DOI: 10.1021/cg1003098
Morphological Investigations of Methane-Hydrate Films Formed on a Glass Surface
2010, Vol. 10 4339–4347
Juan G. Beltran† and Phillip Servio*,‡ †
Department of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, Ontario, Canada, and ‡Department of Chemical Engineering, McGill University, Montreal, Quebec, Canada Received March 9, 2010; Revised Manuscript Received July 13, 2010
ABSTRACT: Methane clathrate formation on water films without previous hydrate formation history was studied. It was found that hydrates form in two layers, each with a clearly distinct morphology. As the clathrate aged, this difference in morphology tended to disappear. Higher driving forces produced smaller hydrate grains and smoother surfaces than lower driving forces within the water film. The converse was true for hydrate formed on the periphery. A third hydrate layer growing outside of the original water boundary was also observed. For the first time, it was shown that this growing front could induce nucleation by creating a “bridge” between segregated water droplets. Hydrate reformation was also studied and it was found that it proceeded in a manner different than that of a system without previous clathrate formation history. For reformation, nucleation occurred within the film and a circular hydrate front(s) advanced toward the periphery of the water film with uniform morphology.
*To whom correspondence should be addressed. E-mail: phillip.servio@ mcgill.ca. Phone: þ1 514 3981026. Fax: þ1 514 398 6678.
Sloan16 on methane and CO2 hydrate grown inside a sapphire tube. Sugaya and Mori17 presented one of the first studies on hydrates formed from bubbles of a fluorocarbon immersed in water; it was found that the surface morphology of the hydrate layer formed at the interface depends strongly on the degree of saturation of the water phase with the guest component. Ohmura and Mori18 expanded the study of Sugaya and Mori17 to include a different type of refrigerant; in addition, it was hypothesized that a concentration driving force could explain the differences in morphology observed at high and low driving forces. The crystal-growth behavior of structure I, structure II, and structure H was described by Smelik and King;19 it was proposed that the characteristic crystal morphology of each structure could be used to identify hydrate types in situ. Uchida and co-workers have published a number of articles on the morphology of CO2 hydrate.7,20-22 The first paper focused on determining the interfacial tension between liquid CO2, water, and hydrate, yet illustrations of their observations were also provided.20 Later, still frames of video recordings of hydrate formation on water droplets suspended in liquid CO2 were presented.7 It was concluded that hydrates formed initially at the interface between the two fluids and proceeded until complete coverage of the water droplet occurred; on some occasions secondary nucleation was also observed inside the water droplet.7 Uchida et al.22 measured the lateral growth rates of CO2 hydrate films on water bubbles to range from 6 to 15 mm/s, and concluded that subcooling was the parameter that highly determined the growth rate of the crystal. Servio and Englezos23 studied the morphology of methane and carbon dioxide hydrates formed on nearly spherical, adjacent water droplets. It was found that the type of hydrate guest did not have an effect on the crystal morphology. It was also reported that hydrate nucleation would occur almost instantaneously in all droplets after one of them had nucleated, and it was suggested that an imperceptible water “bridge”
r 2010 American Chemical Society
Published on Web 08/18/2010
Introduction Clathrate hydrates are nonstoichiometric, crystalline compounds that form when small molecules come in contact with water at appropriate temperatures and pressures. The terms “gas hydrates” and “clathrate hydrates” are now used interchangeably to designate this kind of compound.1 Natural-gas hydrates are abundantly found in the ocean bottom, and to a lesser extent in permafrost regions.2 Conservative estimates suggest that the amount of energy stored in natural hydrates is at least twice that of all other fossil fuels combined.3 In addition, trapping carbon dioxide as a hydrate in the bottom of the ocean has been proposed as an alternative to reduce increasing atmospheric CO2 concentration.4 Hydrates may also play a key role in the global carbon cycle3 and could constitute an alternative means to transport natural gas.5 Hydrate growth at the water-guest interface has been studied by various groups as described below. Observational studies of this kind provide a physical picture of the phenomena that occur upon hydrate crystallization on the water surface. For example, it is generally accepted that in quiescent systems hydrates preferentially form at the water-guest interface and grow in the form of a thin film covering the same.6 Several modeling efforts have built upon this generalization.7-10 Thus, having a clear picture of the events that occur upon crystallization on the water surface determines the type of model you can use for hydrate systems, and it is in this sense that morphology studies increase our understanding of hydrate nucleation, growth, and dissociation mechanisms. Pioneering morphology work was done by Makogon and co-workers11 and by Maini et al.12 Other early observational studies include those of Mori and Mori13,14 on refrigerants, that of Hwang et al.15 on methane hydrate formed from ice supported on stainless-steel discs, and that of Long and
pubs.acs.org/crystal
4340
Crystal Growth & Design, Vol. 10, No. 10, 2010
