1: \documentclass[twocolumn,showpacs,amsmath,amssymb]{revtex4}
2:
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4:
5: \usepackage{graphicx}
6: \usepackage{dcolumn}
7: \usepackage{bm}
8: \usepackage[figuresright]{rotating}
9: \usepackage{wrapfig}
10: \usepackage{subfigure}
11:
12: \def\p{\partial}
13: \def\f{\frac}
14: \def\d{{\rm d}}
15:
16: %\input{tcilatex}
17:
18: \begin{document}
19:
20: \title{Gravity-free hydraulic jumps and metal femtocups}
21:
22: \author{Rama Govindarajan$^{1,*}$, Manikandan Mathur$^{1,\dag}$, Ratul
23: DasGupta$^1$, N.R. Selvi$^2$, Neena Susan John$^2$ and G.U. Kulkarni$^{2,*}$}
24: \affiliation{1. Engineering Mechanics Unit and 2. Chemistry and Physics of
25: Materials Unit and DST Unit on Nanoscience, Jawaharlal Nehru Centre for
26: Advanced Scientific Research, Jakkur, Bangalore 560064, India. \\
27: $\dag$ currently at Dept. of Mechanical Engineering, MIT, Cambridge, MA
28: 02139, USA.
29: }
30:
31: \date{\today}
32:
33: \begin{abstract}
34:
35: Hydraulic jumps created by gravity are seen every day in the kitchen sink.
36: We show that at small scales a circular hydraulic jump can be created in
37: the absence of gravity, by surface tension. The theory is motivated by
38: our experimental finding of a height discontinuity in spreading submicron
39: molten metal droplets created by pulsed-laser ablation. By careful
40: control of initial conditions, we show that this leads to solid femtolitre
41: cups of gold, silver, copper, niobium and tin.
42:
43: \end{abstract}
44:
45: \pacs{47.61.-k, 47.85.Dh, 47.55nd}
46:
47: \maketitle
48:
49: It has long been observed that water flowing horizontally
50: %, especially radially outwards,
51: can display a discontinuity in height \cite{rayl,watson,tani}. This is the
52: {\em hydraulic jump}, seen for example when water from a faucet impinges on the kitchen sink and spreads outwards.
53: %Tidal bores are another standard example \cite{rayl}.
54: %The cause is well-understood, as discussed briefly below, and
55: Gravity is a key ingredient in these well-understood {\em large-scale} hydraulic jumps, as discussed briefly below.
56: %For the first time, so far as we know,
57: In this Letter we show, remarkably, that the shallow water equations support
58: solutions for a {\em gravity-free} hydraulic jump. The driver here is surface
59: tension at the liquid-air (or liquid-vacuum) interface, and jumps may be
60: expected to occur when relevant length scales are submicron.
61: Our theoretical study was prompted by careful
62: experiments showing that molten metal droplets impinging on a solid substrate
63: display such a jump, solidifying into cup-shaped containers of
64: femtolitre capacity. The droplets are created by laser-ablation of a solid
65: metal target. Femtocups made of different metals on various substrates
66: are formed under carefully maintained conditions of
67: laser energy and substrate temperature. Outside this narrow range of
68: parameters we find what one would normally expect: the droplets solidify
69: into lump-shaped structures on the substrate. The ability to make, and
70: subsequently leach out, femtocups
71: at will has potential applications ranging from nanoscale synthetic chemistry
72: to single cell biology.
73:
74: Before describing the experiments and the femtocups, we discuss what causes a
75: gravity-free hydraulic jump. Consider a steady axisymmetric jet of fluid
76: of radius $a$ impinging on a solid plate placed normal to the flow. The
77: density of the surrounding medium is assumed to be negligible. The
78: fluid then spreads radially outwards within a relatively thin film. The
79: dynamics within the film is described by the
80: axisymmetric shallow-water equation \cite{30,bohr1,bush}
81: \begin{equation}
82: u\f{\p u}{\p r} + w \f{\p u}{\p z} = \nu \f{\p^2u}{\p z^2} - g h' +
83: \f{\sigma}{\rho} \f{\d}{\d r}\left[\f{\nabla^2h + (h'^3/r)}{(1 +
84: h'^2)^{3/2}}\right],
85: \label{shallow}
86: \end{equation}
87: where $r$ and $z$ are the radial coordinate and the coordinate perpendicular
88: to the solid wall respectively, with origin on the solid surface at the
89: centre of the impinging jet. The respective velocity components are $u$ and
90: $w$. The total height $h$ of the fluid above the surface is a function of
91: $r$, and a prime thus denotes a derivative with respect to $r$.
