When Cocatalyzed with Carbon Nanostructures - American Chemical

SWNTs were the best cocatalyst, G was the worst cocatalyst, and all five carbons were inactive as a catalyst unless Ti was present. The carbon most li...
0 downloads 0 Views 178KB Size
Download PDF
J. Phys. Chem. B 2006, 110, 17353-17358

17353

Kinetic Behavior of Ti-Doped NaAlH4 When Cocatalyzed with Carbon Nanostructures Jun Wang, Armin D. Ebner, and James A. Ritter* Department of Chemical Engineering, Swearingen Engineering Center, UniVersity of South Carolina, Columbia, South Carolina 29208 ReceiVed: April 26, 2006; In Final Form: June 21, 2006

The effects of SWNTs, MWNTs, AC, C60, and G when used as a cocatalyst with Ti on the dehydrogenation and hydrogenation kinetics of NaAlH4 were investigated for the first time in the important temperature range of 90 to 250 °C. All five carbons exhibited significant, sustaining, and synergistic cocatalytic effects on the dehydrogenation and hydrogenation kinetics of Ti-doped NaAlH4 that persisted through charge and discharge cycling. SWNTs were the best cocatalyst, G was the worst cocatalyst, and all five carbons were inactive as a catalyst unless Ti was present. The carbon most likely was imparting an electronic contribution through the interaction of its facile π-electrons with Ti through a hydrogen spillover mechanism, which explained why one carbon was better than another one in terms of optimal aromatic character, out-of-plane exposure of π-electrons, and interaction of π-bonds with neighboring sheets.

Introduction co-workers1

Ritter and recently reported on the very positive, sustaining, and synergistic effect of graphite as a cocatalyst with Ti on both the dehydrogenation and hydrogenation kinetics of NaAlH4 over a wide range of temperatures from 90 to 250 °C. They demonstrated that the codoping of 2 mol % Ti-doped NaAlH4 with 10 wt % graphite improved the kinetics of both the first and second hydrogen evolution reactions, Ti

3NaAlH4 798 Na3AlH6 + 2Al + 3H2 Ti

2Na3AlH6 798 6NaH + 2Al + 3H2

(1) (2)

by lowering the dehydrogenation temperature over the important range of 90 to 150 °C by as much as 15 °C compared to that without graphite. Similarly, the addition of graphite was shown to improve the dehydrogenation kinetics at 90 and 110 °C, respectively, by factors of 6.5 and 3.0 compared to a sample doped only with Ti. They also showed that the significant effects of graphite (as a codopant) persisted through dehydrogenation/ hydrogenation cycling, and through the addition of aluminum powder, which essentially mitigated the irreversible losses associated with cycling.2 Most importantly, the effect of graphite on the kinetics was shown to be just as pronounced and as sustaining during several hydrogenation cycles as it was during several dehydrogenation cycles. Ritter and co-workers1 also found graphite alone to be essentially inactive as a catalyst for NaAlH4 dehydrogenation, contradicting the results reported in the literature.3 Zaluska et al.3 showed that carbon improved both the dehydrogenation and hydrogenation properties of NaAlH4 when used as a lone dopant with NaAlH4. For example, a sample of NaAlH4 doped with 10 wt % carbon and then ball milled for 2 h attained a hydrogenation state of about 3.0 wt % in 2 h (i.e., ∼1.5 wt % H2/h) at 130 °C and 88 atm, and a dehydrogenation state of about 2.3 wt % in 20 h (i.e., ∼0.12 wt % H2/h) at 90 °C and 1 * To whom all correspondence should be addressed. Phone: (803) 7773590. Fax: (803) 777-8265. E-mail: [email protected].

atm. They showed in some cases and claimed in other cases that these rates were higher than just ball milled samples, and that they compared favorably to and in some cases were even higher than the rates reported for metal catalyzed and ball milled samples of NaAlH4. This claim on the dehydrogenation rate at 90 °C was in excellent agreement with that reported by Wang et al.1 (∼0.1 wt % H2/h) for a sample of NaAlH4 doped with 2 mol % Ti and ball milled for 2 h. The hydrogenation rate of about 1.5 wt % H2/h was about half that observed by Wang et al.1 (3.5 wt % H2/h) for a sample of NaAlH4 doped with 2 mol % Ti, ball milled for 2 h, and charged at 85 atm and 125 °C. These interesting and somewhat contradictory results reported by Zaluska et al.3 and Wang et al.1 warranted further study of the effects of carbon structures as catalysts or cocatalysts on the reactivity of NaAlH4. To this end, Dehouche et al.4 recently reported on the catalytic effects of various forms of carbon nanostructures when they were used as a codopant with Ti/Zr-doped NaAlH4. When they doped NaAlH4 with 1 mol % Ti, 1 mol % Zr, and 2 wt % of single-walled carbon nanotubes (SWNTs), graphite, or AX-21 activated carbon, and then ball milled, the dehydrogenation kinetics improved dramatically. The sample codoped with SWNTs exhibited the best performance, with the samples codoped with graphite and AX-21 activated carbon both exhibiting similar behaviors, but less pronounced enhancements compared to the sample codoped with SWNTs. The effect persisted even when subjecting the sample codoped with SWNTs to over 200 dehydrogenation/hydrogenation cycles between 3.7 and 514 psia at 160 °C. However, since their study was carried out exclusively at 160 °C, they fully realized that the effect of cycling was limited to that associated with the reversibility of only the second hydrogen evolution reaction depicted in eq 2. At 160 °C, they would have had to cycle the sample at around 1200 psia to have any effect on the first hydrogen evolution reaction depicted in eq 1.5 They also claimed that the use of a carbon nanostructure as a codopant would also favorably impact the dehydrogenation and hydrogenation kinetics of the first reaction. Although they had no way to substantiate this supposition, Ritter and co-workers,1 at around the same time, showed this to be true, at least for graphite.

10.1021/jp062574h CCC: $33.50 © 2006 American Chemical Society Published on Web 08/11/2006

17354 J. Phys. Chem. B, Vol. 110, No. 35, 2006

Wang et al.

To further clarify what is known and unknown about the catalytic or cocatalytic effects of carbon nanostructures on the reversible kinetics of the first and second dehydrogenation reactions depicted in eqs 1 and 2, the objective of this article is to report on the effectiveness of five different carbon nanostructures when used as a codopant with Ti-doped NaAlH4 and studied in the important temperature range of 90 to 250 °C. The dehydrogenation and hydrogenation kinetics of NaAlH4 uncatalyzed, catalyzed only with TiCl3, SWNTs, multiwalled carbon nanotubes (MWNTs), C60 fullerene, activated carbon (AC), and graphite (G), and cocatalyzed with Ti and individually with each of these different carbon nanostructures are compared. Results are reported in terms of temperature programmed desorption (TPD), constant temperature desorption (CTD), and constant temperature cycling (CTC) curves obtained with freshly doped and ball milled samples and samples cycled five times. Experimental Section TiCl3 (Aldrich, 99.99%, anhydrous), aluminum powder (Alfa Aesar, 99.97%), SWNTs (Helix Material Solution Inc.), MWNTs (Seldon Laboratories, LLC), AC (MeadWestvaco), C60 (Alfa Aesar, 99.9%), and SFG 75 graphite powder (G) (TIMREX) were used as received. NaAlH4 powder (Fluka, 99.5%) was recrystallized from a 3 M THF (Aldrich, 99.9%, anhydrous) solution, filtered through 0.7 µm filter paper, and vacuum dried. In 10 mL of THF, 1.0 g of the purified NaAlH4 was mixed with the catalyst precursor (TiCl3) to produce a doped sample containing 2 mol % metal (a wet doping procedure). The THF was evaporated while the NaAlH4 and the catalyst were manually mixed with a mortar and pestle for about 30 min, or until the sample appeared dry. The sample was then ball milled for 2 h with a SPEX 8000 high-energy ball mill loaded with a 65 cm3 SS vial containing 0.5 to 1.0 g of powder and a single SS ball (8.2 g) with a diameter of 1.3 cm. After ball milling, carbon and aluminum powder were added directly to the doped and ball milled sample (a dry doping procedure), and ball milling was continued for an additional 2 h. The moles of Ti added to each sample was based on the Na or Al content in NaAlH4; the grams of carbon or Al added to each sample was based on the total mass of the sample after being completely doped (i.e., including the NaAlH4, TiCl3, and carbon and/or Al). All sample handling procedures were performed in a nitrogen glovebox. Thermogravimetric analysis was carried out with a PerkinElmer, TGA 7 Series thermogravimetric analyzer (TGA). The dehydrogenation rates of various doped and ball milled samples of NaAlH4 were measured at atmospheric pressure in flowing helium (∼60 cm3/min) in TPD and CTD modes. For TPD runs, the samples were heated to 250 °C at a ramping rate of 5 deg/ min after purging with He (National Welders, UHP Grade, 99.999%) for 1 min. For CTD runs, a similar procedure was followed except that the samples were heated rapidly to the desired temperature and then held at that temperature for the desired time. Approximately 10 mg of sample were used in each TPD or CTD run. CTC experiments were carried out in a 3000 psi Parr reactor, installed in a fully automated and instrumented pressure and temperature cycling system, to evaluate the dehydrogenation and hydrogenation kinetics and cycling capability of the doped NaAlH4. The cycling procedure consisted of the following steps. While in a N2 glovebox, a doped and ball milled sample of NaAlH4 was loaded into the reactor vessel. The reactor was then sealed and removed from the glovebox and connected to the cycling system. A typical dehydrogenation/hydrogenation cycling experiment was carried out for five cycles in hydrogen

Figure 1. TPD curves at 5 °C min-1 for just doped or undoped and ball milled (0th cycle) samples of NaAlH4 containing (a) 2 mol % Ti and 5 wt % Al, 0 mol % Ti and 5 wt % Al, 0 mol % Ti, 5 wt % Al and 10 wt % carbon (SWNT, AC, MWNT, C60 or G) and (b) 2 mol % Ti and 5 wt % Al, 2 mol % Ti, 5 wt % Al and 10 wt % carbon (SWNT, AC, MWNT, C60 or G).

(National Welders, UHP Grade, 99.999%) by first heating the reactor to 125 °C and carrying out a rudimentary dehydrogenation step at this temperature for 90 min at around 15 psig. This step is referred to as the 0th cycle discharge because it was carried out under varying temperature and pressure conditions. This 0th cycle discharge step was then followed by the 1st constant temperature and pressure hydrogenation step that was carried out for 60 min at 1250 psig and the 1st constant temperature and pressure discharge step that was carried out for 90 min at 15 psig. After the 5th and final hydrogenation half-cycle was completed in that cyclic manner, the reactor was cooled to room temperature, vented to the atmosphere, and transferred to the glovebox to remove the charged sample. This sample was used to carry out TPD and CTD tests as the 5th and final dehydrogenation half-cycle. The ash content of each carbon material was determined with TGA. Approximately 10 mg of carbon was placed in the sample pan, heated to 300 °C at a ramping rate of 10 deg/min, and kept at 300 °C for 1 h. It was then heated to 800 °C at the same ramping rate and kept there for 4 h. The material remaining in the sample pan after that treatment was considered to be the ash content of the carbon. Results and Discussion Figure 1 revealed the effects of Ti and the carbon nanostructures when used separately as single catalysts and together as

Ti-Doped NaAlH4 Cocatalyzed with Carbon Nanostructures cocatalysts on the dehydrogenation kinetics of NaAlH4 in terms of TPD curves. Figure 1a displays TPD curves for doped and ball milled samples of NaAlH4 containing 2 mol % Ti and 5 wt % Al, 0 mol % Ti and 5 wt % Al, and 0 mol % Ti, 5 wt % Al, and 10 wt % carbon, with the carbon being SWNT, AC, MWNT, C60 or G. Figure 1b displays TPD curves for doped and ball milled samples of NaAlH4 containing 2 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % carbon, with the carbon being SWNT, AC, MWNT, C60 or G. From the results in Figure 1a it was clear that none of the carbons, except for C60, exhibited any catalytic effect on the dehydrogenation of NaAlH4. A significant catalytic effect in a TPD curve usually appears as a much lower dehydrogenation temperature, with two distinct steps that are indicative of the first (eq 1) and second (eq 2) reactions taking place at lower and higher temperatures, respectively.1,7 Only the TPD curve for the NaAlH4 containing 2 mol % Ti and 5 wt % Al exhibited this distinctive behavior. In fact, except for C60, only marginal differences were observed between the NaAlH4 sample doped with only 5 wt % Al compared to all those doped with 10 wt % carbon and 5 wt % Al but no Ti. Although C60 exhibited more catalytic activity than the other four carbons, it was still relatively insignificant compared to Ti. These minimal effects of the carbons on the dehydrogenation kinetics of NaAlH4 continued to challenge the effects of carbon reported by Zaluska et al.3 In fact, these results showed very clearly that all five carbons were essentially inactive as a catalyst for NaAlH4. Since most transition metals have a catalytic effect on the dehydrogenation rate of NaAlH4,6 and since a very small amount of Ti, e.g., as low as 0.02 mol %, improves the dehydrogenation rate of NaAlH4,7-9 it was speculated that the slight improvements noticed in the TPD curves shown in Figure 1a might have been due to residual metal impurities in the carbons. To test this hypothesis, the ash content of each carbon was measured. For AC, MWNT, SWNT, C60, and G the ash contents were 7, 4, 2, 0, and 0 wt %, respectively. With C60 and G in Figure 1a exhibiting the best and worst catalytic effects on the dehydrogenation of NaAlH4, with neither of them containing any residual metals, and with the other threes carbons (AC, MWNT, SWNT) having some residual metals while not being substantially more catalytically active than G, no definitive correlation was found between the residual metals in the carbon and the slight catalytic activity. The results in Figure 1b showed that the addition of 10 wt % carbon to the NaAlH4 samples already doped and ball milled with 2 mol % Ti produced a striking effect on the dehydrogenation kinetics of both the first and second hydrogen evolution reactions, no matter the form of the carbon. When Ti was used as a single dopant, hydrogen was released at a significantly lower temperature (∼130 °C), and upon codoping with the various carbons, this temperature was lowered even further by about 20 to 30 °C depending on the carbon. The sample codoped with 10 wt % SWNTs exhibited the best performance, followed by the sample codoped with 10 wt % MWNTs or AC. The sample codoped with 10 wt % C60 or G exhibited the least improvement in the performance. These results clearly showed the synergistic effects associated with the use of essentially any carbon material as a cocatalyst with Ti. Figure 2 revealed the effects of the carbon nanostructures when used as a cocatalyst with Ti on the dehydrogenation kinetics of NaAlH4 before and after cycling in terms of TPD curves. Figure 2a displays TPD curves for doped and ball milled samples of NaAlH4 containing 2 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % SWNT before (0th cycle)

J. Phys. Chem. B, Vol. 110, No. 35, 2006 17355

Figure 2. TPD curves at 5 °C min-1 for doped and ball milled samples of NaAlH4 containing (a) 2 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % SWNT before (0th cycle) and after (5th cycle) five hydrogenation and dehydrogenation cycles and (b) containing 2 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % carbon (SWNT, AC, MWNT, C60 or G) after (5th cycle) five hydrogenation and dehydrogenation cycles.

and after (5th cycle) cycling. Figure 2b displays TPD curves for doped and ball milled samples of NaAlH4 containing 2 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % carbon, with the carbon being SWNT, AC, MWNT, C60 or G, after (5th cycle) cycling. The results in Figure 2a showed that in the temperature region associated with the first hydrogen evolution reaction the sample codoped with SWNTs and Ti exhibited essentially no loss in kinetics and only a slight loss in capacity with cycling. In contrast, in the same region of the TPD curve the sample doped only with Ti exhibited losses in both kinetics and capacity. In both cases, the losses with cycling would have been much greater if Al powder was not added as a dopant.1,2 The other four samples codoped with carbon and Ti cycled similarly to the sample codoped with Ti and SWNTs. In fact, the results in Figure 2b showed that all the samples codoped with carbon and Ti performed the same after cycling except for the SWNTs, which exhibited superior performance in the temperature region associated with the first hydrogen evolution reaction. The sample codoped with SWNTs and Ti exhibited the best synergistic effect on the dehydrogenation kinetics for both the 0th and 5th discharge, by lowering the temperature for dehydrogenation by about 30 °C compared to the sample doped only with Ti. In the worst case, G still lowered the dehydrogenation temperature by about 20 °C compared to the sample doped only with Ti.

17356 J. Phys. Chem. B, Vol. 110, No. 35, 2006 These results clearly showed that the very favorable and synergistic effects of carbon when used as a cocatalyst with Ti persisted even after cycling. In both cases it was clear, however, that those hydrogen capacity losses were due to irreversibilities associated with the first hydrogen evolution reaction, not the second hydrogen evolution reaction. In fact, not much could be stated about the reversibility of the second hydrogen evolution reaction since the dehydrogenation temperature during cycling never exceeded 125 °C. This temperature simply was not high enough to cause significant decomposition of Na3AlH6 according to the second reaction depicted in eq 2, as inferred from the TPD curves in Figure 2 and the CTD curves discussed below. Nevertheless, Dehouche et al.4 showed the synergistic effect of a carbon nanostructure as a cocatalyst with Ti/Zr exclusively on the second hydrogen evolution reaction. Figure 3 revealed the effects of the carbon nanostructures when used as a cocatalyst with Ti on the dehydrogenation kinetics of NaAlH4 before and after cycling in terms of CTD curves. Panels a, b, and c of Figure 3 respectively display the CTD curves obtained at 87, 106, and 124 °C before (0th cycle) and after (5th cycle) cycling for doped and ball milled samples of NaAlH4 containing 2 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % carbon, with the carbon being SWNT, AC, MWNT, C60 or G. These CTD curves clearly delineated the differences between the five carbon nanostructures. The results in Figure 3a showed that at 87 °C the dehydrogenation kinetics of every sample codoped with carbon and Ti increased with cycling, while those for every sample doped only with Ti decreased with cycling. After five cycles, the increase in the dehydrogenation kinetics for the sample codoped with SWNTs was about 30%, with all the other codoped samples exhibiting about 20% increases. However, the results in Figure 3b,c showed that at the two higher temperatures (106 and 124 °C) all the samples exhibited decreases in the discharge kinetics with cycling, but much less so for all the samples codoped with carbon and Ti compared to those doped only with Ti. This was especially true at 124 °C, where significant losses in the discharge kinetics occurred for the sample doped only with Ti. The results in Figure 3 also showed that at all three temperatures and both before and after cycling, the sample codoped with SWNTs and Ti always exhibited the best discharge kinetics. In contrast, the sample codoped with G and Ti always exhibited the least synergistic effect. The samples codoped with the other three carbons were somewhere between these two carbons, depending on the temperature. However, compared to the sample doped only with Ti, the synergistic effects associated with all five carbons were still very significant. The CTD curves in Figure 3, in some cases, also exhibited a very distinct two-slope behavior that was associated with hydrogen release from the first (eq 1) and second (eq 2) dehydrogenation reactions.1,7 The steep slope at short times was primarily due to the first reaction, which occurred at lower temperatures with faster kinetics according to the TPD curves in Figure 2. The gentle slope at long times was primarily due to the second reaction, which occurred at higher temperatures with slower kinetics according to the TPD curves in Figure 2. Hence, at 87 °C the hydrogen released from all the samples was most likely only associated with the first hydrogen evolution reaction (one steep slope). At 106 °C, the hydrogen released from all the samples, except for the Ti-doped sample, was clearly associated with both reactions (two slopes, one steep and one gentle), but more so for the sample codoped with SWNTs. The

Wang et al.

Figure 3. CTD curves at (a) 87, (b) 106, and (c) 124 °C for doped and ball milled samples of NaAlH4 containing 2 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % carbon (SWNT, AC, MWNT, C60 or G) before (0th cycle) and after (5th cycle) five hydrogenation and dehydrogenation cycles.

sample codoped with SWNTs and Ti exhibited similar slopes both before after cycling at about 30 min into the discharge process that were indicative of the slow discharge kinetics associated with the second hydrogen evolution reaction. Recall that cycling should not have any effect on the kinetics of the second hydrogen evolution reaction because the cycling temperature was too low. At 124 °C, the hydrogen released from all the samples before cycling was again clearly associated with both the first and second hydrogen evolution reactions (two slopes, one steep and one gentle). Notice that before cycling all the samples exhibited similar slopes associated with the

Ti-Doped NaAlH4 Cocatalyzed with Carbon Nanostructures second reaction after about 20 min into the discharge process. After cycling, those slopes did not change much, except for the sample doped only with Ti. In that case, the discharge kinetics decreased so much after cycling that the second reaction probably did not contribute much at all to the CTD curve. These results again clearly showed the superior performance of samples codoped with carbon and Ti, no matter the form of the carbon. Figure 4 revealed the effects of the carbon nanostructures when used as a cocatalyst with Ti on the hydrogenation (charge) and dehydrogenation (discharge) kinetics of NaAlH4 during cycling in terms of CTC curves. Figure 4a displays CTC curves during five charge (Po ) 1250 psig) and four discharge (Po ) 15 psig) cycles at 125 °C for doped and ball milled samples of NaAlH4 containing 2.0 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % SWNT. Figure 4b displays CTC curves after the fourth charge and discharge cycle for doped and ball milled samples of NaAlH4 containing 2.0 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % carbon, with the carbon being SWNT, AC, MWNT, C60 or G. The qualitative kinetics of charge and discharge were inferred from the observed linear decreases and increases in pressure over time. The results in Figure 4a showed the distinct influence of the SWNTs when used as a cocatalyst with Ti not only on the discharge kinetics, but also on the charge kinetics. Upon codoping with SWNTs, the time for charging was markedly decreased by a factor of 8, from about 60 min to about 7.5 min. The time for discharge also decreased significantly. The sample codoped with SWNTs and Ti also cycled more consistently after a few cycles than the sample doped only with Ti. These results might represent the best charge kinetics to date at 125 °C for a sample of NaAlH4 doped with as little as 2 mol % Ti, but also with 10 wt % carbon and 5 wt % Al. The results in Figure 4b showed that the sample codoped with SWNTs and Ti once again exhibited the best synergistic effects among the five different carbon nanostructures; however, the effect was much more pronounced during discharge than charge. After cycling the trends were as follows: the sample codoped with SWNTs and Ti always exhibited the best charge and discharge kinetics, the sample codoped with G and Ti always exhibited the least synergistic effect, and the samples codoped with the other three carbons were somewhere between these two carbons, depending on whether it was charging or discharging. These results showed that compared to the sample doped only with Ti, the synergistic effects associated with all five carbons were still exceedingly significant, with SWNTs clearly exhibiting the most cocatalytic activity with Ti. In the previous work by the authors,1 which dealt only with graphite, the observed phenomena were interpreted in terms of some of the unique properties of carbon in general. The carbon might be playing a dual role by serving as a mixing agent manifested through lubrication phenomena (i.e., graphene layer slippage and breakage),10,11 and as a microgrinding agent manifested through the formation of carbide species,12,13 both during high energy ball milling. Carbon might also be imparting an electronic contribution through the interaction of its facile π-electrons with Ti through a hydrogen spillover mechanism.14-16 Although the synergistic effects exhibited by all five carbon nanostructures could be understood and explained in terms of those arguments, they did not explain why one carbon structure was better than another one. The different behavior of the carbon nanostructures, especially that associated with the SWNTs,17-19 which was superior in every case, was understood in terms of their uniquely different

J. Phys. Chem. B, Vol. 110, No. 35, 2006 17357

Figure 4. CTC curves showing qualitative hydrogenation and dehydrogenation rates: a) during five charge (Po ) 1250 psig) and four discharge (Po ) 15 psig) cycles at 125 °C for doped and ball milled samples of NaAlH4 containing 2.0 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % SWNT. The solid symbols correspond to samples containing carbon; the open symbols correspond to samples not containing carbon. (b) CTC curves after the 4th charge and discharge cycle for doped and ball milled samples of NaAlH4 containing 2.0 mol % Ti and 5 wt % Al, and 2 mol % Ti, 5 wt % Al, and 10 wt % carbon (SWNT, AC, MWNT, C60 or G).

π-bonding structures. Graphite has an ideal planar structure. Activated carbon is similar to graphite in that it is comprised of randomly oriented nanosized crystallites of graphite. A SWNT has a tubular structure comprised of a rolled up sheet of graphite. A MWNT also has a tubular structure comprised of several layers of rolled up sheets of graphite. C60 has a spherical structure that is also graphitic. Each of these carbon nanostructures has a different degree of aromatic character due to its unique π-bonding structure, a different degree of out-ofplane exposure of its π-bonds due to its unique geometric curvature, and a different degree of π-bond interaction with neighboring graphene sheets (if present). It is also well understood that a carbon nanostructure with more aromatic character, more out-of-plane exposure of π-electrons, and less interaction of π-bonds with neighboring sheets is more prone to interaction and reaction.17-19 With respect to aromatic character, in a simplified picture of C60 the carbon atoms are in a tensed sp2 state, corresponding to a 3-fold but nonideal planar coordination.18,19 Within its 6-fold rings, there are also two different C-C bond lengths, which impart an asymmetric strain within the structure, which in turn diminishes its aromatic character.18 In contrast, although the carbon atoms in a SWNT are also in a tensed sp2 state,

17358 J. Phys. Chem. B, Vol. 110, No. 35, 2006 corresponding to a 3-fold but nonideal planar coordination, there is only one unique C-C bond length within its 6-fold rings of the tubular structure. This gives rise to a more symmetric strain within its structure, which in turn enhances its aromatic character.17,18 A MWNT has a similar molecular structure to a SWNT, except that the strain becomes increasingly more relaxed with the number of layers making it more and more like graphite. Graphite and activated carbon also have similar molecular structures to MWNTs, except that there is probably no or very little strain, due to them being layered, planar structures. Perhaps MWNTs, AC, and G are slightly less aromatic than SWNTs, and C60 is the least aromatic. With respect to out-of-plane exposure of π-electrons, C60 clearly has the most curvature, but due to the asymmetric strain within its structure the π-bonds are restricted and exhibit more polyenic than aromatic character.18 A SWNT, on the other hand, also has significant curvature, which gives rise to significant out-of-plane exposure of its π-electrons. Although a significant number of π-electrons are present in graphite due to its lack of curvature, most of its π-interactions are restricted to taking place in plane, which limits its out-of-plane exposure of π-electrons. A MWNT may also suffer from this restricted planar π-bonding arrangement due to its diminished curvature at least compared to a SWNT. Activated carbon may have a little more out-ofplane exposure of π-electrons simply because of it being comprised of randomly oriented nanosized crystallites of graphite. With respect to the interaction of π-bonds with neighboring sheets, graphite, activated carbon, and MWNTs are all very similar, as they all have distinct layered structures of grapheme sheets. This layering diminishes their potential for π-bond interactions. Although C60 does not have any nearest neighbor layers for π-electron interactions, its lack of aromatic character restricts its π-electron interactions. SWNTs also do not have any nearest neighbor layers for π-electron interactions which enhances their potential for π-bond interactions. On the basis of these dissimilar features of the different carbon nanostructures, it was surmised that the SWNTs were the best cocatalyst with Ti because it had the most aromatic character, most out-of-plane exposure of π-electrons, and least interaction of π-bonds with neighboring sheets. In other words, a SWNT had optimal characteristics that made its laterally extensive π-electron network more reactive with Ti, perhaps facilitated through nonplaner interaction and the hydrogen spillover mechanism. In contrast, graphite provided the least synergistic behavior as a codopant conceivably because it had slightly less aromatic character, limited out-of-plane exposure of π-electrons, and much more interaction of π-bonds with neighboring sheets. In other words, graphite had characteristics that were less than optimal, which made its π-electrons least reactive with Ti through a hydrogen spillover mechanism. The other three carbon nanostructures with aromatic character, out-of-plane exposure of π-electrons, and/or interaction of π-bonds with neighboring sheets somewhere between those of SWNTs and graphite not surprisingly exhibited synergistic behavior somewhere between that of SWNTs and graphite. In other words, they all had characteristics that gave rise to an intermediate reactivity of their π-electrons with Ti most likely through a hydrogen spillover mechanism. Conclusions In conclusion, the effects of SWNTs, MWNTs, AC, C60, and G when used as a cocatalyst with Ti on the dehydrogenation and hydrogenation kinetics of NaAlH4 were investigated in the important temperature range of 90 to 250 °C. According to temperature programmed desorption, constant temperature de-

Wang et al. sorption, and constant temperature cycling curves, all five carbon nanostructures exhibited significant, sustaining, and synergistic effects on the dehydrogenation and hydrogenation kinetics of Ti-doped NaAlH4. The synergistic effects of all five carbons also persisted through dehydrogenation and hydrogenation cycling. SWNTs were the best cocatalyst, with a sample of NaAlH4 codoped with 2.0 mol % TiCl3, 10 wt % SWNTs, and 5 wt % Al exhibitng perhaps the best dehydrogenation and hydrogenation rates to date. Although G was the worst cocatalyst, the other three carbons exhibited very similar results to G but with their effectiveness being slightly better. However, all five carbons were essentially inactive as a catalyst unless Ti was present. Although C60 exhibited some catalytic activity as a lone catalyst, it was no better than the other carbons when used as a cocatalyst with Ti. The observed phenomena were interpreted in terms of some of the unique properties of carbon in general, and in terms of some of the specific properties associated with each of the carbon nanostructures. Carbon might be playing a dual role by serving as a mixing agent manifested through lubrication phenomena (i.e., graphene layer slippage and breakage), and as a microgrinding agent manifested through the formation of carbide species, both during high energy ball milling. Carbon might also be imparting an electronic contribution through the interaction of its facile π-electrons with Ti through a hydrogen spillover mechanism. The latter notion was used to explain why SWNTs were the best cocatalyst and graphite was the worst cocatalyst, with the synergistic effects of the other carbons lying somewhere in between. It was surmised that the SWNTs had considerable aromatic character, significant out-of-plane exposure of π-electrons, and no interaction of π-bonds with neighboring sheets that led to optimal reactivity of its π-electrons with Ti through a hydrogen spillover mechanism. Acknowledgment. Financial support provided by the U.S. DOE under grant DEFG02-91-ER45439 was greatly appreciated. References and Notes (1) Wang, J.; Ebner, A. D.; Prozorov, T.; Zidan, R.; Ritter, J. A. J. Alloys Compd. 2005, 395, 252-262. (2) Bogdanovic, B.; Felderhoff, M.; Germann, M.; Hartel, M.; Pommerin, A.; Schuth, F.; Weidenthaler, C.; Zibrowius, B. J. Alloys Compd. 2003, 350, 246-255. (3) Zaluska, A.; Zaluski, L.; Stro¨m-Olsen, J. O. J. Alloys Compd. 2000, 298, 125-134. (4) Dehouche, Z.; Lafi, L.; Grimard, N.; Goyette, J.; Chahine, R. Nanotechnology 2005, 16, 402-409. (5) Bogdanovic, B.; Brand, R. A.; Marjanovic, A.; Schwickardi, M.; Tolle, J. J. Alloys Compd. 2000, 302, 36-58. (6) Anton, D. L. J. Alloys Compd. 2003, 356-357, 400-404. (7) Sandrock, G.; Gross, K. J.; Thomas, G. J. Alloys Compd. 2002, 339, 299-308. (8) Bogdanovic, B.; Schwickardi, M. J. Alloys Compd. 1997, 253, 1-9. (9) Weidenthaler, C.; Pommerin, A.; Felderhoff, M.; Bogdanovic, B.; Schueth, F. Phys. Chem. Chem. Phys. 2003, 5, 5149-5153. (10) Shaji, S.; Radhakrishnan, V. J. Mater. Process. Technol. 2003, 141, 51-59. (11) Shaji, S.; Radhakrishnan, V. Mach. Sci. Technol. 2003, 7, 137155. (12) Chang, Y.-H.; Chiu, C.-W.; Chen, Y.-C.; Wu, C.-C.; Tsai, C.-P.; Wang, J.-L.; Chiu, H.-T. J. Mater. Chem. 2002, 12, 2189-2191. (13) Li, J.; Li, F.; Hu, K. J. Am. Ceram. Soc. 2002, 85, 2843-2845. (14) Lueking, A. D.; Yang, R. T. Appl. Catal. A 2004, 265, 259-268. (15) Baumgarten, E.; Maschke, L. Appl. Catal. A 2000, 265, 171-177. (16) Yang, F. H.; Yang, R. T. Carbon 2002, 40, 437-444. (17) Dresselhaus, M. S.; Dresselhaus, G.; Eklund, P. C. Science of Fullerenes and Carbon Nanotubes; Academic Press: New York, 1996. (18) Seifert, G. Solid State Ionics 2004, 168, 265-269. (19) Harris, P. Carbon Nanotubes and Related Structures: New Materials for the Twenty-First Century; Cambridge University Press: Cambridge, U.K., 2001.