Aerogels are typically made from silicon oxide or aluminium

ARTICLES PUBLISHED ONLINE: 17 MAY 2009 | DOI: 10.1038/NCHEM.208

Spongy chalcogels of non-platinum metals act as effective hydrodesulfurization catalysts Santanu Bag1, Amy F. Gaudette2, Mark E. Bussell2 and Mercouri G. Kanatzidis1 * Aerogels are low-density porous materials, made mostly of air, for which hundreds of applications have been found in recent years. Inorganic oxide-based aerogels have been known for a long time, carbon aerogels were discovered in the early 1990s and sulfur- and selenium-based aerogels (chalcogels) are the most recent additions to this family. Here we present new aerogels made of Co(Ni)–Mo(W)–S networks with extremely large surface areas and porosity. These systems are formed by the coordinative reactions of (MoS4)22 and (WS4)22 with Co21 and Ni21 salts in non-aqueous solvents. We show that these low-density sponge-like networks can absorb conjugated organic molecules and mercury ions, and preferentially adsorb CO2 over H2 , which illustrates their high potential as gas-separation media. The chalcogels are shown to be twice as active as the conventional sulfided Co–Mo/Al2O3 catalyst for the hydrodesulfurization of thiophene.

A

erogels are typically made from silicon oxide or aluminium oxide gels that are supercritically dried using carbon dioxide1. Recently, we reported the formation of gels from various germanium-sulfide and -selenide clusters with platinum as the linking metal ion. We named the new gels ‘chalcogels’ because they are chalcogenide species2. The extremely large surface area and low density of aerogels have led to hundreds of applications in recent years, ranging from molecular sieves to collecting dust from the tails of comets1. Chalcogels, however, by virtue of their more polarizable metal–chalcogenide surface promise new applications beyond those of conventional aerogels, including remediation of heavy metals, chemoselective absorption of molecules and macromolecules, and catalysis3,4. Although most oxide-based aerogels have a very wide bandgap, which means they can only absorb light from the ultraviolet region, the chalcogels have a smaller bandgap, which allows them to absorb light from the visible and infrared regions, and thus are potentially useful for applications such as photocatalysis. The success of the platinum-based chalcogels partly arises from the relatively slow ligand-substitution kinetics during reaction with the chalcogenide precursor clusters. This reduces the chances of precipitation and allows a polymeric network to form. Here we show that the chalcogel family is not limited only to platinum-based materials. We have successfully extended this chemistry to other precursor clusters and transition metals. We chose the tetrathiomolybdate(tungstate) anions (MoS4)22 and (WS4)22 as precursors and Co2þ and Ni2þ metal salts as the linking agents. This choice was based on the relevance of these elements to the catalysts involved in the hydrodesulfurization (HDS) process, which is the hydrogenative removal of sulfur from fossil fuels5–7. In addition to other uses, we expected that aerogels composed of Co(Ni)– Mo(W)–S with extremely large surface areas could be promising as novel HDS catalysts. We describe here a sol–gel approach for forming porous chalcogels with such large surface areas, based on Ni–Mo–S, Co–Mo–S, Co–Ni–Mo–S and Co–Mo–W–S compositions. In addition to the large surface area, the surfaces of these chalcogels terminate in sulfur atoms, so they present a unique highly polarizable surface to incoming guest species. Depending on their polarizability the guest molecules are expected to interact more or less strongly with the sulfidic surface. We therefore tested

these chalcogels for their ability to separate a mixtures of gases with different polarizabilities, namely H2 from CO2. We observed a dramatic difference in the interactions with these two molecules and a 15:1 preference for CO2 over H2. We tested a Co–Mo–S chalcogel for the HDS of thiophene and observed that it is an excellent precursor to a highly active catalyst.

Results and discussion We obtained gelatin-like phases that contained Ni–Mo–S, Co–Mo–S and Co–Ni–Mo–S by metathesis reactions between the corresponding metal salts (Ni(NO3)2.6H2O and CoCl2.6H2O) and (NH4)2MoS4 in formamide at room temperature according to equations (1), (2) and (3). ðNH4 Þ2 MoS4 þ NiðNO3 Þ26H2 O ! NiMoS4 þ 2NH4 NO3 þ 6H2 O ð1Þ ðNH4 Þ2 MoS4 þ CoCl26H2 O ! CoMoS4 þ 2NH4 Cl þ 6H2 O ð2Þ ðNH4 Þ2 MoS4 þ xCoCl26H2 O þ ð1  xÞNiðNO3 Þ26H2 O ! Cox Ni1x MoS4 þ 2xNH4 Cl þ 2ð1  xÞNH4 NO3 þ 6H2 O ð3Þ On mixing the precursor solutions, coherent monolithic black gels were obtained in 96 hours. These gels were named chalcogelNi-1, chalcogel-Co-1 and chalcogel-NiCo-1, based on the linking metal ions used. Elemental analysis data showed one metal ion per (MoS4)22 unit for both chalcogel-Ni-1 and chalcogel-Co-1. Even though an equimolar quantity of nickel and cobalt salts was used, chalcogel-NiCo-1 incorporated more cobalt than nickel. This might arise from the difference in reactivity of the precursor salts with (MoS4)22. However, the (Co,Ni)/(MoS4)22 ratio was 1:1 (Table 1). The gelation chemistry could also be extended to the tungsten analogues using a soluble (WS4)22 precursor and transition-metal ions. A representative example is the quaternary

1

Department of Chemistry, Northwestern University, Evanston, Illinois 60208, USA, 2 Department of Chemistry, Western Washington University, Bellingham, Washington 98225, USA. * e-mail: [email protected]

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

1

ARTICLES

NATURE CHEMISTRY

DOI: 10.1038/NCHEM.208

Table 1 | Elemental composition, density and nitrogen physisorption data for chalcogels. Chalcogel with nominal composition* (ICP† and XPS‡ analysis) Chalcogel-Ni-1 ICP: Ni0.89MoS3.88 XPS: Ni0.99MoS3.89 Chalcogel-Co-1 ICP: Co0.98MoS3.73 XPS: Co1.05MoS4.80 Chalcogel-NiCo-1 ICP: Ni0.37Co0.61MoS3.70 XPS: Ni0.55Co0.82MoS4.70 Chalcogel-Co-2 ICP: Co2.47MoW1.29S7.80 XPS: Co2.25MoW0.9S9

S BET k (m2 cm23)

Vp } (cm3 g21)

V N2 # (cm3 g21)

% microporosity**

1120

0.74

0.114

15.4

350, 374 (+15)

Bulk density, skeletal density (g cm23) 0.30, 3.20

340, 338 (+15)

0.19, 3.47

801

1180

0.79

0.120

15.2

528, 537 (+20)

0.20, 3.25

1242

1716

1.19

0.180

15.1

417, 428 (+20)

0.25, 3.83

1138

1597

0.81

0.150

18.5

SBET, SSAXS (m2 g21)

Silica equivalence surface area§ (m2 g21) 824

*Composition normalized per Mo atom. †Inductively coupled plasma analysis was performed by dissolving powder samples in aqua regia. ‡X-ray photoelectron spectroscopy was performed on ground powder samples on copper tape. §Per mol of chalcogel is converted into per mol of SiO2 using respective formula weights. kSBET in m2 g21 is multiplied by the respective skeletal density to give SBET in m2 cm23. } Adsorption total pore volume is measured at a relative pressure (P/P0) of 0.97. #Limiting micropore volume is obtained in the relative pressure region of 5  1026 to 2  1022. **Percentage microporosity is obtained by using the equation (VN2/Vp)  100.

a

b

50 μm

20 μm

c

d

e

f

g

50 μm

Figure 1 | SEM and photographic images of chalcogels. a–c, SEM images of chalcogel-Ni-1 (a), chalcogel-Co-1 (b) and chalcogel-NiCo-1 (c) reveal the spongy nature of these aerogel samples. d–g, Photographs of monolithic hydrogels and aerogels: chalcogel-Ni-1 (hydrogel) (d), chalcogel-Ni-1 (aerogel) (e), chalcogel-Co-1 (hydrogel) (f) and chalcogel-NiCo-1 (hydrogel) (g).

chalcogel-Co-2, in which Co2þ connects between (MoS4)22 and (WS4)22 (Table 1). An example of how these elements can bind to each other in the random networks is found in the molecular cluster (M(MoS4)2)22/32 (M ¼ Ni, Co and other transition-metal ions)8,9. These gels were subjected to supercritical drying with CO2 to give spongy aerogel monoliths (Fig. 1a–c). This process essentially replaces all the solvent molecules with air (hence aerogel), but keeps the gel network intact. Monolithic aerogels of about a centimetre in size can be prepared from the corresponding chalcogels 2

without significant volume loss (Fig. 1d–g), which suggests good rigidity of the interconnection between the secondary particles. It is presumed that primary particles, on the scale of nanometres in size, are formed by the reaction of the linking Co2þ or Ni2þ transitionmetal ions with the (MoS4)22 anion. These primary particles then aggregate to form larger secondary particles during gelation10. Encapsulation of solvent molecules (for example, formamide) within the secondary particles gives rise to larger pores (meso 2–50 nm, and macro .50 nm), and encapsulation in the primary

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY

ARTICLES

DOI: 10.1038/NCHEM.208

a

b

20 nm

50 nm

c

d

Mo L SK

Counts (a.u.)

