1: \documentclass[prl,twocolumn,aps]{revtex4}
2: \usepackage{hyperref}
3: \usepackage{graphicx}
4: \usepackage{subfigure}
5: \usepackage{amsmath}
6: \usepackage[squaren]{SIunits}
7: \begin{document}
8: \title{Pattern formation upon femtosecond laser ablation of transparent dielectrics}
9: \author{Ionu\c{t} Georgescu}%
10: \altaffiliation[Current address ]{Max-Planck-Institut f\"{u}r Physik komplexer Systeme, N\"{o}thnitzer Str. 38, 01187 Dresden, Germany.}%
11: \email{george@pks.mpg.de}
12:
13:
14: \author{Michael Bestehorn}
15: \affiliation{LS Theoretische Physik II, Brandenburg University of Technology, Erich-Weinert Str. 1, 03046 Cottbus, Germany}
16:
17: \author{Floren\c{t}a Costache}
18: \author{J\"{u}rgen Reif}
19: \affiliation{LS Experimentalphysik II, Brandenburg University of Technology, Universit\"{a}tsplatz 3-4, 03044 Cottbus, Germany
20: }
21: \date{\today}
22:
23: \begin{abstract}
24: Costache et al. have reported recently a new type of periodic patterns
25: generated at femtosecond laser ablation of transparent dielectrics (Appl. Surf.
26: Sci. \textbf{186}, 352(2002)). They show features known from other pattern
27: forming systems far from equilibrium, like point and line defects or grain
28: boundaries, and cannot be explained by the classical theory. The present work
29: is an attempt to investigate these pattern by means of a generalized
30: Kuramoto-Shivashinsky equation derived from the Bradley et al. and Cuerno et
31: al. model for ripple formation at ion beam sputtering of surfaces.
32: \end{abstract}
33: \pacs{45.70.Qj, 52.38.Mf, 79.20.Rf}
34:
35: \maketitle
36:
37: %\section{INTRODUCTION}
38:
39: Laser induced periodic surface structures (LIPSS) have been observed for almost
40: 40 years \citep{lipss:univ_phenomenon} on targets made of intrinsic and
41: extrinsic semiconductors, metals and dielectrics, using cw to subpicosecond
42: laser sources with wave lengths varying from the ultraviolet up to the infrared
43: domain (\cite{lipss_1} and references therein,
44: \citep{lipss_2,lipss_3,lipss:uv}). They have therefore been considered to be a
45: universal phenomenon that can occur on any material that absorbs radiation,
46: regardless of its dielectric constant \cite{lipss:univ_phenomenon}.
47:
48: The explanation accepted today on a large scale has been delivered by the group
49: of \citeauthor*{lipss_1} \citep{lipss_1} in \citeyear{lipss_1}, by taking into
50: account the details of the interaction of an electromagnetic wave with the
51: microscopically rough selvedge of a surface. Thus, the periodic surface
52: structures are the result of the interference between the incoming laser light,
53: the light refracted by the bulk and the light scattered by the rough selvedge
54: (or interface). The aspect of the surface is usually a combination of several
55: basic patterns. For $p$-polarized light, these are the $s$-type patterns
56: (Eq.~\eqref{classical:model}) perpendicular to the electric field $\vec{E}$ and
57: the $c$-type patterns parallel to $\vec{E}$; for $s$-polarized light, there are
58: $c$-type patterns perpendicular to $\vec{E}$ and other patterns with no simple
59: dependence on $\theta$, the angle of incidence. Excellent agreement with
60: experimental work that followed \citep{lipss_2,lipss_3} has been found.
61:
62: \begin{equation}
63: s\text{-type: }\Lambda=\frac{\lambda}{1\pm\sin\theta},\ \ \ \ \ c\text{-type: }\Lambda=\frac{\lambda}{\cos\theta}.
64: \label{classical:model}
65: \end{equation}
66: $\lambda$ denotes here the wave length of the laser and $\Lambda$ the
67: periodicity of the patterns.
68:
69: Recently, \citeauthor{newstruct} have reported upon a new type of periodic
70: patterns generated at femtosecond laser ablation of transparent dielectrics
71: \citep{newstruct}. These new structures exhibit fundamental differences when
72: compared to the ``classical'' periodic patterns, such as spacing almost twice
73: as small, independent on the wave length and on the angle of incidence of the
74: laser beam, but correlated with the local incident intensity. Features like
75: point and line defects or grain boundaries can also not be described by the
76: classical theory. They point rather to a self-organizing mechanism.
77: %The numerous bifurcations, the similarities to non-equilibrium patterns, such as sand under shallow, wavy water or the patterns on ion beam sputtered surfaces, point to a self-organizing mechanism.
78:
79: The qualitative and quantitative similarities to the patterns observed in ion
80: beam sputtering \citep{hab:lieb}, as well as the wide acceptance of their
81: theoretical description as proposed by \citeauthor{bradley:harper}
82: \citep{bradley:harper} and \citeauthor{cuerno} \citep{cuerno}, have determined
83: us to use this framework for studying the new structures.
84:
85: %\section{EXPERIMENTAL BACKGROUND}
86: The experiments \citep{newstruct,fc:ii,reif:2beams,arcdischarge} have been
87: carried out under high vacuum ($<\unit{\power{10}{-7}}{\milli\bbar}$) on
88: freshly cleaved single crystal slides of $\mathrm{BaF_2}$ and
89: $\mathrm{CaF_2}$. The laser system generated a pulse of
90: \unit{120}{\femto\second} with a central wave length of \unit{800}{\nano\meter}
91: and intensities of $\unit{1-12\times10^{12}}{\watt\per\centi\meter\squared}$.
92: Additionally, the frequency of the laser radiation could be doubled and the
93: angle of incidence of the laser beam on the target could be varied.
94:
95: The main feature which distinguishes these new structures from the
96: \emph{classical} ones is the spacing. In the classical model the patterns are
97: uniformly distributed across the ablation area, with spacings given by
98: Eq.~\eqref{classical:model}. In this case, however, no dependency of the
99: ripples periodicity on the laser wave length or on the angle of incidence could
100: be observed. Moreover, the periodicity is obviously dependent on the local
101: intensity of the laser beam, as shown in Fig.~\ref{laserbif_p00}. The spacing
102: of the ripples increases here towards the center of the ablation spot, that is,
103: increases with the local intensity. While Eq.~\eqref{classical:model} allows a
104: minimum spacing of $\approx\unit{468}{\nano\meter}$ for the situation shown in
105: Fig.~\ref{laserbif_p00}, the spacing at the boundary of the ablation spot is
106: approximatively \unit{250}{\nano\meter}.
107:
108: %To further check the idea of an interference mechanism leading to surface
109: %ripples, \citeauthor{reif:2beams} have produced a controlled interference by
110: %crossing two non-collinear beams under a small angle and examined the surface
111: %for temporally overlapped pulses \citep{reif:2beams}. While the Bragg's
112: %condition allowed for an interference pattern of spacing
113: %$\Lambda=\lambda/2\sin\theta\approx\unit{15}{\micro\meter}$, whose existence
114: %has also been checked experimentally, no evidence for it has been found in the
115: %surface structure.
116:
117: \begin{figure*}
118: \begin{center}
119: \subfigure[]{\includegraphics[height=4cm]{laserbif_p00}\label{laserbif_p00}}%
120: \hspace{0.5cm}%
121: \subfigure[]{\includegraphics[height=4cm]{laserbif_p45}\label{laserbif_p45}}%
122: \hspace{0.5cm}%
123: \subfigure[]{\includegraphics[height=4cm]{laserbif_p90}\label{laserbif_p90}}%
124: \end{center}
125: \caption{Orientation of the periodic ripple structure with respect to the laser
126: beam polarization. (a) Ablated spot obtained after 9200 shots at
127: $\unit{0.8\times\power{10}{13}}{\watt\per\centi\squaren\meter}$; (c) 5000 shots
128: at $\unit{1.2\times\power{10}{13}}{\watt\per\centi\squaren\meter}$; (a), (b)
129: and (c) \unit{800}{\nano\meter}, \unit{45}{\degree} incidence.}%
130: \label{laserbif_p}
131: \end{figure*}
132:
133: %\section{MODEL AND SIMULATIONS}
134: The model presented by \citeauthor{bradley:harper} \citep{bradley:harper} and
135: further developed by \citeauthor{cuerno} \citep{cuerno} is based on Sigmund's
136: theory of sputtering \citep{sputt:i}. Here, an ion striking a solid will first
137: travel a certain distance $a$, called average depth of energy deposition, and
138: then lose its energy in a cascade of random atomic collisions. For an ion
139: travelling along the $z$-axis, the resulting profile of the average energy
140: deposition in the material ($z\leq h(x,y)$) follows in a good approximation a
141: Gaussian distribution \citep{sputt:i}
142:
143: \begin{equation}
144: F_D(\vec{r})=\frac{\epsilon}{(2\pi)^{3/2}\alpha\beta^2}
145: \times
146: \exp\left[ -\frac{[z + a]^{2}}{2\alpha^2}
147: - \frac{x^2+y^2}{2\beta^2} \right].\label{gauss}
148: \end{equation}
149: $F_D(\vec{r})$ denotes here the energy deposition per unit volume, $\epsilon$
150: is the total energy deposited by the ion and $\alpha$ and $\beta$ are the
151: widths of the distribution parallel and perpendicular to the plane of
152: incidence, respectively. The erosion velocity at a certain point on the surface
153: is then proportional to the total energy deposited there by the ions within the
154: range $\mathcal{R}$ of the distribution \eqref{gauss}
155: \begin{equation}
156: v(O) \cong \Lambda\int_{\mathcal{R}} \Psi(\vec{r}) F_D(\vec{r}) d\vec{r},
157: \label{erosion:rate}
158: \end{equation}
159: where $\Psi(\vec{r})$ is the local flux of ions and $\Lambda$ is a material
160: constant.
161:
162: Eq.~\eqref{gauss} was written in the reference frame of the incoming ion, having the origin at the point of impact $O$ and the $z$-axis pointing away from the surface, along the velocity $\vec{v}$ of the ion. In Eq.~\eqref{erosion:rate} we went over to the local reference frame, with the origin also at $O$ but with the $z$-axis pointing along the surface normal. And finally, there is the reference frame of the laboratory, our destination.
163:
164: To evaluate the erosion velocity \eqref{erosion:rate}, we make
165: the transform $x\rightarrow a\zeta_x$, $y\rightarrow a\zeta_y$ and expand the
166: integrand in powers of $a/R_{X}$ and $a/R_{Y}$. $R_X$ and $R_Y$ are
167: the curvature radii at $O$ and are used to describe the integration region
168: $\mathcal{R}$
169: \begin{equation}
170: h\cong-\frac{1}{2}\left(\frac{x^2}{R_X}+\frac{y^2}{R_Y}\right)
171: \label{taylor:i}
172: \end{equation}
173: with $1/R_X=-\partial^2 h/\partial x^2$ and $1/R_Y=-\partial^2 h/\partial y^2$.
