Crystal and electronic structure, lattice dynamics and thermal

View Online / Journal Homepage / Table of Contents for this issue

Dalton Transactions

Dynamic Article Links

Cite this: Dalton Trans., 2012, 41, 2034

PAPER

www.rsc.org/dalton

Downloaded by Cornell University on 26 January 2012 Published on 19 December 2011 on http://pubs.rsc.org | doi:10.1039/C1DT11747E

Crystal and electronic structure, lattice dynamics and thermal properties of Ag(I)(SO3 )R (R = F, CF3 ) Lewis acids in the solid state† Wojciech Grochala,*a,b Michał Ksawery Cyrański,b Mariana Derzsi,a Tomasz Michałowski,b Przemysław J. Malinowski,b Zoran Mazej,c Dominik Kurzydłowski,b Wiktor Ko´zmiński,b Armand Budzianowskia and Piotr J. Leszczyńskia Received 15th September 2011, Accepted 4th November 2011 DOI: 10.1039/c1dt11747e Trifluoromethansulfonate of silver(I), AgSO3 CF3 (abbreviated AgOTf), extensively used in organic chemistry, and its fluorosulfate homologue, AgSO3 F, have been structurally characterized for the first time. The crystal structures of both homologues differ substantially from each other. AgOTf crystallizes ˚ and c = 32.66(2) A ˚ , while in a hexagonal system (R3¯ space group, No.148) with a = b = 5.312(3) A AgSO3 F crystallizes in a monoclinic system in the centrosymmetric P21 /m space group (No.11) with ˚ , b = 8.1739(14) A ˚ , c = 7.5436(17) A ˚ , and b = 94.599(18)◦ , adopting a unique structure a = 5.4128(10) A type (100 K data). There are two types of fluorosulfate anions in the structure; in one type the F atom is engaged in chemical bonding to Ag(I) and in the other type the F atom is terminal; accordingly, two resonances are seen in the 19 F NMR spectrum of AgSO3 F. Theoretical analysis of the electronic band structure and electronic density of states, as well as assignment of the mid- and far-infrared absorption and Raman scattering spectra for both compounds, have been performed based on the periodic DFT calculations. AgSO3 F exhibits an unusually low melting temperature of 156 ◦ C and anomalously low value of melting heat (ca. 1 kJ mol-1 ), which we associate with (i) disorder of its anionic sublattice and (ii) the presence of 2D sheets in the crystal structure, which are weakly bonded with each other via long Ag–O(F) contacts. AgSO3 F decomposes thermally above 250 ◦ C, yielding mostly Ag2 SO4 and liberating SO2 F2 . AgOTf is much more thermally stable than AgSO3 F; it undergoes two consecutive crystallographic phase transitions at 284 ◦ C and 326 ◦ C followed by melting at 383 ◦ C; its thermal decomposition commences above 400 ◦ C leading at 500 ◦ C to crystalline Ag2 SO4 and an unidentified phase as major products of decomposition in the solid state.

Introduction The trifluoromethansulfonate (triflate) anion, CF3 SO3 - (abbreviated as OTf- ), is contained in nearly seven thousand crystallographically characterized compounds1 and in myriads of others, which have not been analyzed with structural methods.2–4 The CF3 SO3 - anion is thus a common ligand in inorganic, organic and organometallic chemistry. Nevertheless, this anionic derivative of the conjugated triflic superacid is in one way unique: it always acts as a very weakly coordinating anion and thus a very good leaving group. Among all inorganic trifluoromethansulfonate salts the silver(I) derivative, AgSO3 CF3 or AgOTf, is the most frequently utilized a ICM, University of Warsaw, Pawińskiego 5a, 02-106, Warsaw, Poland. E-mail: [email protected]; Fax: +48 22 5540801; Tel: +48 22 5540828 b Faculty of Chemistry, University of Warsaw, Pasteur 1, 02-093, Warsaw, Poland c Department of Inorganic Chemistry and Technology, Joˇzef Stefan Institute, Jamova 39, SI-1000, Ljubljana, Slovenia † Electronic supplementary information (ESI) available: Additional DFT and EGA results.. See DOI: 10.1039/c1dt11747e

2034 | Dalton Trans., 2012, 41, 2034–2047

in chemical synthesis.5 Ag(I)SO3 CF3 shows excellent solubility in water and in a broad spectrum of organic solvents; on the other hand, the heavier halides of Ag(I) (AgX, X = Cl, Br, I) are insoluble in these media. As a consequence, the metathetic (ligand exchange) reactions: AgOTf + RX → R+ OTf- + AgX ↓

(1)

are driven by the lack of solubility of AgX, and are very facile not only for X = Br, I but sometimes even for X = Cl. Reactions of activation of the carbon–halogen bond predominate the practical use of AgSO3 CF3 and they are responsible for the spectacular success of this inorganic reagent in organic chemistry.6–11 AgOTf is also useful for selected technological and industrial applications, such as: (i) the Lewis acid catalyzed intermolecular substitution of hydroxyl groups in carbohydrates to afford polysaccharides or oligosaccharides;12 (ii) the addition, isomerisation or ring closing reactions to afford polycyclic heteroaromatic compounds;13 (iii) initiation of the carbocationic polymerization for a wide variety of monomers such as epoxides, tetrahydrofurans, oxazolines, vinyls, This journal is © The Royal Society of Chemistry 2012


View Online

Fig. 1 Three views of the crystal structure of AgOTf. (A) The crystallographic unit cell (Z = 6). (B) The projection emphasizing the presence of infinite layers composed of edge–sharing [AgO6 ] octahedra. (C) The top view of the infinite layers showing the hexagonal arrangement of [AgO6 ] octahedra (blue hexagon) and positions of the triflate anions below and over the centers of hexagons. S – yellow, O – red, C – brown, Ag – large gray, F – small grey balls.

and lactones;14 and (iv) the synthesis of ethers,15 to mention just a few. In view of the success of AgSO3 CF3 in many synthetic applications it is rather surprising that little is known about the compound itself. Specifically, neither the crystal structure and possible structural phase transitions, nor the reasons underlying the unique reactivity of this compound, have not yet been determined. AgSO3 F16 – the only homologous sibling of AgSO3 CF3 in the AgSO3 (CF2 )n F series (n≥0) reported so far – also remains ill-characterized. The sole synthetic pathway towards AgSO3 F reported up to date relies on reaction between AgCN and HSO3 F, with generation of the extremely toxic volatile HCN.16 Equally, there is no information available from theoretical calculations on the electronic structure, chemical bonding and lattice dynamics in both AgSO3 R (R = F, CF3 ) compounds;17 the assignment of IRactive vibrational modes was tentative only and Raman spectra were not assigned.18 In addition, information given about melting points of AgOTf and AgSO3 F is misleading.19,20 In this contribution we seek to fill blank spots regarding AgSO3 CF3 and its less studied homologue, AgSO3 F. We report convenient synthetic pathways, crystal structures from single crystal and powder data, and physicochemical properties of This journal is © The Royal Society of Chemistry 2012

both compounds, their 19 F and 109 Ag MAS NMR and vibrational spectra (IR, Raman) as well as aspects of their thermal decomposition. We also describe chemical bonding, phase transitions and polymorphism, electronic structure and lattice dynamics in these solids using density functional theory (DFT) calculations.

1. Crystal structure and 19 F and 109 Ag MAS NMR spectra of ambient-temperature ambient-pressure polymorphs of AgOTf and AgSO3 F AgOTf crystallizes in a hexagonal system (R3¯, group No.148) ˚ and c = with six formula units per cell, a = b = 5.312(3) A ˚ 32.66(2) A at 100 K (Fig. 1 and Table 1). It forms a layered structure with silver atoms sandwiched between the triflate anions. Within a single layer, extending in the ab plane, the silver atoms are octahedrally coordinated by oxygen atoms from SO3 groups (there are no terminal oxygens), while the CF3 groups are oriented towards the empty van der Waals spaces – a situation typical for triflate compounds (Fig. 1B).2 The unit cell contains three such layers (Fig. 1A) shifted by (1/3a, 1/3b, 1/3c) with respect to each Dalton Trans., 2012, 41, 2034–2047 | 2035

View Online


Table 1 Comparison of the unit cell parameters for triflates of four monovalent metal cations, MOTf (M = Li, Na, K, Ag); V FU and CNM stand for volume per formula unit (FU) and coordination number of a metal, respectively. Arrangement from the left to the right is according to the increasing V FU

Space group ˚ a/A ˚ b/A ˚ c/A a (◦ ) b (◦ ) g (◦ ) Z (FU) ˚3 V FU /A CNM

Li

Na

Ag (100 K21 )

