Hydrogen Storage in Nanostructured Carbons by Spillover: Bridge

Lines represent data from Goodell38 after 20 adsorption−desorption cycles. All samples were pretreated in situ in the measurement apparatus prior to...
0 downloads 0 Views 605KB Size
11418

Langmuir 2005, 21, 11418-11424

Hydrogen Storage in Nanostructured Carbons by Spillover: Bridge-Building Enhancement Anthony J. Lachawiec, Jr., Gongshin Qi, and Ralph T. Yang* Department of Chemical Engineering, University of Michigan, Ann Arbor, Michigan 48109 Received June 20, 2005. In Final Form: September 3, 2005 The hydrogen storage capacity in nanostructured carbon materials can be increased by atomic hydrogen spillover from a supported catalyst. A simple and effective technique was developed to build carbon bridges that serve to improve contact between a spillover source and a secondary receptor. In this work, a supported catalyst (Pd-C) served as the source of hydrogen atoms via dissociation and primary spillover and AX-21 or single-walled carbon nanotubes (SWNTs) were secondary spillover receptors. By carbonizing a bridgeforming precursor in the presence of the components, the hydrogen adsorption amount was increased by a factor of 2.9 for the AX-21 receptor and 1.6 for the SWNT receptor at 298 K and 100 kPa. Similar results were obtained at 10 MPa, indicating that the enhancement factor is a weak function of pressure. The AX-21 receptor with carbon bridges had the highest absolute capacity of 1.8 wt % at 298 K and 10 MPa. Reversibility was demonstrated through desorption and readsorption at 298 K. The bridge-building process appears to be receptor specific, and optimization may yield even greater enhancement. Using this technique, enhancements in storage of up to 17-fold on other carbon-based materials have been observed and will be reported elsewhere shortly.

Introduction The U.S. Department of Energy (DOE) has established a multistage target for the hydrogen storage capacity in materials intended for fuel cell applications. The targets for on-board hydrogen storage materials are 1.5 kW/kg (4.5 wt %) by 2007, 2 kW/kg (6 wt %) by 2010, and 3 kW/kg (9 wt %) by 2015.1 It is expected that these targets are met at moderate temperatures and pressures for practical implementation. An ideal system would operate near ambient temperature and at low to moderate pressures and possess rapid adsorption kinetics for refueling and desorption kinetics for fuel delivery to the vehicle power plant. The earliest report of hydrogen adsorption by singlewalled carbon nanotubes (SWNTs)2 generated experimental and theoretical interest in the use of nanostructured carbon materials for hydrogen storage, which has continued to increase as researchers attempt to deliver a viable material that meets the DOE targets. Various forms of carbon nanotubes,3-12 graphite nanofibers,13-18 and * To whom correspondence should be addressed. E-mail: yang@ umich.edu. (1) Hydrogen, Fuel Cells, and Infrastructure Technologies Program Multi-Year Research, Development, and Demonstration Plan. U.S. DOE Energy Efficiency and Renewable Energy (EERE) Home Page. http:// www.eere.energy.gov (accessed June 2005), path: /hydrogenandfuelcells/ mypp/pdfs /storage.pdf. (2) Dillon, A. C.; Jones, K. M.; Bekkedahl, T. A.; Kiang, C. H.; Bethune, D. S.; Heben, M. J. Nature 1997, 386, 377-379. (3) Ye, Y.; Ahn, C. C.; Witham, C.; Fultz, B.; Liu, J.; Rinzler, A. G.; Colbert, D.; Smith, K. A.; Smalley, R. E. Appl. Phys. Lett. 1999, 74, 2307-2309. (4) Wang, Q.; Johnson, J. K. J. Phys. Chem. B 1999, 103, 277-281. (5) Liu, C.; Fan, Y. Y.; Liu, M.; Cong, H. T.; Cheng, H. M.; Dresselhaus, M. S. Science 1999, 286, 1127-1129. (6) Chen, P.; Wu, X.; Lin, J.; Tan, K. L. Science 1999, 285, 91-93. (7) Yang, R. T. Carbon 2000, 38, 623-626. (8) Dillon, A. C.; Heben, M. J. Appl. Phys. A 2001, 72, 133-142. (9) Tibbetts, G. G.; Meisner, G. P.; Olk, C. H. Carbon 2001, 39, 22912301. (10) Cheng, H. S.; Pez, G. P.; Cooper, A. C. J. Am. Chem. Soc. 2001, 123, 5845-5486. (11) Simonyan, V. V.; Johnson, K. J. J. Alloys Compd. 2002, 330, 659-665. (12) Lueking, A.; Yang, R. T. J. Catal. 2002, 206, 165-168.

activated carbon and graphite,19-21 have been studied as potential hydrogen adsorbents. A detailed review of the material structure, experimental techniques, and observed capacities is given elsewhere.22 It is well-known that considerable controversy exists in the large amounts of hydrogen storage in these nanostructured carbons. Nonetheless, three common features exist in the reported experiments on hydrogen storage in carbon nanotubes: slow uptake, irreversibly adsorbed species, and the presence of reduced transition metals. The experimental evidence, combined with ab initio molecular orbital calculations of hydrogen atoms on graphite, has led to the proposal of a mechanism for hydrogen storage in carbon nanostructures involving hydrogen dissociation on metal particles followed by atomic hydrogen spillover and adsorption on the nanostructured carbon surface.23 The phenomenon of spillover was first postulated in the early 1960s to explain the room temperature formation of tungsten bronze from the reduction of tungsten oxide mixed with supported platinum.24,25 Fundamental hydrogen spillover concepts were studied in the ensuing (13) Chambers, A.; Park, C.; Baker, R. T. K.; Rodriguez, N. M. J. Phys. Chem. B 1998, 102, 4253-4256. (14) Ahn, C. C.; Ye, Y.; Ratnakumar, B. V.; Witham, C.; Bowman, R. C.; Fultz, B. Appl. Phys. Lett. 1998, 73, 3378-3380. (15) Park, C.; Anderson, P. E.; Chambers, A.; Tan, C. D.; Hidalgo, R.; Rodriguez, N. M. J. Phys. Chem. B 1999, 103, 10572-10581. (16) Gupta, B. K.; Srinivastava, O. N. Int. J. Hydrogen Energy 2001, 26, 857-862. (17) Browning, D. J.; Gerrard, M. L.; Lakeman, J. B.; Mellor, I. M.; Mortimer, R. J.; Turpin, M. C. Nano Lett. 2002, 2, 201-205. (18) Lueking, A. D.; Yang, R. T.; Rodriguez, N. M.; Baker, R. T. K. Langmuir 2004, 20, 714-721. (19) Chahine, R.; Bose, T. K. Int. J. Hydrogen Energy 1994, 19, 161164. (20) Lamari, M.; Aoufi, A.; Malbrunot, P. AIChE J. 2000, 46, 632646. (21) Orimo, S.; Meyer, G.; Fukunaga, T.; Zu¨ttel, A.; Schlapbach, L.; Fujii, H. Appl. Phys. Lett. 1999, 75, 3093-3095. (22) Yang, R. T. Adsorbents: Fundamentals and Applications; Wiley: New York, 2003; pp 305-321. (23) Yang, F. H.; Yang, R. T. Carbon 2002, 40, 437-444. (24) Khoobiar, S. J. Phys. Chem. 1964, 68, 411-412. (25) Sinfelt, J. M.; Lucchesi, P. J. J. Am. Chem. Soc. 1963, 85, 33653367.

10.1021/la051659r CCC: $30.25 © 2005 American Chemical Society Published on Web 10/13/2005

Hydrogen Storage in Nanostructured Carbons

Langmuir, Vol. 21, No. 24, 2005 11419

Figure 1. Hydrogen spillover in a supported catalyst system: (a) adsorption of hydrogen on a supported metal particle; (b) the low-capacity receptor; (c) primary spillover of atomic hydrogen to the support; (d) secondary spillover to the receptor enhanced by a physical bridge; (e) primary and secondary spillover enhancement by improved contacts and bridges.

decade into the early 1970s. A number of comprehensive reviews on hydrogen spillover are available, most recently by Conner26 and Conner and Falconer.27 Despite continued investigations and research to support the theory, the mechanistic details of hydrogen spillover are still poorly understood. Spillover has been defined as the transport of an active species adsorbed on one site to another site that would not typically adsorb the active species at the prevailing conditions.27 By this argument, the adsorbed species can access new sites that are in close proximity or physical contact with the catalytic surface. It is theorized that a catalytic site, such as a metal particle, provides a source of hydrogen atoms by dissociation. The atomic hydrogen then spills over to the receptor via surface diffusion. This phenomenon may be of particular interest in the case of graphite nanofibers, where the spacing of the graphite platelets is restrictive to molecular hydrogen adsorption. The spillover of atomic hydrogen to a nanofiber receptor may promote intercalation of hydrogen into the structure and increase its capacity. (26) Conner, W. C., Jr. In Hydrogen Effects in Catalysis: Fundamentals and Practical Applications; Paa´l, Z., Menon, P. G., Eds.; Marcel Dekker: New York, 1988; Chapter 12. (27) Conner, W. C., Jr.; Falconer, J. L. Chem. Rev. 1995, 95, 759788.

Mixing a catalytic material, or source, with a previously inert or low-capacity receptor has been used to demonstrate hydrogen spillover.28,29 In several instances, the creation of physical “bridges” has been crucially important for spillover from the dissociation sites on metals to the substrate.30-33 This is particularly important when one considers the energy barriers that exist for surface diffusion of hydrogen atoms from the catalytic site to the receptor. The creation of carbon bridges is theorized to occur either as a result of pretreatment techniques31 or by carbonization of a precursor material.33 If the source of atomic hydrogen is a dissociating metal particle on a low-capacity support, the overall hydrogen adsorption may be increased by adding a high-capacity receptor. In this case, transport of atomic hydrogen from the metal particle to the support is primary spillover and transport to the receptor is secondary spillover. Figure 1 depicts the (28) Srinivas, S. T.; Rao, P. K. J. Catal. 1994, 148, 470-477. (29) Lueking, A. D.; Yang, R. T. Appl. Catal., A 2004, 265, 259-268. (30) Boudart, M.; Vannice, M. A.; Benson, J. E. Z. Phys. Chem. (Muenchen) 1969, 64, 171-177. (31) Boudart, M.; Aldag, A. W.; Vannice, M. A. J. Catal. 1970, 18, 46-51. (32) Levy, R. B.; Boudart, M. J. Catal. 1974, 32, 304-314. (33) Fujimoto, K.; Toyoshi, S. In New Horizons in Catalysis: Proceedings of the 7th International Congress on Catalysis; Seiyama, T., Tanabe, K., Eds.; Elsevier: New York, 1981; pp 235-246.

11420

Langmuir, Vol. 21, No. 24, 2005

spillover process in a supported catalyst system. The secondary spillover source is shown in Figure 1a, along with an adsorbed hydrogen molecule. Figure 1b is a schematic of the low-capacity receptor, which could be bundles of SWNTs or a microporous adsorbent such as AX-21. Note that this representation intends to show that the size of the hydrogen molecule precludes it from accessing the interstices in the SWNT bundle or additional micropores of AX-21. The molecular size may also prevent transport of hydrogen through any defects in an individual SWNT surface; thus, the adsorption at endohedral sites is dependent upon diffusion to uncapped ends and further transport inside the tube to these sites. Examples of primary and secondary spillover are shown schematically in parts c and d, respectively, of Figure 1. Dissociation is assumed to take place on the metal particle, and atomic hydrogen spills over to the support. The atoms are transported to the receptor via diffusion across the bridge and can access additional sites on the receptor. Note that, in addition to spillover enhancement by bridges, the creation of more intimate contact between the metal particle and the support may also contribute to primary spillover enhancement, as depicted in Figure 1e. In previous work reported by this laboratory,29 secondary spillover was demonstrated to increase the hydrogen storage capacity by mixing the receptor adsorbent with a small amount of supported catalyst that is capable of dissociating hydrogen. The objective of the current study is to demonstrate that carbon bridges can be effectively built between two carbon materials and that building these bridges can lead to significantly increased spillover and, hence, storage capacity. All measurements were made at ambient temperature and at pressures relevant for practical application to on-board storage systems. Experimental Methods Preparation of Samples. The source of hydrogen atoms for spillover was a commercially available catalyst consisting of 5 wt % palladium supported on active carbon (Strem Chemicals, Inc.). The secondary spillover receptors were AX-21 superactivated carbon (Anderson Development Co.) and SWNTs. The estimated BET surface area of the AX-21 carbon was 2880 m2/g with a median slit pore size of 14 Å measured by the HorvathKawazoe method. It is noted that BET surface area values for microporous materials are presented with the proviso that they may be overestimated since micropores are filled with nitrogen below the traditional relative pressure range that is valid for BET estimation (relative pressures from 0.05 to 0.3).34 The SWNTs used in this study were produced by catalytic decomposition of methane on an iron catalyst supported on a hybrid alumina-silica support, as described by Cassel et al.35 The asproduced SWNTs were purified using washes of hydrofluoric, nitric, and sulfuric acids to remove the catalyst and support. Treatment in hydrofluoric acid (48 wt % in water, Aldrich) was performed at room temperature for 24 h with stirring. After filtration and rinsing with deionized water, the SWNTs were treated three times with nitric acid (70 vol % in water, Mallinckrodt) at 353 K, again with stirring, for 1 h. The SWNTs were filtered and rinsed with deionized water between each treatment. The temperature and agitation conditions were identical for the final treatment, which employed concentrated sulfuric acid (95.9 vol % in water, Fisher Chemical) and was performed for 30 min. The SWNTs were dried in an oven at 373 K overnight prior to further analysis. An ash content analysis, according to the suggested conditions of ASTM D 2866-94, was performed to confirm that the SWNTs contained less than 1.0 wt % metal catalyst after purification. The purified SWNTs had (34) Li, F.; Wang, Y.; Wang, D.; Wei, F. Carbon 2004, 42, 23752383. (35) Cassell, A. M.; Raymakers, J. A.; Kong, J.; Dai, H. J. Phys. Chem. B 1999, 103, 6484-6492.

Lachawiec et al.

Figure 2. Sievert’s apparatus to measure high-pressure hydrogen isotherms. Shaded valves have high-pressure bellows seals. an estimated BET surface area of 820 m2/g with an average cylindrical pore size of 12 Å measured by the Horvath-Kawazoe method. While these receptors have limited ability for hydrogen dissociation, they are capable of receiving spillover hydrogen atoms from the source. To determine the enhancing effect of carbon bridges on secondary spillover, hydrogen adsorption measurements on the source, receptors, and simple physical mixtures of the components were used as a baseline. The source and receptors were ground together with an agate mortar and pestle for 30 min to produce the physical mixtures. The ratio of receptor to source was fixed at 9:1. After grinding, the mixtures were transferred to a quartz tube furnace where they were calcined in flowing helium at 673 K for 2 h. The samples were stored for subsequent pretreatment and hydrogen adsorption measurements. Carbon bridges between the source and receptor were formed with the addition of a hydrocarbon precursor to a physical mixture of the components. In this case, the hydrocarbon precursor was reagent grade D-glucose (Sigma-Aldrich Corp.). The receptor: precursor:source ratio was fixed at 8:1:1 on the basis of the complete carbonization of the precursor; thus, the bridges altered the amount of receptor compared to that in the simple physical mixture but not the source. The receptor/source physical mixture was ground with the precursor for 30 min. This material was transferred to a tubular reactor and heated in flowing helium with a temperature program designed to first melt the precursor and allow it to fill the interstices between the source and receptor and then fully carbonize the precursor to form bridges. For the D-glucose precursor, the temperature was increased at 1 K/min to 453 K (just above the melting point, 426 K) and held for 3 h. This step was designed to allow the glucose to melt thoroughly and to wet and fill the crevices between the receptor and the catalyst particles. In the carbonization step, the temperature was increased at 1 K/min to 673 K and held for 6 h. The material was cooled to room temperature in helium and stored for further pretreatment and hydrogen adsorption measurements. Hydrogen Isotherm Measurements. Hydrogen adsorption at pressures lower than 0.1 MPa (1 atm) was measured with a standard static volumetric technique (Micromeritics ASAP 2010). A custom sample holder was used that was configured to allow in situ pretreatment in a flowing gas stream. Approximately 50-100 mg of sample was used for low-pressure isotherm measurements. The apparatus was calibrated for hydrogen measurements at 298 K using palladium powder (Strem Chemicals, Inc.). Hydrogen adsorption at pressures greater than 0.1 MPa and up to 10 MPa was also measured using a static volumetric technique with a specially designed Sievert’s apparatus. A schematic of the apparatus is shown in Figure 2. The nonideality

Hydrogen Storage in Nanostructured Carbons

Figure 3. Calibration of Sievert’s apparatus with LaNi5 alloy powder: O, adsorption; b, desorption. Lines represent data from Goodell38 after 20 adsorption-desorption cycles. of hydrogen was accounted for by computing the compressibility factor using second and third virial coefficient correlations.36 Excellent agreement was obtained between these values and experimentally determined compressibilities reported in the literature.37 Approximately 300 mg of sample was used for highpressure isotherm measurements. The apparatus was calibrated at 298 K using LaNi5 powder (Cerac, Inc.). A comparison of the adsorption and desorption isotherms obtained with the instrument and literature data38 is presented in Figure 3. All samples were pretreated in situ in the measurement apparatus prior to isotherm measurements. The pretreatment conditions were similar to those of Srinivas and Rao28 and included two steps: reduction in hydrogen (40 sccm) at 523 K for 6 h and degassing in a vacuum at 673 K for a minimum of 8 h. Ultra-high-purity hydrogen (99.999%) and helium (99.999%) were obtained from Cryogenic Gases (Detroit, MI) and used for all pretreatments and measurements. Molecular sieve 3A purifiers were used on each gas stream to ensure purity was maintained in all experiments. Scanning Electron Microscopy. Samples of AX-21, SWNTs, physical mixtures of the spillover source and these receptors, and the source/AX-21 mixture with carbon bridges were examined using scanning electron microscopy (SEM). A Philips XL30 FEG SEM instrument was used to examine the specimens. The SEM accelerating voltage was 30 kV. An X-ray energy-dispersive spectrometry (XEDS) detector (EDAX, Inc.) enabled identification of palladium particles on the carbon support by elemental analysis.

Results and Discussion The low-pressure hydrogen adsorption isotherms for AX-21 and SWNT receptors are presented in Figures 4 and 5, respectively. The isotherm for the 5 wt % Pd-C catalyst shows the expected behavior of palladium hydride formation for pressures less than 12 kPa. The change in curvature at 12 kPa corresponds to an uptake amount of 3.2 cm3 (STP)/g or a hydrogen to metal ratio of 0.65:1, which is in agreement with the sloping upper branches of the isotherm attributed to the palladium hydride β phase.39 A second inflection is observed at approximately 90 kPa for the SWNT receptor. There has been no theoretical basis established for this phenomenon; how(36) McCarty, R. D. In Hydrogen: Its Technology and Implications: Hydrogen Properties; Cox, K. E., Williamson, K. D., Eds.; CRC Press: Cleveland, OH, 1975; Vol. 3, pp 18-19. (37) Smithsonian Physical Tables, 9th ed.; Forsythe, W. E., Ed.; Smithsonian Institution Press: Washington, DC, 2003; pp 260 and 268. (38) Goodell, P. D. J. Less-Common Met. 1984, 99, 1-14. (39) Lewis, F. A. The Palladium Hydrogen System; Academic Press: London, 1967; p 4.

Langmuir, Vol. 21, No. 24, 2005 11421

Figure 4. Low-pressure hydrogen isotherms at 298 K for the AX-21 receptor: O, 5 wt % Pd-C catalyst; ], AX-21/Pd-C/ carbon bridge (8:1:1); 4, AX-21/Pd-C physical mixture (9:1); 0, AX-21. The dotted line is the sum of fractional contributions based on uptake of individual mixture components.

Figure 5. Low-pressure hydrogen isotherms at 298 K for the SWNT receptor: O, 5 wt % Pd-C catalyst; ], SWNT/Pd-C/ carbon bridge (8:1:1); 4, SWNT/Pd-C physical mixture (9:1); 0, SWNT. The dotted line is the sum of fractional contributions based on uptake of individual mixture components.

ever, it is experimentally reproducible and has only been observed for the SWNT receptor. It is an interesting feature for future study as an attempt is made to gain a deeper understanding of the spillover mechanism. For both receptors, the physical mixture and the bridged material exceed the capacity that is expected if the individual contributions of the source and receptor are considered additive (represented by the dotted line in each plot). The enhancement of capacity upon adding bridges is clearly evident in the figures, with the AX-21 receptor demonstrating a higher capacity than the SWNT receptor at 100 kPa. Despite the enhancement due to spillover, the absolute capacities of the materials (0.04 wt % maximum at 100 kPa) are still quite low with respect to the DOE target. The hydrogen adsorption capacity of the AX-21 receptor was studied at high pressuresup to 10 MPa. Figure 6 shows the results of the experiments for a physical mixture and a bridged sample. The dashed line is a fit of the experimental data taken from Zhou and Zhou40 for (40) Zhou, L.; Zhou, Y. Chem. Eng. Sci. 1998, 53, 2531-2536.

11422

Langmuir, Vol. 21, No. 24, 2005

Lachawiec et al. Table 1. Spillover Enhancement for AX-21 and SWNT Receptors with the 5 wt % Pd-C Source at 298 K 100 kPa (1 atm) sample 5 wt % Pd-C SWNT SWNT/5 wt % Pd-C (9:1) SWNT/5 wt % Pd-C/bridge (8:1:1) AX-21 AX-21/5 wt % Pd-C (9:1) AX-21/5 wt % Pd-C/bridge (8:1:1)

Figure 6. High-pressure hydrogen isotherms at 298 K for the AX-21 receptor: ], AX-21/Pd-C/carbon bridge (8:1:1) first adsorption; [, AX-21/Pd-C/carbon bridge (8:1:1) desorption; lightly shaded tilted square, AX-21/Pd-C/carbon bridge (8:1:1) second adsorption; 4, AX-21/Pd-C physical mixture (9:1); 0, AX-21 (AX-21 data from Zhou and Zhou40).

unmodified AX-21. Since the measurement system is designed for a larger pressure range, the data in the lowpressure region are not as sensitive compared with those presented in Figure 4. As the pressure increases, the enhancement due to bridging becomes evident. Another interesting feature is that there is no apparent saturation value for the bridged sample, even at 10 MPa. The unmodified AX-21 data and physical mixture begin to demonstrate a change in curvature indicative of a saturation limit at pressures greater than approximately 6 MPa. The absolute capacity of the AX-21/Pd-C/carbon bridge (8:1:1) material is 1.8 wt % at 10 MPa. The absence of a saturation value suggests that, through optimization of the bridge-building process, there may be additional opportunities to increase the capacity toward the DOE target. Hydrogen was desorbed from the AX-21 bridged sample immediately following the first adsorption. As the data in Figure 6 indicate, there appears to be a hysteresis present. The sample was evacuated to a pressure of 1 Pa (7.5 × 10-3 Torr) for 8 h at 298 K prior to the second adsorption. Data for the second adsorption are shown by the shaded points and are coincident with data from the first adsorption. This indicates that hydrogen adsorption is reversible at 298 K. The reversibility indicates that the strong sites (i.e., sites that form strong bonds with the spillover atomic hydrogen) had already been saturated with hydrogen during the sample pretreatment process (523 K in hydrogen followed by 673 K in helium). It is useful to attempt to quantify the spillover enhancement due to the improved contact between the source and receptor caused by bridges. One method to evaluate spillover in a supported metal catalyst system is to determine the ratio of atomic hydrogen to surface metal atoms (H:MS) exposed to the adsorbate. A value exceeding unity indicates that hydrogen is spilling over to the support. The hydrogen titration method of determining catalyst dispersion,41 which allows computation of surface metal atoms, is difficult or impossible to apply to systems where spillover is occurring. The difficulty arises when an attempt is made to determine the monolayer coverage of hydrogen from a low-pressure chemisoption isotherm, (41) Benson, J. E.; Boudart, M. J. Catal. 1965, 4, 704-710.

BET SA, Q, m2/g cm3 (STP)/g

10 MPa (100 atm) η

Q, cm3 (STP)/g

η

66.7 92.2

1.1

5.0a

940 820

2880

1.8 2.6

1.1

3.8

1.8

1.3 2.6

1.6

4.2

2.9

201

2.7

a A total of 3.2 cm3 (STP)/g is due to adsorption of a stoichiometric amount of hydrogen by Pd (PdH0.6).

which may asymptotically approach a larger adsorbed amount or continue to rise as pressure is increased due to the spillover phenomenon. Spillover can be quantified in systems containing metals that form solid solutions with hydrogen, as the ratio of total adsorbed hydrogen (including that adsorbed by the metal) to total metal atoms (H:MT) exceeds the stoichiometric ratio of the hydride. Primary spillover in the 5 wt % Pd-C system was evaluated from the low-pressure hydrogen adsorption isotherm shown in Figure 4. For this material, the hydrogen adsorption at 100 kPa was 5.0 cm3 (STP)/g. This corresponds to an atomic hydrogen to total metal ratio of 0.95, which is greater than the maximum stoichiometric ratio of 0.77 reported in the literature.39 Primary spillover is occurring on this material. The enhancement factor method proposed by Lueking and Yang29 was used to compare secondary hydrogen spillover to the various receptors. In this case, the enhancement factor was defined for all receptors as

η)

QR QR′

(1)

where QR denotes the adsorbed amount for the receptor in the presence of a source and QR′ denotes the adsorbed amount for the receptor alone. To calculate QR, two assumptions were made. Hydrogen adsorption was assumed to be an equilibrium process in all samples, and the contributions to the composite material from the source and the receptor were assumed to be additive. Mathematically

QR ) QT - xQS

(2)

where QS is the adsorption amount for the source alone, including any amount adsorbed by a metal that forms a solid solution with hydrogen, and x is the mole fraction of the source in the mixture. By this convention, the bridges become part of the receptor and any adsorption on them contributes to an increase in the enhancement factor. While it may be possible for the bridges to adsorb some hydrogen, it is expected that this contribution is negligible compared to their role as pathways for the diffusion of hydrogen from the source to the receptor. It is possible that the adsorption capacity of the source could be altered by the presence of the bridges if significant amounts of micropores are filled with the carbonization product or if the metal particle is covered with carboniza-

Hydrogen Storage in Nanostructured Carbons

Langmuir, Vol. 21, No. 24, 2005 11423

Figure 7. SEM images at 30 kV, spot size 5.0: (a) Pd particle on carbon (commercial catalyst); (b) AX-21 receptor; (c) SWNT receptor, ∼20 nm tube bundles; (d) Pd particle in the AX-21/Pd-C/carbon bridge (8:1:1) sample, apparent bridging between the source and the receptor; (e) Pd particle on the Pd-C source in the AX-21/Pd-C/carbon bridge (8:1:1) sample, intimate contact with the support; (f) Pd particle in the AX-21/Pd-C (9:1) physical mixture, no bridging apparent; (g) Pd particle on the Pd-C source in the AX-21/Pd-C (9:1) physical mixture, no contact enhancement.

tion residue. Since the ratio of source to receptor is low, its overall contribution to the total adsorption amount is also small; therefore, altering the contribution of the source as a consequence of bridge formation is assumed to be inconsequential. If the metal particle is covered with excess

bridge material, it could reduce the surface area of metal available for dissociation. A mild oxidation treatment to remove the carbon residue could improve this situation; however, it would need to be carefully controlled to prevent bridge destruction.

11424

Langmuir, Vol. 21, No. 24, 2005

Enhancement factors for the physical mixtures and bridged samples are shown in Table 1. Note that the addition of the 5 wt % Pd-C source always results in an enhancement to spillover, irrespective of bridge formation. The AX-21 receptor was studied at 100 kPa and 10 MPa to determine the effect of pressure on the enhancement factor. The results for this microporous carbon and the SWNTs at low pressure indicate that the enhancement factor is nearly independent of the receptor for physically mixed samples. In the case of the bridged samples, the AX-21 receptor enhancement factor was a factor of 1.6 times larger than the SWNT receptor. This seems to indicate that atomic hydrogen is accessing more of the available surface area of the AX-21 receptor. The lower enhancement in spillover for SWNTs could indicate that the tube bundles are packed too closely for even atomic hydrogen to adsorb appreciably to the exohedral sites, which have been shown in simulations to be the more thermodynamically favorable positions compared to endohedral, or interior, sites.10 On the basis of the similarity in the adsorption amounts for the individual receptors alone and in enhancement factors for physically mixed samples, it appears that the most effective bridge-building technique may be specific to the receptor. Presumably, the bridging process could act to cap some of the tube ends or fill some of the defects, and thus, there might be limited access to the interior or endohedral sites. It is important to note that the conclusions regarding the enhancement factor are based on high-pressure data for the AX-21 receptor alone. Implicit is the assumption that the lowpressure trend for enhancement with an SWNT receptor continues over the entire pressure range; however, the nanostructure of the SWNT receptor may yield different results at high pressure, and it is worth confirming the assumption experimentally. This is planned as part of further investigations and optimizations of the bridgebuilding technique. SEM was used to investigate the samples for differences in the contacts between Pd particles and the primary support carbon and the receptor carbon. Figure 7a shows a Pd particle on the surface of the Pd-C commercial catalyst used as the secondary spillover source. The metal particle appears lighter compared to the carbon support. Parts b and c of Figure 7 are images of the receptors, AX-21 and SWNTs, respectively. Note the texture on the surface of the AX-21 particle, with porelike structures ranging from 100 to 250 nm in diameter. The SWNTs are arranged in bundles (wirelike structures ∼20 nm in diameter) in a highly disordered array. The AX-21/ Pd-C/carbon bridge sample is shown in Figure 7d,e. In Figure 7d, the bright area in the center of the frame is the Pd particle and appears to be contacting two distinct phases, presumably the AX-21 receptor and carbon support. The texture of the phase beneath the particle does not appear to match the support or AX-21; thus, this might represent the carbonized precursor that has formed the bridge. Figure 7e shows a Pd particle in contact with the commercial support. The contact between this particle and the support appears to be quite intimate and seems improved relative to the contact in (a). As previously mentioned, the precursor may also have filled the interstices between the support and the Pd particle and contributed to the apparent enhancement in contact shown here. The physical mixture of AX-21/Pd-C (9:1) is shown in Figure 7f,g. A 760 nm Pd particle is present on the support; however, this support particle does not appear to have good contact with the surrounding carbon, particularly the phase that looks similar to AX-21 on the

Lachawiec et al.

left-hand side of the image. In Figure 7g, a Pd particle is shown on the surface of the support carbon with noticeably less intimate contact compared to that in (e). While the evidence of bridging and contact enhancement in these images is not conclusive, it seems to support the trends observed for bridge building and the theory postulated for its effectiveness. It is important to note that neither the bridge-building procedure nor the pretreatment of the composite materials has been systematically studied for optimization. There may be improvements to the precursor or preparation procedure that are more effective for a specific receptor. For example, the wetting characteristics of the precursor and the substrates may play a role in the efficiency of the bridging. The pretreatment technique after bridging is another area for investigation. The mild oxidation step proposed to expose more metal surface or remove SWNT caps created by precursor carbonization may also add functional groups to the surfaces, which have been shown to participate in the spillover process42 and influence the physisorption of hydrogen.17,43 The groups from this step would aggregate with the functionalization created by the oxidative environment used for SWNT purification. An alternative to oxidation for opening SWNT ends, ultrasonication of carbon, has enhanced spilloverspresumably by inducing defects on the surface.44 If the process can be tuned to produce results equivalent to those of oxidation, it would be preferable so there is no risk of burning away the bridges during pretreatment. Further research on various aspects of bridge building is ongoing. Using this technique, enhancements in storage of up to 17-fold on other carbon-based materials have been achieved and will be reported elsewhere shortly.45 Conclusions The results from this work have established the importance of bridge building for enhancing hydrogen spillover and increasing storage capacity. While physical mixtures of a primary spillover source and a secondary receptor demonstrate modest capacity increases, adding a bridge to improve the contact between the two components serves to double or triple the capacity. The highest measured capacity was 1.8 wt % at 298 K and 10 MPa for an AX-21/Pd-C/carbon bridge (8:1:1) composite. The results show that pressure has a negligible effect on the enhancement factor up to 10 MPa and the bridge-building technique appears to be receptor-specific. While this work was not intended to study the mechanistic details of spillover enhancement by bridges, the removal of diffusion barriers from a source of hydrogen atoms to a high-capacity receptor is postulated and seems a plausible theory. Acknowledgment. We acknowledge the support of the U.S. Department of Energy (lead: National Renewable Energy Laboratory) for this work under Contract DE FC36 05 GO15078. LA051659R (42) Cheng, Z. X.; Yuan, S. B.; Fan, J. W.; Zhu, Q. M.; Zhen, M. S. In Spillover and Migration of Surface Species on Catalysts, Proceedings of the 4th International Conference on Spillover; Li, C., Xin, Q., Eds.; Elsevier: Amsterdam, 1997; pp 261-266. (43) Poirier, E.; Chahine, R.; Be´nard, P.; Cossement, D.; Lafi, L.; Me´lanc¸ on, E.; Bose, T. K.; De´silets, S. Appl. Phys. A 2004, 78, 961-967. (44) Badenes, P.; Daza, L.; Rodriguez-Ramos, I.; Guerrero-Ruiz, A. In Spillover and Migration of Surface Species on Catalysts, Proceedings of the 4th International Conference on Spillover; Li, C., Xin, Q., Eds.; Elsevier: Amsterdam, 1997; pp 241-250. (45) Yang, R. T.; Qi, G. S.; Lachawiec, A. J., Jr.; Li, Y. W. U.S. Patent Application, 2005.