6122
J . Phys. Chem. 1993,97, 61224725
Interpretation of 13C NMR of Methane/Propane Hydrates in the Metastable/Nonequilibrium Region F. Fleyfel, K. Y. Song, A. Kook, R. Martin, and R. Kobayashi' Department of Chemical Engineering, Rice University, Houston, Texas 77251- 1892 Received: January 15. 1993 Combining macroscopic hydrate experiments (visual rocking cell) with microscopic hydrate experiments (NMR), it is possible to investigate samples of water and methane/propane gas (96% C&) in the metastable/nonequilibrium region of a hydrate formation/decomposition process. The two points investigated are located on the dissociation curve, before and at the point where hydrate particles become invisible. The p r e s s u r e t e m p e r a t u r e p o s i t i o n conditions used during this work obey industrial demands. The gas pressure throughout the experiment varied from 196 to 186 (f0.2)psia, and the temperature ranged from 285 to 278 (f0.002) K. After structure I1 (SII) methane/propane hydrates particles visually disappeared, some microscopic pentagonal-dodecahedra cages (P-D) enveloping the smaller CH, molecules and some microscopic hexakai-decahedra (H-D) cages enveloping the larger C3Hs molecules remained undissociated in the liquid water bulk phase as seen from molecular level probing of the process using NMR. These results confirm the hypothesis regarding the presence of microscopic hydrate particles in the metastable/nonequilibrium region on the dissociation (heating) curve.
Introduction The production, transportation, and processing of natural gas and petroleum depend on the continuous and uninterrupted flow of products. However, thermodynamic departures due to pressure, temperature, and composition can cause unprocessed natural gas to undergo significant phase transitions, one of which is hydrate formation. This process is described by van der Waals and Platteeuwl as being: water
+ gas = clathrate hydrates
(1) The most common hydrate structures are structure I (SI) and structure I1 (SII). (StructureH is well-known but not considered in this work.) Both of these structures are cubic in form and are born out of a combination of cages. Structure I results from a combinationof pentagonal-dodecahedra (P-D) cages and tetrakaidecahedra (T-D) cages, whereas structure I1 is a combination of P-D cages and hexakai-decahedra (H-D) cages. Figures 1-3 show both structures and the cage combinations available that define these structures. In past years, hydrate research evolved around studying thermodynamics of hydrates. Some of the research work probed hydrate phase equilibria, while other research work investigated hydrate heats of diss~ciation.~-~ Not until recently have suggestions regarding shifting the research from time-independent studies (thermo) to time-dependent studies (kinetics) surfaced."1° Sloan and Fleyfel" suggest a mechanism attempting to explain the hydrate formation process from ice. The mechanism is divided into two parts. The first part is a metastable/nonequilibrium area that describesthe nucleation domain. In this domain,hydrate microscopic particles form before the crystal growth part. The latter part is the second part of the mechanism. The complexity of the hydrate formation/decomposition process in the metastable/nonequilibrium region from liquid water restricted the research domain of the hydrate investigators. Therefore, most of the hydrate research has taken place in the crystal growth domain.I2-l4 To fully understand the kinetics of hydrate formation from liquid water, the roles of metastability in the formation and decomposition of gas hydrates and the structure of water must be identified. The P-T curves forming a loop provide a general understanding of the hydrate formation/de"position processes. However, the important questions have remained unanswered. Some of the unanswered questions are as follows: (i) What is the hydrate building block? (ii) At what rate do the
Figure 1. Unit cell of a gas hydrate of structure I (from ref
26).
Figure 2. Unit cell of a gas hydrate of structure I1 (from ref 26).
cages form/de"pose? (iii) Does the formation occur at the interface or in bulk of liquid water? To answer some of these questions, hydrate researchers have introduced spectroscopyas a possible and powerful tool to explore hydrates. In 1949, von Stackelberg15studied both hydrates SI and SI1using X-ray crystallography. Later, Bertie16contributed to the field by applying X-ray powder diffraction and IR to the same bulk grown ethyleneoxide (EO) hydrate sample, determining the structure as well as the absorption bands in mid- and far-IR, building a bridge relating both methods. Fleyfel and Devlinl' investigated CO2 clathrate hydrates with FT-IR spectroscopy using cryogenic molecular deposition techniques. Many used mass spectroscopy to identify the magicnumber water clusters (H20)21H30+,which was explained
0022-365419312097-6722$04.00/0 Q 1993 American Chemical Society
The Journal of Physical Chemistry, Vol. 97, No. 25, 1993 6123
Methane/Propane Hydrates
196 194
5a 192 u)
190 188 186 6
5
8
7
9
10
Temp., C
Figure 5. Pressure-temperature plot representing a CH4/C& hydrate loop and the four formation/decomposition regions. Point A is the final detection point of a hydrate particle.
V
16-HEDRON (HEX AKAIDECAHEDRON)
Figure 3. Polyhedra cages of water moleculesinclathrate ices ofstructures I and 11. Circles represent oxygen atoms and lines the hydrogen bonds between them (from ref 27).
Press
Figure 6. I3C NMR spectrum of CH4 in gas phase at 298 K and 200 psia.
/ Figure 4. Hydrate visual rocking cell installed in the air bath.
to be the P-D cages enveloping an H30+ ion in liquid water. Davidson et a1.21reported some low-temperature 'H NMR results on CH4 hydrates. The temperature ranged from 22 to 4 K. Ripmeester and RatclifP2studied I3CNMR of methane trapped in solid type I hydrate and in mixed methane/propane type I1 hydrates at -80 "C. They showed that hydrate cages and structures influenced the pattern of the chemical shift, forming a dependency between an NMR measurable parameter and physical parameters. The purpose of this work is to show that spectroscopic measurements can be applied not only to low-temperature hydrate work22 but also to systems that obey industrial demands, particularly in the metastable/nonequilibrium regionsrepresented in the P-T formation/decomposition curves. Experimental Procedures The gas mixture used during this work is ultrapure CH4/C& (96%in CH4). This gas is I3C enriched with the same binary mixture composition to improve the NMR carbon signal. The 13Cconcentration is 33% of the total gas mixture. Regular H20 is substituted by D20 in order to get a quick lock on the NMR signal. Once the D20 is loaded in the visual cell, the cell is pressurized with CH4/C& gas mixture. The cell is then cooled by dropping the temperature at a ramping rate of 0.5 OC/h. The temperature is measured by three platinum resistance thermometers (PRTs) calibrated to st3 X l e 3 K vs a NIST certified PRT. The three PRTs are placed on the cell wall, in the air bath, and in the liquid water. The pressure is measured by a Setra
l
~
17.0
Figure 7. psia.
"
"
~
'
PPH 16.0
"
'
~
"
"
16.0
do
NMR spectrum of C3Ha in gas phase at 298 K and 200
system pressure transducer calibrated vs dead weight gauge with 10.1% full range accuracy. The experimentalapparatusis shown in Figure 4. When the formation/decomposition loop was being constructed, we stopped the ramping and the rocking process at 279 K long enough to withdraw a clear water sample from the rocking cell into the high pressure NMR tube, which was also held at 279 K. When the rocking stooped, the hydrate floated to the surface, which left the water clear at the bottom. Once the sample was collected into the NMR tube, the ramping process of the sample in the visual cell started again at 0.5 OC/h in the direction of the decomposition curve to complete the loop. The high-pressure NMR tube, kept at 279 K, was placed inside the IO-" NMR probe of a 300-MHz NMR Bruker instrument, which was also held at 279 K. 13CNMR spectra were collected at 279 K. Later, the temperature in the NMR tube was raised from 279 to 281 K, and 13CNMR spectra were again collected at the latter temperature.
Fleyfel et al.
6724 The Journal of Physical Chemistry, Vol. 97, No. 25,1993
15
10
5
-10
0
PPI4
-19
Figure 8. (a) 4 2 NMR spectrum of a C H ~ / C ~ Hhydrate E sample collected at 187 pia and 279 K in the heating direction. (b) %NMR spectrum of a sample collected at 100 pia and 279 K in the cooling direction.
R d t s and Discussion Ramping Experiments. The ramping experiments using a highpressurevisual rocking cell allowed us to hypothesize the problem concerning hydrates in the metastable/nonequilibrium region. The results of the ramping experiments are plotted as a pressuretemperature loop that describes the hydrate formation/decomposition process, as shown in Figure 5. The ramping experiments are extremely reproducible, particularly in the metastable/ nonequilibriumregion in both the heating and cooling directions. The loop is divided into four different regions. Region I is the cooling curve, where a slight decrease in gas pressure is observed. This decrease is hypothesized to be a result of gas cooling, plus formation of microscopic hydrate building blocks. This metastable/nonequilibrium region is described by Sloan and Fleyfel” as being the nucleation region. Region I1 is where the continuous growth takes place and the solid hydrate phase becomes massive enough to plug pipelines. This behavior is demonstrated by a catastrophic gas pressure drop. Region I11 is the dissociation region, where the system is being warmed up and the hydrates are decomposing. Point A in Figure 5 is the point where the last visible hydrate particle disappeared. In our case, point A is detected at 8.0 OC and 194.03 psia. Beyond point A, no more hydrates are seen by the naked eye. Region IV is located beyond point A. In this region, liquid water presumably returns to its initial state as in region I. The scale on the pressure axis in Figure 5 is enlarged toclearly represent point A. This enlargement caused a gap to appear between the cooling and heating curves. This gap is within 0.2p i a or 0.1% of the actual pressure, resulting from the pressure transducer intrinsic accuracy (hysteresis). N M R Experiment. As a background, CH.+/C’Hu gas mixture spectrum is collected at 298 K. The methane peak shown in Figure 6 has a chemical shift 6 = -8.52ppm. The reported value in the literature for the methane gas chemical shift is 6 = -7.0 p ~ m The . ~ NMR ~ peak corresponding to the c3fIS gas shown in Figure 7 is a doublet. The stronger peak is at 6 = 15.74ppm corresponding to the CH2 group, and the weaker peak is at 6 = 15.6ppmcorrespondingtotheCWIgroups. Theliteraturevalue24 shows that the chemical shiftsfor propane are at 15.9 and at 15.4 ppm for both groups, respectively. As mentioned in the Experimental Section, the clear water sample was drawn into the NMR tube from the visual cell at 279 K and 187 psia. The pressure and temperature inside the NMR
~~
Id 0
1d.S
PPM
id
o
Figure 9. Enlarged I3CNMR plot of W,H* hydrate peak in the trapped H-DCages at 279 K.
tube were held constant, but the ramping process continued in the visual cell to close the loop. Even though the water collected in the NMR tube was clear, the hydrates were still present at 279 K and 187 psia as seen from the location of point B in Figure 5. Figure 8a shows the C&/CJHU mixed hydrate NMR spectrum taken at 187 psia and 279 K. The band located upfield of the spectrum at 6 = -8.8 ppm corresponds to the methane peak in the gas phase discussed earlier. The band at 6 -2.75 ppm matches exactly the value reported by Ripmeester?* who assigns it to the CH4 molecules trapped in the P-D cages of the SI1 clathrate hydrates. The NMR band of the CH4 molecules in the H-D cages is not seen at 6 = 4 . 3 ppm because of the overlap of the intense CH4 gas peak with the weak band of trapped CH, in H-D cages. The band at 6 = 18.07 ppm is assigned for the labeled propane C2 trapped in the large cages of SI1 hydrates, but the propane methyl carbon limes are too weak to be seen, as shown in Figure 9. The propane peak corresponding to the gas phase was not detected because of a low signal-to-noise ratio. The question whether these NMR peaks observed are for gas hydrates or gas in solution is answered by running a separate experiment at conditions below the three-phase ( L H - V ) equilibrium line’ (P = 100 psia, T = 279 K). This particular sample is collected when the cell is being cooled from the initial point, which implies that water is not seeded by microcrystalline hydrate particles, and thus hydrate formation is not a factor to be considered during this experiment. The NMR spectrum of this sample is shown in Figure 8b. This particular spectrum does not show any sign of gas molecules present in the liquid phase, which suggests that the NMR peaks of methane at 6 -2.75 and of propane at 6 = 18.07 shown in Figure 8a represent gas molecules trapped in hydrate cages.
-
-
Methane/Propane Hydrates
The Journal of Physical Chemistry, Voi. 97, No. 25, I993 6725
15
10
5
-5
0 PPM
-10
-15
Figure 10. 13CNMR of a CH4/C3Hs hydrate sample collected a t 194.03 psia and 281 K in the heating direction.
TABLE I: Nuclear Spin Densities of c)4:C& (964%) in the Hydrate Metastable/Nonequilibnum Region T = 279 KJAP = 9 p i a
T = 281 KlAP = 2 psia
XI PI nl XI Pi ni CH4(P-D) 0.371 5.94 X 1021 1.31 X 10'9 0.177 6.3 X 10M 1.4 X 1018 C,Hc(H-D) 0.05 8.05 X 1020 1.79 X 10'' CHd(1) 0.579 9.32 X 1 8 ' 2.07 X lOI9 0.823 2.94 X 1021 6.5 X 1018
By increasing the temperature of the sample in the NMR tube from 279 to 28 1 K, the NMR spectrum showed definite changes, as seen in Figure 10. At point A (281 K and 194 psia) hydrates visually disappeared. The gas-phase peak of the liberated CH4(l) has increased in relative intensity, whereas the peak showing the trapped CH4 in the P-D cages has decreased in relative intensity. This is verified by the relative ratio of the methane gas peak to the peak of methane in the P-D cages. As for the propane peak, it has totally vanished. This behavior confirms the idea that hydrate cages are breaking up. Since H-D cages are more strained than P-D cagesZSdue to the size of the guest molecules, it is obvious that P-D cages will resist the breaking up process more than H-D cages. Integrated signal intensities can be used to get information regarding the nuclear spin densities (pi) of methane and propane enclathrated in small and large hydrate cages, respectively, in a liquid sample. The specific nuclear spin density is
xi
=4
4
(3)
The number of nuclei is (4) Table I shows different nuclear spin densities at various conditions.
Conclusion The results presented in this paper are the first encouraging evidence that one can explore hydrates in the metastable/ nonequilibrium regions. This combination of macroscopic experiments with microscopic experiments is extremely useful and powerful for investigation of clathrate hydrates in their metastable/nonequilibrium stage. Moreover, the presence of a chemical shift/peak intensity dependency on enclathrated and liberated CH4/C,Hg concentrations provided a nondestructive, well-defined technique to significantlyadvance hydrate research. 13C NMR peaks of methane and propane hydrates are detected at 279 K and 187 psia. At 28 1K and 194 psia, hydrate microscopic particles still remained undissociated even though the water cleared. Definite changes in the 13C NMR spectrum were seen: the liberated methane gas peak increased in relative intensity, the methane hydrate peak decreased in relative intensity, and the propane hydrate peak totally vanished. Future work will include much more quantitative kinetic results such as rate constants and
rates of the water +gas = hydrates reaction, plus a deeper insight into the remaining hydrate cages beyond the visual disappearencx of the last particle. Future work will involve studies of regions I and IV, where hydrate particles are invisible.
Acknowledgment. The Gas Research Institute and the Gas ProcessorsAssociation are greatly acknowledgedfor their support of this project under GRI, Contract No. 5091-260-2123. Nomenclature Ai = area under the peak At CAI ni = number of nuclei N = Avogadro's number AP = initial pressure - pressure measured (psia) R = gas-phase constant V = sample volume = 2.22 X 10-3 (L) p = spin density (nuclei/L)
References and Notes (1) van der Walls, J. H.; Platteeuw, J. C. Narure 1959, 183, 462. (2) Hammerschmidt, E. G. Ind. Eng. Chem. 1934, 26, 851. (3) Deaton, W. H.; Frost, E. M.US.Bureau of Mines Monograph 1946, 8.
(4) Handa, Y. P. (a) J. Chem. Thermodyn. 1986,18, 15; (b) Ind. Eng. Chem. 1988, 27, 812. (5) Kobayashi, R. Vapor-Liquid Equilibrium in Binary HydrocarbonWater Systems. Ph.D Dissertation, University of Michigan, 1951. (6) Wilcox, W. I.; Carson, D. B.; Katz, D. L. I d . Eng. Chem. 1941,33, 662. (7) Yamamuro, 0.;Suga, H. J. Therm. Anal. 1989, 35, 2025. (8) Robinson, D. B. Fluid Phase Equilibria 1989, 52, 1. (9) Sloan, E. D. (a) Reuue de L'lnstituf Francais du Perrole 199@,45, 245; (b) Clathrate Hydrate of Natural Gases; Marcel Dekker: New York, 1990. (10) Ward, A. M.;Song, K. Y.; Jett, M.;Kobayashi, R. Thermophysical Laboratory 1989, Invited Paper, NIST. (11) Sloan, E. D.; Fleyfel, F. AIChE J. 1991, 37, 1281. (12) Bishnoi, P. R.; Vysniauskas, A. Kinetics of Gas Hydrate Formation Pt. It.; Final Report to Gas Research Institute, Chicago, 1980a. (13) Holder, G. D.; Manganiello, D. J. Chem. Eng. Sci. 1982, 37, 9. (14) Kamath, V. A.; Godbole, S . P. J . Petr. Tech. 1987, 39, 1379. (15) von Stackelberg, M.Natunvissenschaften 1949, 36, 327. (16) Bertie, J. E.; Othen, D. A. Can. J. Chem. 1972, 51, 1159. (17) Fleyfel, F.; Devlin, J. P. J. Phys. Chem. 1991, 95, 3811. (18) Lin, S.S . Rev. Sci. Instrum. 1973, 44, 516. (19) Hermann, V.; Kay, B. D.; Castleman, Jr., A. W. J . Chem. Phys. 1982, 72, 185. (20) Castleman, Jr., A. W.; Yang, X . J. Am. Chem. Soc. 1989, I 11,6845. (21) Garg, S.K.; Gough, S. R.; Davidson, D. W. J . Chem. Phys. 1975, 63, 1646. (22) Ripmeester, J. A.; Ratcliffe, C. I. J. Phys. Chem. 1987, 92, 337. (23) Jameson, A. K.; Jameson, C. J. Chem. Phys. Lett. 1987,134,461. (24) Breitmaier, E.; Haas, G.; Vcelter, W. Atlas of I3C NMR Data 1979, I, 3. (25) Davidson, D. W. Water, a Comprehensive Treatise; Franks, F., Ed.; Plenum Press: New York, 1972; Vol. 3, Chapter 2. (26) Makogon, Y. F. Hydrates of Natural Gas; Nedra: Moscow, 1914. (27) Davidson, D. W. Gas Hydrates as Clathrate Hydrates. In Gas Hydrates as Clathrate Ices natural Gas Hydrates: Properties, Occurences and Recovery; Cox, J. L., Ed.; Butterworth: Woburn, MA, 1983.