92: The parameters in the problem are the acceleration due to
93: gravity, $g$, the surface tension coefficient, $\sigma$ for the liquid-air or
94: liquid vacuum interface, and the density $\rho$ and the kinematic viscosity
95: $\nu$ of the impinging fluid. For incompressible axisymmetric flow the equation
96: of continuity, in differential and in global form, reads
97: \begin{equation}
98: \f{\partial u}{\partial r}+\f{u}{r}+\f{\partial w}{\partial z}=0 \quad {\rm
99: and} \quad 2 \pi \int_0^{h(r)} ru(r,z)dz = Q,
100: \label{cont}
101: \end{equation}
102: where $Q = \pi a^2u_j$ is the steady inlet volumetric flow rate. A
103: characteristic inlet jet velocity $u_j$ is thus defined.
104: It is reasonable to assume \cite{tani} a parabolic shape in $z$ for the
105: radial velocity
106: \begin{equation}
107: u(r,z)=\zeta(r) (z^{2}-2h(r) z)
108: \label{parabolic}
109: \end{equation}
110: satisfying the no-slip condition at the wall ($z=0$) and the zero shear stress
111: condition at the free surface ($z=h$). The analysis does not hinge on this
112: assumption; any reasonable profile shape will give qualitatively the same
113: results. Using Eq. (\ref{cont}) and the kinematic condition
114: $w=$D$h/$D$t=uh^{\prime }$ at $z=h$, the momentum equation (\ref{shallow})
115: integrated over $z$ from $0$ to $h$ reduces after some algebra to
116: \begin{equation}
117: b\left( \f{h}{r}+h^{\prime }\right) =\f{2r}{R}+\f{h^{3}r^{2}}{F}
118: h^{\prime }-\f{r^{2}h^{3}}{W}\f{d}{dr}\left[ \f{\nabla
119: ^{2}h+(h^{\prime 3}/r)}{(1+h^{\prime }{}^{2})^{3/2}}\right],
120: \label{final}
121: \end{equation}
122: where all lengths are scaled by $a$, and the $O(1)$ positive constant $b = 2/5$ for a parabolic profile. The left-hand side of (\ref{final}) represents
123: inertia, and the three terms on the right hand side appear due to viscosity,
124: gravity and surface tension respectively. The relative importance of the
125: inertial term to each of these is quantified respectively by the Reynolds
126: number $R \equiv u_ja/\nu$ $R$, the Froude number $F \equiv u_j^2/(ga)$, and
127: the Weber number $W \equiv \rho u_j^2a /\sigma$.
128: %For an axisymmetric flow
129: %\begin{equation}
130: %\nabla ^{2}\equiv \f{d^{2}}{dr^{2}}+\f{1}{r}\f{d}{dr}
131: %\label{nabla}
132: %\end{equation}
133: In large-scale flows surface tension has been shown \cite{bush} only to
134: make a small correction to the location of the jump, so the last term is
135: unimportant.
136: This is to be expected, since $F$ in the kitchen sink is of order
137: unity, while $W \sim 10-100$. In contrast, consider $u_j \sim 10$
138: m/s and $a \sim 10^{-7}$m, so $F \sim 10^8$ and $W \sim 10^{-2}$. Here
139: surface tension determines whether and where a jump will occur, whereas it
140: is the {\em gravity} term that may be dropped entirely from the equation.
141:
142: In general, a jump occurs if the pressure gradient becomes increasingly
143: adverse as the flow proceeds downstream, and attains a magnitude large enough
144: to counter the relevant inertial effects. The adverse pressure gradient may
145: be created by gravity, or surface tension, or both.
146: With gravity alone, Eq. (\ref{final}) reduces to
147: \begin{equation}
148: h^{\prime}= \f{2r/R - b h/r}{b - h^3r^2/F}.
149: \label{gravity}
150: \end{equation}
151: It is seen that if $F$ is finite and $h^3r^2$ is an increasing function of $r$,
152: the denominator will go to zero at some $r$, i.e., a jump will occur in the framework
153: of the shallow-water equations \cite{rayl,watson,bohr1,bush}.