174: %\begin{equation}
175: %\frac{1}{R_X}=-\frac{\partial^2 h}{\partial x^2}\text{ and }
176: %\frac{1}{R_Y}=-\frac{\partial^2 h}{\partial y^2}
177: %\end{equation}
178: After evaluating the resulting Gaussian integrals we make the transition to the laboratory coordinate frame $(x, y, h)$ \citep{kpz, cuerno}
179: \begin{equation}
180: \frac{\partial h(x, y, t)}{\partial t}=-v(\phi, R_X, R_Y)\sqrt{1+(\nabla h)^2}
181: \label{lab:coord}
182: \end{equation}
183: and obtain thus for $\alpha=\beta$ \citeauthor{cuerno}'s equation of motion \citep{cuerno}
184: \begin{eqnarray}
185: \frac{\partial h}{\partial t}&=&
186: - v_0 + \mu\frac{\partial h}{\partial x}
187: + \nu_x\frac{\partial^2 h}{\partial x^2}
188: + \nu_y\frac{\partial^2 h}{\partial y^2}
189: + \frac{\lambda_x}{2}\left(\frac{\partial h}{\partial x}\right)^2 +\nonumber\\
190: &&
191: \frac{\lambda_y}{2}\left(\frac{\partial h}{\partial y}\right)^2
192: - K\nabla^2(\nabla^2 h) + \eta(x, y, t).
193: \label{surfev}
194: \end{eqnarray}
195: In Eq.~\eqref{lab:coord} $\phi$ denotes the angle between the incident beam and
196: the surface normal at $O$. It can be expressed as a function of the angle of
197: incidence $\theta$ and the local gradient $\nabla h$ and can be expanded in
198: powers of the latter.
199:
200: Eq.~\eqref{surfev} has the form of the Kuramoto-Shivashinsky (KS) equation
201: \citep{KS:erosion_model}. $v_0$, $\mu$, $\nu_x$, $\nu_y$, $\lambda_x$ and
202: $\lambda_y$ are functions of the angle of incidence $\theta$. The term
203: $-K\nabla^2(\nabla^2 h)$ accounts for the surface self-diffusion and $\eta(x,
204: y, t)$ is a Gaussian white noise, accounting for the stochastic arrival of the
205: ions on the surface. If the surface diffusion is thermally activated, the coefficient K is given by \citep{bradley:harper}
206: \begin{equation}
207: K=D_S\gamma\nu/n^2k_BT,\ D_S=D_{S_0}e^{-Q_a/k_BT}
208: \label{surf:diff}
209: \end{equation}
210: where $D_S$ \citep{neumann} is the surface self-diffusivity, $Q_a$ is the activation energy, $\gamma$ is the surface free energy per unit area and $\nu$ is the areal density of diffusing atoms.
211:
212: In the linear regime, periodical stripes with
213: \begin{equation}
214: |\vec{k}|=\sqrt{-\nu/2K},\ \nu=\min(\nu_x, \nu_y)
215: \label{k}
216: \end{equation}
217: have the highest growth rate. Consequently, for $\nu=\nu_x<0$ the surface will be dominated by ripples along the $\hat{y}$-axis ($\vec{k}\parallel\hat{x}$) and for $\nu=\nu_y<0$ by ripples parallel to the $x$-axis ($\vec{k}\parallel\hat{y}$).
218:
219: In the case of the laser material interaction, we assume that the photons loose
220: their energy in a similar stochastic process. The widths $\alpha$ and $\beta$
221: of the Gauss distribution \eqref{gauss} are equal because the target is
222: isotropic. The incident energy flux $\epsilon\Psi(\vec{r})$ is now given by
223: $a\gamma|\vec{S}_2\vec{n}|$, where $a$ has the meaning of an average
224: penetration depth, $\gamma$ is the absorption coefficient of the material and
225: $\vec{S}_2$ is the Poynting vector of the refracted beam. The product $a\gamma$
226: represents the fraction of the transmitted energy that will be absorbed, that
227: is, in an ionic picture, \emph{scattered} by the material. Using the Fresnel
228: equations, the incident flux can be written in terms of the laser intensity
229: $I_0=|\vec{S}_1|$ and of the transmission coefficient $T$
230:
231: \begin{equation}
232: \epsilon\Psi(\vec{r})\rightarrow a\gamma \frac{|\vec{S}_2\vec{n}|}{|\vec{S}_1\vec{n}|}\frac{|\vec{S}_1\vec{n}|}{|\vec{S}_1|}|\vec{S}_1|=a\gamma I_0 T(\vec{r}, \phi)\cos\phi.