K

P21 /c (No.14) 10.2432 5.0591 9.5592 90 90.00 90 4 123.84 4

P1¯ (No.2) 9.6592 11.1752 11.2768 101.88 105.29 110.50 8 129.77 6

R3¯ (No.148) 5.312(3) 5.312(3) 32.66(2) 90 90 120 6 133.03(80) 6

P21 (No.4) 10.6550 5.9418 14.5700 90 111.34 90 6 143.20 9

˚, c = the R3¯ spacegroup with unit cell vectors (a = b = 5.6055 A ˚ ) comparable to those of AgOTf, and it exhibits flat 31.1417 A layers composed solely of Ca(II) cations. The main difference is that Ca(II) cations occupy only half of the crystallographic positions taken by Ag(I) cations, in accordance with the metal : triflate ratio of 1 : 2. AgSO3 F has been known for over half a century16 but the details of its crystal structure have not been determined up to now. The unit cell of AgSO3 F (Fig. 2 and Table 2) is distinctly different from that of related silver(I) triflate, and also different from those of Li,28 K29 and Cs30 fluorosulfates. AgSO3 F crystallizes as monoclinic in the centrosymmetric P21 /m space group (No.11) with four

other (ABCABC . . . stacking). The [AgO6 ] octahedra are slightly ˚ and deformed with two sets of Ag–O distances: 3 ¥ 2.485(5) A ˚ 3 ¥ 2.504(5) A. The neighbouring [AgO6 ] octahedra are organized into hexagons while sharing edges; each octahedron shares edges with three partners. The shortest Ag(I) ◊ ◊ ◊ Ag(I) separation is quite ˚ , well below the double van der Waals radius short, 3.0676(15) A ˚ ). of Ag (3.44 A The OTf groups are positioned above and below the hexagonal gaps of the 2D silver network (Fig. 1C); each anion is shared between three cations. Since the triflate anion serves as a tridentate ligand, the bonding between Ag(I) and oxide anions from the - OTf moiety may be viewed as predominantly ionic, just like in the majority of alkali and alkaline earth triflates.2 Indeed, the S–O ˚ ) and take values typical for bond lengths are equal (1.440(5) A alkali derivatives where the metal–triflate anion bonding is predominantly ionic in nature.22–24 This feature is nicely corroborated by the DFT calculations (see section 3). The crystal structure of AgOTf is different from that of the related K(I)22 salt but more similar to those of Na(I)23 Li(I)24 by exhibiting of a quasi-2D character (Table 2) with the van der Waals gaps opening between the hydrophobic CF3 groups from adjacent layers. However, the coordination number of a metal is 6 for AgOTf while it is only 4 for LiOTf. AgOTf is probably most similar to triflates of divalent metal cations such as Mg(II), Ca(II), Zn(II),25 Cu(II)26 and Ag(II).27 For example, Ca(OTf)2 also crystallizes in Table 2 Comparison of the unit cell parameters for fluorosulfates of four monovalent metal cations, MSO3 F (M = Li, K, Cs, Ag); V FU and CNM stand for volume per formula unit (FU) and coordination number of a metal, respectively. Arrangement from the left to the right is according to the increasing V FU . The crystal structure of NaSO3 F has not yet been solved

Space group ˚ a/A ˚ b/A ˚ c/A a (◦ ) b (◦ ) g (◦ ) Z (FU) ˚3 V FU /A CNM

Li

Ag (100 K21 )

K

Cs

C2/m (No.12) 8.54 7.62 4.98 90 90.00 90 4 81.02 4

P21 /m (No.11) 5.4128(10) 8.1739(14) 7.5436(17) 90 94.599(18) 90 4 83.17(3) 6

Pnma (No.62) 8.62 5.84 7.35 90 90 90 4 92.50 10

P21 /a (No.14) 7.7243 8.1454 7.7839 90 110.83 90 4 114.43 10

2036 | Dalton Trans., 2012, 41, 2034–2047

Fig. 2 Three views of the crystal structure of AgSO3 F. (A) The crystallographic unit cell (Z = 4). (B) The projection emphasizing the presence of dimerized infinite chains composed of edge- and corner-sharing [AgO5 F] octahedra, running parallel to the crystallographic b axis. (C) The projection emphasizing the layered character of AgSO3 F with cationic and anionic sheets running parallel to the ac diagonal. S – yellow, O – red, Ag – large gray, F – small grey balls. The O/F disorder is emphasized by half-shading of the respective atoms.

This journal is © The Royal Society of Chemistry 2012


View Online

˚ , b = 8.1739(14) A ˚, formula units per cell and a = 5.4128(10) A ◦ ˚ c = 7.5436(17) A, and b = 94.599(18) at 100 K, adopting a unique structure type. Two types of non-equivalent SO3 F ions can be distinguished in the structure; the first one with disordered F and O atoms, the second one without the substitutional disorder (Fig. 2A). The F/O disorder for the first anion is particularly well seen in the single crystal data (at 100 K): the length of bond formed between ˚ ; this would be anomalously long for a S and F/O is 1.494(5) A ˚ ) and unusually short for a typical S–F typical S–O bond (~1.44 A bond (such as e.g. the S–F bond length in the second, ordered ˚ ). The structurally ordered SO3 F ion uses all its anion, ~1.55 A oxygen atoms to coordinate to the silver atom (one O atom is bridging between two different Ag(I) cations, while two other O atoms are each linked to one Ag(I) cation), while fluorine remains terminal - a situation commonly observed for fluorosulfates. Therefore, it comes as a surprise that in the case of the structurally disordered SO3 F ion, apart from all of its oxygen atoms, the fluorine atom is also found in the first coordination sphere of the silver. Here, two disordered O/F atoms are linked to one Ag(I) cation each, and two O atoms are bridging between two Ag(I) cations. The Ag(I) ions are found in a severely deformed octahedral [AgO5 F] coordination. The [AgO5 F] octahedron can be considered as an elongated one due to four short equatorial Ag–O distances ˚ ) and two longer axial Ag–O/F distances with (2.432(3)–2.465(4) A ˚ . The shortest Ag ◊ ◊ ◊ Ag separation average distance of ca. 2.58 A ˚ (much larger than 3.221 A ˚ seen for Ag2 SO4 37 is as large as 3.743 A 38 ˚ or 3.268 A for Ag2 S2 O7 ), excluding any appreciable chemical bonding. The topology of the silver sublattice is quite complex, and unique amongst all compounds of Ag(I) (Fig. 2B–C). The [AgO5 F] octahedra are organized into infinite 1D dimerized chains. Within a single chain, each octahedron shares an edge with one and a corner with another partner. The octahedra from two neighbouring chains share one edge to form a dimerized chain (Fig. 2B–C). These dimerized chains are interconnected through the SO3 F anions in the two remaining directions, thus forming a 3D crystal crystal framework (Fig. 2C). One can, however, distinguish the alternating Ag(I) and SO3 F layers running parallel to the acdiagonal.

Table 3 The chemical shifts, d, in the 19 F MAS NMR spectra of AgSO3 F and AgOTf, as compared to several fluorine compounds31–33 with respect to CCl3 F standard Compound

d (ppm)

SF6 S2 O 5 F 2 AgSO3 F SO2 F2 AgSO3 CF3 CF3 COOH AgF

+57.431 +48.532 +45.6, +37.6 +33.532 -74.6 -78.531 -314.033

Table 4 The chemical shift in the 109 Ag MAS NMR spectra of AgOTf, as compared to several compounds of Ag(I)34–36 with respect to the 9 M aqueous solution of AgNO3 standard Compound

d (ppm)

AgSO3 F AgF AgOTf Ag2 SO4 AgSO3 CH3 Ag(NH3 )2 (SO4 )

NDa -53.0 +66.0 +52.735 +87.235 +606.936

a

not determined.

The fact that half of the F atoms are engaged in weak chemical bonding to Ag(I) makes AgSO3 F different from related LiSO3 F,28 where all F atoms stick towards each other while forming layers. Engagement of F2 in interaction with Ag(I) for AgSO3 F is also reflected in the 19 F NMR spectra (Fig. 3 and Table 3); two resonances are seen: one broader at 37.6 ppm (corresponding to more ionic bridging F2, with disorder) and another one, much narrower, at 45.6 ppm (terminal F1, ordered). Only one sharp resonance is seen at -74.6 ppm for the related AgOTf. Only one sharp signal at +66 ppm is seen in the 109 Ag NMR spectrum of AgOTf (Fig. 4 and Table 4); regretfully, no spectrum could be obtained for AgSO3 F at the same experimental conditions as those for AgOTf despite screening of a broad range of chemical shifts (-400 to +400 ppm).39

Fig. 3 The 19 F MAS NMR spectra of AgOTf and AgSO3 F; rotational sidebands are marked with starlets.


Dalton Trans., 2012, 41, 2034–2047 | 2037


View Online

Fig. 4

The 109 Ag MAS NMR spectrum of AgOTf.