154: However, its precise location may not coincide with this estimate \cite{bohr1}, and
155: in a given experiment the radial extent available may be too small,
156: or the inertia too low, for a jump to occur. Decreasing gravity has been
157: shown to shift the jump location downstream \cite{28}, consistent with
158: Eq. (\ref{gravity}).
159: Now considering surface tension alone, a crude prediction of the
160: existence of a jump may be made by assuming the height upstream to be slowly varying
161: in $r$, i.e., $h^{\prime}<< 1$, and thus setting $h^{\prime\prime}=h^{\prime
162: \prime\prime}=0$. We may then rewrite Eq. (\ref{final}) as
163: \begin{equation}
164: h^{\prime}\simeq \f{2r/R - b h/r}{b - h^3/W}.
165: \label{surface_approx}
166: \end{equation}
167: A jump is now possible if $h$ is an increasing function of $r$, which is a
168: more stringent requirement than in the case of gravity. Note that the second
169: term in the denominator appears due to {\em radial} spreading, i.e., surface
170: tension alone cannot give rise to a one-dimensional jump like a tidal bore.
171:
172: We now solve Eq. (\ref{final}) as an initial value problem beginning at
173: some location $r_i$ and marching downstream. A fourth-order Runge-Kutta
174: algorithm is used. An initial radius $r_i$ somewhat larger than $a$ is chosen,
175: where it is assumed that a parabolic profile has been attained. The
176: initial
177: conditions in $h$ and its derivatives are not known exactly for this complicated
178: problem, and numerical studies are being done to understand the flow in this
179: vicinity. We have, however, repeated the computations with a variety of initial
180: height profiles, and a range $1.2 < r_i < 5$ and $0.1 < h < 1$, and the results
181: do not change qualitatively. Typical solutions are shown in figure
182: \ref{typical}. At a particular radial location $r = r_j$, there is a
183: singularity in the height of the fluid layer. Note that as we approach $r_j$
184: the shallow water equations are no longer valid, even approximately,
185: so the present analysis cannot tell us anything about the actual shape close
186: to or after the jump.
187: \begin{figure}
188: \includegraphics[width=0.25\textwidth]{typical}
189: \vskip3mm
190: \caption{Typical solutions of Eq. (\ref{final}), with $F=\infty$,
191: containing a singularity at $r=r_j$. For demonstration, liquid
192: properties are taken as those of molten silver ($\rho=5000$ kg m$^{-3}$,
193: $\nu=10^{-6}$m/s$^2$ and $\sigma=0.9$Nm$^{-1}$) and $a=5 \mu$m. Solid
194: line: $u_j=5$ cm/s; dashed line: $u_j=80$ m/s. Values for molten
195: tin show similar behavior.}
196: \label{typical}
197: \end{figure}
198: The dependence of the jump location on the inlet jet velocity $u_j$
199: is not monotonic, as seen from figure \ref{jumploc}a. Here the Reynolds number
200: $R$ is varied by changing $u_j$, with other quantities as in figure
201: \ref{typical}, so the Weber
202: number increases as $R^2$, from $3 \times 10^{-9}$ to $90$.
203: For very low $R$ or very high $W$, jumps are unlikely to form within the
204: available radius, i.e., inertia and surface tension must be in the right
205: balance. The Reynolds and Weber numbers are now varied independent of each
206: other (figure \ref{jumploc}b). In the region shown in red $r_j > 60$ so jumps
207: are not predicted. (A higher cut-off does not
208: change answers qualitatively.) Blue color indicates
209: $r_j \sim r_i$, this region merits numerical investigation.
210: Gravity-free hydraulic jumps may be expected in the region shown by
211: intermediate color, seen as a relatively narrow linear patch when $R < 100$.
212: Here the jump location depends only on the ratio $W/R$. For a given $W$, jumps
213: exist for over an order of magnitude variation in $R$.
214: At $R > 100$, undular jumps are seen, which are being investigated further.
215: \begin{figure}
216: \includegraphics[width=0.15\textwidth]{jumploc}
217: \includegraphics[width=0.25\textwidth]{contour}
218: \caption{(a) The location $r_j$ of the singularity as a function of the inlet
219: jet radius, expressed here in terms of the Reynolds number. For $R=0.01-90$
220: ($u_j=0.02-180$m/s for the case considered) the jump radius is of the order of
221: a few microns, as observed in the experiment described below.