233: \label{transmission}
234: \end{equation}
235:
236: In the next step we insert the energy flux \eqref{transmission} and the
237: Gaussian distribution \eqref{gauss} into Eq.~\eqref{erosion:rate} to compute
238: the local erosion rate. To evaluate the integral \eqref{erosion:rate}, we extend the Taylor expansion \eqref{taylor:i} to its complete form
239: \begin{equation}
240: h\cong-\frac{1}{2}\left(\frac{x^2}{R_X}+\frac{y^2}{R_Y}\right) -\frac{xy}{R_{XY}},
241: \label{taylor:ii}
242: \end{equation}
243: where $1/R_{XY}=-\partial^2 h/\partial x\partial y$.
244:
245: We call the last term in Eq.~\eqref{taylor:ii} a ``rotational''
246: correction. In their original model, \citeauthor{bradley:harper}
247: have neglected it and could therefore only describe patterns
248: growing along the $x$- or the $y$-axis. With this term the stability analysis
249: can still be reduced to the case of \citeauthor{bradley:harper}, but in a
250: reference frame rotated by an angle $\Theta$ to the original one.
251: %\footnote{We should polish this one by regarding the laser polarization as an additional degree of freedom.}.
252: Following the notation style of Eq.~\eqref{surfev} and denoting by $\nu_{xy}$ the coefficient of $h_{xy}$, the angle $\Theta$ is determined by
253: \begin{equation}
254: \nu_{xy}\cos2\Theta = (\nu_x - \nu_y)\sin2\Theta.
255: \label{Theta}
256: \end{equation}
257:
258:
259: We perform the transition \eqref{lab:coord} to the coordinate frame of the laboratory and scale the time and the space according to
260: \begin{equation}
261: x=ax',\ y=ay',\ t= \frac{\sqrt{2\pi}a}{\Lambda\gamma I_0}.
262: \label{scaling}
263: \end{equation}
264: to obtain
265: \begin{eqnarray}
266: \dot{h}&=&-v_0+\mu h_x + \nu_x h_{xx} + \nu_y h_{yy} + \nu_{xy} h_{xy} + \frac{1}{2}\lambda_x h_x^2 \nonumber \\
267: && + \frac{1}{2}\lambda_y h_y^2 + \nu_{10100} h_x h_{xx} + \nu_{10010} h_x h_{yy} + \nonumber \\
268: && \nu_{10001} h_x h_{xy} + \nu_{00200}h_{xx}^2 + \nu_{00020}h_{yy}^2 + \nu_{00002} h_{xy}^2 \nonumber \\
269: && + \nu_{00110}h_{xx}h_{yy} + \nu_{00101}h_{xx}h_{xy} + \nu_{00011}h_{yy}h_{xy} \nonumber \\
270: &&-B\Delta^2h.
271: \label{ks:laser}
272: \end{eqnarray}
273: %All the coefficients of Eq.~\eqref{ks:laser} except $B$ have the common factor $F=-\sigma\exp\left(-1/2\sigma^2(n^2-\sin^2\theta)/n^2\right)$. In particular, we write $\nu_{\ldots}$ as $\nu_{\ldots}=-F\Gamma_{\ldots}(\sigma, \varphi, \theta, n)$.
274:
275: The tuples $ijklm$ in $\nu_{ijklm}$ describe
276: the product $h_x^i h_y^j h_{xx}^k h_{yy}^l h_{xy}^m$ these coefficients belong
277: to. Except for $B$, all the coefficients in Eq.~\eqref{ks:laser} have the form $\{,\}=\sigma\exp[-\frac{\sigma^2}{2} \cos^2\theta] f_{\{,\}}(n, \theta, \varphi)$, with $\sigma=a/\alpha$.