between the droplets was the plausible explanation for this phenomenon.23 Furthermore, when hydrates were formed at high driving forces water droplets were observed to become jagged, and numerous needle-like crystals were seen to extend away from the water surface into the gas phase. This was not the case at low driving force, where the hydrate surface was observed to be smooth and shiny.23 Servio and Englezos23 also reported that rough hydrate surfaces appeared when reforming hydrates 24 h after dissociation, but smooth surfaces were seen when the wait period was reduced to 30 min. Perhaps the most successful systematic approaches to correlate the observed clathrate morphology with the corresponding driving force have been put forward by Ohmura et al.6,18,24,25 On the basis of hypothesized concentration differentials between the liquid phase near the newly formed hydrate and the liquid bulk concentration, they were able to match images to their proposed driving force.6,18,24 Very recently, the same group used subcooling as an indicator for the driving force of the system; through this index it was then possible to correlate the driving force with the observed hydrate morphology.25 As a general trend, it was observed that subcoolings g3 K produced sword-like crystals, whereas at smaller subcooling, polygonal faces appeared, the size of the latter increasing with decreasing subcooling.25 Ohmura and Mori18 worked with liquid droplets of refrigerant R-141b (CH3CCl2F) in water and distinguished two stages of hydrate-crystal growth, which were very different from each other. The primary stage was characterized by lateral growth of a thin, fine-grained polycrystalline layer along the surface of each R-141b drop; the secondary stage began typically after 10 min and lasted for several hours: it was characterized by radial growth of plate-like crystals growing from the outer surface of the hydrate shell formed in the primary stage.18 The second stage was never observable with pure water (not presaturated) and/or at a small subcooling. Ohmura and Mori18 also observed that the hydrate surface formed in nonpresaturated water smoothed over time. In subsequent work, Ohmura et al. extended the study done on presaturated water to other hydrate formers including carbon dioxide6 and methane.24 With carbon dioxide, it was observed that when subcooling was greater than 3 K, a hydrate film first grew along the carbon dioxide-water interface; then hydrate crystals with dendritic morphology grew into the liquidwater phase from that hydrate film.6 With subcoolings less than 2 K, it was found that dendritic crystals were replaced by skeletal or polyhedral crystals.6 Similar conclusions were presented for the methane-water system.24 Using laser interferometry, Ohmura et al.26 measured the thickness of a R-134a (CH2FCF3) hydrate film and concluded that initially the film could be as thick as 80 μm. Hirai et al.,27 Freer et al.,8 and Kobayashi et al.28 have studied growth of hydrate films. The former27 and the latter8 measured velocities of the advancing hydrate film for CO2 and methane respectively by successively recording images. Hirai et al.27 reported that lateral growth rates could vary from 0.044 mm/s at 284.2 K and 39.2 MPa to 6.5 mm/s at 278.7 K and 39.2 MPa for CO2 hydrate. Freer et al.8 determined methane hydrate growth rates varying from 20 μm/s at 3.55 MPa and 1 °C to 690 μm/s at 9.06 MPa and 1 °C. Building on many years of experience and extensive compilation of hydrate data from various laboratories, Sloan’s group has presented a conceptual picture for the growth mechanism of hydrates at the hydrocarbon-water interface in quiescent systems.29 In this conceptual picture, a thin porous
Beltr an and Servio
Figure 1. Simplified schematic of the experimental apparatus.
hydrate film propagates across the hydrocarbon-water interface until complete coverage of the same, at which point the hydrate film begins to thicken, eventually fully converting the water to hydrate (in the case of water droplets).29 To summarize, it can be said that it is generally accepted that morphology changes are independent of the guest molecule, but are influenced by the driving force. The latter is understood as the deviation of experimental conditions from those at equilibrium. As mentioned above, there have been successful attempts to correlate a unique morphology with various indices of experimental deviation from equilibrium,6,18,24,25 and a conceptual picture of hydrate growth in quiescent systems has been proposed.29 Furthermore, the memory of the hydrate systems has been identified as a factor that affects hydrate morphology.23 Here we provide new insights into the mechanism of gas hydrate formation in unstirred systems. A novel experimental design, coupled with high-resolution video microscopy, allowed detailed morphological observations of the initial stages of clathrate formation, as well as aging and dissociation of hydrates in water films. We provide evidence of various hydrates morphologies occurring simultaneously (at constant temperature and pressure) and show how the occurrence of these morphologies depends on the memory of the system. Furthermore, the observed layered structure of hydrates formed on water films was found to have interesting implications for hydrate propagation on substrates. Experimental Procedures The experimental design enabled the observation, with the aid of a microscope, of a nearly planar water film resting on a glass slide, immersed in a methane atmosphere under controlled temperature and pressure. The heart of the setup was a 316 stainless steel cell, fitted with two sapphire windows on the top and bottom (Figure 1). The cell had several ports used as follows: to feed gas, to purge gas out of the cell, to insert a thermocouple, and to communicate pressure to a pressure transducer. Temperature was measured with a type K mini thermocouple probe ((1 K) (Omega Engineering, QC, Canada). Pressure was monitored with a Rosemount 3051S pressure transducer