222: (b) Contour plot of jump location in the $R-W$ plane.}
223: \label{jumploc}
224: \end{figure}
225:
226: \begin{figure}
227: \includegraphics[width=0.45\textwidth]{blobs}
228: \caption{AFM image (right) and height profile (left) of a silver blob on
229: silicon substrate kept at room temperature.}
230: \label{blobs}
231: \end{figure}
232:
233: We turn now to our experiments, which show a height discontinuity in spreading
234: drops of molten metal. Since the experimental flow is transient in nature,
235: a detailed comparison with the theoretical results is not possible, but the jump
236: radius is in the right range. At larger scales hydraulic jumps are known to
237: occur even when the incoming flow is in droplets rather than jets \cite{29}.
238: A Q-switched frequency tripled Nd:YAG laser
239: ($\lambda=355$ nm, repetitive frequency, 10 Hz) is focused with
240: pulse energy $E_p$ on a rotating metal disc in a vacuum chamber
241: ($10^{-7}$ torr)
242: and the resultant plume received at a distance of 4 cm on a clean vertical substrate
243: held at a temperature $T_s$, for a duration of 20 min
244: \cite{kulk}. The resulting metallic structures on the substrate
245: are studied by scanning electron microscopy (SEM), atomic force microscopy
246: (AFM) and energy dispersive X-ray analysis (EDAX). Over most of the range of
247: $E_p$ and $T_s$, we expect, and obtain,
248: ill-shaped blobs of solidified metals, see figure \ref{blobs}. However, for a
249: small range of these parameters, there is a strong preference to form
250: cup-like structures of outer diameters $\sim 300$nm to $10 \mu$m, with side
251: walls $\sim 100$nm high, and capacity $\sim 1$ fL (fig. \ref{egcup}).
252: The jump diameter is usually about half the total diameter.
253: Height profiles associated with atomic force micrographs (figure \ref{egcup}b)
254: as well as EDAX spectra (not shown) \cite{kulk} confirm that the
255: central region is raised from the substrate and contains metal.
256: Interestingly, pulsed-laser ablation has been used extensively to produce a
257: variety of structures \cite{eglaser}, but femtocups have not been reported
258: before, although we notice stray instances of similar structures in other
259: studies \cite{stray}.
260: \begin{figure}
261: \includegraphics[width=0.46\textwidth]{egcup}
262: \caption{Microscopy analysis of metal cups (a) SEM image of femtocups of
263: silver on a silicon substrate obtained at $E_p=100$ mJ/pulse and $T_s=1173 K$.
264: A few blobs do exist, as indicated by arrows.
265: (b) Typical height profile of a femtocup from AFM analysis. (c) Tilted field
266: emission SEM image of a tin femtocup on silicon.}
267: \label{egcup}
268: \end{figure}
269:
270: This surprising femtocup structure is consistent with the proposed dynamics:
271: of a droplet spreading out thinly initially and then undergoing a height
272: discontinuity. We obtain femtocups of gold, silver, copper, tin and niobium
273: of repeatable statistics on glass, silicon and
274: graphite (HOPG), see examples in figure \ref{various}a and b, and later in
275: figure \ref{hist}.
276: The solid surface being vertical and the length scales small mean that the effect of
277: gravity is negligible. Inertia on the other hand is considerable, since
278: velocities are high. We do not have a direct estimate of $u_j$, but we
279: may estimate it from earlier measurements in many similar experiments
280: \cite{book} to range between 1m/s to 100 m/s.
281: Also, the range of $R$ and $W$ over which a
282: surface-tension driven hydraulic jump occurs translates to a particular range
283: of $E_p$, since laser fluence determines scales and speeds in the incoming jet,
284: compare figures \ref{egcup}a and \ref{various}c.
285: While the substrate is hotter than the metal's
286: melting point $T_m$, the cup may form initially but cannot solidify and
287: liquid flows back into the cup, so the final object is as seen in the inset of
288: figure \ref{various}d.
289: With $T_s$ far below $T_m$ the tendency to form cups is much reduced (figure
290: \ref{various}d, probably
291: because solidification is too rapid for flow to be completed.
292: Optimal
293: conditions are thus $E_p$ ($\sim$ 100 mJ/pulse for silver) and $T_s$ close to but
294: below $T_m$. Outside the correct range, blobs form rather than cups.