278:
279:
280:
281: %\section{RESULTS}
282:
283: Fig.~\ref{gammasel} shows a 3D plot of $-\nu_x$, $-\nu_y$ and $-\nu_{xy}$
284: against the angle of incidence $\theta$ and the polarization $\varphi$. We
285: chose $\sigma$=4, $B=1.2\cdot10^{-4}$ and $n$=1.47, the refraction index of
286: $\mathrm{BaF_2}$ at $\lambda\sim\unit{750}{\nano\meter}$. As shown in the
287: previous section, the larger of $-\nu_x>0$ and $-\nu_y>0$ determines the
288: prevailing pattern and $\nu_{xy}$ will rotate this pattern by an angle $\Theta$
289: (Eq.~\eqref{Theta}).
290:
291: We see that there is actually only a small region between $\theta_{\text{min}}$ and $\theta_{\text{max}}$, where the ripple orientation changes with the light polarization. In our case $\theta_{\text{min}}$ was \unit{29.64}{\degree} and $\theta_{\text{max}}$ = \unit{34.04}{\degree}. For angles smaller than $\theta_{\text{min}}$ the ripples should always be perpendicular to the surface component of the laser beam, where as for angles larger than $\theta_{\text{max}}$ they should be parallel. The width of this region
292: %$\theta_{\text{max}}-\theta_{\text{min}}$
293: is usually between 1 and 5 degrees, increasing with increasing $n$. Its position varies slightly with $\sigma$, but strongly with $n$. Smaller $n$'s and larger $\sigma$'s will move it towards angles of up to \unit{50-60}{\degree}.
294:
295: %Lower $n$ are motivated by ...
296:
297: \begin{figure}
298: \begin{center}
299: \includegraphics[width=7cm]{gammasel}%
300: \end{center}
301: \caption{Plot of $\nu_x$, $\nu_y$ and $\nu_{xy}$ against the angle of incidence $\theta$ and the polarization $\varphi$. $\sigma = $ 4, $n$ = 1.47}\label{gammasel}
302: \end{figure}
303:
304: \begin{figure}
305: \begin{center}
306: \includegraphics[width=4cm]{43_00}%
307: \hspace{0.3cm}%
308: \includegraphics[width=4cm]{43_30}\\
309: \includegraphics[width=4cm]{43_60}%
310: \hspace{0.3cm}%
311: \includegraphics[width=4cm]{43_90}
312: \end{center}
313: \caption{Surface topography for $\theta$=\unit{31.8}{\degree} and $\varphi$=\unit{0}{\degree}, \unit{30}{\degree}, \unit{60}{\degree} and \unit{90}{\degree} respectively (upper-left to lower-right).}\label{surf:topography}
314: \end{figure}
315:
316: %\section{DISCUSSION}
317:
318: The present work is an attempt to investigate the occurrence of newly
319: discovered laser induced periodic surface structures \citep{newstruct} by means
320: of the Bradley et~al. \citep{bradley:harper} and Cuerno et~al. \citep{cuerno}
321: model. Based on Sigmund's stochastical theory of sputtering \citep{sputt:i},
322: this model has managed to explain in an unified framework most of the dynamic
323: and scaling behaviors observed experimentally at ion bombardment of surfaces.
324: It eventually traces the origin of ripple formation down to the instability
325: caused by the competition between surface roughening (so called negative
326: surface tension) and surface diffusion (the positive surface tension).
327:
328: The use of the Fresnel equations and the inclusion of the $\nu_{xy}(\partial^2
329: h/\partial x \partial y)$ term to the Kuramoto-Sivashinsky equation, have
330: allowed us to confirm the experimentally observed countinous change of the
331: ripple orientation with the laser polarization. However, our model predicts
332: that this should only happen for angles of incidence confined in a small range
333: between $\theta_{\text{min}}$ and $\theta_{\text{max}}$. In our simulations
334: this range corresponded to \unit{29\,-\,34}{\degree}, but could move up to \unit{53\,-\,55}{\degree}.
335: %Experiments at other angles of incidence would be therefore of great help,
336: %since \citeauthor{newstruct} have performed their experiments only at
337: %\unit{0}{\degree} and \unit{45}{\degree} incidence \citep{newstruct, fc:ii}.