Article
Crystal Growth & Design, Vol. 10, No. 10, 2010
(Rosemount, MN, USA) with an accuracy of (0.1% of the span. A copper coil, wound around the high pressure cell and connected to a Neslab RTE740 refrigerated bath (Fisher Scientific, Canada), was used to circulate a mixture of ethylene glycol and water (50/50, V/V) to provide the cooling necessary to reach hydrate forming conditions. A Schott KL 2500 (Optikon Corporation, ON Canada) cold light source fitted with an articulated light pipe illuminated the interior of the reactor through the bottom window. Images were acquired with a PCO.2000 (Optikon Corporation, ON, Canada) high-resolution video camera fitted to a configuration of KC Infinity lenses (Optikon Corp, ON, Canada) that allowed an optical magnification of up to 5. The video camera, the temperature signal, and the pressure signal were connected to a personal computer in order to acquire and analyze the data. The high pressure cell, the video camera, and the light source were mounted on an optical table in order to minimize the effect of environmental vibration. Before starting an experiment a new, precleaned, microscope glass slide was cut to fit inside the high pressure cell (Figure 1). A drop (35 μL) of distilled, deionized water was deposited on the glass slide. Methane gas, 99.99% purity (MEGS, QC, Canada), was then fed to the reactor and the cell content was purged several times to remove any air inside the reactor. After purging, the pressure in the reactor was increased well above the hydrate-liquid-vapor equilibrium line for methane until hydrate nucleation was observed. A hydrate formation experiment was terminated when it became apparent that no liquid water was left on the microscope slide. At this point, it was decided whether an aging or a dissociation experiment would be carried out. An aging experiment consisted of observing the evolution of formed hydrates while keeping the system well within the hydrate stability region at constant temperature and pressure. Dissociation experiments were initiated by slowly decreasing the pressure inside the reactor down to a preset value and maintaining constant pressure thereafter. A dissociation experiment was terminated after Table 1. Experimental Conditions for Hydrate Formation on Water Films without Previous Hydrate Formation History experiment
T/K
p/MPa
growth beyond water boundary
1 2 3 4
274 274 275 275
8.2 8.2 3.6 4.0
yes yes yes yes
4341
having left the reactor for several hours at a pressure well below the gas hydrate stability region. Reformation experiments were carried within minutes of having completed a dissociation experiment, thus performed on water droplets with previous hydrate formation history.
Results Nucleation and Growth. The experimental conditions for hydrate formation without previous clathrate formation history are summarized in Table 1. Figure 2 presents a sequence of still frames of hydrate nucleation and growth (T = 275 K, p = 3.6 MPa) on a water film with no previous hydrate formation history. The intact water surface, prior to nucleation, can be seen in Figure 2a. Nucleation occurred on the periphery of the water film first; hydrate appeared as a whiskery structure that grew toward the center of the water film (Figure 2b). While clathrate growth proceeded from the edge toward the center, keeping a whiskered appearance, numerous nucleation sites became visible as white spots within the water film (Figure 2c). Hydrate propagation continued from the periphery and from the center (Figure 2d), eventually covering the whole water film (about 3 cm2) in less than 15 s (Figure 2e). Following complete hydrate coverage of the water surface, a considerably slower process was observed: the growth of a hydrate layer outside of the original water boundary (Figure 2f). The growth of this layer was observed both at low driving forces and high driving forces (Figure 3 and Table 1). A detailed view of this layer can also be seen to the right of Figure 4. Figure 3 shows water films covered completely by hydrate. In Figure 3b, nucleation and hydrate growth occurred at a much higher driving force (T = 274 K, p = 8.2 MPa) than in Figure 3a (T = 275 K, p = 3.6 MPa). Hydrate growth proceeded in a similar fashion at high and low driving forces; however, the process was much faster at a high driving force where complete water film coverage was achieved in less than 3 s (Figure 3b).
Figure 2. Hydrate formation and growth on a water film with no previous hydrate formation history. T = 275 K, p = 3.6 MPa. (a) Water film before hydrate formation. (b) t = 0 s, nucleation occurs at the periphery of the film; newly formed clathrate appears as a white whiskered layer. (c) t = 2.5 s, growth of the hydrate from the edges toward the center of the water film. New nucleation sites appear in the center of the water film as white spots. (d) t = 6 s, the spotted hydrate and the whiskered hydrate continue to grow. (e) t = 14.5 s, clathrate propagates until the water film is covered completely. (f) t = 42.5 s, bottom right, hydrate extends outside of the original water boundary.
4342
Crystal Growth & Design, Vol. 10, No. 10, 2010
Figure 3. Water films completely covered by hydrate at a low driving force (left) and high driving force (right). Three clearly different regions are appreciable: interior of the film, periphery of the same, and clathrate extending outside of the original water boundary. (a) Nucleation and growth occurred at T = 275 K, p = 3.6 MPa. (b) Nucleation and growth occurred at T = 274 K, p = 8.2 MPa.
Figure 4. Detail the hydrate film formed at high driving force. Translucent layer: hydrate growing outside of the original water boundary. Light gray: hydrate formed in the periphery. Darker gray: hydrate formed within the water film. Hydrate formed at T = 274 K, p = 8.2 MPa.
Comparison of Figure 3, panels a and b reveals three common features. First, the water boundary has a clear effect on the morphology of newly formed hydrates. Second, two types of morphologies can be clearly seen in the water films covered with hydrate: one within the film and another one in the periphery of the same. Third, the clathrate extends beyond the original water boundary. Two clear differences can also be appreciated. First, the high driving force nucleation shows a rather smooth surface (Figure 3b), whereas a striated pattern is observed at a low driving force (Figure 3a). Second, the hydrate annulus on the periphery is about twice the width at a high driving force (Figure 3b) than that at a low driving force (Figure 3a). Figure 4 presents a detailed view of the layers that can be observed after hydrate formation at a high driving force (for lower magnification and broader field of view, see Figure 3b). In Figure 4 the translucent structure corresponds to the hydrate layer that grew outside of the original water boundary (described further in the “Bridge Effect” section), the somewhat darker formation corresponds to the clathrate that crystallized in the periphery of the water film, and the black granular morphology is characteristic of the hydrate that formed within the water film. The hydrate film formed at low driving force is shown in detail in Figure 5. Focus is on the two hydrate morphologies that formed inside the water film in order to show the distinct
Beltr an and Servio
Figure 5. Detail of the hydrate film formed at low driving force. Light gray: hydrate formed in the periphery. Darker gray: hydrate formed within the water film. Hydrate formed at T = 275 K, p = 3.6 MPa.
nature of the hydrate grains. Toward the right of Figure 5 and somewhat shiny is the hydrate formed on the periphery of the film. The large clathrate grains formed in the interior of the cell appear dark gray (Figure 5) (for lower magnification and broader field of view, see Figure 3a). There was a clear effect of the water boundary both at high and low driving force, producing a particular morphology on the film periphery; however, at a low driving force (Figure 5) this annulus was smoother and narrower than at a high driving force (Figure 4). The contrary seemed to be true for the inside of the hydrate film where coarse grains were observed at a low driving force (Figure 5) and finer grains at a high driving force (Figure 4). Aging. By focusing on a particular region near the edge of the water surface, it was possible to observe the evolution of the formed hydrate over a period of 3 days (Figure 6). Figure 6a shows both the interior and the edge of the formed hydrate 20 min after complete coverage of the water surface. Two different morphologies are clearly appreciated: a smooth shiny hydrate formed at the edge, while coarse darker grains were observable away from the original water boundary (Figure 6a). After 20 h the grain boundaries and the difference between the interior and the periphery of the formed hydrate was less evident; grain boundaries were also less pronounced (Figure 6b). Figure 6c shows how edge and interior morphologies were almost indiscernible 51 h past hydrate formation, and in addition grain boundaries were hardly noticeable and slight depressions appeared on the whole surface. After 76 h, small depressions covered the whole hydrate surface and a uniform morphology was seen throughout; grain boundaries were not perceptible (Figure 6d). Reformation. Reformation refers to hydrate crystallization on water droplets with previous hydrate formation history. The experimental conditions for hydrate reformation are summarized in Table 2. Hydrate reformation (i.e., clathrate formation on films with previous hydrate formation history) proceeded in an entirely different manner than that of clathrates formed from water with no previous history of hydrate formation. Methane bubbles were always present prior to reformation, a product of the preceding dissociation (top left, Figure 7b,c). Reformation commenced from a localized point or a few points within the water film (Figure 7a), and advanced as a circular front(s) until reaching the edge of the water surface (Figure 7b,c). As was the case with hydrate formation from
Article
Crystal Growth & Design, Vol. 10, No. 10, 2010
water films with no previous hydrate formation history, a translucent hydrate layer grew out of the original water
Figure 6. Aging effect on hydrate formed on a water film with no previous hydrate formation. T and p were kept at 275 K and 8.0 MPa. t refers to the elapsed time after nucleation was observed. (a) t = 20 min, hydrate that forms near the film edge appears smooth and shiny; toward the interior grain boundaries are evident. (b) t = 20 h, grain boundaries and the edge effect observed in (b) become less evident. (c) t = 51 h, the edge and interior morphologies become almost indiscernible, slight depressions appear on the surface. (d) t = 76 h, only one morphology shows on the hydrate film, grain boundaries have disappeared, and small depressions cover the entire surface. Table 2. Experimental Conditions for Hydrate Reformation on Water Films and Associated Crystal Growth Velocities experiment
T/K
p/MPa
v/μm s-1
growth beyond water boundary
1 2 3 4 5 6 7 8
275 275 275 275 275 275 275 275
8.2 8.4 8.3 8.3 8.3 8.2 8.6 8.0
632 714 742 872 758 910 not determined not determined
yes yes yes yes yes yes yes yes
4343
boundary (bottom right Figure 7c), although at a much slower rate than that within the water surface. The growth of this translucent hydrate layer outside of the original water boundary was observed in all of our reformation experiments (Table 2). Only one type of morphology was evident from reformation experiments (Figure 7) inside the original water boundary, in contrast to the results of formation experiments performed on water with no previous hydrate formation history where two different morphologies were observed (Figure 3). Growth velocities were determined for reformation by considering the position of the advancing interface (dark gray in Figure 7) with respect to a fixed reference point and extracting the instantaneous radial velocity. The reported values in Table 2 constitute an average of the instantaneous velocities over the time span of the collected data. A representative plot is shown in Figure 8 where two sets data are presented: one for each of two fronts that grew simultaneously as shown in Figure 7a. The type of morphology shown in Figure 7 seems similar to the one previously shown by Freer et al.,8 and the velocity magnitude and variability shown in Table 2 are comparable as well. The velocity of the translucent hydrate layer proved to be more difficult to determine due to the irregular geometry of the same. However, this halo’s velocity was comparable throughout the reformation experiments and we estimate it to be on the order of 10 μm/s. The “Bridge Effect”. As mentioned in the sections above, growth beyond the original water boundary was observed for hydrate formation with and without previous clathrate formation history in all of our experiments (Tables 1 and 2; Figures 3 and 7c). In order to further examine the effect of this growing hydrate halo, formation experiments were performed with segregated water droplets. Figures 9 and 10 show hydrate nucleation and growth conducted at 8.2 MPa and 276 K. Clathrate nucleation occurred first on the most extensive water region, and then it spread as hydrate halos formed “bridges” between nucleated and un-nucleated water droplets as described above. Figure 10a shows newly formed hydrate (black) on the broadest, water-covered region that appears in the micrograph’s field of view; an almost imperceptible halo (Figures 9a and 10a), corresponding to an advancing hydrate front, surrounds the black hydrate boundary. As the hydrate halo grows outside of the water boundary, it induces nucleation in a water bubble (Figure 10b). Clathrate propagation within a water droplet that has just nucleated can be seen from Figure 10c,d. Hydrate halos that grow outside of nucleated water boundaries induce
Figure 7. Hydrate reformation on water films. (a) Two hydrate fronts, 2 s after nucleation was observed. T = 275 K, p = 8.6 MPa. (b) Detail of an advancing hydrate front T = 275 K, p = 8.4 MPa. (c) The clathrate covered the observable water region in 4 s. The hydrate halo growing outside of the original water boundary grew in approximately 2 min. T = 275 K, p = 8.4 MPa.
4344
Crystal Growth & Design, Vol. 10, No. 10, 2010
Beltr an and Servio
Figure 8. Representative plot of the position of the advancing hydrate front versus time. Two sets of data are presented: one for each of two fronts that grew simultaneously. Experimental conditions were T = 275 K, p = 8.4 MPa. Growth velocities were determined by considering the position of the advancing interface with respect to a fixed reference point.
Figure 9. Detail of hydrate propagation by the bridge effect. Hydrate appears dark after forming in a water film. (a) A clathrate halo that grew out of the water boundary is about to touch two water bubbles. (b) After one second, the hydrate halo reaches the two water bubbles and induces nucleation.
nucleation in other water droplets (Figure 10c,e,f and enlarged view in Figure 9b). The experimental conditions are summarized in Table 3. Front (dark gray in Figures 9 and 10) velocities were determined in a similar manner as described above for reformation and are shown in Table 3. Shape irregularity and image contrast made it difficult to determine the hydrate halo velocity, but it was estimated to be on the order of 10 μm/s. Dissociation. Dissociation proceeded in a similar manner despite the history of the hydrate film. Figure 11 presents a representative sequence. The intact hydrate surface is shown in Figure 11a; the layered structure is clearly appreciable. The lighter shade corresponds to the original water boundary. To the right of the latter, the hydrate film that extended out of the same boundary can be seen. On the top left, dark granular hydrate formed within the original water film is observed. Hydrate dissociation did not become evident until several seconds after decreasing the pressure from p = 8.2 MPa and T = 274 K to p = 2.0 MPa and T = 273 K. Crystal decomposition started at the periphery and spread toward the center (Figure 11b). Widespread dissociation led to the appearance of a hydrate-water slurry; (Figure 11c). As the slurry thinned down, individual clathrate crystals were observable and retraction of the water film became apparent (Figure 11d). As massive amounts of bubbles appeared the water film continued to retract (Figure 11e). The water film
Figure 10. Still frames of a hydrate propagation video sequence. (a) t = 0 s, hydrate appears dark after forming in a water film. The almost imperceptible halo observed around the dark hydrate corresponds to a growing clathrate film outside of the water boundary. (b) t = 3 s, as soon as the growing hydrate halo touched a new water bubble nucleation occurred. (c) t = 4 s, halo from the bubble that nucleated in (b) induces nucleation in another bubble. (d) t = 8 s, hydrate growth within the water film that nucleated in (c). (e) t = 9 s, nucleation induced in two new water bubbles. (f) t = 11 s, newly induced nucleation. Table 3. Experimental Conditions for Bridge Effect Experiments experiment
T/K
p/MPa
v/μm s-1
1 2
276 276
8.2 8.3
659 853
retraction ceased after about a minute of having reduced the pressure, regaining its original shape. Finally, methane bubble migration and agglomeration near the center of the water film were also apparent (Figure 11f). Discussion When dealing with crystallizing systems, the shape of the interface is not unique but may select any of several possible morphologies that satisfy any thermodynamic constraints as well as all the heat and mass transport constraints,30 thus attributing a specific morphology for a set of conditions can very well be correct but experimental apparatus/procedure specific. Below we identify and discuss several parameters that seemed relevant from our results. The Memory Effect. We have identified hydrate history as a key factor to consider when determining the macroscopic hydrate growth mechanism and morphology in hydrates growing on a water droplet deposited on a glass surface. Evidence for this differing morphology can be clearly appreciated from our results (Figures 2-5). The mechanism for hydrate formation with previous hydrate formation hystory (Figure 7) matches perfectly with the conceptual picture suggested by Sloan and co-workers,29
Article
Figure 11. Dissociation of a layered hydrate film. (a) Intact layered hydrate structure (T = 274 K, p = 8.2 MPa). The bottom right of the corresponds to the original water boundary; directly opposed, black granular hydrate can be observed. (b) Hydrate dissociation became evident 22 s after reducing the pressure to p = 2.0 MPa, T = 273 K. (c) Widespread hydrate dissociation led to the appearance of a hydrate-water slurry 26 s after reducing p. (d) The water-hydrate slurry thinned down until individual hydrate crystals were observable 30 s after reducing p. Retraction of the water film boundary became apparent. (e) 64 s after reducing p massive amounts of methane bubbles appeared and retraction of the water film boundary continued. (f) t = 100 s, water retraction completed and agglomeration of methane bubbles; no hydrate crystals were observable at this point.
whereby a uniform film covers the water surface as it grows. However, our results for systems without memory (Figure 2) suggest an addition to this conceptual picture: water history has to be considered and the concept of one uniform film is not applicable in systems without history. The latter has implications for hydrate modeling since various models have used this uniform film assumption.7-10 Crystal growth velocities were not determined for hydrate formation without memory due to experimental limitations viz. the spatial resolution of the objective used to observe the whole water droplet surface. However, we roughly estimated an average macroscopic velocity by considering the time it took for forming hydrates with or without memory to cover comparable areas (roughly 3 cm2) at equivalent experimental conditions (T ≈ 275 K, p ≈ 8 MPa). Velocities estimated in this way render a figure that helps to compare formation and reformation. These average velocities were found to be comparable for systems with and without memory, and on the order of 2 mm/s. The latter agrees with the values obtained by Tanaka et al.25 and serves as an indirect validation of our estimation technique. Here we have neglected any differences in hydrate film thickness; this is a limitation of our technique. We have also intentionally disregarded the fact that formation without memory did not occur at a constant rate (Figure 2). The fact that both formation and reformation had the same average velocities could imply that both systems had the same sort of transport limitations. However, as seen from
Crystal Growth & Design, Vol. 10, No. 10, 2010
4345
our morphology results, reformation proceeded with a stable uniform front (Figure 7), whereas the dendritic morphology for systems without memory (Figure 2) indicates some sort of transport limitation. By estimating the Prandlt and the Lewis numbers for our water droplet, we can gain some understanding of the limiting process: Pr ≈ 10 and Le ≈ 100; hence, mass transfer has to be the limiting form of transport. However, from Figure 8 we can see velocity is not a function of time for the system without memory so diffusion was not a limiting factor in this case. The presence of numerous initial crystallites in the system without memory (Figure 2) suggests pervasive inhomogeneities in methane concentration within the water droplet. The latter could be exacerbated by the varying growth velocity30 (which we observed in the system without memory). Koga et al.31 have concluded from in situ neutron diffraction experiments that angstrom-scale hydrate “embryos” form at the water-methane surface and until the end of the induction period when such metastable structures turn into macroscopic hydrates. It may very well be possible that the density of these precursors is considerably lower or highly localized for systems without memory, offering a plausible explanation for the numerous nucleation points observed in this system (Figure 2). We speculate that for systems with memory these embryos would be more evenly distributed within the water after dissociation, hence the resulting uniform morphology (Figure 7). Another plausible explanation for the multiple morphologies on systems without memory comes to mind through consideration of a study by Schicks and Ripmeester.32 There, micrographs suggesting the formation of metastable structure II methane hydrate were shown.32 This observation was corroborated with Raman spectroscopy, and it was concluded that the thermodynamically unstable structure II methane hydrate eventually converts to structure I methane hydrate.32 We cannot say from our data whether this effect was present or not, although the fact that only a unique morphology shows for reformation is perhaps an argument against it. Thus, we favor a difference in crystal habit explanation for our results without memory, as opposed to a difference in crystal structure explanation. Driving Force in Systems without Memory. The effect of the driving force on the morphology of the clathrate formed at the edge of the water film (Figures 4 and 5) seems to agree with the results of Servio and Englezos23 where the high driving force produced a rougher hydrate surface. However, the images acquired in this study showed the effect was reversed for the morphology of the interior of the water film: lower driving forces resulted in a rougher surface (Figure 5) than the one observed at high driving forces (Figure 4). The observed morphology in the interior of the hydrate film is more in accord with the results of Tanaka et al.25 The larger width of the edge morphology as well as the smaller grain size at a high driving force are indicative of the higher speed at which hydrate growth occurred. At a lower driving force grains are allowed to grow to a certain extent before meeting each other (Figures 2c,e and 5). At high driving forces, nucleation sites can be numerous33 which would explain the smaller grain size observed at a high driving force (Figure 4). The observed whiskery structures seem to agree well with the results of Uchida et al.,7 where hydrate was first detected as crystal facets growing from the edge of a pendant water bubble in contact with a hydrophilic surface. Although
4346
Crystal Growth & Design, Vol. 10, No. 10, 2010
Uchida et al.7 also reported growth of a secondary hydrate film on a water bubble, their findings contrast with the morphology of the secondary nucleation observed in this work which proceeded in a spotted pattern (Figure 2c,d). Furthermore, the possibility of having two clearly different types of morphologies during clathrate formation contrasts with previous reports, where a unique morphology was attributed to a fixed driving force.24,25 The whiskers that appeared on the periphery of the water droplet (Figure 2) nucleated first, most probably due to reduced surface energy penalty for nucleation at the waterglass interface. The latter is a well-known phenomenon.34 The complex interaction between these initial dendrites and the cellular morphology that develops thereafter in the interior of the droplet produced the characteristic morphology of the periphery (Figures 3-5). The difference in breadth between the border of high driving force and low driving corresponds to the length of the initial dendrites: at high driving force dendrites were much longer. Furthermore, at high driving force, the dendrites were highly branched, which could explain the rougher periphery (Figure 4) when compared to the periphery at a low driving force (Figure 5). Aging. A comparable smoothing phenomenon to that observed in Figure 6 has been reported previously by Sugaya and Mori17 for fluorocarbons, by Uchida et al.7 for CO2 and by Ohmura et al.18 for CH3CCl2F. We propose two explanations for this aging process. First, it is plausible that due to highly localized concentration differentials newly crystallized hydrate forms with variegating concentrations in a hydrate-liquid-vapor environment. As all the water gets converted to hydrate, the system slowly moves toward a uniform methane concentration in a hydrate-vapor environment, with the consequent change in morphology. The second explanation follows a surface energy minimization argument whereby bigger grains engulf the small ones, and thus reduce the system’s energy by eliminating grain boundaries.33 Propagation. Previously, almost simultaneous hydrate nucleation in adjacent water droplets was observed by Servio and Englezos23 who attributed this phenomenon to an imperceptible water “bridge” that allowed communication across the polytetrafluoroethylene surface on which their water droplets were placed. Our results showing a film that grew outside of the original water surface (Figures 2-4 and 7) suggested to us that the “bridge effect” hypothesized by Servio and Englezos23 was probably due clathrate growth on a foreign surface, free from water (glass in this case). This hydrate “halo” only developed once the entire water surface was covered with hydrate and it was present independent of driving force or the history of the system (Figures 2, 3, and 7). By examining the nucleation and propagation of methane hydrate in segregated water droplets, we were able to clearly observe this “bridge effect” (Figures 9 and 10). Our results suggest a mechanism for this process as summarized in Figure 12. We think that once the hydrate covers the complete water surface (dark gray in Figures 9 and 12b), liquid water underneath the hydrate layer is drawn by capillarity toward the naked glass (Figure 12c). As this happens, a hydrate film grows on the advancing water front (Figure 9). The fact that the halo grows in a layered fashion, as it moves outside of the original water surface (right of Figure 4), supports our water migration hypothesis. But the question arises whether water is effectively available for this process. Neglecting curvature, we estimated our initial water droplet thickness to be approximately 150 μm which is considerably
Beltr an and Servio
Figure 12. Hypothesized mechanism for hydrate propagation on a glass surface. (a) Intact neighboring water droplets, deposited on a glass surface. (b) Hydrate (red) covers the larger droplet completely. (c) As water is drawn by capillarity action toward the glass, a hydrate film forms on it (halo). (d) The hydrate halo finds its way to the neighboring molecule and induces nucleation.
thicker than initial hydrate thicknesses measured by Ohmura et al.26 (e80 μm). Furthermore, Kerkar et al.35 have shown that liquid, not hydrate, is the wetting phase, when preparing hydrates in the presence of glass beads. Eventually, the halo finds its way to the nearest water droplet (Figure 12d), inducing nucleation in the new droplet, which in turn develops a halo of its own (Figures 9 and 10). It is tempting to suggest that an extremely thin film of water already present on the glass is responsible for the hydrate propagation between segregated water droplets, but the fact that halo growth is observed in systems without previous hydrate formation is an argument against it. What is more, by comparing the front growth velocity (dark gray in Figure 9, v ≈ 750 μm/s) with that of the halo (translucent in Figure 9, v ≈ 10 μm/s) we can see a limiting transport mechanism is operating in the latter. The halo advances at almost 2 orders of magnitude more slowly than the front, and we attribute this to the slow movement of water on the glass surface. We think that this mode of hydrate propagation could have implications for flow assurance in pipelines, but we are conscious that the present work was done on glass, and thus our conclusions might not be amenable to other surfaces. Conclusion For methane clathrates formed from water films without previous hydrate formation history, it was found that hydrate nucleation occurred on the periphery of a film first; secondary nucleation occurred within the water film, and the crystals thus formed had a completely different morphology than that of the ones formed on the water edge. This difference in morphology tended to disappear over time. It was shown that higher driving forces produced smaller hydrate grains and smoother surfaces than lower driving forces within the water film; the converse was true for hydrate formed on the periphery. Furthermore, a third clathrate layer that grew outside of the original water boundary was observed. Images suggested that this third layer could grow on water-free glass and thus was capable of communicating nucleation from one water covered region to another creating a hydrate bridge. Clathrate reformation proceeded in a manner different than that of hydrates formed from water without previous hydrate formation history: reformation occurred within the film and a circular hydrate front(s) advanced toward the periphery of the water film, creating a clathrate with uniform morphology. Dissociation progressed in a similar manner despite whether the hydrate was formed from water with or without previous clathrate formation: decomposition started on the periphery and proceeded until gas bubbles were observed within the film; moreover, as dissociation became ubiquitous within the film, water receded until the water film regained its original shape and gas bubbles agglomerated in the center of the water film. Acknowledgment. The financial support from the Natural Sciences and Engineering Research Council of Canada
Article
Crystal Growth & Design, Vol. 10, No. 10, 2010
(NSERC), the Canada Research Chair program (CRC), Le Fond Quebecois de la Recherche sur la Nature et les Technologies (FQRNT), the EUL Chemical Engineering fund at McGill University, and the DGLEPM of the Canadian Forces is greatly appreciated. Anonymous reviewers are recognized for their insightful comments; their reviews helped to improve the manuscript’s quality. Supporting Information Available: A short video showing an example of the bridge effect. The video does not play in real time, and the resolution has been reduced significantly. Please refer to the manuscript for higher resolution images. This material is available free of charge via the Internet at http://pubs.acs.org.
References (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
Englezos, P. Ind. Eng. Chem. Res. 1993, 32, 1251–1274. Sloan, E. D. Nature 2003, 426, 354–359. Kvenvolden, K. A. Terra Nova 2002, 14, 302–306. Brewer, P. G. Ann. N. Y. Acad. Sci. 2000, 912, 195–199. Brewer, P. G.; Orr, F. M.; Friederich, G.; Kvenvolden, K. A.; Orange, D. L.; McFarlane, J.; Kirkwood, W. Geology 1997, 25, 407–410. Ohmura, R.; Shimada, W.; Uchida, T.; Mori, Y.; Takeya, S.; Nagao, J.; Minagawa, H.; Ebinuma, T.; Narita, H. Phil. Mag. 2004, 84, 1–16. Uchida, T.; Ebinuma, T.; Kawabata, J.; Narita, H. J. Cryst. Growth 1999, 204, 348–356. Freer, E. M.; Selim, M. S.; Sloan, E. D. Fluid Phase Equilib. 2001, 185, 65–75. Mori, Y. J. Cryst. Growth 2001, 223, 206–212. Mochizuki, T.; Mori, Y. H. J. Cryst. Growth 2006, 290, 642–652. Makogon, Y. F.; Dunlap, W. D. Hydrates of Hydrocarbons; Penwell Publishing Company: Tulsa, OK, 1997. Maini, B. B.; Bishnoi, P. R. Chem. Eng. Sci. 1981, 36, 183–189. Mori, Y.; Mori, T. AIChE J. 1989, 35, 1227–1228.
4347
(14) Mori, Y. H.; Mori, T. Int. J. Refrig. 1989, 12, 259–265. (15) Hwang, M. J.; Wright, D. A.; Kapur, A.; Holder, G. D. J. Inc. Phenom. Mol. 1990, 8, 103–116. (16) Long, J. P.; Sloan, E. D. Int. J. Thermophys. 1996, 17, 1–13. (17) Sugaya, M.; Mori, Y. H. Chem. Eng. Sci. 1996, 51, 3505–3517. (18) Ohmura, R.; Mori, Y. H. J. Cryst. Growth 1999, 196, 164–173. (19) Smelik, E. A.; King, H. E. Am. Mineral. 1997, 82, 88–98. (20) Uchida, T.; Kawabata, J. Energy 1996, 22, 357–361. (21) Uchida, T.; Ebinuma, T.; Narita, H. J. Cryst. Growth 2000, 217, 189–200. (22) Uchida, T.; Ikeda, M.; Takeya, S.; Ebinuma, T.; Nagao, J.; Narita, H. J. Cryst. Growth 2002, 237, 383–387. (23) Servio, P.; Englezos, P. AIChE J. 2003, 49, 269–276. (24) Ohmura, R.; Matsuda, S.; Uchida, T.; Ebinuma, T.; Narita, H. Cryst. Growth Des. 2005, 5, 953–957. (25) Tanaka, R.; Sakemoto, R.; Ohmura, R. Cryst. Growth Des. 2009, 9, 2529–2536. (26) Ohmura, R.; Kashiwazaki, S.; Mori, Y. H. J. Cryst. Growth 2000, 218, 372–380. (27) Hirai, S.; Tabe, Y.; Kamijo, S.; Okazaki, K. Therm. Sci. Eng. 2000, 8, 1–5. (28) Kobayashi, I.; Ito, Y.; Mori, Y. H. Chem. Eng. Sci. 2001, 56, 4331– 4338. (29) Sloan, E. D.; Koh, C. A. Clathrate Hydrates of Natural Gases, 3rd ed.; Chemical Industries: CRC Press, 2008. (30) Tiller, W. The Science of Crystallization: Macroscopic Phenomena and Defect Generation; Cambridge University Press: Cambridge, UK, 1991. (31) Koga, T.; Wong, J.; Endoh, M.; Mahajan, D.; Gutt, C.; Satija, S. Langmuir 2010, 26, 4627–4630. (32) Schicks, J. M.; Ripmeester, J. A. Angew. Chem., Int. Ed. 2004, 43, 3310–3313. (33) Mullin, J. W. Crystallization; Elsevier: Amsterdam, 2001. (34) Kashchiev, D.; Firoozabadi, A. J. Cryst. Growth 2002, 241, 220– 230. (35) Kerkar, P.; Jones, K.; Kleinber, R.; Lindquist, B.; Tomov, S.; Feng, H.; Mahajan, D. App. Phys. Lett. 2009, 95, 1–3.