2. Chemical bonding as viewed from vibrational spectroscopy and theoretical calculations of phonon spectrum The IR absorption spectroscopy provides valuable data about the bonding mode of SO3 F- and OTf- anions, which is especially valuable if structural information is missing,3 or when a compound shows structural complexity, as seen here for AgSO3 F. Complete assignment of the normal modes for solid materials is usually facilitated by the theoretical calculations of normal mode frequencies and occasionally of their IR and Raman intensities.27,40,41 While the MIR spectra have previously been reported for both AgSO3 R (R = F,42 CF3 18 ) homologues, the information on the FIR and Raman spectra is still missing. Moreover, only a tentative assignment of selected MIR–active modes has been given without any theoretical support. Therefore, we have conducted a combined experimental and theoretical study to get complete information about the lattice dynamics of AgSO3 F and AgOTf and we were able to assign all bands appearing in the spectra. The MIR and FIR absorption spectra of AgOTf are shown together with the Raman spectrum in Fig. 5; an analogous set of spectra for AgSO3 F is presented in Fig. 6. The corresponding numerical data are gathered in Tables 5 and 6. The values of the DFT-calculated wavenumbers together with group symmetry assignments and IR and Raman activity are also given for all modes. 2.1.

Assignment of vibrational spectra of AgOTf

The MIR spectrum for AgOTf is in fair agreement with the previously published data (Fig. 5).18 It is predominated by a broad band in the 1100–1400 cm-1 range with several distinct maxima and shoulders; the corresponding bands in the Raman spectrum are of medium intensity. There appears a separate group of IR bands in the 1000–1100 cm-1 range; the corresponding is a very strong band at 1048 cm-1 that predominates the Raman spectrum. According to DFT calculations the highest frequency bands in the 1200–1250 cm-1 range originate from antisymmetric and symmetric stretching of the CF3 moieties, while those in the 1000–1200 cm-1 range 2038 | Dalton Trans., 2012, 41, 2034–2047

are assigned to modes involving antisymmetric and symmetric stretching of SO3 groups. The bands above 1250 cm-1 appearing in the IR spectrum are either combination modes, or they are attributed to impurities (cf. Table 5). The antisymmetric (1187, 1190 cm-1 ) SO3 stretching modes fall ~110 cm-1 below the values found for related Ag(II) triflate (1272, 1278 cm-1 ); the symmetric SO3 stretching modes (1010, 1050 cm-1 ) are also redshifted by ~40–70 cm-1 as compared to Ag(II)(OTf)2 (1058, 1122 cm-1 ).27 Simultaneously the antisymmetric CF3 stretching modes (1206, 1224 cm-1 ) have frequencies nearly identical to those measured for Ag(OTf)2 (1219 cm-1 ).27 This comparison reflects an increased degree of covalence of the Ag–O bonding and concomitant localization of the double S O bonds (with the strength of CF3 bonding virtually unaffected) as the oxidation state of silver increases from (I) to (II). None of the doubly degenerate Eg and Eu bands splits in two, which indicates a genuine local C3 symmetry of anion. Thus, ligand dentity deduced from IR spectra (tridentate - OTf) is consistent with the structural data for AgOTf. The strong Raman band at 774 cm-1 is assigned to a totally symmetric C–S stretching mode coupled to umbrella deformation of the CF3 group. An analogous Raman active but much weaker band is seen also at 774 cm-1 for Ag(OTf)2 .27 The 774 cm-1 Ag mode is very prolific in giving rise to overtones; at least three distinct Eu modes coming from the combination of this mode with various IR-active modes, are seen in the IR spectrum. We notice that the umbrella deformations of the SO3 group do not yield pronounced IR or Raman bands, in contrast to those of the CF3 group. The FIR region of the infrared spectrum (30–600 cm-1 ) also carries valuable information about structure and bonding. The two medium IR bands at 518 cm-1 and 580 cm-1 can both be assigned to Eu OSO and FCF deformations, while the Raman bands are at virtually identical wavenumbers to the corresponding Eg sets. The remaining deformation modes yield only weak IR or Raman active bands. The modes at 182 and 190 cm-1 assigned from DFT calculations to the hindered Au and Ag rotations of the SO3 moieties are silent in both types of spectra (albeit not forbidden by group theory). Their counterparts for the CF3 groups fall at much lower wavenumbers (43 and 51 cm-1 , respectively) consistent with This journal is © The Royal Society of Chemistry 2012


View Online

Fig. 5 The IR and Raman spectra of AgOTf together with the DFT-predicted wavenumbers (scaling factor 1.05).

Fig. 6 The IR and Raman spectra of AgSO3 F together with the DFT-predicted wavenumbers (scaling factor 1.08).


Dalton Trans., 2012, 41, 2034–2047 | 2039

View Online Table 5 The assignment of modes active in the MIR, FIR and Raman spectra of AgSO3 CF3 based on the theoretical DFT calculations of normal phonon modes.a s – strong, m – medium, w – weak, v – very, sh – shoulder, b – broad; as – antisymmetric, s – symmetric (within a functional group), NM – not measured IR

Raman DFTb Symmetry Assignment


1437 w 1351 w 1291 m 1266 sh

— — — — 1224 w 1225 1206 vs 1220 1183 w 1209 — 1194 1181 s 1181 — 1178 1048 vs 1042 1043 m 1024 sh 1009 m 1002 965 vw — 866 vw — 774 m 778 — 777 731 vw — 675 sh — 657 w 648 644 w — 620 w 614 580 m 571 579 m 571 560 vw 560 sh — 536 vw — 518 vw 512 518 m 511 484 vw — 472 vw — 353 w 345 347 w 338 339 sh 325 324 w 318 222 vw 214

Eu Eu — Eg Eu Ag Au Eu Eg Ag — — Au — Eu Ag Au — — Ag — Au Eu Eg — — Eg Eu — — Eu Eg Ag Au Eg

—

214

Eu

190 182 146 126 93 88 56 51

Au Ag Au Ag Eg Eu Eg Ag

52 w

43

Au

37 w NM

39 35

Eu Ag

NM

27

Eg

— — — 119 w NM 89 m NM NM

impurityc combination mode (774 R + 580 IR) combination mode (774 R + 518 IR) impurityc as CF3 stretching as CF3 stretching s CF3 stretching s CF3 stretching as SO3 stretching as SO3 stretching s SO3 stretching impurityc , d impurityc , d s SO3 stretching impurityc combination mode (774 R + 89 IR) CS stretching + CF3 umbrella CS stretching + CF3 umbrella impurityc impurityc SO3 umbrella impurityc SO3 umbrella F–C–F + O–S–O deformations F–C–F + O–S–O deformations impurityc impurityc F–C–F + O–S–O deformations F–C–F + O–S–O deformations impurityc impurityc O–S–C–F deformations O–S–C–F deformations s [AgO6 ] stretching + CS stretching as [AgO6 ] stretching + CS stretching [AgO6 ] stretching + O–S–C–F deformations [AgO6 ] stretching + O–S–C–F deformations hindered rotations of SO3 hindered rotations of SO3 as [AgO6 ] stretching + deformation [AgO6 ] breathing lattice modes (Ag motion in plane) lattice modes (Ag motion in plane) lattice modes hindered rotations of CF3 + lattice modes hindered rotations of CF3 + lattice modes lattice modes lattice modes (Ag motion out of plane) lattice modes

The mutual exclusion principle applies for centrosymmetric R3¯ (S6 ) point group but it may be bent due to application of selection rules other than dipole moment-related. b The calculated wavenumbers were scaled by the factor of 1.05. c Several very weak bands or shoulders coming from impurities can be identified from their varying intensities while comparing the spectra taken for different batches of the compound. d The bands in this region are sometimes attributed to S-OH stretching;43 note that remnants of triflic acid might come from synthesis of AgOTf. a

weak van der Waals interactions between adjacent CF3 groups. Increase of temperature will likely lead to their free rotation. 2040 | Dalton Trans., 2012, 41, 2034–2047

Table 6 The assignment of modes active in the MIR, FIR and Raman spectra of AgSO3 F based on the theoretical DFT calculations of normal phonon modes.a s – strong, m – medium, w – weak, v – very, sh – shoulder, br – broad; as – antisymmetric, s – symmetric (within a functional group), NM – not measured IRa

Ramana

1395 vw 1315 sh 1295 vs 1247 vs 1230 sh 1215 sh 1163 sh 1095 vw 1070 m 1062 m 1057 sh 1040 sh 1020 sh 967 vw 880 w 837 vw 797 m 745 m 710 sh 608 sh 598 w 588 m 579 m 572 m 564 m 559 w 500 vw, br 431 vw 399 vw 189 w, br

1310 vw 1301 vw 1230 vw 1217 vw

1076 vs 1062 sh

DFTb 1332 1321 1291 1261 1232, 1230 1215, 1215 1079 1074 1066 1063

1045 vw 1021 vw 969 vw 803 m 791 sh 752 w

598 w 589 w 578 w 572 vw 563 sh 559 w 484 vw 434 vw 420 vw 408 vw 398 vw

161 w, br

800 791 753 730 596 593 582, 582 573, 573 567, 567 556, 556 550, 550 437, 437 418, 417 406, 405 396, 396 186

150

132 w, br

129 sh

126, 125

98 w, br 78 w, br 52 sh

NM NM NM

102 77 55

Assignment impurityc as SO3 stretching as SO3 stretching as SO3 stretching as SO3 stretching as SO3 stretching as SO3 stretching impurityc s SO3 stretching s SO3 stretching s SO3 stretching s SO3 stretching impurityc , d impurityc , d impurityc , d impurityc , d impurityc , d S–F terminal stretching S–F terminal stretching S–F bridging stretching S–F bridging stretching impurityc O–S–O deformations O–S–O deformations O–S–O deformations O–S–O deformations O–S–O deformations O–S–O deformations O–S–O deformations impurityc impurityc O–S–F deformations O–S–F deformations O–S–F deformations O–S–F deformations 216–110 hindered rotations of SO3 F tetrahedra & [AgO5 F] stretching 216–110 hindered rotations of SO3 F tetrahedra & [AgO5 F] stretching 216–110 hindered rotations of SO3 F tetrahedra & [AgO5 F] stretching 109–31 lattice modes 109–31 lattice modes 109–31 lattice modes

a

The mutual exclusion principle does not hold due to factual O/F ordering; the true local symmetry is P21 (point group C2 ). b The calculated wavenumbers were scaled by the factor of 1.08. c Ag2 SO4 is the most frequently found crystalline impurity and it its presence is occasionally revealed in the Raman spectrum via a presence of 969 cm-1 band. d The bands in this region are sometimes attributed to S-OH stretching and deformations;45 note that remnants of fluorosulfuric acid might come from synthesis of AgSO3 F. Some IR bands in this region are also seen for Ag2 S2 O7 .

The weak Raman band at 126 cm-1 is assigned to the totally symmetric stretching (breathing) of the [AgO6 ] octahedron. The analogous band for Ag(OTf)2 (with its much shorter Ag–O bonds) This journal is © The Royal Society of Chemistry 2012


View Online

is considerably stiffened (242 cm-1 )27 as expected. Other stretching and deformation modes of the [AgO6 ] octahedron are found above the totally symmetric one, and they mix up considerably with the OSCF deformations. All the lattice modes involving motions of both Ag(I) cations and triflate anions fall below 100 cm-1 ; two of these are nicely resolved in FIR with a weak sharp band at 37 cm-1 and an unusually broad (half-width ~50 cm-1 ) band at 89 cm-1 of medium intensity. The latter is assigned to the sliding (in plane) motion of cationic sublattice with respect to an anionic one (Eu ); judging from both the half-width and wavenumber of this band one expects good anisotropic ionic conductivity for the 2D lattice of AgOTf even at room temperature (note, 298 K is an equivalent of 207 cm-1 , so even the overtone of the 89 cm-1 mode can be easily thermally excited). Group theory predicts the following 36 vibrational modes for the rhombohedral representation of the unit cell of AgSO3 CF3 (Z = 2): C vibr = 9 Ag (R) + 9 Eg (R) + 9 Au (IR) + 9 Eu (IR), two of which are acoustic (Au + Eu ) and the remaining ones are optical modes. Summarizing this section we would like to stress that, based on the DFT calculations, we have successfully assigned 22 optical modes (6 Ag , 5 Eg , 4 Au , 7 Eu ), thus eliminating errors given in the previous assignment.18 The 12 remaining optical modes (3 Ag , 4 Eg , 4 Au , 1 Eu ) do not have a detectable IR or Raman intensity or fall below the spectral range measured here in Raman spectrum (< 100 cm-1 ). The theoretical wavenumbers (scaled by the factor of 1.05) show excellent linear correlation with the experimental values (see ESI†). 2.2.

Assignment of vibrational spectra of AgSO3 F

Having analyzed the vibrational spectra of AgOTf let us now turn to the more complicated case of its lighter homologue, AgSO3 F. The mutual exclusion principle should apply for the centrosymmetric P21 /m space group (point group C2h ) but the mirror plane, m, is in fact introduced only by the O/F disorder. Since a true SO3 F- anion does not have the respective local mirror plane at the sulfur atom (F–S–O), the local symmetry must be lower; the more adequate space group for vibrational analysis is P21 (point group C2 ). This is additionally confirmed by simultaneous appearance of selected bands in IR and Raman spectra – this would not happen if the centrosymmetric C 2h point group applied. Since all modes (of A and B symmetry) are now both IR- and Raman-active, we skip the group theory assignments in Table 6 and we label only the most important modes. The MIR spectrum for AgSO3 F is in good agreement with the previously published data (Fig. 6).44 A broad ‘sulfate band’ appears in the 1100–1350 cm-1 range, centered at 1247 cm-1 , with several distinct maxima and shoulders; the corresponding bands in the Raman spectrum are very weak. According to DFT calculations these bands are assigned to eight modes involving antisymmetric stretching of all SO3 moieties. On the other hand, there appears a separate group of bands in the 1000–1100 cm-1 range most of which are both IR and Raman active. These bands are assigned to four modes arising from symmetric stretching of the SO3 moieties; one of these (at 1076 cm-1 ) yields a very strong band, which predominates the Raman spectrum. The most pronounced IR bands in the SO3 stretching range (1062, 1247 cm-1 ) are upshifted by about 40–60 cm-1 as compared to their This journal is © The Royal Society of Chemistry 2012

counterparts for AgOTf (1009, 1206 cm-1 ) reflecting the fact ˚) that the shortest (terminal) S–O bonds for AgSO3 F (~1.41 A are slightly shorter than the uniform bonds seen for AgSO3 CF3 ˚ ). (~1.44 A The bands in the 730–880 cm-1 range are assigned to 4 distinct S– F stretching modes; two of these yield quite strong IR bands. The bands coming from terminal SF groups (SO3 F- anions without disorder) exhibit higher stretching frequency (791, 803 cm-1 ) than those from the bridging SF groups (disordered SO3 F- anions, 745 and 752 cm-1 ). The OSO and OSF deformation modes appear as two very characteristic groups of bands (multiplets) in the 550–600 cm-1 and 390–440 cm-1 ranges, respectively. The former ones are much more IR-active than the latter, while their Raman intensities are comparable. The weakly resolved FIR bands (110–220 cm-1 ) assigned to modes coming from hindered rotations of the SO3 F tetrahedra (combined with stretching of the AgO bonds) fall in a similar frequency range as for AgOTf. They are followed by lattice modes having a large contribution from heavy Ag atoms (30–110 cm-1 ). It is tempting to associate the presence of a nearly continuous spectrum of thermal-energy absorbing lattice modes with an unusually low melting point of AgSO3 F (see section 4). The as yet unassigned very weak bands or shoulders (see Table 5) come from impurities, as deduced based on their varying intensities when comparing spectra taken for different batches of freshly prepared AgSO3 F. The most pronounced among these are IR bands at 880 cm-1 and 1040 cm-1 , which probably originate from Ag2 S2 O7 and HSO3 F, respectively, and Raman bands at 969 and 484 cm-1 from traces of Ag2 SO4 . Group theory predicts the following 72 vibrational modes for the unit cell of AgSO3 F (Z = 4), three of which are two acoustic modes and the remaining ones are optical. Considering optical modes there are 12 SO stretching modes, 4 SF stretching modes, 20 OSO and OSF deformation modes, 12 hindered rotations of SO3 Fanions, and 21 lattice modes (including stretching and deformation modes of the [AgO5 F] octahedra). In summary of this section we would like to point out that, based on the DFT calculations, we have successfully assigned all stretching and deformation modes of four SO3 F- anions as well as 4 hindered rotations, and 3 lattice modes. The remaining modes do not have a detectable IR or Raman intensity or fall below the spectral range measured in Raman spectrum (< 100 cm-1 ). The theoretical wavenumbers (scaled by the factor of 1.08) show excellent linear correlation with the experimental values (see ESI†).

3. Phase transitions, melting and thermal decomposition of AgSO3 F and AgOTf Thermogravimetry (TGA) and calorimetry (DSC) coupled with evolved gas analysis (EGA) are useful tools for characterization of inorganic compounds. They allow for the detection of crystallographic phase transitions, melting, evaporation, as well as thermal decomposition processes. 3.1. TGA–DSC analysis for AgOTf The TGA/DSC profiles of AgSO3 CF3 in the temperature range from 30–500 ◦ C are shown in Fig. 7. The corresponding EGA results (Q-MS & FT-IR of gases) are shown in ESI.† Two Dalton Trans., 2012, 41, 2034–2047 | 2041


View Online

Fig. 7 The TGA (black, left vertical axis) and DSC (gray, right vertical axis) profiles for AgSO3 CF3 in the 30–500 ◦ C range; scanning speed 10 ◦ C min-1 .

endothermic phase transitions are seen at 284 ◦ C (with associated heat, Q = 18.4 ± 1 J g-1 = 4.7 kJ mol-1 ) and 326 ◦ C (Q = 8.8 ± 0.1 J g-1 = 2.3 kJ mol-1 ) followed by melting at 383 ◦ C (Q = 20.6 ± 0.5 J g-1 = 5.3 kJ mol-1 , also confirmed by visual observation of a sample).27 The first endothermic peak has been sometimes erroneously interpreted as melting of AgOTf (286 ◦ C19 ), while it is clearly a solid-solid phase transition. Observation of temperaturedriven polymorphism of AgOTf is not surprising. Indeed, the related triflates of monovalent alkali metals exhibit polymorphic phase transitions at rather low temperatures (Cs: P21 to Cmcm at 107 ◦ C; Rb: Cm to P21 at 48 ◦ C, P21 to Cmcm at 188 ◦ C; K: (?) to P21 at -63 ◦ C; Li: P21 /c to Cmca at 156 ◦ C). Moreover, the NMR studies for triflates of Na and Li indicate a great degree of mobility of the triflate anions even below room temperature and these compounds are very good ionic conductors.46 Since a detailed understanding of the temperature-induced polymorphism of AgOTf falls off the scope of the current work we have measured only short scans of the high–temperature XRD patterns of AgOTf at the temperatures close to 260 ◦ C, 310 ◦ C and 350 ◦ C to identify the high-temperature phases (see ESI†). The results indicate that novel polytypes of AgOTf retain the trigonal lattice typical of the low-temperature form and they probably exhibit free rotations of CF3 and CF3 +SO3 groups, respectively. The latent heat associated with these phase transitions suggests that they are of the first-order. Once melted, AgOTf starts losing mass appreciably; the mass loss is as large as 42.0% between 400 ◦ C and 550 ◦ C and the reaction product is light gray. Simultaneously the EGA (see ESI†) indicates evolution of species with m/z of 28 (CO), 44 (CO2 ), 64 (SO2 ), 69 (CF3 ), 48 (SO), 32 (O2 , S), 16 (O), 12 (C), 83 (SO2 F), 99 (SO3 F), 31 (CF), 49 (SF); these are supposedly fragmentation products at high temperatures of volatile SO2 (CF3 )2 . Small amounts of H2 SO4 (98), H2 SO (50), HSO2 CF3 (13447 ) and H2 SOCF3 (119), originating from reactions of evolved gas with traces of omnipresent water vapour, can also be detected. It may thus be deduced that thermal decomposition of AgOTf leads 2042 | Dalton Trans., 2012, 41, 2034–2047

mostly to formation of silver(I) sulfate (with a theoretical mass loss of 39.3%): 2 Ag(I)SO3 CF3 → Ag2 (SO4 ) + SO2 (CF3 )2 ↑

(2)

The XRD of the solid residue of decomposition (collected at 500 ◦ C and cooled down to room temperature, ESI†) indicates that Ag2 SO4 is indeed one crystalline product; another unidentified crystalline phase is also present. The MIR and FIR spectra of the solid residue (ESI†) consistently suggest the presence of Ag2 SO4 but three bands are of unknown origin. The amount of Ag2 SO4 product cannot be large at 424 ◦ C as indicated by the absence of the characteristic DSC peak corresponding to orthorhombic → hexagonal phase transition; Ag2 SO4 supposedly forms at temperatures closer to 500 ◦ C. 3.2. TGA–DSC analysis for AgSO3 F The TGA–DSC profiles of AgSO3 F in the temperature range from 30–500 ◦ C are shown in Fig. 8. The corresponding EGA results (Q-MS & FT-IR of gases) are provided in ESI.† The thermal behaviour of AgSO3 F is completely different from that of related AgOTf. A small amount (ca. 2.7 wt.%) of HSO3 F occluded in samples of AgSO3 F during synthesis, is liberated in the 30– 150 ◦ C temperature range as revealed by the endothermic peak at 106 ◦ C. Dry AgSO3 F melts at 156 ◦ C as confirmed also by visual observation of a sample. This is actually an unusually low melting temperature as for the salt of a rather small cation, Ag(I); for example, sulfide melts at temperatures as high as 825 ◦ C,48 related sulfate at 652 ◦ C,48 iodide at 556 ◦ C,48 fluoride at 435 ◦ C,48 and nitrate at 212 ◦ C.48 The heat of melting of AgSO3 F is ca. 4.6 ± 0.3 J g-1 (~1 kJ mol-1 ), which is one order of magnitude smaller than the respective values for sulfide (14.1 kJ mol-148 ), iodide (9.4 kJ mol-148 ) or nitrate (11.5 kJ mol-148 ). To our best knowledge, among compounds of silver(I) only trinitromethanide, AgC(NO2 )3 , melts at a lower temperature (98 ◦ C)49 than AgSO3 F. This suggests that crystal lattice disorder, present in AgSO3 F, facilitates melting (and This journal is © The Royal Society of Chemistry 2012


View Online

Fig. 8

The TGA (black, left vertical axis) and DSC (gray, right vertical axis) profiles for AgSO3 F in the 30–500 ◦ C range; scanning speed 10 ◦ C min-1 .

supposedly it leads to increased anionic conductivity, as well). Another possible reason for the low melting point could be the presence of 2D sheets in the crystal structure, weakly bonded with each other via long Ag–O(F) contacts (ESI†). Thermal decomposition of AgSO3 F starts at ~270 ◦ C with concomitant liberation of SO2 F2 as seen in MS (large peaks at m/z of 48 and 64 are seen, corresponding to SO and SO2 , the typical fragmentation products50 ). Decomposition is mildly endothermic as indicated by small broad DSC peaks extending from 230 ◦ C to 400 ◦ C with a distinct maximum at 316 ◦ C. The mass loss between 170 ◦ C and 500 ◦ C is 20.3%,51 i.e. an equivalent of less than 12 mol of SO2 F2 per one mol of AgSO3 F (theoretical mass loss for reaction (1) is 24.6%). It may be anticipated that thermal decomposition of AgSO3 F leads to Ag2 SO4 : Ag(I)SO3 F → 1/2 Ag(I)2 (SO4 ) + 1/2 SO2 F2 ↑

(3)

The product of thermal decomposition of AgSO3 F at 500 ◦ C and cooled down to room temperature is white; its XRD and IR spectra (ESI†) indicate that this is indeed mostly Ag2 SO4 with some residual AgSO3 F. Thus, Ag(I)SO3 F decomposes thermally in a similar way as do fluorosulfates of Sr and Ba.52 A sharp endothermic peak is seen at 424 ◦ C, close to the value that was previously erroneously assigned as a melting temperature of AgSO3 F.20 In fact, this peak corresponds to an orthorhombic-tohexagonal phase transition of Ag(I)2 SO4 , reported to take place at 425 ◦ C.53 In conclusion, the main channel of the thermal decomposition of both AgSO3 R derivatives studied in this work may be described by a generalized reaction equation: 2 Ag(I)SO3 R → Ag2 (SO4 ) + SO2 R2 ↑ (R = F, CF3 )

The combined TGA–DSC results discussed in the previous section suggest that AgSO3 F melts without undergoing any preceding phase transitions, in contrast to its AgOTf congener. As we will show in this section, polymorphism of AgSO3 F cannot, however, be entirely excluded. But before we analyze two hypothetical polytypes of AgSO3 F we will address the issue of O/F disorder in the polytype known from experiment. Single crystal XRD diffraction at 100 K testifies to substitutional disorder of O/F atoms in one of two independent SO3 Fanions present in the structure of AgSO3 F. To understand this disorder we have generated two ordered structures with F and O atoms localized. Two possible ways of ordering the O/F atoms within an experimental unit cell (with two disordered SO3 Fanions) lead to monoclinic cells of the following symmetries: (i) P1¯ and (ii) P21 (Fig. 9 A and B, respectively). Energies of the optimized cells differ only by 2 meV/FU = 0.2 kJ mol-1 (which is smaller than the error of the computational method), while their densities differ only marginally (4.125 and 4.134 g cm-3 , respectively; Table 7). Calculations performed for various supercells with even more possibilities for O/F localization lead to the same result: negligible energy and volume differences between the optimized models. The energy differences between various ordered models do not exceed 10% of the thermal energy at room temperature (25 meV). The calculations thus suggest that there is Table 7 The DFT-calculated energies per formula unit, E FU , and densities, d, for AgSO3 F in four different polymorphic types after total optimizations (VASP results). Energies are referred to that of the most stable form calculated, P21

(4)

Additional decomposition channels open up at high temperature and the solid residues do not contain pure Ag2 SO4 ; amorphous AgF is also present (as judged from FIR), together with other unidentified phases. This journal is © The Royal Society of Chemistry 2012