295: \begin{figure}
296: \includegraphics[width=0.40\textwidth]{various}
297: \caption{SEM images with different metals, laser energy and substrate
298: temperature. (a) Cu femtocups on Si. $E_p$=100 mJ/pulse, $T_s = 300$ K; (b)
299: Nb on Si, $E_p$=100 mJ/pulse, $T_s=1273$ K; (c) Ag on Si, $E_p=60$ mJ/pulse.
300: The number density of well-formed cups is lower. (d) Ag on Si, $E_p=100$
301: mJ/pulse, $T_s=773$ K much less than $T_m=1234$ K. The cups are not
302: well-formed. Inset, $T_s=1273$ K $> T_m$. Only patches are observed.}
303: \label{various}
304: \end{figure}
305:
306: That the jump is directly
307: related to droplet dynamics is confirmed by varying the substrate orientation
308: with respect to the incoming jet (see schematic in Fig. \ref{angle}).
309: As $\theta$, the inclination of the substrate away from the normal, is
310: increased, the structures become increasingly elliptical, especially beyond
311: $40^\circ$, in accordance with the azimuthal variation of $R$ and $W$.
312: \begin{figure}
313: \includegraphics[width=0.40\textwidth]{angle}
314: \caption{Elliptical femtocups with inclined jets. Inset: Sample SEM image of tin
315: obtained at $\theta=40 \pm 1.5^\circ$, defining $a$ and $b$. The ratio $b/a$
316: increases with $\theta$, each data point is an average on several cups. The
317: major axis lies along the direction of maximum flow (dashed arrows).}
318: \label{angle}
319: \end{figure}
320:
321: Since the experiment includes additional complexity in the form of
322: solidification, we estimate relative time-scales of jump formation
323: $t_j$ and solidification of a droplet $t_c$. For the experimental values of
324: substrate thickness, $t_c$ ranges from $\sim 3 \times 10^{-4}$s on silicon to
325: $\sim 10^{-2}$s on glass (taking into consideration conduction, radiation
326: and latent heat), while $t_j \sim r_j/u_j \sim 10^{-6}$s or less.
327: Contact-line freezing can give rise to an increase in
328: height in the vicinity, typically amounting to a small percentage of the
329: height in the central region. Contrast this to our jump where the height at
330: the rim is several-fold larger than that in the central region. In spite of
331: this, and the disparity in
332: time scales, we cannot rule out a role for local freezing at the contact line
333: \cite{sonin}. We do notice
334: a dependence on the substrate of the size
335: distribution of femtocups (figure \ref{hist}) and also some visual differences
336: in the shape of the femtocup. Our ongoing numerical study, including a
337: non-uniform temperature profile and its effects, is therefore aimed at
338: a better representation of the experiments. Also being addressed are the
339: experimental finding of radial striations in the femtocups under certain
340: conditions, and the theoretical finding of undular hydraulic jumps (similar to
341: \cite{bowles} in other conditions) when $R$ is
342: greater than about $25/W$ (figure \ref{floral}).
343: \begin{figure}
344: \includegraphics[width=0.45\textwidth]{hist}
345: \caption{Histograms and SEM images of tin femtocups on (a) glass and (b)
346: silicon, deposited simultaneously.}
347: \label{hist}
348: \end{figure}
349: \begin{figure}
350: \centering
351: %\begin{minipage}{1.0\textwidth}
352: %{\hskip-4.5in
353: \includegraphics[width=0.40\textwidth]{floral}
354: %\hskip2mm
355: %\includegraphics[width=0.17\textwidth]{undular_scaled}}
356: %\end{minipage}
357: \caption{(a) Striations in the structure. (b) An undular hydraulic jump
358: at $R=5000$, $W=0.005$.}
359: \label{floral}
360: \end{figure}
361:
362: In summary, for the first time, a hydraulic jump solely driven by surface
363: tension is shown to occur. Experimentally we show evidence for such jumps in
364: submicron high inertia droplets of molten metals spreading radially outwards
365: on a substrate. The detailed shape in the vicinity of the jump and the transient
366: problem including the solidification process is being studied numerically.
367:
368: We are grateful to Prof. G Homsy, Prof. CNR Rao and Prof. R Narasimha for
369: useful discussions. RG and NSJ acknowledge support from DRDO (India)
370: and CSIR (India) respectively.