Counts (a.u.)

Mo L SK

C

Co

C Cu Ni

Cu

Co O

O

Co 0

2

4 6 Energy (keV)

Ni Cu

Cu

Cu 8

Ni

10

0

2

4 6 Energy (keV)

8

10

Figure 2 | TEM images and STEM-XEDS spectra of aerogels. a,b, Typical TEM images of chalcogel-Co-1 (inset shows diffraction pattern) (a) and chalcogelNi-1 (b). c,d, STEM-XEDS spectra of the spot (white circles in a and b) analyses for chalcogel-Co-1 (c) and chalcogel-Ni-1 (d) (peaks of Cu are from the TEM grid).

particles produces smaller pores (microsize, ,2 nm). A homogeneous sponge-like aggregation of globular objects in the aerogel samples is evident in the scanning electron microscopy (SEM) images (Fig. 1a–c) of chalcogel-Ni-1, chalcogel-Co-1 and chalcogelNiCo-1. The bulk densities of chalcogel-Ni-1 and chalcogel-Co-1 are 0.30 g cm23 and 0.19 g cm23, respectively, consistent with the aerogel nature of these materials. The transmission electron microscopy (TEM) images of the aerogel samples clearly show the random porous agglomeration of nanoparticles (Fig. 2a,b). The images show pores in the mesoand macroregions between the secondary particles, whereas within the secondary particles microporosity was detected. Electron diffraction of these particles revealed diffuse scattering, which indicates their amorphous nature (Fig. 2a inset). Scanning TEM–X-ray energy-dispersive spectroscopy (STEM-XEDS) spot analysis (Fig. 2c,d) of these aerogel samples showed that Co, Ni and Mo–S coexist in a single phase with no evidence of phase segregation. The open porosity of these aerogel systems is revealed by their exceptionally high Brunauer–Emmett–Teller (BET)11 surface area, as determined by nitrogen physisorption and small-angle X-ray scattering (SAXS) measurements11,12. Nitrogen-adsorption isotherms gave BET surface area (SBET) values of 350 m2 g21, 340 m2 g21 and 528 m2 g21 for chalcogel-Ni-1, chalcogel-Co-1 and chalcogel-NiCo-1, respectively. To our knowledge, these are the highest known surface areas for any single Ni–Mo–S or Co–Mo–S phase (Table 1). The type-IV adsorption branch with a combination of H1- and H3-type hysteresis loops in the nitrogen

adsorption isotherms verifies that the secondary interparticle aggregation possesses mesoporosity. Barrett–Joynes–Halenda (BJH)11 plots of pore-size distribution suggest a broad range of pores, consistent with the aerogel nature (Fig. 3a,b and Supplementary Fig. S1). The intraparticle microporosity was determined from the Dubinin–Radushkevich equation in the low relative-pressure region (5  1026 to 2  1022) using limiting micropore volume (VN2) and found to be around 15% for all samples (Table 1 and Supplementary Fig. S2)11. The difference in void-fraction values calculated from total pore volume as determined by single-point adsorption near the saturation pressure and that measured from the monolithic aerogel samples validates the presence of macroporosity in these chalcogels (see Supplementary Information). The surface areas of the chalcogels described here are compared by converting the SBET values into their respective silica-equivalent ones (Table 1). The silica-equivalent SBET for chalcogel-NiCo-1 is 1,242 m2 g21. Surface-area values per unit volume give a better sense of the area of the surface folded in a tiny space (Table 1). Surface areas determined by SAXS experiments (Supplementary Fig. S3) from the Porod region (SSAXS) agree well with those obtained from the nitrogen physisorption data, within experimental error (Table 1)13. The SSAXS calculation is based on electron-density variation along the air–solid interface, and thus accounts for both open and closed porosity. The closeness of the SSAXS and SBET results suggests that almost all pores are accessible for the guest nitrogen molecules. The average particle size, mass fractality (Supplementary Fig. S4), average length of pore chord and

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

3

ARTICLES 600 500 400

b 1,200 0.035 0.030 0.025 0.020 0.015 0.010 0.005

300

BJH adsorption dV/dD pore volume

1

Pore volume (cm3 g−1 nm−1)

Volume adsorbed (cm3 g−1)

700

1,000 Volume adsorbed (cm3 g−1)

Pore volume (cm3 g−1 nm−1)

a

NATURE CHEMISTRY

10 100 Pore diameter (nm)

200

800

600

BJH adsorption dV/dD pore volume

0.05 0.04 0.03 0.02 0.01 0.00 1

400

10 100 Pore diameter (nm)

200

100

Adsorption Desorption

Adsorption Desorption

0

DOI: 10.1038/NCHEM.208

0 0.0

0.2

0.4 0.6 0.8 Relative pressure (P/P0)

c 10

1.0

0.0

d

(NH4)2 MoS4

2.18

0.2

0.6 0.8 0.4 Relative pressure (P/P0)

1.0

1.5 Chalcogel-Ni-1

2.37

5 3.54

Reduced PDF

Reduced PDF

1.0

7.03 6.24

0.5 3.54 4.84

0 0.0

−5

−0.5 5

10

20

15

5

r (Å)

10

15

20

r (Å)

e

f

30

ne (μmol g−1)

25

NHCOMe

20 15

NH

10

N

5

CO2Me

N HN CO2Me

0.0

0.5

1.0 1.5 Ce (μmol l−1)

CO2Me

2.0

Figure 3 | Nitrogen adsorption–desorption isotherms, PDF analysis and adsorption of porphyrin I on chalcogel. a,b Nitrogen adsorption–desorption isotherms of chalcogel-Ni-1 (a) and chalcogel-NiCo-1 (b) at 77 K (solid circles, adsorption data; open circles, desorption data; each inset shows pore-size (V–D) distribution plot calculated from the adsorption isotherm by the BJH method). c,d, Reduced atomic PDFs G(r) of (NH4)2MoS4 (c) and chalcogel-Ni-1 (d) as a function of interatomic distance r. e, Adsorption isotherm of porphyrin I on chalcogel-NiCo-1 at 25 8C (Ce is the equilibrium concentration of porphyrin I and ne is the amount of porphyrin I adsorbed; inset shows the structure of porphyrin I). f, Photographs of porphyrin I solution before and after adsorption on chalcogel-NiCo-1 and a control experiment with mesoporous SBA-15 (BET surface area of 900 m2 g21).

average pore diameters of these aerogels were calculated from SAXS data (Supplementary Table S1). The average pore diameter value of 24.6 nm obtained for chalcogel-Ni-1 further supports the mesoporous nature of these materials. The local structure of the chalcogel framework was probed by the atomic pair distribution function (PDF) technique14. Figure 3c,d shows the PDF plots as a function of interatomic distance (r) for 4

polycrystalline (NH4)2MoS4 and chalcogel-Ni-1. The strong correlation vector at 2.37 Å in chalcogel-Ni-1 corresponds to Mo–S or Ni–S bonds. By comparison, polycrystalline (NH4)2MoS4 exhibits the same correlation peak at 2.18 Å, consistent with the average crystallographic Mo–S bond distance15. The shift of the first correlation vector by 0.2 Å suggests a lengthening of the Mo–S bond, which indicates Mo reduction from þ6 to þ5 or þ4. Similar increases in

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY

ARTICLES

DOI: 10.1038/NCHEM.208

Table 2 | CO2 and H2 adsorption capacity of aerogels. H2 adsorption capacity† (mmol g21; vol %) 1.15; 8248 1.02; 7933 2.15; 15,535 1.30; 11,157

CO2 adsorption capacity* (mmol g21; vol %) 0.88; 6312 0.84; 6533 1.5; 10,839 1.4; 13,785

Chalcogel Chalcogel-Ni-1 Chalcogel-Co-1 Chalcogel-NiCo-1 Chalcogel-Co-2

Adsorption capacity as volume percentage was calculated based on unit volume per unit inorganic backbone (hence skeletal density) of aerogel sample. *CO2 adsorption capacity was measured at 258 K and atmospheric pressure. †H2 adsorption capacity was measured at 77 K and atmospheric pressure.

1.6

b Quantity CO2 adsorbed (mmol g−1)

Absolute pressure (mmHg)

a 1,000

100

10 Adsorption isotherm at 77 K Adsorption isotherm at 87 K 1 0.0

c

1.2 1.0 0.8 0.6 0.4 Adsorption isotherm at 258 K

0.2

Adsorption isotherm at 268 K 0.0

0.5

1.0 1.5 H2 loading (mmol g−1)

2.0

0

2.5

150

300

450

600

750

Absolute pressure (mmHg)

0.6

20

d

H2 adsorption CO2 adsorption

0.5

16 Selectivity (CO2 / H2)

Adsorbed amount (mmol g−1)

1.4

0.4 0.3 0.2

12

8

4

0.1 0.0

0 0

200

400 600 Pressure (mmHg)

800

0

100

200 300 Ptotal (mmHg)

400

500

Figure 4 | Adsorption isotherms of H2 and CO2 and measure of selectivity. a, H2-adsorption isotherms of chalcogel-NiCo-1 at 77 K and 87 K. b, CO2-adsorption isotherms of chalcogel-NiCo-1 at 258 K and 268 K. c, Single-component adsorption isotherms of CO2 and H2 at 273 K on chalcogel-Co-1. d, Selectivity of CO2 over H2 in chalcogel-Co-1, as predicted by the ideal adsorbed solution theory for equimolar mixtures of CO2 and H2 at 273 K (Ptotal is the total pressure of the gas mixture).

bond length were observed previously in Co2Mo2(m3-S)4-structured thiocubane16 and M(MoS4)22–/32 (M ¼ Zn, Ni, Pd, Pt and thiomolybdenyl, Mo==S) complexes9. The Mo–S bond length in the corresponding thiocubane complex is 2.32–2.36 Å, and that in (PPh4)2(Ni(MoS4)2) is 2.15 and 2.23 Å for the respective terminal and bridging Mo–S bonds. For (Mo2S12)22 and (Mo3S13)22 anions, the Mo–S bond length ranges from 2.37 to 2.47 Å and from 2.26 to 2.40 Å, respectively17,18. The correlation peak observed at 3.54 Å in both (NH4)2MoS4 and the aerogel sample is attributed to the nearest-neighbour S...S distance15. The presence of local order in chalcogel-Ni-1 is evident from the three well-defined correlation peaks in the PDF. The lack of correlation peaks above 6 Å in the aerogel sample indicates a random Ni–Mo–S network and the absence of any long-range translational symmetry. The binding energies, as measured by X-ray photoelectron spectroscopy (XPS), of the elements present in the chalcogel systems

studied here suggest the partial reduction of Mo from the þ6 to a þ5/þ4 oxidation state (Supplementary Table S2 and Figs S5– S10). In support of this, XPS data also showed the presence of polysulfide ions (Sx22), as did the infrared spectrum of chalcogel samples (Supplementary Fig. S11). The formation of polysulfide species on the reduction of Mo6þ to Mo5þ/4þ is well known in thiomolybdate chemistry9. Analysis of the peak intensities (after correction with sensitivity factors) of the elements in those aerogels matches well with the expected stoichiometry. The magnetic data for these aerogels showed that both chalcogel-Ni-1 and chalcogel-Co-1 are paramagnetic in nature (Supplementary Fig. S12). The effective magnetic moment of chalcogel-Ni-1 is only 1.72 mB per mol of aerogel and suggests that both square planar diamagnetic and octahedral–tetrahedral paramagnetic Ni2þ centres are present. Similarly, low-spin octahedral and square pyramidal Co2þ species in chalcogel-Co-1 are evident from the

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

5

ARTICLES

NATURE CHEMISTRY

6

a

12,000

10,000 Thiophene HDS activity (nmol Th per g catalyst s−1)

Chalcogel-Co-1 8,000

6,000

4,000

Sulfided Co−Mo/Al2O3

2,000

0 0

300

200

10

Pore volume (cm3 g−1 nm−1)

b

Volume adsorbed (cm3 g−1)

20

0.025

30 Time (h)

40

50

60

BJH adsorption dV/dD pore volume

0.020 0.015 0.010 0.005 0.000 1

100 10 Pore diameter (nm)

100 Adsorption Desorption 0 0.0

0.2

0.4 0.6 0.8 Relative pressure (P/P0)

1.0

c Co9S8 MoS2

Intensity (a.u.)

corresponding magnetic measurement of 2.06 mB per mol of aerogel (Supplementary Information). Thermogravimetric analysis of the supercritically dried chalcogel materials studied here showed 8–10% weight loss up to 160 8C (Supplementary Fig. S13). This might be caused by the evaporation of physisorbed or chemisorbed solvent. Above 180 8C, the weight loss became rapid and continued until 400 8C. Weight loss in this step arises from loss of sulfur in the inorganic framework of the chalcogel materials. However, a significant amount of surface area was retained even after heating at 350 8C. The BET surface area of chalcogel-Co-1, for example, changed to 75% and 66% of its original value after heating at 320 8C and 350 8C, respectively, in a nitrogen flow (Supplementary Fig. S14). The chalcogel materials of large surface area described here preferentially absorb Hg2þ from water. From 10 ml of mercurycontaminated (754.8 ppm) water, 10 mg of chalcogel-NiCo-1 removed Hg2þ to give a concentration of below 0.1 ppm within 48 hours. These aerogels also showed efficient adsorption of aromatic molecules, such as porphyrin I. From 16 ml of an ethanolic solution of porphyrin I (11.28 mmol l21), 5 mg of chalcogel-NiCo-1 removed almost all the porphyrin within 20 hours at room temperature to give a final concentration of 2.16 mmol l21. The adsorption isotherm of porphyrin I on chalcogel-NiCo-1 did not reach a plateau with the concentrations we used (the limited solubility of porphyrin I meant we were unable to obtain higher concentrations) and increased with increasing concentration values (Fig. 3e). This indicates the porous structures and polarizable surfaces of these spongy aerogels could also be used to remove organic toxins from industrial spills. As a comparison, control experiments with mesoporous silica SBA-15 (ref. 19) showed almost negligible porphyrin uptake (the final concentration measured was 10.74 mmol l21) (Fig. 3f and Supplementary Fig. S15). We also tested the absorption of CO2 and H2 by these chalcogel materials. Table 2 presents the adsorption capacities of different aerogels towards pure molecular CO2 and H2 at atmospheric pressure. Chalcogel-NiCo-1 shows the highest adsorption capacity of the aerogel samples. At 258 K, 1 g of this aerogel adsorbed almost 1.5 mmol of CO2 at atmospheric pressure, whereas its H2adsorption capacity at 77 K and atmospheric pressure was 2.15 mmol g21 (Fig. 4a,b). Given that the inorganic frameworks of these chalcogels are made of heavier elements (and hence a higher skeletal density), 1 litre of chalcogel-NiCo-1, for example, can hold up to 108 litres (0.4 mol per mol of aerogel) of CO2 at 258 K and 155 litres (0.6 mol per mol of aerogel) of H2 at 77 K and atmospheric pressure, respectively. Analysis of the H2-adsorption isotherms of chalcogel-NiCo-1 at two different temperatures (77 K and 88 K) showed 5.2 kJ mol21 (from 5.27 to 5.17 kJ mol21) isosteric heat of adsorption (Qs) up to 25 cm3 of H2 loading per gram of sorbent weight. The same aerogel’s isosteric heat of adsorption for CO2 at 258–268 K varied from 26.5 to 10.7 kJ mol21 depending on the amount of gas loading (Supplementary Fig. S16). At 273 K and atmospheric pressure, the CO2-adsorption capacity of these aerogels is significantly higher than that for H2 (Fig. 4c), which suggests these materials might effectively separate H2 from CO2 , an essential industrial process. A mixture of CO2 and H2 is produced in the water–gas shift reaction and thus it is desirable to separate CO2 to obtain clean H2. From the single-component CO2 and H2 isotherms and using the ideal adsorbed solution theory20 a high selectivity of 15 was obtained for CO2 over H2 with chalcogel-Co-1 at atmospheric pressure and 273 K (Fig. 4d). Detailed work to understand fully the gas-absorption behaviour is in progress. A preliminary investigation of the HDS catalytic properties of chalcogel-Co-1 was carried out and indicated that the chalcogel material is a highly promising catalyst. Thiophene-HDS activity data for chalcogel-Co-1 as well as a sulfided Co–Mo/Al2O3 catalyst (Co/Mo ¼ 1) are plotted in Fig. 5a. The chalcogel-Co-1 exhibited