338:
339: The role of the local laser intensity can also be explained. According to
340: Eq.~\eqref{surf:diff}, for local temperatures $T_{\text{local}}$ below $Q_a$,
341: $K$ is a monotonically increasing function of $T_{\text{local}}$ and thus of
342: the laser intensity $I_{\text{local}}$. A first estimation yields $Q_a >
343: \unit{0.53}{\electronvolt}$
344: %
345: \footnote{$Q_a$ is additively composed of $H_{Fa}$ and $H_{Ma}$, the formation
346: and migration energy of an adatom \citep{neumann}. We have numerically evaluated $H_{Ma}$
347: for the (110) surface of $\mathrm{BaF_2}$ and
348: obtained \unit{0.53}{\electronvolt} \citep{feynman}.}
349: %
350: , much larger than the melting point of $\mathrm{BaF}_2$
351: ($\unit{1280}{\celsius}\,\approx\,\unit{0.13}{\electronvolt}$). The periodicity
352: $2\pi/|\vec{k}|$ of the ripples (Eq.~\eqref{k}) is thus an increasing function
353: of $I_{\text{local}}$, just like in Fig.~\ref{laserbif_p00}.
354:
355: At normal incidence, all patterns with the wave vector satisfying
356: $|\vec{k}|=\sqrt{-\nu/2B}$ can occur with equal probability. The symmetry
357: break has therefore been looked for and traces of it have been found in the
358: 3$^{\mathrm{rd}}$ order terms. However, further investigation is needed, as we have not
359: been able to simulate any ripples yet.
360:
361: %The direct comparison of the simulated wave lengths and times with the
362: %experiment is not yet possible. Not only is there a great number of unknown
363: %parameters, $a$, $\alpha$ (or $\sigma=a/\alpha$), $\Lambda$ and $K$, but the lack of a comprehensive study of their time behaviour makes it impossible.
364:
365:
366: It is well known that the spacing of the structures described by the KS
367: equation increases with time according to $\lambda\sim t^{\gamma}$. The value
368: of $\gamma$ varies with different experimental conditions
369: \citep{ionhexagons,hab:lieb,hab:bol} and different numerical simulations
370: \citep{kpz, KS:erosion_model, KS:numerical_analysis:2D}. Considering that the
371: exposure times varied across the experiments, an unknown value of $\gamma$
372: makes the direct quantitative comparison of the numerical simulations with the
373: experiment impossible. A comprehensive study of the time behavior of the
374: structures is therefore a further requirement for their understanding.
375:
376: \citeauthor{ionhexagons} have recently shown that the compound nature of a
377: material can considerably influence the scaling behavior of the surface
378: structures in the early stages of ion sputtering \citep{ionhexagons}.
379: Considering the extreme conditions of the laser ablation, the loss of atoms is
380: with high probability different for Ba and F, so that a spacially varying
381: concentration of adatoms on the surface would arise.
382: \citeauthor{adatoms_gradient} have shown that surface diffusion driven by a
383: concentration gradient of the adatoms is able to generate structure coarsening
384: \citep{adatoms_gradient}. According to \citeauthor{ionhexagons}, this type of
385: coarsening would be one of the causes for the deviation from the KS predicted
386: scaling behavior they have observed in the early stages of $\mathrm{Ar}^+$
387: sputtering of InP surfaces.
388:
389: Further experiments with rotating targets or circular polarized light could
390: also confirm the common nature of this new type of LIPSS and of the ion
391: sputtering processes. Studying the topography of simultaneously rotated and
392: $\mathrm{Ar}^+$ ion sputtered InP surfaces, \citeauthor{ionhexagons} have been
393: able to find structures of highly hexagonal symmetry \citep{ionhexagons}.
394:
395:
396: %\begin{acknowledgments}
397: I. Georgescu would like to thank Mario de Menech for helpful discussions.
398: %\end{acknowledgments}
399:
400:
401: %\nocite{*}
402: \bibliographystyle{apsrev}
403: \bibliography{literature}
404:
405: \end{document}
406: