HETEROGENEOUS NUCLEATION OF CLATHRATES FROM SUPERCOOLED THF/WATER MIXTURES AND THE EFFECT OF AN ADDED CATALYST P. W. Wilson* and A. D. J. Haymet Scripps Institution of Oceanography UC Diego, USA ABSTRACT The statistics of liquid-to-crystal nucleation are measured for clathrate-forming mixtures of tetrahydrofuran and water using an automatic lag time apparatus (ALTA). We measure the nucleation temperature where a single sample is repeatedly cooled, nucleated and thawed. This is done for a series of tetrahydrofuran concentrations and in several different sample tubes since the nucleation is heterogeneous and occurring on the tube wall. The measurements are also done at the same concentrations and tubes but with an added catalyst, a single crystal of silver iodide. We discuss the need for this type of measurement if the true nucleation temperature of the clathrate is to be found. Comparisons are also made with our high pressure data on real-world clathrate formers. Keywords: ALTA, THF, nucleation, supercooling, statistics INTRODUCTION The lag-time before a supercooled sample nucleates to a solid is a stochastic function, strongly dependent on the degree of supercooling and is often called the induction time. One important, and interesting, crystallization process is that of gas hydrates from aqueous solution. Clathrate-hydrates are solid compounds formed between water molecules and small gas molecules. The statistics of liquid-to-crystal nucleation are measured here for clathrate-forming mixtures of tetrahydrofuran (THF) and water using an automatic lag time apparatus (ALTA) [1,2,3,4,5]. We measure the nucleation temperature using this apparatus in which a single sample is repeatedly cooled, nucleated and thawed. This is done for a series of tetrahydrofuran concentrations and in several different sample tubes since the nucleation is heterogeneous and so occurring on the tube wall. The measurements are also done at the same concentrations and tubes but with an added catalyst, a single crystal of silver iodide. MATERIALS AND METHODS The measurements described here are made on a purpose built automated nucleation device that we have termed automatic lag time apparatus (ALTA). The water used is Ultrapur reagent grade water (Merck, Germany) filtered through a 0.2 μm filter. The THF is reagent grade from Chem. Supply, S.A., Australia. For the silver iodide experiments the crystals are made freshly every few weeks and are typically 0.5 mm across the longest axis. Sample volumes of 300 μl have been used in each case and these are placed in purpose built sample tubes made from borosilicate glass with an outside diameter of 5.0 mm and a length of 65 mm (similar to a shortened NMR tube). These glass tubes are inserted with a snug fit into a hole drilled into an aluminium sample holder in the ALTA. The aluminium block is cooled by thermoelectric units on either side, and the temperature control is by a PID package built Proceedings of the 6th International Conference on Gas Hydrates (ICGH 2008), Vancouver, British Columbia, CANADA, July 6-10, 2008. into a Genie software package (Advantech Inc.) which controls the experiment via a multipurpose DAQ card and PC interfacing. A cooling rate of typically 4.5 K min. -1 is used and freezing of the samples is detected optically due to a sudden lowering in the optical transmission of a laser beam passed through the sample tube. Once freezing has been detected the software causes heating of the tube to +20 C where it is then held for five minutes to ensure complete melting of any ice/hydrates, before the same tube and sample is cooled again. This cycle is repeated more than 300 times in any single measurement run to gather statistically valid data about the nucleation temperature of the sample in that tube. Figure 1. The experimental arrangement of ALTA, where samples in glass tubes are repeatedly cooled till frozen, warmed to +20ºC and cooled again. Figure 2. The experimental protocol where the sample is linearly cooled until it nucleates and freezes. The cooling rate is typically between 2 and 4.5 K min. -1 . Freezing of the samples is detected optically due to a sudden lowering in the transmission of a laser beam passed through the sample. RESULTS The data presented here has been chosen simply to illustrate the typical results generated by ALTA and to help clarify our description of measuring and defining the SCP. The primary data collected in this experiment are the time, ti, and the temperature, Ti, at which nucleation occurs, as a function of run number “i”. Figure 3 shows these data collected from 900 consecutive heating / cooling cycles on a single sample which was cooled at a rate 0.018 K second -1 . We call this type of plot a “Manhattan” and it neatly demonstrates the stochastic nature of the nucleation process. Although we are using the same sample from run to run, the sample does not freeze at the same temperature on each run. Instead, there exists a distribution of temperatures, spanning approximately 4 K in this case, over which nucleation occurs. This width of temperatures is an important quantity, which is reproducible and inherent. It must be considered when determining the nucleation temperature of any solution. Thus, making only a few measurements of the nucleation temperature introduces a source of error into reporting a value for that temperature. Manhatten 0 50 100 150 200 250 300 350 400 1 45 89 13 3 17 7 22 1 26 5 30 8 35 2 39 6 44 0 48 4 52 8 57 2 61 5 65 9 70 3 74 7 79 1 83 5 87 9 Run # T im e ( s ) Figure 3. Typical ALTA raw data, in the form of a “Manhattan”, shows the freezing time for 900 runs on the same sample in the same tube. The stochastic nature of nucleation is evident. A simpler and hence better way to evaluate these data is to calculate a “survival curve”. We denote by N0 the total number of repetitions on the same sample. Simple statistical analysis shows that to determine unambiguously the width and location of the SCP, N0 should be of order 200-300. The survival curve is simply the fraction unfrozen of samples as a function of time or scaled time, and we denote this F(t), F(t) = N(t) / N0 , where F(t) equals unity at the time zero, when the sample cools below its equilibrium melting point, at which point all of the samples are unfrozen, and evolves to the value zero at some finite time later, when all of the samples have frozen. Since the supercooled temperature decreases linearly with time,t, we may convert the time axis to a scaled time by multiplying by the cooling rate, . A typical set of survival curves is shown in Figure 4. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 5 10 15 20 25 30 Delta T (K) F ra c ti o n u n fr o z e n 19% THF 10% THF 5% THF Figure 4. The survival curves for three runs, each at a different THF concentration are shown. We define the T50 (50% of the fraction of sample runs are unfrozen at that temperature) as the kinetic freezing point or nucleation temperature. If now we take the first derivative of the survival curve we get a shape as shown in Figure 5. The peak is very close to the T50 and can be taken as the nucleation temperature for that sample. The width of the peak illustrates well the spread of possible nucleation temperatures for that sample. Figure 5. The first derivatives of the s curves give not only the nucleation probability but also the spread of nucleation temperatures for that solution in that container. If now we look at some data for THF in two different tubes at two different concentrations, as shown in Figure 6, we see that the nucleation temperatures vary markedly from tube to tube and this must be a clear warning not to draw invalid conclusions from measurements of a given sample without taking into account the stochastic nature of nucleation. S-Curves 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 0.900 1.000 0 5 10 15 20 25 30 35 Delta T F r a c ti o n Alta 5 19% Alta 4 19% Alta 5 10% Alta 4 10% Figure 6. Four survival curves, two at 10 % THF and two at 19 % THF, in each of two different tubes. On the other hand, if the same solution is measured in the same tube multiple times then the results look like those in Figure 7. We see that the nucleation temperatures have the same spread, even over 1200 runs. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Delta T F ra c ti o n 1st 300 Runs 2nd 300 Runs 3rd 300 Runs 4th 300 Runs Figure 7. This plot shows four back to back series of 300 runs each on the same 10 % THF sample in the same tube. The nucleation temperature has not changed significantly over the three weeks required to collect all this data. If now we add a catalyst (silver iodide) to the solution, we see that using the T50s is a valid way to determine if the catalyst has had any effect on the average nucleation temperature. Figure 8 shows that over various concentrations silver iodide has reduced the average nucleation temperature of that sample in that container. 0 5 10 15 20 25 30 0% 10% 20% 30% 40% 50% concentration o f THF Δ T 50% P ure A lta 4 A gI A lta 4 Figure 8. Plot of average T50 for each THF concentration, both with and without added AgI. We have now begun to examine the statistics of nucleation for natural gas hydrates at elevated pressures. A new instrument has been developed (largely based on previous versions of the ALTA) which is able to make the same statistical measurements of nucleation for natural gases under pressure. The new instrument is named the High Pressure Automatic Lag Time Apparatus (HP ALTA). Preliminary data for methane at various pressures is shown in Figure 9. 0 1 2 3 4 5 6 7 8 9 10 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 Tem perature (K ) P r e s s u r e ( M P a ) M ethane H ydrate Phase Boundary H P ALTA R esu lts W ater Phase Boundary Figure 9. The diagram shows that the equilibrium formation temperature for methane hydrate. The results obtained from the HP ALTA indicate that the instrument is detecting hydrate formation and not ice. CONCLUSION We have shown that it is essential to use the ALTA method on gas samples if the true nucleation temperatures and probabilities are to be found. It is also essential to use an ALTA type method to determine the effects of any additives such as kinetic inhibitors. ACKNOWLEDGEMENTS We gratefully acknowledge the support of Chevron Texaco and BP for partially funding the ALTA instruments. REFERENCES [1] Heneghan, A. F., Wilson, P. W. and A. D. J. Haymet. Statistics of heterogeneous nucleation of supercooled water, and the effect of an added catalyst. Proc. Natl. Acad. Sci. 2002; 99; 9631- 9634. [2] Heneghan, A. F., Wilson, P. W., Wang, G. and A. D. J. Haymet. Liquid-to-Crystal Nucleation: Automated Lag-Time Apparatus to study supercooled liquids. J. Chem. Phys. 2001; 115; 7599-7603. [3] Wilson, P. W., Lester, D. and A. D. J. Haymet. Heterogeneous nucleation of Clathrates from supercooled tetrahydrofuran (THF) / water mixtures, and the effect of an added catalyst. Chemical Sciences Engineering. 2005; 60; 2937- 2941. [4] Wilson, P. W. and A. D. J. Haymet. Nucleation from a supercooled binary mixture studied by cross polarisers. J. Physical Chemistry A, 2005, 109, 50, 11354 – 11357. [5] Wilson, P. W., Lester, D. and A.D.J. Haymet. Heterogeneous nucleation of Clathrates from supercooled tetrahydrofuran (THF) / water mixtures. Fifth International Conference on Gas Hydrates, Trondheim, Norway, June, 2005