DOI: 10.1038/NCHEM.208

Chalcogel-Co-1

Sulfided Co–Mo/Al2O3

20

30

40

50

60

70

80

Bragg angle (2θ)

Figure 5 | HDS-activity data for chalcogel-Co-1 as well as adsorption– desorption isotherms and XRD pattern for the tested material. a, Thiophene HDS-activity data for chalcogel-Co-1 (solid circles) and a sulfided Co–Mo/Al2O3 catalyst (open circles) at 643 K. b, Nitrogen adsorption–desorption isotherms of HDS-tested chalcogel-Co-1 at 77 K (solid circles, adsorption data; open circles, desorption data; inset shows pore-size distribution plot calculated from the adsorption isotherm by the BJH method). c, Powder XRD patterns for the HDS-tested chalcogel-Co-1 and sulfided Co–Mo/Al2O3 catalysts.

excellent stability over the 48-hour testing period and was over twice as active as the conventional sulfided Co–Mo/Al2O3 catalyst at the reaction temperature of 643 K. Given that the chalcogel material was essentially presulfided in an as-prepared form, similar HDS activities were measured for samples of chalcogel-Co-1 that had

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY

ARTICLES

DOI: 10.1038/NCHEM.208

been pretreated in flowing helium only or in an H2S/H2 mixture. After the HDS tests, samples of the chalcogel-Co-1 material were analysed by nitrogen physisorption (Fig. 5b) and powder X-ray diffraction (XRD) (Fig. 5c). The BET surface area of the HDStested chalcogel-Co-1 was 142 m2 g21 and the total pore volume was 0.5 cm3 g21, compared with 340 m2 g21 and 0.7 cm3 g21, respectively, for the freshly prepared material. Only a slight decrease in HDS activity was observed during the 48-hour test period, which suggests that the decrease in surface area occurred rapidly at the beginning of the HDS measurement. The powder XRD patterns for the tested chalcogel-Co-1 and the conventional sulfided Co– Mo/Al2O3 catalyst are compared in Fig. 5c. As indicated on the figure, XRD peaks can be assigned to MoS2 and Co9S8 crystallites, the thermodynamically stable sulfide phases under HDS conditions. Additional testing is underway to characterize more fully the HDS catalytic properties of these chalcogel materials. The Ni(Co)MoS combination of elements is relevant to the HDS process, which is the removal of organosulfur compounds from fossil fuels. On a global scale this is a process that all crude oil must go through. Cobalt- and nickel-promoted molybdenum sulfides are the catalysts of choice for this5,6. The rapidly increasing severity of environmental legislation to promote cleaner fuels means it is of paramount importance to design new porous materials that consist of nanostructured Ni–Mo–S and Co–Mo–S compositions21,22. The wedding of porosity with catalytically relevant components in a single-phase system can lead to better control and significant enhancement of their catalytic HDS properties. The chalcogels reported here are a new class of materials that have dual functionality of support materials (for example, large surface area) and an active inorganic component (for example, Ni(Co)MoS). Their continuous covalent networks with large open pores are widely accessible to large organic molecules. The experimental evidence presented here on the capture of heavy metal ions, adsorption of organic molecules and preferential uptake of gaseous species suggests these materials are promising for a broad set of applications such as separation processes, environmental remediation and catalysis. The generality of the synthetic procedure dictates that a number of other active materials that possess a high internal surface area could also be developed using soluble precursor and transition-metal ions.

Methods The starting metal salts Ni(NO3)2.6H2O, CoCl2.6H2O, (NH4)2WS4 , porphyrin I, ICP standards of the elements of interest and solvents, like formamide, used in this synthesis were purchased from Aldrich. The HgCl2 for heavy metal adsorption studies was bought from Alfa Aesar. Freshly prepared (NH4)2MoS4 was used for all syntheses. General procedure for chalcogel synthesis. All chalcogels reported here were synthesized in formamide in a glove box filled with inert nitrogen at room temperature. Solvents (including ethanol and water, which were used for the solventexchange process) were deoxygenated by bubbling nitrogen through them for about an hour before they went into the glove box. Chalcogel-Ni-1 and chalcogel-Co-1. To prepare chalcogel-Ni-1 and chalcogel-Co-1, a 1:1 molar ratio of building block to linking metal ion was used. In a typical preparation, 0.2 mmol of (NH4)2MoS4 (0.052 g) and Ni(NO3)2.6H2O (0.058 g) were each dissolved in 2 ml of formamide and shaken well. Then the nickel nitrate solution was added slowly to the (NH4)2MoS4 solution. The dark red solution of (NH4)2MoS4 turned black on the addition of a pale green solution of Ni(NO3)2.6H2O. This final solution was shaken well and kept on a plastic Petri dish for ageing (beware that the above mixture immediately forms precipitate if any of the precursor solution is even slightly warm). After one week, the resultant black gel was soaked in water for a day and the water drained out to remove formamide. This solvent-exchange process was repeated at least 3–4 times followed by 4–5 washings with ethanol over 3–4 days. Finally, well-formed monolithic hydrogels were cut into small pieces, stacked into a supercritical drying basket, placed in a closed container within the glove box and taken out for supercritical drying before use. After supercritical drying, aerogels were obtained in high yield (.90%) and immediately placed in a glove box before any surface characterization was made. The same procedure was also applied to the other chalcogel systems. The only exception was the

total volume of added solvent. Chalcogel-Co-1 was made using 0.2 mmol (0.047 g) of CoCl2.6H2O with 0.2 mmol (0.052 g) of (NH4)2MoS4 and total solvent volume was 5 ml instead of 4 ml. (Warning: CoCl2.6H2O salt reacts very quickly with MoS22 4 . During mixing of these two precursor solutions prolonged shaking was avoided.) Chalcogel-NiCo-1 and chalcogel-Co-2. To prepare chalcogel-NiCo-1, 0.2 mmol of Ni(NO3)2.6H2O and CoCl2.6H2O were dissolved together in 5 ml of formamide and added to 0.4 mmol of (NH4)2MoS4 dissolved in 5 ml of formamide. ChalcogelCo-2 was made by mixing solutions of (NH4)2MoS4 and (NH4)2WS4 (0.2 mmol each dissolved in 5 ml of formamide) and adding Co2þ (0.4 mmol of CoCl2.6H2O dissolved in 5 ml of formamide) solution to the mixture. Porphyrin I adsorption study. A solid sample of porphyrin I (C40H47N5O7) was dissolved in ethanol to make a stock solution (11.28 mmol l21). In a typical adsorption study, 5 mg of an aerogel sample was stirred at room temperature for 20 hours with 16 ml of the stock solution. After stirring, the ultraviolet–visible spectrum of the supernatant ethanol solution was checked to quantify the amount of porphyrin I adsorbed (Supplementary Fig. S15a). Six standard solutions in the same concentration range were prepared to make a calibration curve for quantitative analysis (Supplementary Fig. S15b). Heavy metal adsorption experiment. A stock solution of 755 ppm Hg2þ was prepared by accurately measuring the required amount of solid HgCl2 and dissolving it in deionized water. To ensure the correctness of the concentration of the stock solution, we analysed it by inductively coupled plasma atomic emission and optical emission spectroscopy. For the experiments on the adsorption of heavy metal ions, 10 ml of the stock solution was added to 10 mg of the chalcogel sample, stirred for 48 hours at room temperature, centrifuged and then the supernatant solution was analysed for the final metal ion concentration. Under identical conditions, a control experiment was performed for Hg2þ adsorption using non-porous polycrystalline MoS2. The result showed that 10 mg of MoS2 reduced the concentration of Hg2þ from 754.8 ppm to 646 ppm in 10 ml of solution. Thiophene HDS-activity measurements. Thiophene HDS-activity measurements were carried out using an atmospheric pressure flow reactor according to a method described previously23,24. Samples of the chalcogel-Co-1 (0.100 g) were pretreated in one of two ways: degassed in helium (60 ml min21) at room temperature for 30 minutes, or sulfidation by heating from room temperature to 673 K in 1 hour in a 60 ml min21 flow of 3 mol % H2S/H2 and then keeping at 673 K for 2 hours. The Co–Mo/Al2O3 catalyst (Co/Mo ¼ 1; 9.7 weight % CoO, 18.6 weight % MoO3) was pretreated only using the H2S/H2 procedure. After pretreatment, the gas flow was maintained at or switched to 60 ml min21 helium, the temperature was adjusted to the reaction temperature of 643 K and the flow was switched to the 3.2 mol% thiophene–hydrogen reactor feed (50 ml min21). The gas effluent was sampled at 1-hour intervals and the final measurement taken after 48 hours on stream. The total product peak areas from the chromatogram were used to calculate the thiophene HDS activities (nmol Th per g catalyst per s) for the catalysts.

Received 21 January 2009; accepted 3 April 2009; published online 17 May 2009

References 1. Hu¨sing, N. & Schubert, U. Aerogels – airy materials: chemistry, structure, and properties. Angew. Chem. Int. Ed. 37, 22–45 (1998). 2. Bag, S., Trikalitis, P. N., Chupas, P. J., Armatas, G. S. & Kanatzidis, M. G. Porous semiconducting gels and aerogels from chalcogenide clusters. Science 317, 490–493 (2007). 3. Bag, S., Arachchige, I. U. & Kanatzidis, M. G. Aerogels from metal chalcogenides and their emerging unique properties. J. Mater. Chem. 18, 3628–3632 (2008). 4. Mohanan, J. L., Arachchige, I. U. & Brock, S. L. Porous semiconductor chalcogenide aerogels. Science 307, 397–400 (2005). 5. Topsøe, H., Clausen, B. S. & Massoth, F. E. in Catalysis: Science and Technology (eds Anderson, J. R. & Boudard, M.) Vol. 11, 1–312 (Springer, 1996). 6. Prins, R., Debeer, V. H. J. & Somorjai, G. A. Structure and function of the catalyst and the promoter in Co–Mo hydrodesulfurization catalysts. Catal. Rev. Sci. Eng. 31, 1–41 (1989). 7. Chianelli, R. R., Daage, M. & Ledoux, M. J. Fundamental studies of transition metal sulfide catalytic materials. Adv. Catal. 40, 177–232 (1994). 8. Pan, W. H. et al. Syntheses and characterization of the cobalt bis(tetrathiomolybdate) trianion, Co(MoS4)32 2 . Inorg. Chim. Acta 97, 17–19 (1985). 9. Mu¨ller, A., Diemann, E., Jostes, R. & Bogge, H. Transition-metal thiometalates – properties and significance in complex and bioinorganic chemistry. Angew. Chem. Int. Ed. Engl. 20, 934–955 (1981). 10. Brinker, C. J. & Scherer, G. W. The Physics and Chemistry of Sol–Gel Processing (Academic Press, 1990). 11. Gregg, S. J. & Sing, K. S. W. Adsorption, Surface Area and Porosity (Academic Press, 1982).

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

7

ARTICLES

NATURE CHEMISTRY

12. Ehrburger-Dolle, F. et al. Nanoporous carbon materials: comparison between information obtained by SAXS and WAXS and by gas adsorption. Carbon 43, 3009–3012 (2005). 13. Faire´n-Jime´nez, D. et al. Surface area and microporosity of carbon aerogels from gas adsorption and small- and wide-angle X-ray scattering measurements. J. Phys. Chem. B 110, 8681–8688 (2006). 14. Billinge, S. J. L. & Kanatzidis, M. G. Beyond crystallography: the study of disorder, nanocrystallinity and crystallographically challenged materials with pair distribution functions. Chem. Commun. 749–760 (2004). 15. Lapasset, J., Chezeau, N. & Belougne, P. New refinement of crystal structure of ammonium thiomolybdate. Acta Crystallogr. B 32, 3087–3088 (1976). 16. Halbert, T. R., Cohen, S. A. & Stiefel, E. I. Construction of heterometallic ‘thiocubanes’ from M2S2(m-S)2 core complexes: synthesis of Co2M2S4(S2CNEt2)2(CH3CN)2(CO)2 (M ¼ Mo, W) and structure of the Co2Mo2(m3-S)4 cluster. Organometallics 4, 1689–1690 (1985). 17. Mu¨eller, A., Nolte, W. O. & Krebs, B. (NH4)2[(S2)2Mo(S2)2Mo(S2)2].2H2O, a novel sulfur-rich coordination compound with two nonequivalent complex anions having the same point group but different structures: crystal and molecular structures. Inorg. Chem. 19, 2835–2836 (1980). 18. Mu¨eller, A. et al. Crystal structure of (NH4)2MoS(S2)6 containing the novel isolated cluster (Mo3S13)22. Z. Naturforsch. B 34, 434–436 (1980). 19. Zhao, D. Y. et al. Triblock copolymer syntheses of mesoporous silica with periodic 50 to 300 angstrom pores. Science 279, 548–552 (1998). 20. Myers, A. L. Equation of state for adsorption of gases and their mixtures in porous materials. Adsorption 9, 9–16 (2003).

8

DOI: 10.1038/NCHEM.208

21. Song, C., Klein, M., Johnson, B. & Reynolds, J. Catalysis in fuel processing and environmental protection. An introduction. Catal. Today 50, 1 (1999). 22. Dhas, N. A., Ekhtiarzadeh, A. & Suslick, K. S. Sonochemical preparation of supported hydrodesulfurization catalysts. J. Am. Chem. Soc. 123, 8310–8316 (2001). 23. Sawhill, S. J., Phillips, D. C. & Bussell, M. E. Thiophene hydrodesulfurization over supported nickel phosphide catalysts. J. Catal. 215, 208–219 (2003). 24. Layman, K. A. & Bussell, M. E. Infrared spectroscopic investigation of CO adsorption on silica-supported nickel phosphide catalysts. J. Phys. Chem. B 108, 10930–10941 (2004).

Acknowledgements These studies were supported primarily by the Nanoscale Science and Engineering Initiative of the National Science Foundation, and the HDS studies were supported by the National Science Foundation. We are thankful to P. J. Chupas for collection of the PDF data.

Author contributions S.B. and M.G.K. designed and conducted the research, HDS experiments were performed by A.F.G. and M.E.B., and S.B. and M.G.K. wrote the paper.

Additional information Supplementary information accompanies this paper at www.nature.com/naturechemistry. Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/. Correspondence and requests for materials should be addressed to M.G.K.

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2009 Macmillan Publishers Limited. All rights reserved.

Supplementary information

Supporting Online Material for

doi: 10.1038/nchem.208

Spongy chalcogels of non-platinum metals act as effective hydrodesulphurization catalysts

Santanu Bag1, Amy F. Gaudette2, Mark E. Bussell2, Mercouri G. Kanatzidis1* 1 2

Department of Chemistry, Northwestern University, Evanston, IL 60208, USA, Department of Chemistry, Western Washington University, Bellingham, WA 98225 USA

*To whom correspondence should be addressed. E-mail: [email protected]

This PDF file includes: Method Characterization Figs. S1 to S16 References

1

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Method

Synthesis of (NH4)2MoS4. About 8 g of MoO3 was dissolved in 100 ml of conc NH4OH with slight heating and stirring. To this clear colorless solution H2S was bubbled continuously (Caution. H2S is very toxic gas and should be bubbled in well ventilated hood). The solution became orange and then increasingly deep red in color. After about 20 minutes huge amounts of dark orange red crystals began to deposit. The H2S was bubbled for another one and half hour to complete the reaction. Then the product was filtered, washed with ethanol and ether and dried under vacuum (Caution: The filtration should be done as soon as possible to avoid any surface oxidation). Shinny orange red needle shaped crystals were obtained with almost 90% yield (based on starting MoO3) and stored in N2 filled atmosphere. Characterization

Pair distribution function analysis. Diffraction experiments for pair distribution function analysis (PDF) were performed at the Advanced Photon Source (APS) located at Argonne National Laboratory, Argonne, Illinois (USA) using the high energy X-rays with the powder samples packed in 1 mm glass capillary. For data collection, an X-ray energy of 77.528 keV (h = 0.15992 Å) was used to record diffraction patterns to high values of momentum transfer while eliminating fluorescence from the sample. The two dimensional images were integrated within Fit 2D to obtain the one dimensional powder diffraction pattern, masking areas obscured by the beam stop arm. The PDFs, G(r) = 4ʌr[ȡ(r)ïȡo] where ȡ(r) and ȡo are the instantaneous and average densities, were extracted using PDFgetX2 (S1), subtracting the contributions from the sample environment and background to the measured diffraction intensities. Corrections for multiple scattering, X-ray polarization, sample absorption, and Compton scattering were then applied to obtain the structure function S(Q). Direct Fourier transform of the reduced structure function F(Q) = Q[S(Q) ï 1] yielded G(r), the pair distribution function as described previously (S2). Small angle X-ray scattering measurements. Small angle X-ray scattering (SAXS) was carried out with a Bruker Hi-star two-dimensional area detector (512X512 pixels) coupled with an

Elliott GX3 micro-focusing Cu rotating anode operated at 40kV and 13mA (0.01
2

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

custom built cell. The sample cell made of Kapton windowed (~4mm width) cell in ~40—m thick zirconium metal foil (1cmX2cm). The sample to detector distance and the center of the beam were precisely determined by calibration with Ag-behenate diffraction standard (d001=5.838 nm). The diffraction intensities of two-dimensional (2-D) collected images were integrated to yield one-dimensional (1-D) diffraction patterns as a function of the wave vector q with the Fit 2D program. The wave vector q is defined as q=4ʌ/Ȝ sin(ș), where 2ș is the scattering angle. Scattering data were corrected for dark current and empty cell scattering. Supercritical drying. Critical point drying of the chalcogels was done with a Bal-Tec CPD 030 (Balzers) instrument. Before placing the gel filled custom-built drying basket into the CPD chamber, the chamber was purged a couple of times with N2. In this supercritical drier, the sample was soaked with liquid carbon dioxide and flushed 5 to 6 times over a period of 4 hours at 10ºC to completely exchange the ethanol from the material. Depending on the amount of gel, the soaking time and exchange time vary. Finally, a black aerogel was obtained after supercritical drying at 40ºC. (Caution. During supercritical drying, CO2 gas should be vented off very slowly. Rapid CO2 ventilation would cause collapse of pore structure). Density measurements. The skeletal densities of the aerogel materials were measured by a Micromeritics AccuPyc 1340 gas pycnometer (1cc model) using ultra high purity (UHP) helium gas. About 200 mg sample was taken for measurement. Analysis run cycles were continued until a standard deviation of ±0.0002 cm3 was obtained for the sample volume. Bulk densities were measured by cutting a rectangular shaped monolithic aerogel and measuring its dimension by slide calipers and subsequently weighing out the aerogel sample in electronic balance (accurate upto 5 decimal place). Typical monolithic aerogel sample dimensions are 0.7 cm X 0.6 cm X 0.1 cm. At least two different samples for the same aerogel were measured.

Nitrogen physisorption measurements. Nitrogen adsorption and desorption isotherms were measured at 77 K on a Micromeritics ASAP 2020 system. For each measurement, about 200 mg of samples were taken. Before measurement, samples were degassed at 348 K under vacuum (<10-4 mbar) for overnight. Low pressure incremental dosing of 3 cc/g STP and 45 s

equilibration were applied as analysis conditions. BET transform plot was obtained in the 0.05 to 0.3 relative pressure (P/Po) regions and correlation coefficient of 0.99999 was obtained in every case.

3

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

SUPPLEMENTARY INFORMATION

doi: 10.1038/nchem.208

Hydrogen and carbon dioxide physisorption measurements. All physisorption measurements were performed on a Micromeritics ASAP 2020 system. Hydrogen adsorption at 77 K and 87 K was measured using liquid nitrogen and liquid argon respectively. Cryogenic water bath (50:50 vol% water-ethylene glycol mixture) controlled by NESLAB RTE10 Digital Plus (Thermo Electron Corporation) chiller system was used for carbon dioxide measurement. After free space measurement with He gas, sample tube was evacuated manually at 348 K for 4 hours to remove any trapped He. Free space was measured separately for each temperature studied. For hydrogen adsorption low pressure incremental dosing of 1 cc/g STP and 120 s equilibration were employed. The respective parameters for carbon dioxide adsorption were 0.8 cc/g STP and 120 s. XPS analysis. X-ray photoelectron spectroscopy was acquired on an Omicron ESCA Probe (Taunusstein, Germany) equipped with an Aluminum KĮ X-ray source. Samples were analyzed at pressures between 10-9 and 10-8 torr with a pass energy of 29.35 eV and a take-off angle of

45°. All peaks were referenced to the signature C1s peak for adventitious carbon at 284.6 eV to account for the charging effects. Powder aerogel samples were mounted on copper tape on aluminum sample holder and placed in the pretreatment chamber of the spectrometer and then outgassed overnight before transfer to the analysis chamber. The areas of the peaks were computed after fitting of the experimental spectra to Gaussian/Lorentzian curves and removal of the background (Shirley function). Surface atomic ratios were calculated from the peak area ratios normalized by the corresponding atomic sensitivity factors. Instrument atomic sensitivity factors were recalibrated for Mo, S, Ni and Co using powdered MoS2, MoO3, Ni(NO3)2.6H2O and Co(NO3)2.6H2O samples. Infrared spectroscopy. FT-IR spectra were recorded on a Nicolet 750 Magna-IR series II spectrometer with 2 cm-1 resolution.

SEM images. Scanning electron microscopy images of the aerogel samples were taken with Hitachi S-3400N VP-SEM. Powdered aerogel samples were gently placed on carbon tape and taken in instrument chamber for image capture. TEM images. TEM samples were prepared by suspending the aerogel sample in ether and then casting on holey carbon coated Cu grid. High-resolution transmission electron micrograph (TEM) was obtained with a JEOL 2100F instrument (field emission) operating at 200 kV.

4

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

SUPPLEMENTARY INFORMATION

doi: 10.1038/nchem.208

Magnetic Measurements. Magnetic susceptibility measurements were made on powder samples using a Quantum Design MPMS SQUID magnetometer. Temperature dependent susceptibility measurements were made with an applied field of 2000 Oe from 2 to 300K with both field cooled and zero field cooled modes. Inductively Coupled Plasma-Atomic Emission (Optical Emission) [ICP-AES(OES)]

analysis. Accurate determinations of Mo, Ni, Co and S concentrations and Hg2+ ion were performed by ICP-AES using a VISTA MPX CCD SIMILTANEOUS ICP-OES instrument. Standards of the elements of interest were prepared by diluting commercial (Aldrich or GFS chemicals) 1000 ppm ICP-standards of these elements. Ten calibration standards from 1 ppm to 12 ppm were made. The calibration was linear with errors around 3%. The samples were also diluted before the measurements, so that their concentrations can fall within the range of calibration. The ICP-AES intensity was the result of three (30 seconds) exposures. For each sample, three readings of the ICP-AES intensity were recorded and averaged. The standards

were reanalyzed after each analysis of the samples. To help stabilization of Hg2+ in solution and to avoid contamination of the plasma by trace mercury amounts, solution of Au (of about 10 times higher concentration than Hg) was added to the standards and Hg-containing samples. To determine the elemental compositions of chalcogel samples, powder aerogels were first suspended in water and then aqua-regia was added slowly to the samples in closed volumetric flasks which were kept overnight for sonication until all solid was dissolved. These samples were diluted before the measurements, so that their concentrations can fall within the range of calibration. For each chalcogel, samples from three different lots were analyzed independently to have better statistics of results. Gas selectivity using the Ideal Adsorbed Solution Theory (IAST) approach (S3). The single-component adsorption isotherms were described by fitting the data with the following virial-type equation (S3): p=

v exp(c1v+c 2 v 2 +c3 v3 +c 4 v 4 ) K

(1)

Where p is the pressure in Torr, v is the adsorbed amount in mmolg-1, K is the Henry constant in

mmolg-1Torr-1 and ci are the constants of the virial equation.

5

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

SUPPLEMENTARY INFORMATION

doi: 10.1038/nchem.208

The free energy of desorption at a given value of temperature and pressure of the gas is obtained from the analytical integration of eq. (1): p

n 1 2 3 4 G(T,p)=RT ³ dp=RT(v+ c1v2 + c2 v3 + c3 v 4 + c 4 v5 ) p 2 3 4 5 0

(2)

The free energy of desorption is a function of temperature and pressure G(T, p) and describes the minimum work (Gibbs free energy) that required to completely degas the adsorbant surface. For

a binary mixture of component i and j the eq. (2) yields the individual pure loadings vi0 and vj0 at the same free energy of desorption:

G i0 (vi0 )=G 0j (v0j )

(3)

The partial pressure of component i and j in an ideal adsorption mixture given: py i = p i0 (v i0 )x i

(4)

py j = p0j (v0j )x j

(5)

where yi (=1-yj) and xi (=1-xj) is the molar fraction of component I in the gas phase and the

adsorbed phase respectively and pi0, pj0 is the pure component pressure of i and j, respectively.

Having solved the eq. (3)-(5) and eq. (1), the selectivity for the adsorbates i and j (si,j) and the total pressure (p) of the gas mixture were obtained from eq. (6) and eq. (7), respectively. 0 x i /yi p j Si,j = = x j /y j pi0 j

p= ¦ pi0 x i i

(6) (7)

6

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Nitrogen Physisorption Data & Pore Size Distribution Plots of Chalcogels.

Figure S1. Nitrogen adsorption desorption isotherms (77K) of (a) Chalcogel-Co-1 and (c) Chalcogel-Co-2. (b, d) Respective pore size distribution plots of these chalcogels calculated from adsorption isotherm by the BJH (Kruk-Jaroniec-Sayari thickness curve and Faas Correction) method.

7

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Dubinin Radushkevich (DR) Plots for Microporosity Calculation.

Figure S2. Dubinin Radushkevich (DR) plots of (a) Chalcogel-Ni-1, (b) Chalcogel-Co-1, (c) Chalcogel-NiCo-1 and (d) Chalcogel-Co-2 in the low relative pressure region (5X10-6 to 2X10-2).

Calculation of Void Fraction from Nitrogen Physisorption Measurement. The fraction of void space (ij) in aerogel samples were calculated using ij = (Vo.ȡs)/(1+Vo.ȡs); where Vo is the pore volume near saturation pressure and ȡs is the skeletal density and compared with ij’ [=1-(ȡ / ȡs)] (where ȡ =bulk density) calculated from measured bulk density of aerogel sample.

8

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

SUPPLEMENTARY INFORMATION

doi: 10.1038/nchem.208

Small Angle X-Ray Scattering (SAXS). The SAXS curve usually displays fractal behavior as expressed by the intensity (I(q)) vs scattering vector, q plot according to the equation, I(q)= k. q -_ where q=4ʌsinș/Ȝ, ș is the scattering angle, Ȝ is the wavelength, k is a constant, Į is related to the fractal dimension. It follows Porod’s law at a large value of the q vector and a limiting behavior is observed: lim q A 'I(q)=K/q 4 (K is Porod constant) (S4, S5). The specific surface area (SSAXS) is then determined using the method of the invariant (where b is the background intensity) Q=

q max K 2 ³ (I(q) < b)q dq + q max q min

§ · § K · K ¸ ¸¸ = ʌ u ij u ¨¨ S = ʌ u ij u (1 < ij)¨¨ ¸ ¨ ȡ uQ ¸ © ȡuQ ¹ © s ¹

(8) (9)

where the Porod constant K=limq A ' {I(q)q 4} , ij is the volume fraction of framework voids, ȡ, ȡs are the bulk and skeletal mass density respectively (where ȡ/ȡs=1-ij), and Q is the invariant

parameter defined as the q2 weighted integrated intensity I(q) of the scattering curve according to eq. 8. The Q value was determined from eq (8) where the upper limit, qmax was evaluated by arbitrarily choosing a q value and fitting with

' K 2 ³ (I(q) < b)q dq ~ (S4, S5). q min q max

Average chord length (Ɛp) (S6) of the pores of the aerogel samples were determined by Ɛp=Ɛ/(1ij) (where Ɛ=4Q/ʌK and (1-ij) is the volume fraction of solid part) and listed in Table S1. Average pore diameters (dp) were calculated by plotting I(q)-1/2 vs ș2(in radians) (S7) and using eq. 10.

dp =

3a 2(1-q )

(10)

Where ‘a’ can be obtained from the ratio of the slope to intercept in the I(q)-1/2 vs ș2(in radians) plot.

9

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Figure S3. Small angle X-ray scattering I(q) vs q graphs of (a) Chalcogel-Ni-1 and (b) Chalcogel-Co-1. Determinations of their Porod’s constant value (K) are shown in (c) and (d).

10

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

SUPPLEMENTARY INFORMATION

doi: 10.1038/nchem.208

Figure S4. Fractal behaviors of (a) Chalcogel-Ni-1 and (b) Chalcogel-Co-1 are shown. Table S1. Average particle size (Dparticle), mass fractality, average chord-length (Ɛ), average pore diameter (dp) and tortuosity calculations from SAXS data.

Chalcogel

Dparticle (nm)

Mass Fractality

Chalcogel-Ni-1

8.8 to 21.7 12.9 to 23.0 5.2 to 20.2 6.5 to 13.4

2.85 & 1.14 2.60 & 2.10 2.45 & 1.85 2.60 & 1.33

Chalcogel-Co-1 Chalcogel-NiCo-1 Chalcogel-Co-2 [a] [b]

Ɛp=Ɛ/(1-ij) (nm) [a]

Ɛs=Ɛ/ij (nm) [b]

dp (nm)

Tortuosity factor

32.2

3.34

24.6

1.31

3.24

58.9

3.43

45.8

1.29

2.17

35.0

2.31

26.9

1.30

2.27

34.9

2.43

27.1

1.29

Ɛ=

(nm) 3.03

Average chord-length of pore Average chord-length of solid part

11

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

X-Ray Photoelectron Spectra (XPS) of Chalcogel Samples.

The electronic states and possible coordination environments of the elements present in the title chalcogel systems were probed with X-ray photoelectron spectroscopy and the measured binding energies are listed in Table S2. The supercritically dried Chalcogel-Ni-1, Chalcogel-Co1, Chalcogel-NiCo-1 and Chalcogel-Co-2 showed spin coupled Mo doublets in the 3d(Mo)-2s(S) region (Fig. S5a, S7a, S8a and S10a). Peaks at 227.1 eV, 226.8 eV, 227.0 eV and 226.7 eV of respective Chalcogel-Ni-1, Chalcogel-Co-1, Chalcogel-NiCo-1 and Chalcogel-Co-2 arise from S 2s binding energy; characteristic of S2-. Deconvolution of 3d peaks reveals at least two different

molybdenum species are present. The Mo 3d5/2 peaks at 229.6 eV and 230.2 eV of Chalcogel-Ni1suggest partial reduction of Mo from the +6 to a +5/+4 state. Similar reduction of formal positive charge of Mo is also observed in Co/Ni supported MoS2 (S8). The Mo 3d binding energy values obtained in these aerogels fall between +4 and +6 oxidation states in a sulfur environment (S9). The tiny shoulder peak at 236.1 eV which is about 6.6-7.0% of the entire Mo peak might be from some Mo (+6) environment (S10). The 2p3/2 binding energy of Ni in Chalcogel-Ni-1 is significantly higher than reported literature values for nickel sulfides (S11) which suggests charge transfer from Ni to Mo centre. However, the formation of any species like NiSO4 or any unreacted Ni(NO3)2.6H2O was ruled out because of the absence of any trace peak at 856.9 eV (Fig. S5d) (S11). The O 1s peak observed at 531.0 eV and 532.3 eV (Fig. S6a) in Chalcogel-Ni-1 confirms that no surface oxidation occurred; instead incorporation of solvents like formamide/ethanol might give those peaks (S12, S13). MoO3, NiO, Ni2O3 have characteristic O 1s peaks at 530.4 eV, 529.6 eV and

530.0 eV respectively. It is more probable that vacant coordination sides of Ni2+/Co2+ are

occupied by solvent molecules during their synthesis and solvent exchange process and they form a part of the frameworks. C 1s peak observed in Chalcogel-Co-1 (Fig. S6c) confirms presence of solvent molecules. In addition to the peak at 284.6 eV, characteristic peaks of –C=O and –C-N-C at 286.4 and 288.8 eV were detected (S14). This is also consistent with the IR spectroscopic observation. Sulfur 2p peaks (2p3/2 and 2p1/2), which are generally spin coupled with ǻE of 1.2 eV, were observed at 161.9 eV and 163.0 eV for Chalcogel-Ni-1 (Fig. S5b).

These values are consistent with Ni2+ metal ion binding to sulfur . Typical S 2p3/2 for NiS has

binding energy of 161.5 eV. Due to delocalization of negative charge from sulfide to Mo in NiS-Mo, the value increased little bit. A tiny additional S 2p3/2 peak at 164.1 eV was built up might

12

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

be due to some Sx2-. The possibility of arial oxidation of sulfide to sulfate was excluded due to absence of any 2p3/2 sulfate peak at around 168.7 eV (S15). Analysis of the Mo (3d) and S (2p) peak intensities (after correction with sensitivity factors) of Chalcogel-Ni-1 agrees a S/Mo

atomic ratio of 3.89, very close to that in the starting [MoS4]2- unit. Ni/Mo atomic ratio of 0.99 obtained from XPS data matches well with the starting stoichiometry. Similar XPS results were obtained for Chalcogel-Co-1. Co 2p3/2 peak at 779.6 eV indicates charge transfer from Co centre (Fig. S7c,d). This value is higher than the conventional Co-Mo-S HDS catalyst suggesting greater interaction with Mo centre (S9). Determination of the spin-orbit splitting (ǻE2p1/2-2p3/2) parameter value of 15.0 eV indicates more of diamagnetic cobaltic compounds compared to 16.0 eV for starting cobaltous CoCl2.6H2O (S10). Nevertheless, magnetic susceptibility data confirms paramagnetic behavior and hence charge transfer proposition is further validated. Co/Mo ratio of 1.05 obtained from Co (2p3/2) and Mo (3d) XPS peak intensities is also in agreement with complete metathesis reaction.

13

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Figure S5. X-ray photoelectron spectra of Chalcogel-Ni-1. (a) Mo(3d)-S(2s) region, (b) S 2p region, (c) Ni whole 2p region, (d) Deconvoluted Ni 2p3/2 peak.

14

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Figure S6. X-ray photoelectron spectra of (a) Chalcogel-Ni-1 and (b) Chalcogel-Co-1 in O 1s region. (c) XPS of Chalcogel-Co-1 in C 1s region.

15

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Figure S7. X-ray photoelectron spectra of Chalcogel-Co-1. (a) Mo(3d)-S(2s) region, (b) S 2p region, (c) Co whole 2p region, (d) Deconvoluted Co 2p3/2 peak.

16

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Figure S8. X-ray photoelectron spectra of Chalcogel-NiCo-1. (a) Mo(3d)-S(2s) region, (b) S 2p region, (c) Ni whole 2p region, (d) Deconvoluted Ni 2p3/2 peak.

Figure S9. X-ray photoelectron spectra of Chalcogel-NiCo-1. (a) Co whole 2p region, (b) Deconvoluted Co 2p3/2 peak.

17

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Figure S10. X-ray photoelectron spectra of Chalcogel-Co-2. (a) Mo(3d)-S(2s) region, (b) S 2p region, (c) Co whole 2p region and (d) O 1s region. (Due to overlapping with Mo 4p peaks, no W 4f peak is clearly identified).

18

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

SUPPLEMENTARY INFORMATION

doi: 10.1038/nchem.208

Table S2. XPS Binding energy values for supercritically dried chalcogel samples.

Elements Mo[a] S

3d5/2 3d3/2 2s 2p

Chalcogel-Co-1

Chalcogel-NiCo-1

Chalcogel-Co-2

229.6, 230.2

229.3, 230.0

229.5, 230.0

229.3, 230.1

232.8, 233.5 227.1 161.9, 163.0, 164.1 854.4

232.5, 233.1 226.8 162.0, 163.1, 164.0 779.6

232.7, 233.3 227.0 162.3, 163.4, 164.0

232.6, 233.4 226.7 161.8, 162.9, 164.0 779.6

854.5 (Ni), 779.9 (Co) 2p1/2 871.7 794.6 872.0 (Ni), 794.9 794.6 (Co) ¨E 17.3 15.0 17.5 (Ni), 15.0 15.0 (Co) [a] Due to high sample charging, XPS peaks of (NH4)2MoS4 became broader and only an average peak was observed. Thus direct comparison with (NH4)2MoS4 cannot be made. M (Ni, Co)

2p3/2

Chalcogel-Ni-1

Infrared Spectroscopy.

Vibrational frequencies of the aerogel samples showed that the characteristic peaks from the starting building block were retained. Due to completely amorphous nature of the inorganic framework and a number of different binding modes present in those samples, the peaks became much broader. Peaks in the region of 400 to 500 cm-1 are due to Mo-S vibrational modes (S16).

A strong adsorption at 534 cm-1 might be due to S-S stretching mode of some S22- environments present in those systems (S16).

Figure S11. Comparison of vibrational frequencies of Chalcogel-Ni-1 (red line) with those of starting (NH4)2MoS4 (black line).

19

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Magnetic Properties.

The magnetic properties of these aerogels were measured from 3K to 300K to estimate the chemical states of the magnetically active metal centres and their mutual interactions. The data for Chalcogel-Co-1obey Curie-Weiss law in the 80K-300K range from which —eff of 2.06 —B (with Weiss constant ș of 27.64 K) was obtained per mol of aerogel. This value is close to one unpaired electron only effective magnetic moment of 1.73 —B suggesting low spin octahedral or square pyramidal Co2+ species.

The Mo atoms some of which are in a oxidation state of +5 also contribute to —eff resulting in an increased value. 1/Ȥm vs T plot of Chalcogel-Ni-1 in the 99K-260K temperature range (zero field cooled data) follows Curie-Weiss law from which —eff of 1.61 —B was obtained per mol of aerogel. The same plot in the range of 2K-60K temperature region yields —eff of 1.72 —B per mol of aerogel. This value is close to one electron spin only effective magnetic moment of 1.73 —B .

Given that square planar Ni2+ (d8) is diamagnetic and both tetrahedral and octahedral Ni2+ centre contains two unpaired electrons, the one electron spin only effective magnetic moment suggests mixture of square planar diamagnetic and octahedral or tetrahedral paramagnetic centers are present simultaneously.

Figure S12. Magnetic susceptibility data of (a) Chalcogel-Co-1 and (b) Chalcogel-Ni-1 as a function of temperature (applied field 2000 Oe).

20

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Thermal Stability.

Figure S13. TGA curves of aerogel samples under nitrogen flow. Temperature dependent surface area analysis.

Figure S14. Nitrogen adsorption isotherms of supercritically dried Chalcogel-Co-1 heated under N2 at different temperatures.

21

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Porphyrin I Adsorption Study on Chalcogel-NiCo-1.

Figure S15. (a) UV-Vis spectra of ethanolic porphyrin I solution before (red line) and after adsorption on SBA-15(green line) and on Chalcogel-NiCo-1(blue line). (b) Calibration curve used for quantification of porphyrin I.

22

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

Heat of adsorptions of hydrogen and carbon dioxide on Chalcogel-NiCo-1.

Figure S16. Isosteric heat of adsorptions of (a) hydrogen and (b) carbon dioxide on ChalcogelNiCo-1.

23

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.

doi: 10.1038/nchem.208

SUPPLEMENTARY INFORMATION

References. S1. Qiu, X., Thompson, J. W. & Billinge, S. J. L. J. Appl. Cryst. 37, 678 (2004). S2. Egami, T. & Billinge, S. J. L. Underneath the Bragg Peaks: Structural Analysis of Complex Materials (Pergamon Press, Amsterdam, 2003).

S3. Myers, A. L. Adsorption 9, 9 (2003). S4. Ehrburger-Dolle, F. et al. Carbon 43, 3009 (2005). S5. Fairen-Jimenez, D. et al. J. Phys. Chem. B 110, 8681(2006). S6. Dieudonne, P., Delord, P. & Phalippou, J. J. Non-Cryst. Solids 225, 220 (1998). S7. Longman, G. W., Wignall, G. D., Hemming, M. & Dawkins, J. V. Colloid. Polym. Sci. 252, 298 (1974). S8. Harris, S. & Chianelli, R. R. J. Catal. 98, 17 (1986). S9. Dhas, N. A., Ekhtiarzadeh, A. & Suslick, K. S. J. Am. Chem. Soc. 123, 8310 (2001). S10. Chin, R. L. & Hercules, D. M. J. Phys. Chem. 86, 3079 (1982). S11. Cid, R., Atanasova, P., Cordero, R. L., Palacios, J. M. & Agudo, A. L. J. Catal. 182, 328 (1999). S12. Jones, C. & Sammann, E. Carbon 28, 509 (1990). S13. Fierro, G., Ingo, G. M. & Mancia, F. Corrosion 45, 814 (1989). S14. Ramanathan, T., Fisher, F. T., Ruoff, R. S. & Brinson, L. C. Chem. Mater. 17, 1290 (2005). S15. Laurent, E. & Delmon, B. J. Catal. 146, 281 (1994). S16. Weber, Th., Muijsers, J. C. & Niemantsverdriet, J. W. J. Phys. Chem. 99, 9194 (1995).

24

nature chemistry | www.nature.com/naturechemistry

© 2009 Macmillan Publishers Limited. All rights reserved.