371:
372: $*$ To whom correspondence should be addressed, rama@jncasr.ac.in,
373: kulkarni@jncasr.ac.in
374:
375: \begin{thebibliography}{10}
376:
377: \bibitem{rayl}
378: L Rayleigh, {\em Proc. R. Soc. Lond.} A, {\bf 90}, 324-328 (1914).
379:
380: \bibitem{watson}
381: E.J. Watson, {\em J. Fluid Mech.} {\bf 20}, 481-499 (1964).
382: %
383: %\bibitem{elleg98}
384: C. Ellegaard {\em et al.}, {\em Nature}, {\bf 392}, 767-768 (1998).
385: %
386: %\bibitem{bohr2}
387: T. Bohr, V. Putkaradze, \& S. Watanabe, {\em Phys. Rev. Lett.}, {\bf 79},
388: 1038-1041 (1997).
389: %\bibitem{jkb}
390: S.B. Singha, J.K. Bhattacharjee \& A. Rai {\em Eur. Phys. J.} B, {\bf 48},
391: 417-426 (2005).
392: %\bibitem{bush06}
393: J.W.M. Bush, J.M.J. Aristoff \& A. E. Hosoi, {\em J. Fluid Mech.}, {\bf 558},
394: 33-52, (2006).
395:
396: \bibitem{tani}
397: I. Tani, {\em J. Phys. Soc. Japan}, {\bf 4}, 212-215, (1949).
398:
399: \bibitem{bohr1}
400: T. Bohr, P. Dimon \& V Putkaradze {\em J. Fluid Mech.} {\bf 254}, 635-648 (1993).
401:
402: \bibitem{bush}
403: J.W.M. Bush \& J.M.J. Aristoff, {\em J. Fluid Mech.} {\bf 489}, 229-238 (2003).
404:
405: \bibitem{30}
406: L.D. Landau \& E.M. Lifshitz, {\em Fluid Mechanics
407: (Course of Theoret. Phys., Vol. 6)}, Pergamon press, U.K., (1987).
408:
409: \bibitem{kulk}
410: G.U. Kulkarni {\em et al.} preprint, NS John, PhD thesis, (2006).
411:
412: \bibitem{eglaser}
413: M. Terrones, {\em et al.}, {\em Nature}, {\bf 388}, 52-55 (1997). A. Thess,
414: {\em et al.}, {\em Science}, {\bf 273}, 483-487 (1996). X. Duan \& C.M.
415: Lieber, {\em Adv. Mater.} {\bf 12}, 298-302 (2000). Y. Zhang, K. Suenaga, C.
416: Colliex \& S. Iijima, {\em Science}, {\bf 281}, 973-975 (1998).
417: UK Gautam {\em et al.}, {\em J. Am. Chem. Soc.}, {\bf 127}, 3658-3659 (2005).
418:
419: \bibitem{stray}
420: C.J.K Richardson {\em et al. Mat. Res. Soc. Symp.} {\bf 617},
421: J7.4.1-J7.4.6 (2000).
422: S.J. Henley, M.N.R. Ashfold \& S.R.J. Pearce,
423: {\em Appl. Surf. Sci.} {\bf 217}, 68-77 (2003).
424:
425: \bibitem{28}
426: C.T. Avedisian \& Z Zhao {\em Proc. R. Soc. Lond.} A 456,
427: 2127-2151 (2000).
428:
429: \bibitem{29}
430: S. Chandra \& C.T. Avedisian {\em Proc. R. Soc. Lond.} A,
431: 432, 13-41 (1991).
432:
433: \bibitem{book}
434: D.B. Chrisey \& G.K. Hubler {\em Pulsed Laser Deposition of Thin Films}
435: Wiley, New York (1994) and references therein.
436:
437: \bibitem{sonin}
438: S. Schiaffino \& A. A. Sonin, {\em Phys. Fluids} {\bf 9}, No. 8, 2227-2233;
439: and 2217-2226 (1997).
440:
441: \bibitem{bowles}
442: R.I. Bowles \& F.T. Smith, {\em J. Fluid Mech.}, {\bf 242}, 145-168.
443: H. Steinruck, W. Schneider \& W. Grillhofer, {\em Fluid Dyn. Res.}, {\bf 33}
444: 41-55 (2003).
445:
446: \end{thebibliography}
447:
448: \end{document}
449: