1: \documentclass[aps,twocolumn,prl,showpacs]{revtex4}
2: \usepackage{amssymb}
3: \usepackage{epsfig}
4: \usepackage{graphicx}
5: \usepackage{amssymb}
6: \usepackage{wasysym}
7:
8: \begin{document}
9:
10: \title{Autocatalytic chemical smoke rings}
11:
12: \author{Michael C. Rogers and Stephen W.
13: Morris}
14: \affiliation{Department of Physics, University of
15: Toronto, 60 St. George St., Toronto, Ontario, Canada M5S 1A7}
16:
17: \date{\today}
18:
19: \begin{abstract}
20:
21: Buoyant plumes, evolving free of boundary constraints, may develop
22: well-defined mushroom shaped heads. In normal plumes, overturning
23: flow in the head entrains less buoyant fluid from the surroundings as the head rises,
24: robbing the plume of its driving force. We consider here a new type
25: of plume in which the source of buoyancy is an autocatalytic chemical
26: reaction. The reaction occurs at a sharp front which separates reactants from less
27: dense products. In this type of plume, entrainment assists the reaction, producing new
28: buoyancy which fuels an accelerating plume head. When the head has grown to a
29: critical size, it detaches from the upwelling conduit, forming an accelerating, buoyant
30: vortex ring. This vortex is analogous to a rising smoke ring.
31: A second-generation head then develops at the point of detachment.
32: Multiple generations of chemical vortex rings can detach from a single
33: triggering event.
34:
35: \end{abstract}
36:
37: \pacs{47.20.Bp, 47.70.Fw}
38:
39: \maketitle
40:
41: Hydrodynamic flows driven by chemical reactions have been studied since the
42: discovery of fire, the oldest technology of humankind~\cite{combust}.
43: Exothermic combustion reactions produce buoyant flames in the form of rising,
44: reacting plumes and thermals which are often highly turbulent~\cite{fires}. In this Letter we examine
45: a much simpler, yet still very rich, new system in which relatively gentle
46: autocatalytic reaction fronts give rise to laminar plumes and thermals in the form of free vortex rings.
47: Unlike conventional non-reacting plumes and thermals arising from localized
48: heat sources, these flows have an internal source of buoyancy which is enhanced by subsequent entrainment, leading to flow acceleration similar to combustion. They provide a new and especially elegant model system in which to study instabilities in reacting flows.
49:
50: Many autocatalytic chemical reactions exhibit sharp fronts. The well-studied reaction we employed was the iodate oxidation of arsenous acid~\cite{hanna}. The short range diffusion of the autocatalyst, here the iodide ion, limits
51: the reaction to a very thin front separating reacted and unreacted solution.
52: The front region of active reaction is typically only a few tens of microns thick,
53: and is by far the smallest lengthscale in the system.
54: Under gravity, hydrodynamic convection can be driven by buoyancy due to the density jump
55: across the front. Density change is created both by thermal expansion due to the
56: slight exothermicity of the reaction, and by the partial molal density decrease of
57: the product solution~\cite{pojman}. Ascending fronts are hydrodynamically unstable because they put less dense products below denser unreacted solution~\cite{hanna,pojman,vasquez_01,maser,wilder,huang,carey,bockmann,dewit, edwards, leconte}.
58:
59:
60: In our plume experiments, we increase the viscosity
61: of the water solution by adding glycerol. An ascending front is allowed to escape from
62: a capillary tube into a much larger volume, so that boundaries no longer
63: constrain the shape of the front. In the larger volume, a rising plume with
64: a well-defined head is formed. During the early stages of the evolution of the
65: head, its growth is due to both a volume flux
66: of upwelling product solution and by the entrainment of reactant solution.
67: Later, the plume head detaches and becomes a rising vortex ring, analogous to a
68: chemically reacting smoke ring. The upwelling conduit that is left behind
69: forms a new head which also eventually detaches, leading to generations of
70: accelerating vortex rings. This surprising behaviour is in contrast to conventional buoyant vortex rings,
71: which do not accelerate. It may have application as a laboratory analog of plumes in internally heated fluids, such as the Earth's mantle~\cite{mantle, griffiths_01}.
72:
73: %
74: % background on previous IAA convection experiments
75: %
76: The propagation of iodate-arsenous acid (IAA) reaction fronts has enjoyed
77: considerable theoretical
78: and experimental attention. It has been well-studied in capillary
79: tubes~\cite{pojman,vasquez_01,maser}, in thin
80: slots~\cite{wilder,huang,carey,bockmann,dewit},
81: and in the presence of a superposed
82: flow~\cite{edwards, leconte}. In sufficiently narrow vertically oriented capillary
83: tubes, convection is suppressed and the sharp reaction front is observed to
84: propagate at constant speed in either direction. If the tube radius $a$ is
85: increased beyond a critical size, the front deforms and upward propagation
86: is accompanied by convection~\cite{vasquez_02}. Within a simple thin front
87: approximation~\cite{edwards}, the relative importance of buoyancy is
88: described by a dimensionless quantity
89: \begin{equation}
90: S=\frac{\delta g a^3}{\nu D_c},
91: \end{equation}
92: where $g$ is the acceleration due to gravity, $\nu$ is the kinematic
93: viscosity, $D_c$ is the diffusion constant of the autocatalyst, and
94: $\delta=(\rho_u/\rho_r)-1$ is the dimensionless density jump
95: between the reacted ($\rho_r$) and unreacted ($\rho_u$) solutions.
96: In practice, for the IAA reaction, $\delta \sim 10^{-4}$. The critical tube
97: radius for convection corresponds to a critical value $S_c \sim 90$. At
98: slightly larger values $S > S_c$, a transition to an axisymmetric mode of
99: convection can be identified~\cite{pojman}. In the plume experiments
100: described below, a very large tube was used, corresponding to $S \approx 10^{8}$.
101:
102: In all previous studies of the IAA reaction~\cite{hanna, pojman, vasquez_01, maser, wilder, huang, carey, bockmann, dewit, edwards, leconte}, convection effects were severely
103: constrained by the viscous interaction with nearby solid boundaries. We set out to find the reaction-driven flow phenomenology for an {\it unbounded} solution in which the reaction is initiated at a point, as nearly as possible. Two experimental difficulties immediately arise. For a water solution, the IAA reaction produces fronts that are
104: sufficiently buoyant that the resulting plumes ascend quickly to the upper
105: boundaries of any tank of managable size. These plumes also had Reynolds numbers, $Re \sim 7$
106: and therefore had a rather complex internal
107: flow structure. We therefore explored the reaction in a water-glycerol solution in which the increased viscosity acts to limit the
108: Reynolds number to $Re \sim 0.1$, and to allow for an effectively longer evolution time within the experimental tank. These experiments reveal plumes with well-defined heads which eventually pinch-off to form essentially free vortex rings.
109:
110: The second difficulty concerns initiating the front at a localized point. When the front is released from a very narrow capillary tube with diameter $0.9$ mm into a much larger volume, we found that the front did not propagate. This ``front death" phenomenon~\cite{front_death} is apparently due to the requirement of a minimum critical concentration of autocatalyst near the initiation region, which therefore cannot be too small. To our knowledge, this effect has not been observed in the IAA reaction before. This constraint is easily overcome by using a large enough capillary tube to launch the reaction into the unbounded region. A diameter of $ 2.7$ mm proved to be sufficient.
111:
112: %
113: % experimental section
114: %
115: Reactant solutions were prepared using
116: reagent grade chemicals and distilled water with 40\% glycerol by volume.
117: Iodate stock solutions were prepared by dissolving ${\rm KIO_3}$ powder
118: in distilled water. Arsenous acid stock solutions
119: were prepared from ${\rm As_2O_3}$ powder. These stock solutions were diluted so that the working
120: solution contained [${\rm IO_3^-}$]=0.005M and [As(III)]=0.020M.
121: These concentrations were not varied in the experiments we report here. Congo red indicator was used to visualize the reaction fronts. Congo red has a pH range of 3.0 to 5.0, where the acid form is blue and the base form is red.
122: The reaction front leaves in its wake a product solution with pH of $\sim 2.7$ so that
123: the upwelling blue products are easily visible within the red unreacted solution.
124:
125: The apparatus is shown in Fig.~\ref{schematic}.
126: The reaction tank was a large glass cylinder sealed by large rubber stoppers. A capillary
127: tube entered through a hole
128: in the lower stopper. The outside end of the capillary
129: tube was sealed with a short rubber tube clamped at one end which formed
130: the initiation volume. The
131: rubber tube was filled by a porous plug made of loosely packed cotton. Reactions
132: were initiated by inserting a thin needle into the rubber
133: tube and then injecting a very small amount of product solution
134: into the plug. The plug served to quench any
135: hydrodynamic disturbance of the reactant solution during reaction initiation. The reaction proceeded up the capillary tube, which is large enough that $S > S_c$ and some convection is already present. It then escapes into the larger tank.
136: The apparatus was illuminated from behind and still images of the
137: evolving front were captured using a digital camera. The temperature was constant at $24.0 \pm 0.2^\circ$~C.
138:
139: \begin{figure}
140: \includegraphics[height=5.4cm]{apparatus_new.eps}
141: \caption{\label{schematic}
142: A schematic of the apparatus. The main volume is a cylinder 32~cm long and 8.9~cm in diameter. The capillary tube has an inner diameter of 2.7~mm. }
143: \label{apparatus}
144: %\vspace{-0.40cm}
145: \end{figure}
146:
147:
148: The various stages in the evolution of the resulting free autocatalytic plume are shown in Fig.~\ref{plumes}.
149: Initially the plume rises out of the capillary tube and its
150: head remains roughly spherical and grows slowly. This stage of growth is shown in Fig.~\ref{plumes}a. In this initial stage, there is essentially no entrainment
151: of reactant solution into the plume head.
152:
153:
154: %Entrainment begins once
155: %$\alpha$ reaches a critical value of $\alpha_{1c} = 2.6 \pm 0.1$.
156: At a well-defined point, entrainment sets in. Fig.~\ref{plumes}b shows a plume head well past the onset of entrainment. This shows the familiar mushroom-shaped head for a plume in which the surrounding fluid is being drawn into the head by a single overturning vortex ring.
157:
158: Once the entraining autocatalytic plume head reaches a second critical size,
159: %$\alpha_{2c}=4.6 \pm 0.1$,
160: the plume head begins to pinch-off from the upwelling conduit.
161: The pinch creates a bottleneck in the conduit which then swells as it fills with
162: rising product solution. During the swelling below the bottleneck, what was
163: formerly the head of a starting plume becomes an essentially free vortex
164: ring. As shown in Fig.~\ref{plumes}c,
165: the vortex ring detaches from the conduit, and is eventually connected only by a
166: very thin filament of product solution. The swelling in the conduit develops into a new, second generation plume head. In Fig.~\ref{plumes}d, a new pinch-off process has
167: started again below the second
168: generation plume head. Yet another bottleneck is formed when the second generation
169: head pinches-off and becomes a vortex ring. The new bottleneck swells and
170: the process is repeated. We have observed four generations of pinch-off and
171: subsequent vortex ring formation in autocatalytic plumes. The
172: fourth generation plume head eventually reaches the top boundary of
173: the reaction vessel, closing the possibility of further pinch-off~\cite{movie}.
174:
175: >From the digital images of plume evolution, we determined the height and width of the
176: ascending buoyant plume as a function of time. Height
177: was measured from the top of the front to where it exited the capillary
178: tube, and the width refers to the maximum width across the head or
179: vortex ring. It is useful to define the dimensionless ratio $\alpha = w_h / w_c$, where $w_h$ is the width of the head and $w_c$ is the width of the conduit. Only the first generation head and vortex ring were considered in this analysis.
180:
181: We found that the initial stage of growth ends and entrainment begins once
182: $\alpha$ reaches a critical value of $\alpha_{c1} = 2.6 \pm 0.1$. Similarly, once
183: the entraining plume head reaches a second critical size, $\alpha_{c2}=4.6 \pm 0.1$,
184: the plume head begins to pinch-off from the upwelling conduit.