4. The O/F disorder and possible polymorphism of AgSO3 F: DFT view

Ordered experimental models

Hypothetical polytypes

P1¯

P21

LiSO3 F-type AgOTf-type

+2.0 4.134

31.8 3.590

E FU /meV 0.0 d/g cm4.125

-10.5 3.738

Dalton Trans., 2012, 41, 2034–2047 | 2043


View Online

Fig. 9 AgSO3 F in four optimized unit cells as described in the text and in Table 7: (A) experimental unit cell with ordering of the O/F atoms enforced in one possible way (P1¯). (B) As (A), but with a different ordering of O/F atoms (P21 ). (C) and (C¢) Two projections of the LiSO3 F–type. (D) and (D¢) Two projections of the AgOTf–type.

no strong preference for any particular ordering, which, we feel, helps to explain the appearance of the O/F disorder in crystalline AgSO3 F. Polymorphism of AgOTf, typical of many triflates,22,54,55 raises a natural question of whether its lighter homologue, AgSO3 F, might also be found in different crystalline forms. To address this question we have again applied the DFT methods and we have optimized unit cells of AgSO3 F in two prototypical structures: LiSO3 F- and AgOTf-type (Fig. 9C–C¢ and D–D¢, respectively, and Table 7). The first one was generated by Li→Ag substitution56 while the latter by removing all F atoms from the CF3 groups and subsequent C→F substitution. AgSO3 F optimized in the LiSO3 Ftype structure shows the presence of discrete [O2 AgO2 Ag O2 ] units (Fig. 9C¢); the van der Waals gap opens between the F atoms in the structure (Fig. 9C). It turns out that this hypothetical polymorph of AgSO3 F is destabilized by nearly 32 meV/FU with respect to the ordered P21 polytype found experimentally. This can be understood when recalling a small coordination number (CN = 4) of a metal, M, and the presence of unusual [O]2 bridges – both features are highly atypical for Ag(I) in its compounds. The LiSO3 F- is thus disfavoured as compared to the experimentally found polymorph. The outcome of the calculations for the AgOTf-type of AgSO3 F, however, is quite surprising. The AgOTf–type is favoured by ~10.6 meV/FU = 1 kJ mol-1 with respect to the experimentally found structure (at T = 0 K). Simultaneously, its volume per FU is over 10% larger than the calculated volume of the ordered experimental polytype (P21 ). This suggests that the AgOTf–type should be 2044 | Dalton Trans., 2012, 41, 2034–2047

entropically stabilized with respect to the experimentally found ˚ 3) polymorph. The calculated volume difference per FU (8.5 A -1 translates to stabilization of approximately 4.5 kJ mol in entropy term at ambient (p,T) conditions.57 In addition, our calculations of phonon frequencies at gamma point indicate that the hypothetical AgOTf–type polymorph of AgSO3 F is dynamically stable, which suggests that its synthesis should be viable. Or to put it in another way, the polytype of AgSO3 F described in this contribution could be only a metastable form of this compound.

Conclusions and prospect We report here the synthesis, crystal structures and full assignment of vibrational spectra for two related oxo- salts of monovalent silver, fluorosulfate and triflate. The fluorosulfate adopts a unique structure type and it exhibits an anomalously low value of melting temperature (156 ◦ C), and of melting heat (1 kJ mol-1 ). We attribute these anomalies to a quasi-2D structural character of this compound as well as to partial disorder of fluorosulfate anions in its lattice. The related silver(I) triflate crystallizes in a hexagonal unit cell, and it shows two consecutive first-order phase transitions at elevated temperature (associated by small latent heat), followed by melting at 383 ◦ C. The ionic conductivity of AgOTf in its various polymorphic forms and especially that of AgSO3 F below its melting point, are supposedly large. The AgSO3 R salts thus await exploration using impedance spectroscopy. Both AgOTf and AgSO3 F decompose thermally at elevated temperatures yielding Ag2 SO4 and unidentified secondary product(s). This journal is © The Royal Society of Chemistry 2012

View Online

Experimental


Synthesis and single crystal growth AgOTf (99%) has been purchased from Aldrich. AgOTf may also be obtained in high yield by action of HOTf on AgF. Use of acetate or trifluoroacetate instead of fluoride precursor was found undesirable due to various side reactions occurring in superacidic media. AgSO3 F has been prepared via reaction between AgF (99.9% trace metal basis, Aldrich) and HSO3 F (purified by triple distillation, Aldrich). Also in this case, use of acetate or trifluoroacetate precursors is discouraged. Our new synthetic pathway is much more convenient that the one described earlier utilizing AgCN.16 The crude product was rinsed three times with a 1 : 1 mixture of trifluoroacetic acid and trifluoroacetic anhydride and then dried at RT under vacuum. Yield: 81.3%. Samples of dry AgSO3 F may be obtained by heating to 170 ◦ C and quenching of a melt. Single crystals of AgSO3 CF3 (thin pellets) and AgSO3 F have been grown from triflic and fluorosulfuric acid, respectively, by slow evaporation from FEPR or TeflonR containers in an Ar atmosphere. The single crystals of AgSO3 CF3 are thin and fragile; they tend to stick to one another and they easily exfoliate into thinner pieces. Single crystal X-ray diffractometry and structure solution and refinement (100 K) All measurements of each crystal were performed on a KM4CCD k-axis diffractometer with graphite-monochromated Mo-Ka radiation. The crystal was positioned at 62 mm from the CCD camera. 1000 frames were measured at 0.6◦ intervals with a counting time of 10 s. The data were corrected for Lorentz and polarization effects. Analytical correction for absorption was applied.58 Data reduction and analysis were carried out with the Oxford Diffraction programs.59 The structure was solved by direct methods60 and refined using SHELXL.61 The refinement was based on F 2 for all reflections except those with very negative F 2 . Weighted R factors wR and all goodness-of-fit S values are based on F 2 . Conventional R factors are based on F with F set to zero for negative F 2 . The F o 2 >2s(F o 2 ) criterion was used only for calculating R factors and is not relevant to the choice of reflections for the refinement. The R factors based on F 2 are about twice as large as those based on F. Scattering factors were taken from Tables 6.1.1.4 and 4.2.4.2 in ref. 62 The measured crystal of AgSO3 F is twinned via pseudomerohedry with the twin lattice to be orthorhombic. Thus the twin obliquity is equal to 4.60(2)◦ what is the obtuse beta angle of the monoclinic cell. In such a case and regarding the centrosymmetric space group of the crystal the twin elements could be either rotation axes parallel to [100] and [001] or reflection planes (100) and (001). The final data were not corrected for the twinning effect – such an approach did not improve the final refinement. Only the values of residual remaining electron densities were slightly lower whereas all other refinement parameters were worse in contrast to the non corrected for the twinning data. Because in this type of twinning reflections from different twin domains were overlapped at different stages, the final refinement parameters including discrepancy factors are slightly above those characteristic for good quality single crystals. In addition there is substitutional disorder This journal is © The Royal Society of Chemistry 2012

in the structure. Atoms F2 and O5 occupy the same fractional position and they were refined with occupancy ratio 0.5 : 0.5 with the same anisotropic displacement parameters. The data quality for the second structure presented here, AgSO3 CF3 , is not high but still acceptable. This is due to the crystal quality – its small size and the presence of other tiny specimens giving additional noise to the diffraction pattern. In addition, the CF3 groups in the crystal lattice might be disordered. Thus the intensity of higher angle reflections is rather small resulting the GooF parameter to be slightly out of the desired range. Further details of the crystal structures of both compounds at 100 K may be obtained from Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (fax: (+49)7247-808-666; email: crysdata@fizkarlsruhe.de) on quoting the CSD numbers 423464 (AgSO3 F) and 423465 (AgSO3 CF3 ). Powder X-ray diffractometry (298 K) The crystallinity and purity of AgSO3 R (R = F, CF3 ) and products of their thermal decomposition were investigated using powder diffractometry. Each sample was sealed inside a 0.3 mm quartz glass capillary. Three different machines were used. (1) D8 Discoverer BRUKER diffractometer with CuKa1 and CuKa2 radiation at an intensity ratio of ca. 2: 1, angle range from 3◦ to 120◦ , and a typical running time of 16 to 40 h. (2) D8 Advanced BRUKER diffractometer with CuKa1 and CuKa2 radiation at an intensity ratio of ca. 2: 1, angle range from 5◦ to 120◦ , and a typical running time of 16 h. (3) Panalytical diffractometer X’Pert Pro with CoKa1 and CoKa2 radiation at an intensity ratio of ca. 2: 1, angle range from 5◦ to 120◦ , and a typical running time of 12 h. The sequence of high-temperature-resolved scans for AgOTf was collected using the high-temperature probe: at 30 ◦ C, then at 260 ◦ C (below the first phase transition), 310 ◦ C (after the first phase transition), 350 ◦ C (after the second phase transition but before melting), and again at 30 ◦ C (after cooling). Structure refinement from the powder data For AgOTf only, lattice constants were refined due to insufficient quality of powder data. For AgSO3 F the best quality powder pattern was collected for a sample obtained by heating Ag(SO3 F)2 at 290 ◦ C for 7 min. The light brown powder contained mostly AgSO3 F and was contaminated only with Ag3 (SO3 F)4 (27 wt.%). The multiphase Le Bail decomposition and Rietveld structure refinement were performed in Jana2006.63 The background was smooth but its correct description required use of 29 Legendre polynomials. We used single crystal structures as initial models for refinement; for AgSO3 F the substitutional disorder for atoms F2 and O5 has been described by allowing for different fractional positions for both atoms and setting their occupancies to 0.5. Soft distances and angles restraints were applied for the SO3 F- anion. The S– ˚ for oxygen atoms bonded O distances were restrained at 1.46 A ˚ for oxygen weakly strongly to two Ag+ cations and at 1.40 A ˚ ; the S–F distance was coordinated to the Ag+ cation at > 2.7 A ˚ . Since the OSO angles found in other fluororestrained at 1.55 A sulfates (e.g. Sn(SO3 F)2 )64 are slightly larger while the OSF angles are slightly smaller than those found in a perfect tetrahedron, the respective values were restrained at 113.5◦ and 105◦ . The Dalton Trans., 2012, 41, 2034–2047 | 2045

View Online

uncertainties of bond distances and angles were obtained with application of Bérar’s correction.65 Further details of the crystal structure of AgSO3 F at 298 K may be obtained from Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (fax: (+49)7247-808-666; email: crysdata@fizkarlsruhe.de) on quoting the CSD number 423474.


FT-MIR & FIR spectroscopy All solid samples were characterized by MIR & FIR absorption spectroscopy using a Vertex 80v BRUKER vacuum FT-IR spectrometer. We used a modified gas-tight cuvette designed originally for liquids. Spectra were measured for thin layer of powder samples attached to the surface of AgCl (MIR) or polyethylene windows (FIR). Alternatively, conventional KBr pellets were used for the MIR region. Raman spectroscopy High resolution Raman spectra were recorded using Horiba Jobin Yvon LabRam-HR 800 Raman micro-spectrometer with 532 nm diode laser exciting line. Full power at laser head was 71 mW, meanwhile the full power on the surface of the sample was 36 mW; the latter could be varied from 0.36 mW to 36 mW with use of a gray (neutral) filter. Longer expositions (around 8 min) with full laser power focused on a small spot (approximately 1 mm2 ) lead to photochemical and/or thermal decomposition of AgOTf and formation of Ag2 SO4 , while AgSO3 F is stable in the beam. 19

F and 109 Ag NMR spectroscopy

The 19 F and 109 Ag NMR spectra for samples closed in a zirconia rotor have been obtained using a 700 MHz NMR spectrometer (Varian NMR Systems) using magic angle spinning (MAS) at 24 kHz/5 kHz for 19 F and 109 Ag. For 19 F NMR a typical pulse duration was 1 ms, at an acquisition time of 20 ms and a repetition delay of 10 s for 4 independent scans. The 109 Ag NMR spectra of AgSO3 CF3 and AgF (see ESI†) were obtained by means of cross-polarization (CP) between 109 Ag and 19 F nuclei39 with g(109 Ag)B1 /2p = g(19 F)B1 /2p = 39.6 kHz. The pulse duration was 2.7 ms (109 Ag 90◦ pulse) with an acquisition time of 30 ms and a repetition delay of 5 s for 4 scans. The contact time for was 5 ms.

Theoretical calculations The solid state DFT calculations were performed using Vienna ab initio Simulation Package (VASP) formally for p = 0 Pa and T = 0 K.66–68 We used the general gradient approximation (GGA) and the projector augmented wave (PAW) method69 with the Perdew, Burke, and Ernzerhof functional revised for solids (PBEsol).70 The wave functions were expanded in plane waves up to the kinetic energy cutoff of 600 eV providing very good energy convergence. All structural models were fully relaxed (atomic positions and cell parameters) with the electronic and ionic convergence criteria set to 10-7 and 10-5 eV, respectively, and k-mesh spacing via the ˚ -1 . Results Monkhorst–Pack scheme with uniform spacing of 0.4 A of optimizations are shown in the ESI. For the calculation of normal modes in Gamma point, forces on atoms were minimized ˚ -1 per atom. Next, the Hellmann-Feynman down to 0.001 eV A forces were generated while displacing each atom in the cell per ˚ in x,y,z directions. The density of electronic states and time by 0.3 A electronic band structure (see ESI†) were calculated with denser ˚ -1 . All optimizations were performed on 1 ¥ 1 ¥ 1 mesh of 0.2 A unit cells except for F/O ordered cells of AgSO3 F, where larger supercells (211, 221 etc.) were also tested. In the case of Ag(I)OTf, the phonons, DOS and band structure were calculated for the rhombohedral representation (Z = 2). For AgSO3 F the ordered P21 cell (which best reproduced the experimental unit cell vectors, cf. ESI†) was used for phonon and single point calculations. Partial atomic charges were calculated with the GGA-PBE method using CASTEP.71 PHONON72 was used to verify symmetry labels for the calculated phonon modes at C, VESTA73 for structure drawings.

Acknowledgements The project ‘Quest for superconductivity in crystal–engineered higher fluorides of silver’ is operated within the Foundation for Polish Science ‘TEAM’ Programme co-financed by the EU European Regional Development Fund. This work has been partly supported by the Slovenian Research Agency (ARRS) within the research program P1-0045 Inorganic Chemistry and Technology. DFT calculations have been performed at ICM supercomputers within grant G34-10. The X-ray structures were measured in the Crystallography Unit of the Physical Chemistry Laboratory at the Chemistry Department of the University of Warsaw. MKC gratefully acknowledges Dr Łukasz Dobrzycki (University of Warsaw) for his kind assistance.

TGA/DSC/EGA measurements Thermal decomposition (mass loss, heat flow) was monitored using a STA (simultaneous thermal analyzer) 409 from Netzsch in the temperature range 30–450 ◦ C. STA permits a simultaneous thermogravimetric analysis (TGA) and differential scanning calorymetry (DSC) of the samples with concomitant evolved gas analysis (EGA). A typical heating rate of 10 K min-1 in a 99.999% Ar stream (flow: 50 mL min-1 ) and Alumina (Al2 O3 ) crucibles were used. The evolved gases were analyzed with a Q-MS 403 C Aeolos Netzsch mass spectrometer (connected to STA via a quartz capillary) and a Vertex 80v BRUKER FT-IR spectrometer (its gas cuvette connected to STA via a teflon tube). Both transfer lines were preheated to 200 ◦ C to avoid condensation of low-boiling residues. 2046 | Dalton Trans., 2012, 41, 2034–2047

Notes and references 1 Inspection of the CCSD database revealed 6828 crystal structures containing triflic anion (accessed June 2011; database actualized in February 2011). 2 R. Dinnebier, N. Sofina, L. Hildebrandt and M. Jansen, Acta Crystallogr., Sect. B: Struct. Sci., 2006, 62, 467–473. 3 G. A. Lawrence, Chem. Rev., 1986, 86, 17–33. 4 F. Mistry, PhD Thesis, University of British Columbia, Canada (1987). Accessed at: https://circle.ubc.ca/bitstream/2429/1814/1/ ubc_1993_fall_phd_mistry_fred.pdf. 5 Search of the Web of Science database using three keywords: AgSO3 CF 3 , silver trifluoromethanesulfonate and silver triflate, leads to ca. 750 (only seldom overlapping) hits. On the other hand AgSO3 F and silver fluorosulfate are represented by as few as 10 hits. 6 E. Bell-Loncella and C. A. Bessel, Inorg. Chim. Acta, 2000, 303, 199– 205.



View Online 7 A. Xia and P. R. Sharp, Polyhedron, 2002, 21, 1305–1310; Y. Ding, H. W. Roesky, M. Noltemeyer and H. G. Schmidt, Organometallics, 2001, 20, 1190–1194. 8 J. L. Cross, T. W. Crane, P. S. White and J. L. Templeton, Organometallics, 2003, 22, 548–554. 9 A. M. Neculai, D. Neculai, G. B. Nikiforov, H. W. Roesky, Ch. Schlicker, R. Herbst-Irmer, J. Magull and M. Noltemeyer, Eur. J. Inorg. Chem., 2003, 3120–3126. 10 J. Cipot, D. Wechsler, R. McDonald, M. J. Ferguson and M. Stradiotto, Organometallics, 2005, 24, 1737–1746. 11 J. R. Bleeke, P. V. Hinkle and N. P. Rath, Organometallics, 2001, 20, 1939–1951. 12 M. H. Fisher, R. L. Tolman, US Patent 4134973; 4156720; 4203976. 13 K. Y. Lin, M. D. Matteucci, US Patent 6007992; 6028183; 6800743; 6951931. 14 O. Forooq, US Patent 5124417. 15 R. B. Chapman, J. L. Andershak, S. A. Shackelford, US Patent 4855508. 16 E. Hayek, A. Czaloun and B. Krismer, Monatsh. Chem., 1956, 87, 742–748. 17 In the only theoretical paper on AgSO3 CF3 found by us, the authors try to understand complexation of this compound by prismands: P. Saarenketo, R. Suontamo and K. Rissanen, Organometallics, 2002, 21, 5473–5485. ¨ 18 H. Burger, K. Burczyk and A. Blaschette, Monatsh. Chem., 1970, 101, 102–119. 19 See for example MSDS form taken from www.aldrich.com. 20 MSDS form for previously sold product no. S460559 from rare chemicals library. 21 The room temperature data from powder XRDP: AgOTf a = b = ˚ , c = 32.812(3) A ˚ , V = 806.96(14) A ˚ 3 ; AgSO3 F a = 5.3290(5) A ˚ , b = 8.2135(17) A ˚ , c = 7.6742(16) A ˚ , b = 94.259(5)◦ , 5.3954(11) A ˚ 3. V = 339.15(12) A 22 G. Korus and M. Jansen, Z. Anorg. Allg. Chem., 2001, 627, 1599–1605. 23 N. Sofina, E. M. Peters and M. Jansen, Z. Anorg. Allg. Chem., 2003, 629, 1431–1436. 24 M. Bolte and H. Lerner, Acta Cryst. E, 2001, 57, m231-m232. 25 M. Tremayne, P. Lightfoot, M. A. Mehta, P. G. Bruce, K. D. M. Harris, K. Shankland, C. J. Gilmore and G. Bricogne, J. Solid State Chem., 1992, 100, 191–196. 26 R. Dinnebier, N. Sofina, L. Hildebrandt and M. Jensen, Acta Crystallogr., Sect. B: Struct. Sci., 2006, 62, 467–473. 27 P. J. Malinowski, Z. Mazej, M. Derzsi, Z. Jagliˇcić, J. Szydłowska, T. Gilewski and W. Grochala, CrystEngComm, 2011, 13, 6871–6879. ˇ ak and M. Kosiˇcka, Acta Crystallogr., Sect. B: Struct. Crystallogr. 28 Z. Z´ Cryst. Chem., 1978, 34, 38–40. 29 K. O’Sullivan, R. C. Thompson and J. Trotter, J. Chem. Soc. A, 1967, 2024–2027. 30 P. Roegner and K. J. Range, Z. Naturforschung B, 1993, 48, 688–690. 31 C. H. Dungan and J. R. Van Wazer, Compilarion of Reported F19 NMR Chemical Shifts, Interscience, New York, 1970. 32 G. Franz and F. Neumayr, Inorg. Chem., 1964, 3, 921–922. 33 Z. Mazej, E. Goreshnik, Z. Jagliˇcić, B. Gaweł, Wiesław Łasocha, D. ´ D. Kurzydłowski, P. Malinowski, W. Ko´zminski, ´ Grzybowska, T. Jaron, ´ J. Szydłowska, P. J. Leszczynski and W. Grochala, CrystEngComm, 2009, 11, 1702–1710. 34 K. D. Becker and E. von Goldammer, Chem. Phys., 1980, 48, 193–201. 35 G. H. Penner and W. Li, Inorg. Chem., 2004, 43, 5588–5597. 36 G. A. Bowmaker, R. K. Harris, B. Assadollahzadeh, D. C. Apperley, P. Hodgkinson and P. Amornsakchai, Magn. Reson. Chem., 2004, 42, 819–826. 37 B. N. Mehrotra, T. Hahn, W. Eysel, H. Roepke and A. Illguth, Neu. Jahr. Mineral., 1978, 408–421. ´ 38 P. J. Malinowski, M. Derzsi, A. Budzianowski, P. J. Leszczynski, B. Gaweł, Z. Mazej and W. Grochala, Chem.–Eur. J., 2011, 17, 10524– 10527. 39 Observation of 109 Ag NMR signals in conventional experiments is extremely difficult given the low gyromagnetic ratio (g ) of 109 Ag as well as extremely long spin–lattice relaxation times. To our knowledge


the spectra presented here are the first evidence of a successful 109 Ag F CP MAS experiments (most experiments up to date were 109 Ag 1 H CP MAS, see: G. H. Penner and X. Liu, Prog. Nucl. Magn. Reson. Spectrosc., 2006, 49, 151–167. ´ P. J. Malinowski, M. Derzsi, Z. Mazej, Z. Jagliˇcić, P. J. Leszczynski, T. Michałowski and W. Grochala, Eur. J. Inorg. Chem., 2011, 2499–2507. M. Derzsi, P. J. Malinowski, Z. Mazej and W. Grochala, Vib. Spectrosc., 2011, 57, 334–337. P. C. Leung and F. Aubke, Inorg. Chem., 1978, 17, 1765–1772. R. K. Khanna, J. C. Pearl and R. Dahmani, Icarus, 1995, 115, 250–257. T. Michałowski, P. J. Malinowski, M. Derzsi, Z. Mazej, Z. Jagliˇcić, P. J. ´ Leszczynski and W. Grochala, Eur. J. Inorg. Chem., 2011, 2508–2516. R. K. Khanna, J. C. Pearl and R. Dahmani, Icarus, 1995, 115, 250–257. ¨ ¨ L. van Wullen, N. Sofina, L. Hildebrandt, C. Muhle and M. Jansen, ¨ Solid State Ionics, 2006, 177, 1665–1672; L. van Wullen, N. Sofina and M. Jansen, ChemPhysChem, 2004, 5, 1906–1911. Molecular peak at m/z = 133 is seen in MS of pure SO2 (CF3 )2 : H. Oberhammer, G. D. Knerr and J. M. Shreeve, J. Mol. Struct., 1982, 82, 143–146. Handbook of Inorganic Chemicals, Ed., P. Patnaik, McGraw-Hill, New York, 2003. Hawley’s Condensed Chemical Dictionary (15th Edition), Ed., R. J. Lewis Sr., Wiley, New York, 2007. There are no peaks from SOx Fy species in the mass spectra. At temperatures as high as 300 ◦ C fluorine from SO2 F2 fluorinates the quartz capillary through which gases travel between TGA/DSC apparatus and the mass spectrometer; this is indicated by a strong peak at m/z = 85 assigned to SiF3 + . For some samples even smaller mass losses down to 12% are observed. E. L. Muetterties and D. D. Coffmann, J. Am. Chem. Soc., 1958, 80, 5914–5918. A. A. A. El-Rahmana, M. M. El-Desokyb, A. El-Wahab and A. ElSharkawy, J. Phys. Chem. Solids, 1999, 60, 119–127. L. Hildebrandt, R. Dinnebier and M. Jansen, Inorg. Chem., 2006, 45, 3217–3223. L. Hildebrandt, R. Dinnebier and M. Jansen, Z. Anorg. Allg. Chem., 2005, 631, 1660–1666. The structure of related NaSO3 F is not yet known. H. D. B. Jenkins and L. Glasser, Inorg. Chem., 2006, 45, 1754–1756. CrysAlis RED, Oxford Diffraction Ltd., Version 1.171.28. Analytical numeric absorption correction using a multifaceted crystal model based on expressions derived by R. C. Clark and J. S. Reid. CrysAlis CCD, Oxford Diffraction Ltd., Version 1.171.28; CrysAlis RED, Oxford Diffraction Ltd.,Version 1.171.28. G. M. Sheldrick, Acta Crystallogr., Sect. A: Found. Crystallogr., 1990, 46, 467–473. G. M. Sheldrick, SHELXL93, Program for the Refinement of Crystal ¨ Structures, Univ. of Gottingen, Germany. International Tables for Crystallography Vol. C., Ed., A. J. C. Wilson, Kluwer, Dordrecht, 1992. V. Petriˇcek, M. Duˇsek, L. Palatinus, Jana2006, The crystallographic computing system, Institute of Physics, Praha, Czech Republic, 2006. D. C. Adams, T. Birchall, R. Faggiani, R. J. Gillespie and J. E. Vekris, Can. J. Chem., 1991, 69, 2122–2126. J.-F. Bérar and P. Lelann, J. Appl. Crystallogr., 1991, 24, 1–5. ¨ G. Kresse and J. Furthmuller, Comput. Mater. Sci., 1996, 6, 15–50. ¨ G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter, 1996, 54, 11169–11186. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775. ¨ P. E. Blochl, Phys. Rev. B: Condens. Matter, 1994, 50, 17953–17979. J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou and K. Burke, Phys. Rev. Lett., 2008, 100(136406), 1–4. S. J. Clark, M. D. Segall, C. J. Pickard, P. J. Hasnip, M. J. Probert, K. Refson and M. C. Payne, Z. Kristallogr., 2005, 220, 567–570. ´ K. Parlinski, Software PHONON, Cracow (2008). K. Momma and F. Izumi, J. Appl. Crystallogr., 2008, 41, 653–658. 19

40 41 42 43 44 45 46 47 48 49 50

51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

Dalton Trans., 2012, 41, 2034–2047 | 2047