Dan Zhao, Daqiang Yuan and Hong-Cai Zhou*
Department of Chemistry and Biochemistry, Miami University, Oxford, Ohio 45056, USA. E-mail: zhouh@muohio.edu; Fax: +1 513 529-0452; Tel: +1 513 529-8091
First published on 24th June 2008
The theoretical and experimental hydrogen storage studies on metal–organic frameworks (MOFs) have been reviewed. Seven distinct factors influencing hydrogen uptake capacity in MOFs have been classified and discussed. Based on existing studies, some possible future developments have been proposed.
Dan Zhao |
Daqiang Yuan |
Hong-Cai Zhou | Hongcai “Joe” Zhou obtained his BSc degree from Beijing Normal University and PhD from Texas A&M University. After a postdoctoral stint at Harvard, he joined the faculty of Miami University, Oxford in 2002. He will move back to Texas A&M as a full professor in the coming fall. His research focuses on hydrogen/methane storage and gas separation that are relevant to clean energy technologies. |
Energy spent on transportation accounts for a significant part of total energy consumption. It is estimated that in industrialized countries, one-third of all the energy generated annually is consumed in transportation.2 For a modern vehicle with a driving range of 400 km per tank of fuel, about 8 kg of hydrogen is needed for a combustion engine-driven automobile and 4 kg for a fuel-cell-driven one.3 Although these gravimetric requirements are far less demanding than that of gasoline (24 kg), hydrogen is notoriously difficult to compress for on-board storage. Volumetrically, even liquid hydrogen has a much smaller combustion heat than that of gasoline (8960 MJ m−3vs. 31170 MJ m−3). In the gas phase, 4 kg of hydrogen occupy 45 m3 of space at room temperature and 1 atm.3
In order to facilitate the research and application of hydrogen as an energy carrier, the US Department of Energy (DOE) has set the targets for on-board hydrogen storage systems: 6.0 wt% and 45 g L−1 by the year 2010, and 9.0 wt% and 81 g L−1 by 2015.4 These targets should be reached at ambient temperature (from −40 to 85 °C) and applicable pressure (less than 100 atm). Note that these are the goals for the system including container and any necessary accessories, the hydrogen storage capacity of the material itself should be even higher.
A safe and effective hydrogen storage technique has become the bottleneck for a possible hydrogen economy. High pressure or cryogenic liquid hydrogen tanks have been certified worldwide and demonstrated in some prototype fuel-cell vehicles. However, their limited storage densities prevent them from reaching the DOE targets. For example, high pressure tanks can reach a pressure of 10000 psi (680 atm) with a 2.35 safety factor (23500 psi burst pressure).4 However, the heavy weight of the system offsets the gain in gravimetric storage density under pressurized conditions, and the volumetric density is far from that of liquid hydrogen (70.8 g L−1). Cryogenic liquid hydrogen tanks, on the other hand, can be used to improve the volumetric hydrogen storage capacity. However, about 20% of the recoverable energy is needed to liquefy the hydrogen and another 2% is spent to keep the tank cool.1 The hydrogen storage capacity of the aforementioned tanks is demonstrated to be between 3.4 to 4.7 wt% gravimetrically and 14 to 28 g L−1 volumetrically.4
In solid-state storage systems, a hydrogen atom/molecule either forms a strong chemical bond to a solid support (chemisorption) or interacts weakly with a sorbent (physisorption).
In chemisorption, dihydrogen molecules split into hydrogen atoms upon contacting the solid support. The highly reactive hydrogen atoms can form chemical bonds with the solid, leading to the formation of metal hydrides or chemical hydrides, depending on the nature of the solid support. Due to the short bonds between hydrogen and the solid, some hydride compounds can reach a relatively high hydrogen storage capacity. However, this strong bonding also leads to severe kinetic and thermodynamic problems during the charging and discharging procedures. Complete charging may take several hours, and the hydrogen-releasing temperature is typically very high (300 °C or higher).5 Alanates are among the most-explored hydride compounds for hydrogen storage. For example, an alanate treated with titanium doping and ball milling showed more than 5 wt% hydrogen storage capacity with a hydrogen-releasing temperature only modestly higher than ambient.6 In another example, Balde et al. showed that by reducing the particle size of NaAlH4, the hydrogen desorption temperature and activation energy can be decreased from 186 °C and 116 kJ mol−1 for the largest particles to 70 °C and 58 kJ mol−1 for the smallest particles, respectively.7 However, heat management and reversibility of these systems remain a concern.8
The physisorption method, on the other hand, stores hydrogen in the molecular form in a sorbent with a large surface area. The most frequently-studied sorbents are activated carbons, carbon nanostructures, zeolites, porous polymers, and metal–organic frameworks (MOFs). Because of the weak sorbent–sorbate interaction, physisorption-based hydrogen storage systems show fast kinetics with a charging time of minutes. However, the same weak interaction results in gravimetric hydrogen uptake of a sorbent at ambient temperature and applicable pressure of typically less than 2 wt%.
Materiala | SA/m2 g−1 | Pore volume/cm3 g−1 | H2 uptake at 77 K, 1 atm/wt% | Maximum H2 uptake/wt% (g L−1) | ∆Hads/kJ mol−1 | Reference | ||
---|---|---|---|---|---|---|---|---|
BET Langmiur | 77 K | 298 K | ||||||
a 1,4-ndc = 1,4-naphthalenedicarboxylate; 2,6-ndc = 2,6-naphthalenedicarboxylate; 2,4-pdc = 2,4-pyridinedicarboxylate; 2-pmc = 2-pyrimidinecarboxylate; 2-pymo = 2-pyrimidinolate; 4,4′-bpe = 4,4′-trans-bis(4-pyridyl)-ethylene; 4,4′-bipy = 4,4′-bipyridine; 4,6-pmdc = 4,6-pyrimidicarboxylate; 4-pymo = 4-pyrimidinolate; 5-bbdc = 5-tert-butyl-1,3-benzenedicarboxylate; 6-mna = 6-mercapto-3-pyridinecarboxylate; abdc = 2-aminobenzene-1,4-dicarboxylate; abtc = azobenzene-3,3′,5,5′-tetracarboxylate; adc = 4,4′-azobenzenedicarboxylate; atdc = 9,10-anthracenedicarboxylate; bbdc = 2-bromobenzene-1,4-dicarboxylate; bdc = 1,4-benzenedicarboxylate; bdp = 1,4-benzenedi(4′-pyrazolyl); bdt = 1,4-benzeneditetrazolate; bhtc = biphenyl-3,4′,5-tricarboxylate; bpdc = 4,4′-biphenyldicarboxylate; bptc = 3,3′,5,5′-biphenyltetracarboxylate; bpydc = 2,2′-bipyridyl-5,5′-dicarboxylate; bpytc = 4,4′-bipyridine-2,6,2′,6′-tetracarboxylate; btc = 1,3,5-benzenetricarboxylate; btt = 1,3,5-benzenetristetrazolate; cbbdc = 1,2-dihydrocyclobutabenzene-3,6-dicarboxylate; cyclam = 1,4,8,11-tetraazacyclotetradecane; dabco = 1,4-diazabicyclo[2.2.2]octane; dbbd = 6,6′-dichloro-2,2′-dibenzyloxy-1,1′-binaphthyl-4,4′-dibenzoate; dccptp = 3,5-dicyano-4-(4-carboxyphenyl)-2,2′:6′4″-terpyridine; debd = 6,6′-dichloro-2,2′-diethoxy-1,1′-binaphthyl-4,4′-dibenzoate; detkb = 2,2′-diethoxylbiphenyl-3,3′,5,5′-tetra-kis(benzoate); dhbdc = 2,5-dihydroxyterephthalate; dipyni = N,N′-di-(4-pyridyl)-1,4,5,8-naphthalenetetracarboxydiimide; dpac = 4,4′-dipyridylacetylene; fma = fumarate; hfipbb = 4,4′-(hexafluoroisopropylidene)bis(benzoate); meim = 2-methylimidazole; ntb = 4,4′,4″-nitrilotrisbenzoic acid; ntc = naphthalene-1,4,5,8-tetracarboxylate; ox = C2O42−; oxdc = oxydiacetate; pbpc = pyridine-3,5-bis(phenyl-4-carboxylate); p-cdc = 1,12-dihydroxy-carbonyl-1,12-dicarba-closo-dodecaborane; pda = p-phenylenediacrylate; pdc = pyridine-3,5-dicarboxylate; phim = benzimidazolate; pic = 3-picoline; pydc = pyrene-2,7-dicarboxylate; pyen = 3,3′-(1E,1′E)-(ethane-1,2-diylbis(azan-1-yl-1-ylidene))bis(methan-1-yl-1-ylidene)dipyridin-4-olate; pyz = pyrazine; qptc = quaterphenyl-3,3‴,5,5‴-tetracarboxylate; sbtc = trans-stilbene-3,3′,5,5′-tetracarboxylate; sip = 5-sulfoisophthalate; tatb = 4,4′,4″-s-triazine-2,4,6-triyltribenzoate; tcppda = N,N,N′,N′-tetrakis(4-carboxyphenyl)-1,4-phenylenediamine; tfbdc = tetrafluoroterephthalate; tmbdc = tetramethylterephthalate; tpb-3tz = 1,3,5-tri-p-(tetrazol-5-yl)phenylbenzene; tptc = terphenyl-3,3″,5,5″-tetracarboxylate; trz = 1,2,4-triazole; ttdc = thieno[3,2-b]thiophene-2,5-dicarboxylate; tz = 3,5-bis(trifluoromethyl)-1,2,4-triazolate; tzi = 5-tetrazolylisophthalate; (refer to ESI1 for more details).b 1.2 bar.c 1.17 bar.d absolute hydrogen uptake. | ||||||||
Co(bdp) | 2670 | 0.93 | 3.1, 30 bar | 10 | ||||
Co(2,6-ndc)(4,4′-bipy)0.5 | 115 | 0.10 | 0.72 | 11 | ||||
Co(ox)(4,4′-bipy) | 0.1 | 12 | ||||||
Co(pyz)[Ni(CN)4] | 127 | 7.2 | 13 | |||||
Co(pyz)[Pd(CN)4] | 122 | 7.8 | 13 | |||||
Co(pyz)[Pt(CN)4] | 138 | 7.6 | 13 | |||||
Co3(bdc)3(dabco) | 360 | 538 | 0.195 | 1 | 14 | |||
Co3(bpdc)3(4,4′-bipy) | 922 | 0.38 | 1.98 | 6.8 | 15 | |||
Co3(D2-Htcppda)2 | 504 | 16 | ||||||
Co3(ndc)3(dabco) | 1502 | 2293 | 0.822 | 2.45 | 4.2 (35.3), 40 bar | 0.89 (4.48), 17.2 bar | 14 | |