185:
186: \begin{figure}
187: \includegraphics[height=6.2cm]{plumes_photo_sequence.eps}
188: \caption{\label{plumes}
189: A sequence of images showing the evolution of the plume structure. (a) 202~s, (b) 328~s, (c) 458~s and (d) 554~s after release.}
190: \label{plumes_photo}
191: %\vspace{-0.40cm}
192: \end{figure}
193:
194: Typical width and height data for first generation
195: head formation of an autocatalytic plume is shown in Fig.~\ref{plumedata}. We observed that the head accelerates during its ascent. The acceleration of the plume head continues
196: after it has pinched-off, even though it is almost entirely disconnected
197: from the conduit. This acceleration is in contrast to the constant velocity
198: rise behavior previously observed in conventional plumes~\cite{moses}. The
199: continued acceleration of the head once it pinches-off and becomes a thermal
200: is also unique to autocatalytic plumes; normal vortex rings expand and slow down after
201: they have pinched-off~\cite{turner_01}.
202:
203:
204: \begin{figure}
205: \includegraphics[height=8cm]{ht_width_plot_fits.eps}
206: \caption{\label{plumedata} (Color online)
207: The growth of the plume as a function of time, up to the completion of the first pinch-off, showing (a) the width of the head, and (b) its height. Colored lines in (a) show the nearly linear evolution of the width in the three regimes indicated. Vertical arrows indicate the positions of images (a), (b) and (c) from the previous figure. }
208: \label{plume_data}
209: %\vspace{-0.40cm}
210: \end{figure}
211:
212: Fig.~\ref{plumedata}a shows that the width of the plume goes through
213: three distinct regions of approximately linear growth.
214: The initial, and slowest, growth rate obtains when $\alpha < \alpha_{c1}$ and there is
215: no entrainment into the plume head and it remains roughly spherical.
216: A transition to a second, increased growth regime occurs
217: once entrainment begins and $ \alpha_{c1} < \alpha < \alpha_{c2}$. Finally, the third and fastest growth regime commences once $\alpha > \alpha_{c2}$ and the plume head detaches and becomes a thermal. In experiments on conventional plumes
218: the growth rate of the head has been determined~\cite{moses} to scale with time $t$ as $t^\beta$, where $\beta=0.54\pm 0.05$. This is clearly not the case for autocatalytic plumes.
219:
220: %
221: % review of the connection with previous plume experiments
222: %
223: To our knowledge, our experiments mark the first study of buoyant plumes
224: that are directly driven by an autocatalytic reaction. Of course, buoyant
225: plumes~\cite{moses,morton,turner_02,turner_03,griffiths_01} and buoyant
226: vortex rings~\cite{turner_01,turner_03,lundgren_01} have been the subjects of extensive previous
227: study. Refs.~\cite{fay,list,gebhart} review much of the older work. Some clarification of nomenclature is useful here. A {\it plume} is a continuous region of buoyancy supplied by a point source. A {\it starting plume} is a plume with a well-defined, advancing head. These cases of buoyancy driven flow may be distinguished from a {\it thermal}, which is a
228: freely evolving buoyant object disconnected from its source. Thermals
229: may take the form of vortex rings, similar to smoke rings.
230:
231: There is great diversity in past experiments on {\it
232: passive} plumes, where we use {\it passive} to mean flows that do not react internally to produce buoyancy. In typical experiments, passive plumes are formed from
233: buoyant fluid that has been injected into a less-buoyant medium at a point source. The buoyancy may be due to a temperature or composition difference or both. In
234: this case, however, all the buoyancy is derived from processes external to the experiment
235: and arrives with the volume of injected fluid. Subsequent stirring can only dilute and reduce this buoyancy. In the case of autocatalytic plumes, the volume of buoyant fluid injected is negligible and all of the buoyancy is generated by the conversion of dense fluid into buoyant fluid at the thin reaction front. The volume of buoyant fluid is steadily increasing, particularly in the head.
236:
237: Symmetrical plume heads similar to those we observe can be found
238: in passive starting plumes~\cite{griffiths_01}. Moreover, the heads
239: of passive buoyant plumes can be observed to pinch-off to form buoyant thermals in the form of vortex rings. The formation thermals by pinch-off has been the subject of theoretical and
240: of experimental interest in passive systems~\cite{lundgren_02,fabris, gharib, shusser}.
241: The formation of discrete thermals is a characteristic of systems in
242: which the ratio of the viscosities of the buoyant fluid
243: and its surroundings is of order 1~\cite{jellinek}.
244:
245:
246: There is a critical difference between the passive vortex rings
247: formed by pinch-off and those that we observe in autocatalytic reactions. As the head
248: or vortex ring rises, it entrains surrounding unreacted fluid, stirring it into the
249: plume head and accelerating the production of new buoyancy. This feedback mechanism
250: is unique to autocatalytic plumes. In our plumes, this stirring is laminar, rather than turbulent, and presumably involves the extreme stretching of the thin reaction front within the vortex ring.
251:
252:
253: Autocatalytic plumes may have a useful analogy to important geophysical processes.
254: Conventional plumes have been the subject of numerous laboratory analog experiments
255: aimed at understanding how material upwelling from deep in the mantle forms oceanic and
256: continental hotspots~\cite{griffiths_01}. In addition to plumes, instabilities in the bottom layer
257: of the mantle are also likely to give rise to thermals. Such thermals may incorporate internal sources of buoyancy due to the heat supplied by the breakdown of radionuclides. Internally heated thermals are one plausible explanation for ocean island volcanism~\cite{griffiths_02}. The buoyancy-producing capabilities of autocatalytic thermals could serve as an interesting laboratory analog of this process.
258:
259:
260: We have described a study of buoyant, three-dimensional plumes driven by the iodate-arsenous acid reaction. In a water solution that was made more viscous with the
261: addition of glycerol, we found that a rising plume with a well-defined
262: head was formed. The plume head grew to a critical size and subsequently
263: detached from the upwelling conduit to form a buoyant vortex ring. The
264: entrainment of reactant solution into the autocatalytic chemical plume head
265: and subsequent vortex ring produced additional buoyancy by assisting the
266: chemical reaction, leading to an acceleration of the vortex ring. This behavior reverses the usual role of entrainment, which normally acts to reduce buoyancy in conventional passive plumes and thermals. Autocatalytic chemical smoke rings offer some simple insights into combustion-driven flow and certain geophysical processes.
267:
268: We thank M. Menzinger, A. De Wit, and A. M. Jellinek for helpful discussions and constructive comments. This research was supported by the Natural Science and Engineering Research Council of Canada.
269:
270: %
271: % references
272: %
273:
274: \begin{references}
275:
276: %1
277: \bibitem{combust} J. Warnatz, U. Maas, and R.W. Dibble, {\it Combustion},
278: 3rd ed., Springer (2001).
279:
280: %2
281: \bibitem{fires} S. R. Tieszen, Ann. Rev. Fluid Mech. {\bf 33}, 67 (2001).
282: {\bf 84}, 4357 (2000).
283:
284: %3
285: \bibitem{hanna} A. Hanna, A. Saul, and K. Showalter, J. Am. Chem. Soc.
286: {\bf 104}, 3838 (1982).
287:
288: %4
289: \bibitem{pojman} J. A. Pojman, I. R. Epstein, T. J. McManus, and
290: K. Showalter, J. Phys. Chem. {\bf 95}, 1299 (1991).
291:
292: %5
293: \bibitem{vasquez_01} D. Vasquez, J. Wilder, and
294: B. Edwards, Phys. Fluids A {\bf 4}, 2410 (1992).
295:
296: %6
297: \bibitem{wilder} J. W. Wilder, D. A. Vasquez and B. F. Edwards,
298: Phys. Rev. E {\bf 47}, 3761 (1993).
299:
300: %7
301: \bibitem{huang} J. Huang, D. A. Vasquez, B. F. Edwards and P. Kolodner,
302: Phys. Rev. E {\bf 48}, 4378 (1993).
303:
304: %8
305: \bibitem{maser} J. Maser, D. A. Vasquez, B. F. Edwards, J. W. Wilder and
306: K. Showalter, J. Phys. Chem. {\bf 98}, 6505 (1994).
307:
308: %9
309: \bibitem{carey} M. R. Carey and S. W. Morris and P. Kolodner,
310: Phys. Rev. E {\bf 53}, 6012 (1996).
311:
312: %10
313: \bibitem{bockmann} M. Bockmann and S. C. M{\"u}ller,
314: Phys. Rev. Lett. {\bf 85}, 2506 (2000).
315:
316: %11
317: \bibitem{dewit} A. De Wit, Phys. Rev. Lett. {\bf 87}, 054502 (2001).
318:
319: %12
320: \bibitem{edwards} B. F. Edwards,
321: Phys. Rev. Lett. {\bf 89}, 104501 (2002).
322:
323: %13
324: \bibitem{leconte} M. Leconte, J. Martin, N. Rakotomalala, and D. Salin,
325: Phys. Rev. Lett. {\bf 90}, 128302 (2003).
326:
327: %14
328: \bibitem{mantle} G.F. Davies, M.A. Richards, J. Geology, {\bf100}, 151 (1992).
329:
330: %15
331: \bibitem{griffiths_01} R. W. Griffiths, I. H. Campbell, Earth Planet. Sci.
332: Lett. {\bf 99}, 66 (1990).
333:
334: %16
335: \bibitem{vasquez_02} D. A. Vasquez, B. F. Edwards, and J. W.
336: Wilder, Phys. Rev A {\bf 43}, 6694 (1991).
337:
338: %17
339: \bibitem{front_death} E. Jakab, D. Horv{\'a}th, J. H. Merkin, S. K. Scott, P. L. Simon and A T{\'o}th, Phys. Rev. E., {\bf 66}, 016207 (2002).
340:
341: %18
342: \bibitem{movie} For a movie of this process, see\\ http://www.physics.utoronto.ca/nonlinear/movies/plume.html
343:
344: %19
345: \bibitem{moses} E. Moses, G. Zocchi, A. Libchaber, J. Fluid. Mech. {\bf
346: 251}, 581 (1993).
347:
348: %20
349: \bibitem{turner_01} J. S. Turner, Proc. R. Soc. Lond. A {\bf 239}, 61
350: (1957).
351:
352: %21
353: \bibitem{morton} B. R. Morton, G. Taylor, and J. S. Turner, Proc. R. Soc.
354: Lond. A
355: {\bf 234}, 1 (1956).
356:
357: %22
358: \bibitem{turner_02} J. S. Turner, J. Fluid. Mech. {\bf 13}, 356 (1962).
359:
360: %23
361: \bibitem{turner_03} J. S. Turner, {\it Buoyancy Effects in Fluids},
362: Cambridge University Press (1973).
363:
364: %24
365: \bibitem{lundgren_01} T. S. Lundgren, N. N. Mansour, J. Fluid. Mech. {\bf 224},
366: 177 (1991).
367:
368: %25
369: \bibitem{fay} J. A. Fay, Ann. Rev. Fluid. Mech., {\bf 5}, 151 (1973).
370:
371: %26
372: \bibitem{list} E. J. List, Ann. Rev. Fluid. Mech., {\bf 14}, 189 (1982).
373:
374: %27
375: \bibitem{gebhart} B. Gebhart, D. S. Hilder, M. Kelleher, Adv. Heat Transfer
376: {\bf 16}, 1 (1984).
377:
378: %28
379: \bibitem{lundgren_02} T. S. Lundgren, J. Yao, N. N. Mansour, J. Fluid. Mech. {\bf 224},
380: 177 (1992).
381:
382: %329
383: \bibitem{fabris} D. Fabris and D. Leipmann, Phys. Fluids {\bf 9}, 2801
384: (1997).
385:
386: %30
387: \bibitem{gharib} M. Gharib, E. Rambod, and
388: K. Shariff, J. Fluid Mech. {\bf 360}, 121 (1998).
389:
390: %31
391: \bibitem{shusser} M. Shusser and M. Gharib, J. Fluid Mech. {\bf 416}, 173
392: (2000).
393:
394: %32
395: \bibitem{jellinek} A. M. Jellinek, A. Lenardic, M. Manga, Geophys. Res. Lett. {\bf 29}, 10.1029/2001GL014624.
396:
397: %33
398: \bibitem{griffiths_02} R. W. Griffiths, Earth Planet. Sci.
399: Lett. {\bf 78}, 435 (1986).
400:
401:
402: \end{references}
403: \end{document}