J. Chem. SOC., Faraday Trans. I , 1989, 85(7), 1537-1544 Study of Ultramicroporous Carbons by High-pressure Sorption Part 1.-N,, CO,, 0, and He Isotherms Jacob E. Koresh,* T. H. Kim and W. J. Koros Department of Chemical Engineering, University of Texas at Austin, Austin, T X 78712, U.S.A. Adsorption-desorption isotherms for nitrogen, carbon dioxide, oxygen and helium on as-received TCM- 128 ultramicroporous carbon are reported for pressures up to 60 atmt at 35 "C. Evidence is presented that suggests that there are regions in the carbon that are composed of tiny hydrophobic constrictions in series, which are hardly penetrated, and more open pores inside. At room temperature, water molecules cannot penetrate these constrictions in reasonable time due to a clustering effect, while the much bigger, but unclustered nitrogen and carbon dioxide molecules do penetrate these constrictions at a measurable rate.These regions are responsible for unexpected hysteresis observed for nitrogen and carbon dioxide at 35 "C and the unusually large amount of helium adsorbed. In earlier papers, we have reported studies under a variety of conditions of a carbon sorbent referred to as TCM- 1 28.'-11 Room-temperature water sorption equilibria were measured in as-received and activated samples of this material as a means of probing the total available pore volume, including tiny pores not easily accessible to larger penetrants such as nitrogen and o ~ y g e n . ~ - ~ * ~ ' Low-temperature sorption kinetics and equilibria for various gases were also measured for the as-received and activated material^.^ Pore regions that were detectable with water sorption in the unactivated as-received TCM-128 carbon at room temperature were not penetrated by nitrogen at a detectable rate at - 80 "C.Two TCM-128 samples with slightly different degrees of activation showed no difference in nitrogen sorption uptake at -80 "C. On the other hand, sorption levels were two orders of magnitude smaller at - 196 "C for the less-activated sample compared with the slightly more activated material.2 This molecular sieving results from the inability of the low-temperature nitrogen to penetrate the rigid-pore system of the less-activated carbon at an observable rate. Based on previous studies, basic features of the morphology of the TCM-128 material have been post~lated.l*~ Specifically, the material is believed to consist of rather open pores with periodic constrictions that occur in series.The intervening passageways between the constrictions are believed to be sufficiently large to offer little resistance to movement, so the essential resistance to transport consists of the series of constrictions. This morphology explains the observed coexistence of a molecular sieving ability and a high effective degree of molecular mobility in the TCM. Because of the paucity of data on the as-received substrate material, we chose to focus our attention on it to establish a basepoint for later comparisons. The present study extends our characterization of TCM- 128 to include high-pressure t 1 atm = 101 325 Pa.15371538 Study of Ultramicroporous Carbons by High-pressure Sorption pressure transducer A pressure transducer B w Fig. 1. Schematic of the volumetric high-pressure adsorption systei m. sorption/desorption equilibria data for nitrogen, carbon dioxide, oxygen and helium in the as-received material near ambient temperature. The data lead us to an improved understanding of the complex morphological nature of this ultramicroporous material. In this respect, the different size penetrants serve as ultrafine probes of the structure of the material. Experimental Materials A fibrous carbon cloth TCM-128 was supplied by Carbone Lorraine, France, and was used as received. The material was obtained from the same sample lot as the material that was considered earlier in our sorption measurements with water vapour.The nitrogen, carbon dioxide, oxygen and helium gases used were obtained from Linde at a purity of greater than 99.9% and were used as received. Adsorption Cell Gas sorption measurements were made up to 60 atm at 35 "C using a volumetric system described before12 but with some modification. The system is schematically shown in fig. 1. The carbon cloth was placed in a chamber of volume VB. After an evacuation for 24 h, valve B was closed and gas was introduced into chamber A through valve A at the desired pressure. The amount of gas introduced was calculated by: n = - ' A 'A zRT where z , the compressibility factor, was calculated for each pressure from a PVT data source.13 During the experiment, valve B was opened to allow gas into the sample for sorption, or to remove a controlled amount for incremental desorption.The pressure in the sample chamber was recorded continuously as a function of time until the rate of change of pressure decay became undetectable in all cases except for the higher pressure (> 35 atm) nitrogen adsorption runs. In these cases, a very protracted long-term uptake was apparent, as will be described later. In these high-pressure nitrogen runs, the experiments were arbitrarily terminated when the change in pressure over a 24 h period was less than 0.5% of that observed over the first 24 h period of the incremental sorption run. Typically, this meant that run times for each high-pressure incremental nitrogen sorption point were carried out for 6 days.The amount of gas adsorbed in each incremental sorption or desorption step was calculated by subtracting the remaining amount of gas in each volume (A and B) fromJ . E. Koresh, T. H. Kim and W. J. Koros 1539 60 1 .' 04 I I I 0 2 0 4 0 60 Platm Fig. 2. High-pressure nitrogen adsorption-desorption isotherm on TCM- 128 as-received carbon at 35 "C. 4, desorption; ., adsorption. the cumulative amount injected or removed. Since the system has a separate pressure transducer for each chamber, a material balance is maintained on all gas in the cell; therefore, equilibrium calculations are not affected by the length of time, or the extent to which valve B is opened. Results and Discussion Nitrogen Isotherms Fig. 2 shows the high-pressure adsorption and desorption isotherms for nitrogen at 35 "C on the as-received carbon TCM-128 after evacuation at 35 "C for 24 h.The iso- therm has a type I shape, without reaching a plateau, thereby indicating that pore volume filling is not complete over the pressure range studied. The as-received TCM- 128 is an ultramicroporous adsorbent and does not adsorb measurable amounts of nitrogen at liquid nitrogen temperature over experimentally accessible time scales.' Moreover, it hardly adsorbs the tiny CO, molecule at -78 "C,' but does adsorb water7.' as well as nitrogen near room temperature as shown in fig. 2. Under these conditions, it is reasonable to assume a mass density for adsorbed water equivalent to the density of water in the liquid state (e.g. 1.0 g ~ m - ~ ) . For comparison, it is useful to assume that the mass density for adsorbed nitrogen is equivalent to the density of nitrogen in the liquid state at - 196 "C (0.808 g cme3).The nitrogen-accessible pore volume calculated in this fashion from the sorption level at 60 atm is 0.066 cm3 g-' (TCM- 128) as compared with 0.120 cm3 g-' (TCM- 128) determined from the apparent plateau value of the water isotherm on the same batch of TCM-128.7i10 This is an interesting result, since the nitrogen isotherm is clearly still not saturated at 60 atm, so more capacity exists for nitrogen if higher pressures could be reached. Moreover, note that correction of the liquid density for the ambient temperatures used here will lead to even higher estimates of accessible pore volumes for the nitrogen, so the use of the 0.808 g cm-3 value provides a conservatively low estimate of the pore volume accessible to nitrogen at 35 "C.Thus, the fractional saturation of the pore volume (estimated from water sorption plateau) by nitrogen at 60 atm and 35 "C is, therefore, about 0.55.1540 Study of Ultramicroporous Carbons by High-pressure Sorption Further perspective on the large extent of penetration of the as-received TCM- 128 by the high-pressure nitrogen at this ambient temperature can be offered by considering earlier nitrogen adsorption studies at 20 and -78 "C on an active carbon (AC) with less discriminating constrictions in the pore network than the TCM- 128? The nitrogen uptake at 60atm and 20 "C in the AC corresponds to filling an apparently smaller fraction of the pore volume, as in the TCM; note that the AC pore volume was sensed by the -78 "C nitrogen sorption uptake plateau instead of by the room temperature water uptake plateau as in the TCM.Under equivalent pressure conditions on any adsorbent, however, nitrogen at -78 "C, clearly adsorbs much less than at liquid- nitrogen conditions where a high affinity constant allows saturation of all pore volume that is not 'closed' by molecular sieving exclusion. For example, on slightly activated TCM-128 at - 196 "C, the asymptotic nitrogen uptake was more than three times higher than at -78 "C (see fig. 2 in ref. 2). It was shown earlier that essentially the same pore volume was available to nitrogen at liquid nitrogen temperature in a slightly activated TCM-128 sample as was available to water at room temperature on the as-received TCM (comparing fig.2 in ref. 2 with fig. 1 in ref. 10). The kinetic restrictions to penetration of the standard pore regions by nitrogen were eliminated by the slight activation. Without the slight activation, essentially no nitrogen penetration of the TCM-128 was possible and the pore volume for the as-received TCM could be assessed only from the plateau level of the water isotherm. Therefore, we can conclude that on an equivalent basis, the 60 atm nitrogen uptake in the present study corresponds to very significantly higher fractional filling of the total volume as compared with that occurring in the AC under similar pressure and temperature conditions. At least two hypotheses are possible to explain these results.In fact, phenomena related to both hypotheses may actually be responsible for the observed results; however, they are presented independently here for ease of discussion. The first hypothesis relies upon the conventional wisdom that a more energetically favourable thermodynamic environment may exist in the TCM- 128, since the attractions from both pore walls experienced by adsorbates in the fine-pored TCM produces a higher effective affinity constant compared with the more open AC in which dual wall interactions of individual penetrants are not anticipated. l5 A second viable hypothesis is more speculative, and suggests that nitrogen at 35 "C and at high concentrations or pressures can penetrate pore volume in the TCM- 128 that is not accessible to water at the same temperature.In this case, since the water is excluded from certain regions in the TCM- 128, one tends to underestimate the available pore volume and overestimate the fractional saturation of the total volume. This situation should not be present in the more thoroughly oxidized and open AC where water has essentially full access to all regions. The above hypothesis is based on the fact that a potential clustering mechanism exists for the hydrogen-bonding water molecule which is not a factor for nitrogen. Clearly, an individual water molecule has a smaller sieving dimension than a corresponding individual nitrogen molecule ;Is however, when clustering occurs, this simple picture no longer applies. Even in hydrophobic polymers, clustering of water is well known to cause reduced effective water diffusion ~0efficients.l~ In the case of carbons, polar surface oxide groups provide a more compatible environment for the individual water molecules, and thereby should suppress water cluster formation. Kinetic factors related to the tremendously slower penetration of clustered water, compared with the unclustered species, may make water penetration through tiny hydrophobic constrictions unobservable under realistic time scales. Mechanistically, the reduced mobility can be described in terms of the additional energy of activation needed for dissociation to individual molecules to be able to fit through the smallest constrictions.Overcoming a hydrogen bond strength of the order of 5 kcal mol-1 whichJ .E. Koresh, T. H. Kim and W. J. Koros 1541 is typical of water in such a process would account for a 3500-fold lower diffusion coefficient of the clustered material. Clearly in this case, one would not observe such extremely slow penetration of tiny hydrophobic pores. Hydrophobic tiny Rores could easily arise from defects in unoxidized graphite interlayers whose 3.4 A effective spacing is smaller than the standard constrictions which presumably have oxide surfaces. Evidence for the presence of oxidized surfaces on the standard constrictions has been reported.' Specifically, low-temperature (300 "C) activation of TCM led to CO, emission with a negligible resultant weight loss and caused an increase of four orders of magnitude in the rate of adsorption of CO, at -78 "C.These results suggest that, while the internal surfaces of the standard pores responsible for most of the sorption capacity have oxide groups, the standard molecular sieving constrictions must also have oxide groups, since the gentle activation which removes small amounts of these oxides led to such a large increase in the uptake rates of CO,. The hysteresis shown for the nitrogen isotherms reflects a gradual merging of the desorption curve with the adsorption curve. This form of hysteresis differs from that observed when desorption curves drop rapidly to meet the adsorption curve at a certain characteristic pressure, depending upon bottle necks or constriction dimensions. Hysteresis similar to that shown in fig. 2 for the as-received TCM-128 is not typical of nitrogen or CO, high-pressure isotherms at ambient temperature with microporous carbons of zeolites.1 4 7 18. l9 For an earlier room- temperature high-pressure methane isotherm on an activated TCM-128, with larger pore dimensions that were still in the microporous range, no hysteresis was observed at all. Such a hysteresis response, referred to as low-pressure hysteresis, has been reported by Bailey et al. for sorption of strongly interacting organic vapour on high- and low-activated carbons.20 In the present study, long equilibration times were allowed (typically 5 days, even though the samples appeared to be essentially at equilibrium after only 2 days). During desorption, a secondary process was observed at high pressures and long times, which suggests a protracted time-scale for diffusive uptake into regions whose access is controlled by tiny constrictions.The secondary process was observed as a slow but easily detectable readsorption of nitrogen after the simple desorption process had apparently been completed for high-pressure desorption steps. The rate of readsorption moderated by uptake through the tiny constrictions decreased as the pressure decreased, and it completely stopped at 40 atm. At this point, therefore, it appears that true equilibrium saturation had been achieved in both the regions, accessible through only tiny pores as well as through the larger pores. During subsequent desorption, we waited for extremely long periods of time beyond that when the majority of the desorption process occurred. Usually we waited for 2-3 days for each point to reach equilibrium and have waited an extra 3 days for some points and found that the additional amount desorbed is negligible (about 4% of the total amount desorbed at that pressure) and far from closing the gap between the adsorp- tion-desorption curves.In summary, we believe that the restricted regions have essentially reached desorption equilibrium, and in this case, the true equilibrium adsorption levels of both the restricted and the more open pore regions are well represented by the desorption curve below 40 atm. Conversely, the sorption uptake in regions whose access is moderated by dif- fusion through standard pores is approximately represented by the adsorption curve up to 40 atm. Based on the above discussion, above 40 atm, presumably both sorption and desorption isotherms represent a complex mixture of uptake into the standard and 'closed porosity' regions determined by the pressure and exposure time of the sample to high-pressure gas.52 FAR I1542 Study of Ultramicroporous Carbons by High-pressure Sorption 0 4 8 0 2 0 4 0 60 Platm Fig. 3. High-pressure carbon-dioxide adsorption-desorption isotherm on TCM- 128 as-received carbon; inset lower pressures isotherm. H, adsorption ; +, desorption. Carbon Dioxide Isotherms The adsorption4esorption isotherms for CO, in fig. 3 show a hysteresis that is similar to that seen for nitrogen; however, the desorption curve does not meet the adsorption curve. As in the preceding discussion of nitrogen, the CO, hysteresis is believed to reflect uptake into regions at high pressures whose access is limited by movement through tiny constrictions.Unlike N,, for the smaller CO, molecule these regions are able to saturate during the extended high-pressure adsorption runs prior to beginning the desorption process. This fact is reflected by the observation that for carbon dioxide there was no sign of the readsorption phenomenon at the higher desorption pressures mentioned for desorption runs with N, for pressures between 60 and 40 atm. This observation is reasonable, since the ratio of the diffusion coefficients (and hence equilibration times) in the standard pores relative to those in the tiny constrictions would be nearer unity for the smaller CO, compared to N,. At about 55-60 atm, both the sorption and desorption isotherms indicate that saturation of the entire pore volume has occurred.By following the same approach as in the N, case, and assuming the liquid-like density of 1.031 g cm-3 for CO, measured at -20 "C21 along with the pore saturation value taken from fig. 3, an accessible pore volume of 0.146 cm3 g-l results. This estimate is more than 20 % higher than the pore volume calculated from the plateau amount sorbed in the water isotherm at 25 "C. Again, this is a conservative estimate, since the adsorbed density at 35 "C may be somewhat less than that of liquid CO, at - 20 "C. As in the case of N,, the foregoing discussion suggests that carbon dioxide molecules can penetrate into regions which are not accessible to water over the time scale of a typical sorption experiment and thus giving the more accurate higher pore volume.In the case of the water sorption measurements, sorption equilibration appeared to be reached in 3-4 h, so a much slower (> 3000 times) penetration of clustered water wouldJ. E. Koresh, T. H. Kim and W. J. Koros 1543 60 n O 40 v 3 0 0 20 n c D v 5 0 0 10 2 0 3 0 0 20 4 0 6 0 Platm P l a n Fig. 4. Fig. 5. Fig. 4. High-pressure oxygen adsorption-desorption isotherm on TCM- 128 as-received carbon. m, adsorption ; +, desorption. Fig. 5. High-pressure helium adsorption4esorption isotherm on TCM- 128 as-received carbon. A, adsorption; + , desorption. not be apparent. For non-clustering penetrants such as nitrogen and carbon dioxide, experiments that are only four to five times longer than necessary for saturation of the standard pore volume allow observation of at least measurable extents of saturation of the regions whose access is moderated by diffusion through tiny constrictions.Oxygen and Helium Isotherms High-pressure adsorption-desorption isotherms on TCM- 128 carbon are shown in fig. 4 and 5 at 35 "C for oxygen and helium, respectively. Unlike the two previous penetrants, no hysteresis is apparent. It is likely that measurements at higher pressures with oxygen would have shown results between those of nitrogen and carbon dioxide, since oxygen's minimum dimension falls between those of nitrogen and carbon dioxide. We performed the oxygen measurements only up to 25 atm, due to concerns about safety in our external gas handling equipment. Based on the preceding discussions, it is likely that the oxygen results represent adsorption equilibria only into the standard pores with essentially no contribution from the tiny pores probed by both nitrogen and carbon dioxide at high pressures and long exposure times.The helium data are interesting in terms of the extremely large uptake that is apparent for this low condepibility penetrant. We presume that the extremely small minimum dimension (ca. 2.6 A) of helium allows it to probe essentially the entire accessible volume of the solid; however, due to helium's low sorption affinity constant, saturation would require studies at higher pressures than our equipment can presently reach. Comparable high-pressure adsorption levels of helium on active carbon were observed previously only at liquid nitrogen temperature.14 52-21544 Study of Ult ram icroporous Carbons by High-pressure Sorption Conclusions Unexpected hysteresis, observed in the adsorption-desorption isotherms for nitrogen and carbon dioxide, can be explained in terms of the existence of tiny constrictions that moderate uptake into these regions in the as-received TCM-128.These new features were able to be observed due to the different experimental techniques used in this study which combined high pressures and ambient temperature capabilities with protracted run times. Comparison of the nitrogen, carbon dioxide and helium data in this study with earlier water adsorption measurements in the same material suggests that these gases have access to regions not available to water.This phenomenon can be explained in terms of clustering of water near tiny hydrophobic constrictions, moderating access to these regions. Mechanistically, the clustering adds significant activation energy to the diffusion coefficient for water, thereby greatly suppressing the ability of this otherwise tiny molecule to penetrate hindered environments. References 1 J. Koresh and A. Soffer, J. Chem. Soc., Faraday Trans. I , 1980, 76, 2457. 2 J. Koresh and A. Soffer, J. Chem. Soc., Faraday Trans. 1, 1980, 76, 2472. 3 J. Koresh and A. Soffer, J. Chem. Soc., Faraday Trans. 1, 1980, 76, 3005. 4 J. Koresh and A. Soffer, J. Chem. Soc., Faraday Trans. 1, 1980, 76, 2507. 5 J. Koresh, J. Colloid Interface Sci., 1982, 88, 398. 6 J. Koresh and A. Soffer, J. Colloid Interface Sci., 1983, 92, 517. 7 S. S. Barton and J. E. Koresh, J. Chem. SOC., Faraday Trans. 1, 1983, 79, 1147. 8 S. S. Barton and J. E. Koresh, J. Chem. Soc., Faraday Trans. I , 1983, 79, 1165. 9 S. S. Barton and J. E. Koresh, J. Chem. SOC., Faraday Trans. I , 1983, 79, 1173. 10 J. E. Koresh, A. Soffer and H. Tobias, Carbon, 1985, 23, 571. 1 1 S. S. Barton, M. J. B. Evans, J. E. Koresh and H. Tobias, Carbon, 1987, 25, 663. 12 W. J. Koros and D. R. Paul, J. Polym. Sci. : Polym. Phys. Edn, 1976, 14, 1903. 13 F. Din, Thermodynamic Function of Gases (Butterworths, London, 1961), vol. 3. 14 A. von Antropoff, Kolloid Z., 1952, 129, 1 1 ; Kolloid Z., 1954, 137, 105, 108. 15 D. A. Everett and J. C. Powl, J. Chem. SOC., Faraday Trans. I , 1976, 72, 619. 16 D. W. Breck, Zeolite Molecular Sieves (Wiley, New York, 1974), p. 636. 17 V. T. Stannett, G. R. Ranade and W. J. Koros, J. Membrane Sci. 1981, 10, 219. 18 S. S. Barton, J. R. Dacey and D. F. Quinn, in Fundamentals of Adsorption, ed. A. L. Myers and 19 P. G. Menon, Chem. Rev., 1968, 68, 277. 20 A. Bailey, D. A. Cadenhead, D. A. Everett and A. J. Miles, Trans. Faraday SOC., 1971, 67, 231. 21 Handbook of Chemistry and Physics (CRC Press, Boca Raton, Florida, 64th edn, 1983-84), p. C-219. G. Belfort, Proceedings of Engineering Foundation Conference, West Germany, May (1983), p. 65. Paper 81012755; Received 30th March, 1988