Banerjee for the technical help in SEM
and Nupur Biswas for the AFM. Thanks are also due to the referee
for critically reading the manuscript and useful comments.
References

Fig. 6. (a) A typical EDX analysis across a pit morphology showing the relative intensities of Ge and Si atoms inside and outside of the pit, respectively; (b) the variation
of Si concentrations inside and outside of the pits as a function of uence.

experiments [28] show that the activation energy and the prefactor of the Arrhenius equation associated with Si diffusion are
greater than those for Ge self-diffusion [29]. This means a slower
diffusivity of Si in Ge than that of Ge in Ge. In fact, the rst-principles
calculations of surface diffusion barriers show that Ge diffuses 103
times faster than Si on a compressive Ge surface at room temperature [30]. Such a large disparity in Si and Ge surface mobility seems
to be the reason for enrichment of Si outside the pits rather than
inside.
4. Summary
To summarize, we have observed formation of pits of different
shapes on Ge surface by high uence keV energy Si ion sputtering.

[1] J.W. Mayer, L. Eriksson, J.A. Davies, Ion Implantation in Semiconductors, Academic Press, New York, 1970.
[2] G. Dearnaley, J.H. Freeman, R.S. Nelson, J. Stephen, Ion Implantation, NorthHolland, Amsterdam, 1973.
[3] H. Rysell, H. Glawischnig, Ion Implantation Techniques, Springer-Verlag, Berlin,
1982.
[4] Suan Abedrabbo, D.-E. Arafah, S.A. Salem, J. Electron. Mater. 34 (2005) 468.
[5] J. Weber, M.I. Alonso, Phys. Rev. B 40 (1989) 5683.
[6] Yu.N. Parkhomenko, A.I. Belogorokhov, N.N. Gerasimenko, A.V. Irzhak, M.G.
Lisachenko, Semiconductors 38 (2004) 572.

L. Vzquez, R. Cuerno, J.A. Snchez-Garca, M. Castro, R. Gago,


[7] J. Munoz-Garca,
in: Z. Wang (Ed.), Lecture Notes on Nanoscale Science and Technology 5,
Springer, New York, 2009.
[8] W.L. Chan, E. Chason, J. Appl. Phys. 101 (2007) 121301.
[9] D. Ghose, J. Phys. Condens. Matter 21 (2009) 224007.
[10] C. Hofer, S. Abermann, C. Teichert, T. Bobek, H. Kurz, K. Lyutovich, E. Kasper,
Nucl. Instrum. Methods B 216 (2004) 178.
[11] R.M. Bradley, J.M.E. Harper, J. Vac. Sci. Technol. A 6 (1988) 2390.

M. Castro, R. Cuerno, Phys. Rev. Lett. 96 (2006) 086101.


[12] J. Munoz-Garca,
[13] S. Facsko, T. Bobek, A. Stahl, H. Kurtz, T. Dekorsy, Phys. Rev. B 69 (2004)
153412(R).
[14] Q. Wei, X. Zhou, B. Joshi, Y. Chen, Kun-Dar Li, Q. Wei, K. Sun, L. Wang, Adv.
Mater. 21 (2009) 2865.
[15] G. Ozaydin, K.F. Ludwig Jr., H. Zhou, R.L. Headrick, J. Vac. Sci. Technol. B 26
(2008) 551.
[16] J. Tersoff, Phys. Rev. B 39 (1989) 5566.
[17] F. Bechstedt, W.A. Harrison, Phys. Rev. B 39 (1989) 5041.
[18] J.F. Ziegler, IBM Research, SRIM-2000.40 (PC version), Yorktown Heights, NY,
1999.
[19] N.V. Medhekar, W.L. Chan, V.B. Shenoy, E. Chason, J. Phys. Condens. Matter 21
(2009) 224021.
[20] C.H. Lam, F.G. Shin, Phys. Rev. E 58 (1998) 5592.
[21] Q. Wei, J. Lian, L.A. Boatner, L.M. Wang, R.C. Wing, Phys. Rev. B 80 (2009)
085413.
[22] J. Vllner, B. Ziberi, F. Frost, B. Rauschenbach, J. Appl. Phys. 109 (2011)
043501.
[23] M.J. Nobes, J.S. Colligon, G. Carter, J. Mater. Sci. 4 (1969) 730.
[24] A.D.G. Stewart, M.W. Thompson, J. Mater. Sci. 4 (1969) 56.
[25] A.R. Bayly, J. Mater. Sci. 7 (1972) 404.
[26] O. Auciello, R. Kelly, R. Iricibar, Radiat. Eff. 46 (1980) 105.
[27] R.J. Borg, G.J. Dienes, An Introduction to Solid State Diffusion, Academic Press,
London, 1988.
[28] H.H. Silvestri, H. Bracht, J. Lundsgaard Hansen, A. Nylandsted Larsen, E.E. Haller,
Semicond. Sci. Technol. 21 (2006) 758.
[29] M. Werner, H. Mehrer, H.D. Hochheimer, Phys. Rev. B 32 (1985) 3930.
[30] L. Huang, F. Liu, G.-H. Lu, X.G. Gong, Phys. Rev. Lett. 96 (2006) 016103.

Das könnte Ihnen auch gefallen