Cu(2-pymo)2 | 350 | 0.86b | 17 | |||||
Cu(4-pymo)2 | 65 | 0.03b | 17 | |||||
Cu(bdc)(dabco)1/2 | 1461 | 0.52 | 1.8 | 18 | ||||
Cu(bdt) | 200 | 0.66c | 19 | |||||
Cu(dccptp)(NO3) | 268 | 0.113 | 1.34 | 1.91 (23.0), 20 bar | 6.12 | 20 | ||
Cu(fma)(4,4′-bpe)1/2 | 0.8 | 21 | ||||||
Cu(hfipbb)(H2hfipbb)1/2 | 0.23 | 1.0 (16), 48 bar | 22 | |||||
12 | ||||||||
Cu2(C2h-tcppda) | 504 | 1.2 | 23 | |||||
Cu2(D2-tcppda) | 626 | 1.4 | 23 | |||||
Cu2(detkb) | 733 | 0.42 | 0.8 | 24 | ||||
Cu2(qptc) | 2932 | 1.138 | 2.24 | 6.07 (35.6), 20 bar | 25 | |||
Cu2(tptc) | 2247 | 0.886 | 2.52 | 6.06 (39.4), 20 bar | 25 | |||
Cu2.9Mn1.1Cl(btt)3/8 (Cu) | 1695 | 1778 | 2.02b | 6.0–8.5 | 26 | |||
Cu3[(Cu4Cl)3(tpb-3tz)8]2 | 1120 | 1200 | 2.8 (11.4), 30 bar | 0.5 (2.03), 65 bar | 8.2 | 27 | ||
Cu4Cl(btt)3/8 | 1710 | 1770 | 2.42b | 4.2 (38), 30 bar | 9.5 | 28 | ||
5.7 (53), 90 bard | ||||||||
Cu6O(tzi)3(NO3) | 2847 | 3223 | 1.01 | 2.4 | 9.5 | 29 | ||
CUK-1, Co3(2,4-pdc)2(OH)2 | 630 | 0.26 | 1.60 | 30 | ||||
CUK-2, Co2(6-mna)2 | 420 | 0.17 | 0.66 | 30 | ||||
Dy(btc) | 655 | 1.32 | 31 | |||||
Er2(pdc)3 | 427 | 32 | ||||||
Fe3(OH)(pbpc)3 | 1200 | 1.6 | 3.05(33.1), 20 bar | 33 | ||||
Fe3O(tfbdc)3 | 635 | 0.9 | 34 | |||||
Fe4O2(btb)8/3 | 1121 | 1835 | 0.69 | 2.1 | 35 | |||
FMOF-1, Ag2[Ag4(tz)6] | 810.5 | 0.324 | 2.33 (41), 64 bar | 36 | ||||
HKUST-1, Cu3(btc)2 | 1154 | 1958 | 3.6 (31.6), 50 bar | 0.35 (3.08), 65 bar | 37 | |||
2.27 | 3.6 (31.6), 10 bar | 38 | ||||||
1944 | 2260 | 3.3 (29), 77 bar | 39 | |||||
1507 | 2175 | 0.75 | 2.54 | 5.8–6.6 | 40 | |||
0.4 | 1.44 | 6.0–7.0 | 12 | |||||
1239 | 0.62 | 2.18 | 6.1 | 41 | ||||
HKUST-1 + Pt/Ac | 1.12, 100 bar | 42 | ||||||
In3O(abtc)3/2(NO3) | 1417 | 0.5 | 2.61b | 6.5 | 43 | |||
IRMOF-1, MOF-5, Zn4O(bdc)3 | 3534 | 4170 | 5.2 (31), 45 bar | 39 | ||||
3080 | 4.3 (25.5), 30 bar | 0.45 (2.67), 60 bar | 4.1 | 44 | ||||
572 | 1014 | 0.28 | 1.6 (9.49), 10 bar | 0.2 (1.19), 67 bar | 45 | |||
2885 | 1.18 | 1.15 | 46 | |||||
3362 | 1.32 | 47 | ||||||
3800 | 4400 | 7.1 (42.1) 40 bar | 48 | |||||
10.0 (66), 100 bard | 48 | |||||||
2296 | 3840 | 4.7 (27.9), 50 bar | 0.28 (1.66), 65 bar | 37 | ||||
1.32 | 4.3 (25.5), 30 bar | 4.8 | 49 | |||||
4.5 | 9 | |||||||
IRMOF-1 + Pd | 958 | 0.39 | 1.86 | 46 | ||||
IRMOF-1 + Pt/Ac | 3.0, 100 bar | 50 | ||||||
IRMOF-2, Zn4O(bbdc)3 | 1722 | 2544 | 0.88 | 1.21 | 40 | |||
IRMOF-3, Zn4O(abdc)3 | 2446 | 3062 | 1.07 | 1.42 | 40 | |||
IRMOF-6, Zn4O(cbbdc)3 | 2476 | 3263 | 1.14 | 1.48 | 4.9 (34.9), 32 bar | 40 | ||
2804 | 3300 | 4.8 (32), 45 bar | 39 | |||||
IRMOF-8, Zn4O(ndc)3 | 1818 | 3.6 (20.9), 15 bar | 0.4 (2.32), 30 bar | 6.1 | 44 | |||
1466 | 1.5 | 47 | ||||||
IRMOF-8 + Pt/Ac | 1.8, 100 bar | 51 | ||||||
4.0, 100 bar | 50 | |||||||
IRMOF-9, Zn4O(bpdc)3 | 1904 | 2613 | 0.9 | 1.17 | 40 | |||
IRMOF-11, Zn4O(hpdc)3 | 1984 | 2340 | 3.5 (27), 34 bar | 5.1–9.1 | 39 | |||
1911 | 1.62 | 47 | ||||||
IRMOF-13, Zn4O(pydc)3 | 1551 | 2100 | 0.73 | 1.73 | 40 | |||
IRMOF-18, Zn4O(tmbdc)3 | 1501 | 0.89 | 47 | |||||
IRMOF-20, Zn4O(ttdc)3 | 3409 | 4346 | 1.53 | 1.35 | 6.7 (43.9), 70 bar | 40 | ||
4024 | 4590 | 6.7 (34), 80 bar | 39 | |||||
JUC-48, Cd3(bpdc)3 | 880 | 0.19 | 2.8 (20.0), 40 bar | 1.1 (7.84), 100 bar | 52 | |||
MAMS-1, Ni8(5-bbdc)6 | 0.6 | 53 | ||||||
α-[Mg3(HCO2)6] | 150 | 0.043 | 0.6 | 54 | ||||
Mg3(ndc)3 | 520 | 0.78 | 55 | |||||
190 | 0.46c | 7.0–9.5 | 56 | |||||
MIL-53(Al), Al(OH)(bdc) | 1100 | 1590 | 0.59 | 2.1 | 3.8 (37.0), 16 bar | 57 | ||
MIL-53(Cr), Cr(OH)(bdc) | 1100 | 1500 | 0.56 | 1.8 | 3.2 (33.2), 16 bar | 57 | ||
MIL-53 + Pt/Ac | 0.63, 50 bar | 58 | ||||||
MIL-96, Al3O(btc)3 | 1.91 (27.9), 3 bar | 59 | ||||||
MIL-100(Cr), Cr3OF(btc)2 | 2700 | 1 | 3.3 (23.0), 25 bar | 0.15 (1.04), 73 bar | 5.6–6.3 | 60 | ||
MIL-101(Cr), Cr3OF(bdc)3 | 5500 | 1.9 | 2.5 | 6.1 (26.1), 60 bar | 0.43 (1.84), 80 bar | 9.3–10.0 | 60 | |
MIL-101 + Pt/Ac | 1.43, 100 bar | 42 | ||||||
1.14, 50 bar | 58 | |||||||
MIL-102, Cr3OF(ntc)3/2 | 42.1 | 0.12 | 1.0 (16), 35 bar | 0.05 (0.8), 35 bar | 6 | 61 | ||
Mn(HCO2)2 | 240 | 0.9 | 62 | |||||
Mn(ndc) | 191 | 0.068 | 0.57 | 63 | ||||
Mn3(bdt)8Cl2 | 530 | 0.82c | 6.0–8.8 | 19 | ||||
Mn3(bdt)3 | 290 | 0.97c | 6.3–8.4 | 19 | ||||
Mn3[(Mn4Cl)3(tpt-3tz)8]2 | 1580 | 1700 | 3.7(25.0), 25 bar | 0.5 (3.38), 65 bar | 7.6 | 27 | ||
4.5 (37), 80 bard | ||||||||
Mn4Cl(btt)3/8 | 2100 | 2.2b | 5.1 (43), 50 bar | 10.1 | 64 | |||
6.9 (60), 90 bard | ||||||||
Mn4Cl(btt)3/8 (Co) | 2096 | 2268 | 2.12b | 5.6–10.5 | 26 | |||
Mn4Cl(btt)3/8 (Cu, Mn) | 1911 | 2072 | 2.00b | 5.6–9.9 | 26 | |||
Mn4Cl(btt)3/8 (Fe) | 2033 | 2201 | 2.21b | 5.5–10.2 | 26 | |||
Mn4Cl(btt)3/8 (Li) | 1904 | 2057 | 2.06b | 5.4–8.9 | 26 | |||
Mn4Cl(btt)3/8 (Ni) | 2110 | 2282 | 2.29b | 5.2–9.1 | 26 | |||
Zn0.7Mn3.3Cl(btt)3/8 (Zn) | 1927 | 2079 | 2.10b | 5.5–9.6 | 26 | |||
MOF-177, Zn4O(btb)2 | 4750 | 5640 | 7.6 (32), 66 bar | 4.4 | 65 | |||
11.4 (49), 78 bard | ||||||||
4746 | 5640 | 7.5 (32), 70 bar | 39 | |||||
4526 | 1.25 | 47 | ||||||
MOF-177 + Pt/Ac | 1.5, 100 bar | 5.8–11.3 | 66 | |||||
MOF-505, Cu2(bptc) | 1670 | 1830 | 0.68 | 2.59 | 4.02 (37.3), 20 bar | 25 | ||
0.63 | 2.47 | 67 | ||||||
MOF-74, Zn3(dhbdc)3 | 950 | 1070 | 2.3 (27), 26 bar | 39 | ||||
783 | 1132 | 0.39 | 1.77 | 8.3 | 40 | |||
870 | 2.8 (32.9) 30 bar | 68 | ||||||
NaNi3(OH)(sip)2 | 743 | 0.94 | 9.4–10.4 | 69 | ||||
Ni(4,4′-bipy)[Ni(CN)4] | 234 | 1.84 | 7.5 | 13 | ||||
Ni(4,4′-bipy)[Pd(CN)4] | 220 | 7 | 13 | |||||
Ni(cyclam)(bpydc) | 817 | 0.37 | 1.1 | 70 | ||||
Ni(dpac)[Ni(CN)4] | 398 | 2.24 | 6 | 13 | ||||
Ni(ox)(4,4′-bipy) | 0.16 | 12 | ||||||
Ni(pyz)[Ni(CN)4] | 124 | 1.76 | 7.2 | 13 | ||||
Ni2(4,4′-bipy)3(NO3)4 | 0.181 | 0.987 | 71 | |||||
Ni2(4,4′-bipy)3(NO3)4 | 0.149 | 0.653 | 71 | |||||
Ni2(dhbdc) | 1083 | 0.41 | 1.8 (21.5), 70 bar | 0.3 (3.58) 65 bar | 72 | |||
Ni3(btc)2(3-pic)6 | 0.63 | 71 | ||||||
Ni3(OH)(pbpc)3 | 1553 | 1.99 | 4.15 (43.9), 20 bar | 33 | ||||
PCN-5, Ni3O(tatb)2 | 225 | 0.13 | 0.63 | 73 | ||||
PCN-6, Cu3(tatb)2 | 3800 | 1.45 | 1.9 | 74 | ||||
PCN-6′, Cu3(tatb)2 | 2700 | 1.045 | 1.35 | 74 | ||||
PCN-9, Co4O(tatb)8/3 | 1355 | 0.51 | 1.53 | 10.1 | 75 | |||
PCN-10, Cu2(abtc) | 1407 | 1779 | 0.67 | 2.34 | 4.33 (33.2), 20 bar | 4–7 | 76 | |
PCN-11, Cu2(sbtc) | 1931 | 2442 | 0.91 | 2.55 | 5.05(37.8), 20 bar | 4–7 | 76 | |
PCN-13, Zn4O(adc)3 | 150 | 0.1 | 0.41 | 77 | ||||
PCN-17, Yb(tatb)8/3(SO4)2 | 820 | 0.34 | 0.94 | 78 | ||||
Pd(2-pymo)2 | 600 | 1.29b | 17 | |||||
rho-ZMOF, Cd(2-pmc)2 | 1168 | 0.474 | 1.16 | 8.7 | 79 | |||
sod-ZMOF, In(4,6-pmdc)2 | 616 | 0.245 | 0.9 | 8.4 | 79 | |||
Sc(bdc)3 | 721 | 0.332 | 1.5 | 80 | ||||
Sm2Zn3(oxdc)6 | 718.8 | 0.31 | 1.19 (18.6), 34 bar | 0.54 (8.4), 35 bar | 81 | |||
TUDMOF-1, Mo3(btc)2 | 1280 | 2010 | 0.67 | 1.75 | 82 | |||
UMCM-150, Cu3(bhtc)2 | 2300 | 3100 | 1 | 2.1 | 5.7 (36), 45 bar | 7.3 | 83 | |
Y2(pdc)3 | 676 | 32 | ||||||
ZIF-11, Zn(phim)2 | 1676 | 0.582 | 1.37 | 84 | ||||
ZIF-8, Zn(meim)2 | 1.3 | 3.3 (35.6), 30 bar | 0.13 (1.40), 30 bar | 4.5 | 49 | |||
1630 | 1810 | 0.636 | 1.29 | 3.1 (33.4), 55 bar | 84 | |||
Zn(adc)(4,4′-bpe)1/2 | 100 | 0.62 | 85 | |||||
Zn(bdc)(4,4′-bipy)1/2 | 946 | 0.8 | 86 | |||||
Zn(bdc)(dabco)1/2 | 0.73 | 2.1 | 5.0–5.3 | 87 | ||||
1794 | 0.65 | 2.1 | 18 | |||||
Zn(2,6-ndc)(4,4′-bpe)1/2 | 303 | 0.2 | 0.80 | 2.0 (17.9), 40 bar | 0.3(2.7), 65 bar | 88 | ||
Zn2(1,4-bdc)(tmbdc)(dabco) | 1100 | 1670 | 0.59 | 2.08 | 89 | |||
Zn2(1,4-bdc)2(dabco) | 1450 | 2090 | 0.75 | 2.01 | 89 | |||
Zn2(1,4-ndc)2(dabco) | 1000 | 1450 | 0.52 | 1.7 | 89 | |||
Zn2(bdc)2(dabco) | 1603 | 2420 | 0.86 | 1.8 | 90 | |||
Zn2(bpytc) | 312.7 | 0.146 | 1.08(16.7), 4 bar | 0.057(0.88), 4 bar | 5.12 | 91 | ||
Zn2(detkb) | 252 | 0.2 | 24 | |||||
Zn2(2,6-ndc)2(dipyni) | 802 | 0.30 | 0.93 | 5.6 | 92 | |||
Zn2(2,6-ndc)2(dipyni) + Li+ | 756 | 0.34 | 1.63 | 6.1 | 92 | |||
Zn2(tfbdc)2(dabco) | 1070 | 1610 | 0.57 | 1.78 | 89 | |||
Zn2(tmbdc)2(4,4′-bipy) | 1120 | 1740 | 0.62 | 1.68 | 89 | |||
Zn2(tmbdc)2(dabco) | 920 | 1400 | 0.5 | 1.85 | 89 | |||
Zn3(bdc)3[Cu(pyen)] | 0.257 | 1.10 | 12.29 | 93 | ||||
Zn3(bdt)3 | 640 | 1.46c | 6.8–8.7 | 19 | ||||
Zn3(bpdc)3(4,4′-bipy) | 792 | 0.33 | 1.74 | 7.1 | 15 | |||
Zn3(OH)(p-cdc)5/2 | 152 | 2.1 | 7 | 94 | ||||
Zn3(tatb)2(HCOO) | 1100 | 1.3 | 95 | |||||
Zn4(trz)4(1,4-ndc)2 | 362.1 | 0.84 | 96 | |||||
Zn4(trz)4(2,6-ndc)2 | 584.1 | 1.11 | 96 | |||||
Zn4O(D2-tcppda)3/2 | 2095 | 0.8 | 16 | |||||
Zn4O(dbbd)3 | 396 | 0.13 | 0.98 (10), 48 bar | 97 | ||||
Zn4O(debd)3 | 502 | 0.2 | 1.12 (10.3), 48 bar | 97 | ||||
Zn4O(ntb)2 | 1121 | 0.51 | 1.9 | 98 | ||||
Zn7O2(pda)5 | 0.17 | 1.01 (4.35), 71.4 bar | 99 |
Fig. 1 Illustration of MOF synthesis and chemical composition. |
Although the DOE targets require hydrogen uptake measurements carried out at ambient temperature (from −40 to 85 °C) and applicable pressure (up to 100 atm), most reported hydrogen uptake measurements in MOFs are carried out at 77 K and 1 atm. This is largely because of the availability of that condition in commercially available gas adsorption equipment. Although far from DOE-target conditions, hydrogen uptake data obtained at low temperature and pressure are still useful in the initial exploration and comparison of hydrogen uptake capacities in different MOFs. As high pressure hydrogen uptake measurement equipment becomes more widely available, more and more data of high pressure hydrogen uptake in MOFs have been reported. At room temperature, the binding energy between hydrogen and the MOFs is comparable to thermal vibration energies, which leads to very poor hydrogen uptake (typically less than 1 wt%) and makes the comparison and discussion of hydrogen uptake capacity ambiguous and difficult. At the cryogenic condition, however, the relatively strong interaction between hydrogen and MOFs (as compared to thermal energies) gives rise to greatly enhanced hydrogen uptake. This enhancement is so pronounced that some saturation hydrogen uptake data obtained at 77 K have reached or even surpassed the DOE targets.39,48,64 Thus the comparison of saturation hydrogen uptake data at 77 K is more reliable and instructive in determining the MOFs' hydrogen uptake capacity.
By applying the Clausius–Clapeyron equation to two sets of hydrogen adsorption data collected at different temperatures (typically 77 K and 87 K), the isosteric heat of adsorption (∆Hads) can be deduced; this is an important criterion in judging how strongly hydrogen binds the MOFs.40,64 In order to guarantee the validity of these ∆Hads values, data sets should be collected at more than two different temperatures.
Nabs = Nex + ρbulkVpore |
A large isosteric heat of adsorption would lead to a steep rise in the adsorption isotherm within the low pressure range (less than 3 atm), which at first was deemed to be helpful for hydrogen storage. However, in the DOE targets, the minimum delivery pressure is 3–4 atm for fuel-cells and 35 atm for internal combustion engines, which means that any hydrogen that is stored at a pressure less than 3 atm would not be fully utilized.4 Furukawa et al. proposed the deliverable capacity as another criterion for judging material's hydrogen storage uptake, based on the absolute adsorbed amount of hydrogen in the sorbent materials from 1.5 to 100 bar.65 The same idea has been addressed by other research groups.87,107
In order to directly judge the sorbent materials' gas uptake capacities, Zhou et al. introduced the “effective adsorption” concept, which compares the amount of gas held in a container with and without the sorbent materials.49 According to their high pressure hydrogen uptake measurements at room temperature, the effective hydrogen adsorption in MOF-5 and ZIF-8 are nearly zero and negative respectively, indicating no positive contribution to hydrogen storage compared with a high pressure tank. However, the high pressure hydrogen uptake measurements at 77 K conducted by Mueller et al. show that the container filled with MOF-5 takes up higher amounts of hydrogen than the empty container.108 Among the MOFs they tested, HKUST-1, a MOF composed of copper ions and 1,3,5-benzenetricarboxylate,109 reaches an effective adsorption of +44%, with a volumetric hydrogen uptake capacity of 18.5 g L−1. They also pointed out that because of the many volume-limited fuel-cell applications and the low density of MOFs, the volumetric hydrogen uptake capacity should also be addressed along with the gravimetric one.
Fig. 2 Correlation between surface area (red: Langmuir; blue: BET) and total pore volume. |
There is a well-established positive relationship between the surface area and the hydrogen uptake in carbon-based sorbents.110,111 Note that the linker portion of most MOFs is composed of aromatic ring motifs, which give rigidity to the framework. Such chemical composition is very similar to that of carbon materials, which are largely composed of sp2-hybridized carbon atoms. A positive, roughly linear relationship between specific surface area and hydrogen uptake in MOFs can be observed by plotting the surface areas versus the 77 K saturation hydrogen uptake data (Fig. 3).39,104 The slope of the linear relationship is 1.45 × 10−3 wt% (m2 g−1)−1 for the Langmuir surface area and 1.92 × 10−3 wt% (m2 g−1)−1 for the BET surface area, which is comparable to the theoretical value for carbon (2.28 × 10−3 wt% (m2 g−1)−1).112 Although it has been suggested that adsorption in MOFs occurs through a pore-filling mechanism rather than layer formation, grand canonical Monte Carlo (GCMC) simulations performed on a series of MOFs have verified the validity of the BET theory in determining the surface areas of MOFs.113
Fig. 3 Correlation between surface area and saturation hydrogen uptake at 77 K (red: MOFs, Langmuir method; blue: MOFs, BET method; black: sp2 carbon, theoretical). |
Among MOFs, MOF-177 ranks highest for gravimetric hydrogen uptake, with a value of 7.6 wt% at 77 K and 70 bar.65 It also has a very large surface area of around 4500 m2 g−1. In the case of MIL-101, although it possesses the highest surface area (Langmuir surface area is ∼5900 m2 g−1),103 the high pressure hydrogen uptake at 77 K is not the highest.60 However, according to the author, the sample has not been fully activated due to the small opening within the cage structure.
It can be safely concluded that in most cases, the saturation hydrogen uptake in MOFs at 77 K is mainly determined by the surface area and the pore volume regardless of the chemical composition.114 Theoretically, a minimum surface area of ∼1400 m2 g−1 is needed for a material to reach an excess adsorption of 6 wt% (or ∼2100 m2 g−1 for 9 wt%).68
In the case of HKUST-1, both neutron powder diffraction and inelastic neutron scattering (INS) data support that hydrogen molecules are adsorbed into the smaller cage before the larger one, indicating a stronger interaction between hydrogen and the smaller pore.117,118 This conclusion is confirmed indirectly by desorption studies of hydrogen in various MOFs (HKUST-1, MIL-53, MOF-5, and IRMOF-8), where the hydrogen desorbed first from the larger pore, then the smaller pore with increasing temperature.119
One way to reduce pore size is by introducing bulky groups in the ligands. Farha et al. used a bulky carborane ligand to construct a MOF with 2.1 wt% of hydrogen uptake at 77 K and 1 atm.94 Pan et al. constructed a microporous MOF with a ligand decorated with the bulky trifluoromethyl group.22 Due to the curved internal surfaces and the reduced pore size, the room temperature hydrogen uptake capacity of this MOF is close to 1 wt% at 48 atm, which is comparable to the best-performing carbon nanotubes they examined. The same strategy was adopted by Yang et al., who generated a MOF with volumetric hydrogen uptake capacity of 41 g L−1 at 77 K and 64 bar, close to the DOE 2010 volumetric target of 45 g L−1.36 However, one should bear in mind that the improved gravimetric hydrogen uptake by this strategy is partially counteracted by the increased framework density given by the decorating groups.40
Because catenation will decrease the free volume, whether catenation is helpful for hydrogen uptake is determined by the compromise between the increased hydrogen density within the pores and the decreased free volume from catenation.121 The GCMC simulation on IRMOF-9 and IRMOF-10 suggested that catenation is not a promising option to increase the hydrogen uptake at high pressure due to the reduced pore volume.121 At low pressure and 77 K, however, the experimental study of catenated IRMOF-11 showed higher hydrogen uptake compared to non-catenated IRMOFs with the same topology, which has been supported by GCMC simulation.40,122
Using oxalic acid as a template molecule, Ma et al. have generated a catenated MOF, PCN-6, and its non-catenated counterpart, PCN-6′, making the study of catenation's effect on hydrogen uptake in MOFs as an independent criterion possible.74 In this case, the low pressure and 77 K gas sorption results showed that catenation leads to a 41% increase in Langmuir surface area and 133% of enhancement in volumetric hydrogen uptake (29% in gravimetric). The high pressure and 77 K hydrogen uptake measurements done by Dincă et al. on one pair of ligand-directed catenated and non-catenated MOFs indicated 41% increase in surface area and 32% increase in gravimetric hydrogen uptake by catenation.27
Up to now, catenation has shown some improvement in hydrogen uptake in some MOF systems at 77 K. The study of catenation's effect on MOFs' hydrogen uptake capacity at ambient temperature is being undertaken by our group and will be published soon.
In contrast to using a larger aromatic ring, the VASP ab initio computer calculations indicated that there is little effect on the hydrogen uptake capacity if the aromatic ring is substituted with halogen, and the added mass would be detrimental to the hydrogen uptake capacity.126 This is confirmed experimentally. A systematic study by Chun et al. on the influence in hydrogen uptake capacity brought by the modulation of the organic ligands resulted in no direct relationship between the hydrogen uptake capacity and the chemical composition of the organic ligands.89 On the contrary, the authors suggested that the shape and size of channels instead of the ligands' chemical nature should be responsible for the hydrogen uptake trend in their study. The same conclusion is drawn by Rowsell et al., in which the low pressure hydrogen uptake measurements were done on a series of IRMOFs.40
The unsaturated metal sites are not exclusive to copper-containing MOFs. Inspired by the entatic state in biological systems, Ma et al. constructed a MOF, PCN-9, in which cobalt atoms are five-coordinate with square pyramidal geometry, leading to a hydrogen adsorption heat of 10.1 kJ mol−1.75 The neutron powder diffraction study done on MOF-74 revealed the strong interaction between hydrogen and the exposed Zn2+ ions and indicated a strong correlation between the existence of unsaturated metal sites and the high hydrogen surface packing density.68 Dincă et al. constructed a MOF with both exposed Mn2+ coordination sites and free Mn2+ within the channel.64 Neutron powder diffraction data showed direct hydrogen binding at the unsaturated Mn2+ within the framework, with the maximum isosteric heat of adsorption 10.1 kJ mol−1. The absolute hydrogen uptake is 6.9 wt% at 77 K and 90 bar with the density of the stored hydrogen 85% of that of liquid hydrogen. By replacing the coordinated Mn2+ with Cu2+, a more robust MOF is generated, which can be fully desolvated to expose a larger number of open metal sites.28 A slightly decreased heat of adsorption of the generated copper MOF as compared to its manganese counterpart was explained by Jahn–Teller distortion of the coordination environment of the Cu2+ ions. Another explanation using spin state has been given by a computational study, which demonstrated that binding energy can be tuned in a range of about 10 to 50 kJ mol−1 using different transition metal ions.129 Another theoretical study pointed out that the interaction is not of the expected Kubas-type but only comes from classical Coulomb interaction.130 Using the same MOF, Dincă et al. replaced the free Mn2+ cation with other cations to generate a series of isostructural MOFs.26 There is an adsorption heat difference of 2 kJ mol−1 between the weakest and strongest hydrogen-binding MOFs; among them the Co2+-exchanged MOF exhibits the highest heat of adsorption, 10.5 kJ mol−1.
A combined DFT and GCMC simulation study on MOF-505 showed that open metal sites have favorable impact on hydrogen adsorption in MOFs at low pressures, and the hydrogen molecule is inclined to expose the negative lobe of its quadrupole to the exposed copper atoms, which act as Lewis acids.131 According to another simulation study, if the open metal sites could be incorporated on the organic linkers, the metal–hydrogen dissociation energy could go as high as 84 kJ mol−1, a potential route to achieve reversible sorption at ambient conditions.132
Although there are still some arguments about whether the unsaturated metal sites are the main reason for the increased interaction between hydrogen and the framework,63 the combination of unsaturated metal sites with the appropriate pore size and geometry discussed above gives rise to some MOFs with strong hydrogen binding energies. In the case of Mg3(O2C–C10H6–CO2)3, where the Mg2+ centers are unsaturated and the pore dimensions are constricted, the hydrogen isosteric heat of adsorption reaches 9.5 kJ mol−1.56 Chen et al. immobilized unsaturated metal sites within ultramicropores to generate a mixed zinc/copper MOF with an isosteric heat of hydrogen adsorption of up to 12.29 kJ mol−1, the highest reported so far.93 With the increasing amount of hydrogen adsorbed, the heat decreases and reaches a plateau where all the open metal sites become saturated by the hydrogen. Even more interestingly, because the pore size is so small, the quantum effects in the sorption of H2 and D2 are observable, demonstrating potential applications for isotope separation.
In addition to cations, anions can also be helpful in hydrogen adsorption. For example, charge-separated ammonium fluorides are calculated to have enhanced binding energy towards molecular hydrogen.137
Because it is well demonstrated that the hydrogen molecule can be dissociated into monoatomic hydrogen by certain heavy transition metals (e.g. Pt), making use of this “dissociation/spillover” in a MOF-based hydrogen-storage system leads to hydrogen uptake enhancement, increasing adsorbed hydrogen by a factor of 3.3 for MOF-5 and 3.1 for IRMOF-8.51 In the latter exploration, by using a carbon bridge to facilitate the secondary spillover, the enhancement factor for IRMOF-8 has been increased to 8, resulting in a hydrogen uptake of 4 wt% at room temperature and 10 MPa, the highest among all the MOFs, with the entire process completely reversible.50 The spillover effect has been reproduced by Liu et al.58 Their results show the storage capacity of 1.14 and 0.63 wt% for MIL-101 and MIL-53 at 5.0 MPa and 293 K, which is greatly increased from that of pristine samples (0.37 wt% and none). In another approach, in which palladium was doped into MOF-5 via solution infiltration, the hydrogen adsorption capacity is increased by 62% to 1.86 wt% at 77 K and 1 atm.46 However, according to the authors, the increase at low pressures does not necessarily imply a higher capacity at high pressures.
Besides the sample quality and activation conditions, sample size can also potentially affect the accuracy of the measurement.106 Too small a sample size would lead to larger uptake while too large a sample size may need more time for the sorption to reach equilibrium. For the volumetric method, an appropriate sample size is typically about 100 mg.
In our own study, we have observed that a sample with small particle size has a larger hydrogen uptake than one with larger particle size. The possibility of the surface area difference caused by the particle size has been ruled out because the external surface area increase due to smaller particle size is neglectable compared to the much larger internal surface area. One possible explanation would be that in larger particles, the longer diffusion path limits access into the interior of the particle, either for the guest molecules coming out from the frameworks during activation or the hydrogen molecules going into the frameworks during adsorption. Besides, the chemical difference between the terminal and the inner parts of the particle should also be considered. It is possible that more unsaturated metal sites would be exposed at the surface of smaller particles than that of the larger one, leading to increased hydrogen uptake.
Solvent exchange is also a crucial step in sample activation. By replacing the high-boiling-point and strongly-bound solvent or guest molecules (e.g. amide) with low-boiling-point and weakly-bound molecules (e.g. dichloromethane, chloroform, and methanol), the void inside the MOFs could be evacuated under moderate condition without framework collapse.
Sample preparation is also important for material stability. For example, MOF-5, containing the Zn4O motif, is proven to be unstable upon contact with moisture.42,45,48,66,138 Kaye et al. modified the previously reported method to obtain a sample of MOF-5 with the highest surface area among the reported data, in which the exposure to water and air was minimized.48 Their sample adsorbed 7.1 excess wt% hydrogen, and the absolute hydrogen uptake climbed to 11.5 wt% at 170 bar, with a volumetric storage density of 77 g L−1, which is greater than the density of liquid hydrogen (70.8 g L−1).
The nature of the primary interaction between hydrogen and MOFs is also unclear. Theoretically, the interaction forces between molecular hydrogen with any system include weak van der Waals forces, electrostatic interactions, orbital interactions, and non-classical sigma bonding (metal–dihydrogen complexes or the “Kubas complex”).133 In the case of MOFs, INS data indicated that there are two hydrogen-binding sites in MOF-5, with the stronger binding site associated with the metal–oxide cluster and the weaker one with the organic linker.9 These conclusions have been supported by both neutron powder diffraction and molecular dynamics simulation.141,142 The IR spectroscopic study conducted by Bordiga et al. demonstrates that the interaction between hydrogen and MOF-5 is largely due to van der Waals interactions with the internal wall structure and to weak electrostatic forces associated with the metal–oxide cluster.143 The isosteric heat of adsorption for hydrogen uptake in most MOFs lies in the range of 3.5 to 6.5 kJ mol−1, and these values tend to decrease with increasing amount of hydrogen due to the formation of a hydrogen monolayer on available surfaces.1
Clearly, the inherently weak interaction between hydrogen and MOFs would not meet the interaction requirement discussed above. Although the introduction of unsaturated metal sites into MOFs is an effective way to increase the hydrogen binding energy, the enhancement of hydrogen uptake in this way is limited due to the short-range nature of this interaction. In the case of HKUST-1, the enhancement is only around 1 wt% if each copper open site can bind one hydrogen molecule, and the enhanced uptake is limited to a very narrow pressure range (below 0.3 bar),128 leading to poor delivery capacity.65 Spillover might be a plausible method to strengthen the hydrogen binding were it not for the unpredictable hydrogen uptake enhancement factor. In contrast with the initial impressive report (enhancement factor of 8 for IRMOF-8),50 other studies only show a moderate effect (enhancement factor of 2.08–3.2).42,58,66 In addition, the usage of expensive and environmentally harmful heavy transition metals would greatly limit large scale application of this method.
Introducing charges into the MOFs would be a good method to increase the interaction. Since there is neither charge nor dipole moment in the dihydrogen molecule, the highest-energy interactions between a point charge and hydrogen are through the quadrupole moment, which is ∼3.5 kJ mol−1 at 3 Å separation, and via charge-induced dipole interaction, with an energy ∼6.8 kJ mol−1 at 3 Å separation.133 Calculations done by Garberoglio et al. find that electrostatic charges on MOFs would substantially increase the hydrogen uptake at 77 K and low pressure, but the effect on high pressure uptake is only marginal, with minimal effect on room temperature hydrogen uptake.144 Eddaoudi et al. have prepared ionic MOFs that show high hydrogen uptake and isosteric heat of adsorption, which, according to their explanation, is due to the narrow pore and highly localized charge density.43,79 The corresponding calculation study confirmed the speculation that polarization interactions are significantly enhanced by the presence of a charged framework with confined pores, which makes these MOFs excellent hydrogen storage candidates.145 The same conclusion has been drawn based on theoretical study of other charged carbon materials. GCMC simulation on charged single-walled carbon nanotubes demonstrates that a hydrogen uptake increase of ∼10%–20% for 298 K and 15%–30% for 77 K is achievable in realistically charged (0.1 e per carbon atom) nanotubes compared to uncharged ones.146 In order to achieve the DOE targets, however, the charges on the nanotubes need to be unrealistically large, which is both theoretically and experimentally impossible. A more optimistic conclusion is given by the first-principles calculations on charged fullerenes, in which the binding strength for hydrogen could be enhanced to a desirable range for potential near ambient applications with a maximum storage capacity of up to ∼8.0 wt%.147
For hydrogen storage applications, however, the opposite trend in hysteresis would be more useful, leading to improved usable storage capacity due to a clean release (Fig. 5). There is no such sorption isotherm existing among the currently known isotherm types. In order to reach this optimized sorption isotherm, other methods besides pressure can be used to trigger the gas delivery, such as raising the temperature or exposure to UV light.107
The DOE targets for on-board hydrogen storage pose a formidable challenge to those who are interested in solving such a fundamental but rewarding problem. The on-board hydrogen storage goal can only be achieved if theorists and experimentalists work together to find revolutionary systems based on basic studies including those reviewed above.
Footnote |
† Electronic supplementary information (ESI) available: Abbreviation, full name and chemical structure of the ligands listed in Table 1. See DOI: 10.1039/b808322n |
This journal is © The Royal Society of Chemistry 2008 |