Supplementary Figure 1. Different views of the experimental setup at the ESRF beamline ID15B involving the modified MM200 Retsch mill: (left) side-on

Transcription

1 Supplementary Figure 1. Different views of the experimental setup at the ESRF beamline ID15B involving the modified MM200 Retsch mill: (left) side-on and (right) almost parallel to the incident beam.

2 Supplementary Figure 2. Corresponding to reaction monitoring depicted in Figure 2a of the main manuscript. Combined Pawley and Rietveld and fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 32 μl of aqueous solution of acetic acid as the grinding liquid (c = 2.5 M), after 1 minute and 45 seconds grinding. Observed crystalline phases are ZnO and ZIF-8. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.35%, R wp = 0.20 %, R p = 0.12 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 4.41 %, R p (-bgr) = 5.58 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

3 Supplementary Figure 3. Corresponding to reaction monitoring depicted in Figure 2a of the main manuscript. Combined Pawley and Rietveld and fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 32 μl of aqueous solution of acetic acid as the grinding liquid (c = 2.5 M), after 44 minutes and 24 seconds grinding. Observed crystalline phase is ZnO. The tick marks represent calculated reflection positions for the crystalline phase. R exp = 1.51%, R wp = 0.08 %, R p = 0.07 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) =17.86 %, R p (-bgr)= %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

4 Supplementary Figure 4. Corresponding to reaction monitoring depicted in Figure 2b of the main manuscript. Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 32 μl of aqueous solution of acetic acid (c = 1.25 M), after 6 minutes and 6 seconds grinding. Observed crystalline phases are ZnO and ZIF-8. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.30%, R wp = 0.12 %, R p = 0.09 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) =3.15 %, R p (-bgr)= 3.79 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

5 Supplementary Figure 5. Corresponding to reaction monitoring depicted in Figure 2b of the main manuscript. Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 32 μl of aqueous acetic acid (c = 1.25 M), after 42 minutes and 41 seconds grinding. Observed crystalline phase is ZnO. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.36%, R wp = 0.13 %, R p = 0.10 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 26.9 %, R p (-bgr)= 50.9 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

6 Supplementary Figure 6. Corresponding to reaction monitoring depicted in Figure 2c of the main manuscript Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 64 μl of aqueous acetic acid (c = 1.25 M) and 50 mg of of crystalline silicon as an internal diffraction standard, after 5 minutes and 4 seconds of grinding. Observed crystalline phases are ZnO and ZIF-8. The tick marks represent calculated reflection positions for the crystalline phases phases. Agreement factors: R exp = 1.28%, R wp = 0.17 %, R p = 0.13 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 2.70 %, R p (-bgr)= 3.32 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

7 Supplementary Figure 7. Corresponding to reaction monitoring depicted in Figure 2c of the main manuscript Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 64 μl of the aqueous solution of acetic acid as the grinding liquid (c = 1.25 M), after 50 minutes and 1 second of grinding. Observed crystalline phases are ZnO and ZIF-8. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.30%, R wp = 0.10 %, R p = 0.08 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 6.13 %, R p (-bgr)= 7.42 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

8 Supplementary Figure 8. Time-resolved X-ray diffractogram for the LAG reaction of ZnO and HMeIm in the presence of water as the grinding liquid, revealing amorphisation of initially formed ZIF-8 and crystallisation into dia-zn(meim) 2.

9 Supplementary Figure 9. Corresponding to reaction monitoring depicted in Figure 4a of the main manuscript. Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 40 μl of aqueous solution of acetic acid as the grinding liquid (c = 2.5 M) and 50 mg of crystalline silicon as an internal diffraction standard, after 5 min grinding. Observed crystalline phases are ZnO, silicon and ZIF-8. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.23%, R wp = 0.19 %, R p = 0.12 %, gof = 0.15 For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 5.83 %, R p (-bgr)= 5.86 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

10 Supplementary Figure 10. Corresponding to reaction monitoring depicted in Figure 4a of the main manuscript. Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 40 μl of aqueous acetic acid (c = 2.5 M) and 50 mg of crystalline silicon as an internal diffraction standard, after 35 minutes of grinding. Observed crystalline phases are ZnO, kat-zn(meim) 2 and silicon. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.35%, R wp = 0.27 %, R p = 0.19 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 7.71 %, R p (-bgr)= 9.90 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

11 Supplementary Figure 11. Corresponding to reaction monitoring depicted in Figure 4a of the main manuscript. Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 40 μl of aqueous acetic acid (c = 2.5 M) and 50 mg of crystalline silicon as an internal diffraction standard, after 60 minutes of grinding. Observed crystalline phases are dia-zn(meim) 2 and silicon. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.36%, R wp = 0.20 %, R p = 0.13 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 5.28 %, R p (-bgr)= 5.24 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

12 Supplementary Figure 12. Corresponding to reaction monitoring depicted in Figure 4b of the main manuscript. Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 40 μl of aqueous acetic acid (c = 2.5 M) and 50 mg of crystalline silicon as an internal diffraction standard, after 5 minutes of grinding. Observed crystalline phases are ZIF-8, ZnO and silicon. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.32%, R wp = 0.20 %, R p = 0.13 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 5.61 %, R p (-bgr)= 5.76 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

13 Supplementary Figure 13. Corresponding to reaction monitoring depicted in Figure 4b of the main manuscript. Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 40 μl of aqueous acetic acid (c = 2.5 M 3 ) and 50 mg of crystalline silicon as an internal diffraction standard, after 40 minutes of grinding. Observed crystalline phases are ZnO, dia-zn(meim) 2 and silicon. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.41%, R wp = 0.21 %, R p = 0.13 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 7.30 %, R p (-bgr)= 7.72 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

14 Supplementary Figure 14. Time-resolved diffractogram for LAG conversion of ZnO and HMeIm in the presence of 2.5 M acetic acid without added silicon, demonstrating the formation of the new phase (kat) at ca. 50 minutes milling. Calculated PXRD patterns for crystalline phases are given on top of the time-resolved diffractograms. Quantitative plot for the evolution of each phase in the reaction mixture are shown below the diffractogram. Due to its porous nature and disorder of included guest molecules, ZIF-8 could not be included in Rietveld analysis and, therefore, changes in the amount of this phase are shown only with variations in the intensity of its (011) reflection, while changes in the amounts of other, non-porous phases are represented with variations in their phase scale factors. In order to keep the liquid-to-solid ratio (η) after addition of silicon consistent with initial experiments, the volume of aqueous acid was 40 L. Three characteristic ZnO reflections are marked with #.

15 Supplementary Figure 15. Corresponding to reaction monitoring depicted in Supplementary Figure 14. Combined Pawley and Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 40 μl of aqueous acetic acid (c = 2.5 M), after 10 minutes of grinding. Observed crystalline phases are ZnO and ZIF-8. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.32%, R wp = 0.21 %, R p = 0.15 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 6.35 %, R p (-bgr)= 6.67 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

16 Supplementary Figure 16. Corresponding to reaction monitoring depicted in the Supplementary Figure 14. Rietveld fit of the reaction mixture for the LAG reaction of ZnO (0.8 mmol) and HMeIm (1.6 mmol) with 40 μl of aqueous acetic acid (c = 2.5 M), after 70 minutes and 23 seconds grinding. Observed crystalline phases are dia- Zn(MeIm) 2 and kat-zn(meim) 2. The tick marks represent calculated reflection positions for the crystalline phases. Agreement factors: R exp = 1.36%, R wp = 0.20 %, R p = 0.13 %, gof = For comparison, we also provide agreement factors with background subtracted: R wp (-bgr) = 5.28 %, R p (-bgr)= 5.24 %. The observed (blue) and calculated (red) patterns are difficult to discern due to a very good overlap. The difference curve is shown in grey. The contributions of individual phases are inserted above the difference curve and are colour coded according to designations given in upper right corner.

17 a b c d e f g h Supplementary Figure 17. Powder X-ray diffraction patterns of selected samples prepared by milling in a stainless steel milling assembly: (a) kat-zn(meim) 2 prepared by 30 minutes milling in the presence of aqueous acetic acid (c=1.25 M); (b) dia-zn(meim) 2 prepared by 90 minutes milling in the presence of aqueous acetic acid (c=1.25 M); (c) amorph-zn(meim) 2 prepared by 35 minutes milling and (d) 50 minutes milling in the presence of water; (e) amorph-zn(meim) 2 prepared by 30 minutes milling in the presence of aqueous acetic acid (c=2.5 M) ; (f) simulated pattern for kat-zn(meim) 2 ; (g) simulated pattern for dia-zn(meim) 2 (CSD OFERUN01) and (h) simulated pattern for ZIF-8 (CSD OFERUN).

18 Supplementary Figure 18. Time-resolved X-ray diffractogram for the recrystallisation of amorph-zn(meim) 2 into ZIF-8 by milling 200 mg of mechanochemicaly prepared amorphous material with 100 μl DMF. All X-ray reflections are consistent with the ZIF-8 structure (CSD VELVOY).

19 Supplementary Figure 19. (a) Final Rietveld refinement plot for the kat-zn(meim) 2, achieved by using geometry restraints on bond distances, valence angles, and planarity restraints for the methylimidazolate fragments. Space group: P 4 2c, a = (1) Å, c = (1) Å, R wp = 7.08 %, R p = 5.43 %, gof = 1.38, R exp = 5.11 %. (b) Rietveld refinement plot for the katsenite structure model obtained after structure optimisation at 0 K. The refinement included the unit cell parameters but the fractional coordinates of the structure model remained unaltered. a = (1) Å, c = (1) Å, R wp = 8.08 %, R p = 6.26 %, gof = 1.57, R exp = 5.14 %. For both (a) and (b), the high angle region of the diffractogram is enlarged to reveal more detail

20 Supplementary Figure 20. The kat network with different coloring for each type of vertex (top). The augmented form where the vertices of the original net are replaced by their "vertex figures" (bottom).

21 Supplementary Figure 21. (a) Six different kinds of essential rings found in the kat-zn(meim) 2 framework and (b)-(d) three different kinds of tiles (bottom) found in the kat-zn(meim) 2 framework and their skeletons (top).

22 Weight (%) Weight (%) Weight (%) Weight (%) a mg b mg Residue: 33.98% (1.747mg) 40 Residue: 33.62% (2.656mg) Temperature ( C) Temperature ( C) c mg d mg Residue: 31.42% (4.270mg) 40 Residue: 32.41% (1.878mg) Temperature ( C) Temperature ( C) Supplementary Figure 22. TGA thermograms of: (a) a sample of kat-zn(meim) 2 prepared by 30 minutes milling in a stainless steel milling assembly after brief drying in air at room temperature; (b) a sample of dia-zn(meim) 2 prepared by 90 minutes milling in a stainless steel milling assembly after brief drying in air at room temperature; (c) a sample of amorph-zn(meim) 2 prepared by 50 minutes milling in a stainless steel milling assembly after brief drying in air at room temperature; (d) a sample of amorph-zn(meim) 2 prepared by 30 minutes milling in a stainless steel milling assembly after brief drying in air at room temperature. All thermograms were recorded in air.

23 a b c d e wavenumber (cm -1 ) Supplementary Figure 23. Overlay of FTIR-ATR spectra of: (a) kat-zn(meim) 2 prepared by 30 minutes milling in the presence of aqueous acetic acid; (b) dia-zn(meim) 2 prepared by 50 minutes milling in the presence of aqueous acetic acid; (c) amorph-zn(meim) 2 prepared by 35 minutes milling, and (d) 50 minutes milling in the presence of water; (e) amorph-zn(meim) 2 prepared by 30 minutes milling in the presence of aqueous acetic acid.

24 Supplementary Figure 24. SEM images of: (a)-(c) a kat-zn(meim) 2 sample sprepared by 30 minutes milling in the presence of aqueous acetic acid (c=1.25 M) and (d)-(f) a sample of dia-zn(meim) 2 prepared by 90 minutes milling in the presence of aqueous acetic acid (c=1.25 M).

25 Supplementary Figure 25. SEM images of: (a)-(c) a sample of amorph-zn(meim) 2 prepared by 35 minutes milling in the presence of water and (d)-(f) a sample of amorph-zn(meim) 2 prepared by 50 minutes milling in the presence of water.

26 Supplementary Figure 26. SEM images of a sample of amorph-zn(meim) 2 prepared by 30 minutes milling in the presence of aqueous acetic acid (c=2.5 M).

27 Supplementary Table 1. Topological analysis of the kat structure. Number of Vertices: 4 Vertex cn x y z Symbolic Wyckof Symmetry order Zn /2, y, 1/4 4 h 2 2 Zn x, y, z 8 n 1 1 Zn /2, 1/2, 0 2 f -4 4 Zn , 1/2, 1/4 2 d Coordination Sequences Verte x CS 1 CS 2 CS 3 CS 4 CS 5 CS 6 CS 7 CS 8 CS 9 CS 10 Cum 10 Vertex Symbol Zn Zn Zn Zn Number of Edges: 4 Supplementary Note1 Modelling and refinement of diffraction data As previously described 1,2, crystalline phases with fully known crystal structures were modeled using the Rietveld method while phases with only partially known crystal structures were modeled using the structureless Pawley method. The background was modeled using a shifted Chebyshev polynomial with additional broad peaks to account for the amorphous "humps" arising from the amorphous PMMA vessel. As described previously 2 due to diffraction form the two portions of the sample residing on two sides of the cylindrical vessel wall, the peaks at higher angles are split. Therefore, the most strongly diffracting crystalline components of the reaction mixtures, zinc oxide and silicon, were modelled by introducing two separate zinc oxide and silicon phases, respectively. The two phases were identical except in a shift applied to peak positions of one of them. The shift, which applied to both zinc oxide and silicon was defined by a single function proportional to tan(2θ). Such modelling was not applied for other, less strongly diffracting organic or metal-organic crystalline phases. No structure model was refined during modeling of in situ collected patterns. The instrument contribution to peak widths was estimated by collecting a pattern where the vessel contained two corundum balls and was set into oscillation, as reported before. 2

28 Supplementary Note 2 Crystal structure determination The powder X-ray diffraction pattern of kat-zn(meim) 2 was indexed and the structure solved and refined against laboratory powder X-ray diffraction data. Indexing revealed a tetragonal unit cell (V = (6) Å 3 ) and the most likely space group was determined to be P 4 2c which was finally confirmed by a successful structure solution. Unit cell volume corresponded to 16 formula units Zn(MeIm) 2. Volume of this unit cell is larger than twice the volume of the unit cell of dia-zn(meim) 2 ( Å 3 = 3832 Å 3 ), in accordance with observed chemical behaviour, BET measurements, and sequence of structural transformations. Solid-state NMR revealed four independent methylimidazolate ligands. Assuming Z = 8 (8 is also the multiplicity of the general position in this space group), the asymmetric unit must consist of four independent MeIm - fragments and a total of two zinc atoms balancing the charge. MeIm - fragments were assumed to all lie on general positions, while some of zinc atoms were suspected to lie on special positions. Structure was solved by simulated annealing in direct space treating MeIm - as rigid bodies. Zinc atoms were either treated as free atoms or connected to a nitrogen atom in the MeIm - rigid group. Each rigid group had three translational and three rotational degrees of freedom. Occupancies of Zn atoms were included into global optimisation but limited in the range This allowed a correction to the occupancy of Zn atoms close to a special position but not exactly on it and, also, the determination of the correct number of independent Zn atoms in the unit cell. One of the four Zn atoms was found to lie on a general position while three others were on special positions. Multiplicities of the positions of Zn atoms add up to a total of 16 in the unit cell, confirming Z = 8. All MeIm - fragments lie on general positions. Structure solution was recognized when a three-dimensional network was formed with all Zn atoms in tetrahedral environments and all MeIm - units bridging pairs of Zn atoms. This model was introduced into rigid-body Rietveld refinement with additional geometry restraints imposed on coordination environment of Zn atoms. Before the final Rietveld refinement the structural model was optimised at 0 K in space group P1 using quantum mechanics, specifically the VASP code with the projector augmented wave method. The plane wave cut-off was 500 ev and a -point only k-mesh has been used. Atomic forces were relaxed below 10 mev/ Å. We have used two different exchange-correlation functionals: a semi-local one without the van der Waals interaction (PBE), 3 as well as a version of vdw-df type functional 4 which includes nonlocal correlations (i.e. van der Waals interaction from first principles), namely the one which uses optb88 exchange functional. 5,6 Resulting relaxed structures obtained with semi-local (PBE) and non-local (optb88-vdw) functional are similar but different in size. The GGA relaxed structure has the volume of the unit cell of Å 3 (a = b = Å, c = Å) while the vdw relaxed structure has the volume of Å 3 (a = b = Å, c = Å). The P1 optimized structure was searched for higher symmetry and found to exactly match the previously chosen space group. Optimisation also revealed a distorted tetrahedral geometry around zinc atom Zn3 (tetrahedral angles in the range ). The final Rietveld refinement was accomplished using geometry restraints on bond distances, angles as well as appropriate planarity restraints on MeIm - fragments. Values for restraining geometric parameters were taken from the optimized structure and allowed to vary by

29 0.02 Å for bond distances and 3 for bond angles. The distorted tetrahedral geometry around Zn3 was slowly restrained to a more regular tetrahedral coordination. Supplementary Note 3 Relative energies of crystal structures calculated using methods of quantum mechanics DFT calculations with self consistently implemented van der Waals functional 7,8 give the following order of stability of the compounds, in agreement with expectations based on differences in the density of tetrahedral centres: dia [0], kat [+7 mev/atom, kcal mol -1 ] and ZIF-8 [+12.7 mev/atom, kcal mol -1 ]. However, DFT calculations using GGA type of the exchange correlation functional yields results that are clearly not consistent with the experiment - claiming the ZIF-8 structure as the most stable one: ZIF-8 [0], kat [+4.7 ev/atom, kcal mol -1 ] and dia [+7.3 mev/atom, kcal mol -1 ]. Because of the very different topologies of structures of dia-zn(meim) 2, kat-zn(meim) 2 and ZIF-8, it is not likely that the total energies should be so close in value. Exactly in these situations 9 the difference between results obtained with different functionals may not only be quantitative but also qualitative in nature, showing that vdw-df, as a nonlocal functional, is a much better choice than the (still mostly used) semi-local ones of GGA type 10.

30 Supplementary References 1. Friščić, T. et al. Real-time and in situ monitoring of mechanochemical milling reactions. Nature Chem. 5, (2013). 2. Halasz, I. et al. In situ and real-time monitoring of mechanochemical milling reactions using synchrotron X-ray diffraction. Nat. Protoc. 9, (2013) Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, (1996). 4. Dion, M., Rydberg, H., Schröder, E., Langreth, D. C. & Lundqvist, B. I. Van der waals density functional for general geometries. Phys. Rev. Lett. 92, (2004). 5. Mittendorfer, F., Garhofer, A., Redinger, J., Klimeš, J., Harl, J. & Kresse, G. Graphene on Ni(111): Strong interaction and weak adsorption. Phys. Rev. B 84, (R) (2011). 6. Klimeš, J., Bowler, D. R. & Michaelides, A. Van der Waals density functionals applied to solids. Phys. Rev. B 83, (2011). 7. Dion, M., Rydberg, H., Schröder, E., Langreth, D. C. & Lundqvist, B. I. Van der waals density functional for general geometries. Phys. Rev. Lett. 92, (2004). 8. Langreth, C., Dion, M., Rydberg, H., Schröder, E., Hyldgaard, P. & Lundqvist, B. I. Van der Waals density functional theory with applications. Int. J. Quantum Chem. 101, (2005). 9. Lazić, P., Alaei, M., Atodiresei, N., Caciuc, V., Brako, R. & Blügel, S. Density functional theory with nonlocal correlation: A key to the solution of the CO adsorption puzzle. Phys. Rev. B 81, (2010). 10. Lazić, P., Atodiresei, N., Caciuc, V., Brako, R., Gumhalter, B. & Blügel, S. Rationale for switching to nonlocal functionals in density functional theory. J. Phys.: Condens. Matter 24, (2012).