Revising the Highest Oxidation States of the 5d Elements: The Case of Iridium(+VII)

The quest for the highest achievable oxidation states of the transition-metal elements is of fundamental interest, not least because complexes in very high oxidation states may serve as oxidation agents.1–3 The maximum oxidation states for the early transition metals follow the group number up to Group 8 (that is, [OsO₄] and [RuO₄], the lack of evidence for an existence of [FeO₄] marks the exception). The trends for the later transition metals tend to be less clear-cut (see Figure 1 in Ref. 4). In the 5d series, the trend of the experimentally suggested maximum oxidation states appears to be irregular (see filled circles and dotted line in Figure 1).

Figure 1. Maximum oxidation states of the 5d transition metals: (•) highest experimentally known values, (▴) probably incorrect experimental assignments, (□) suggested most likely values.

The highest experimentally known oxidation states after osmium are represented by the hexafluorides of iridium ([IrF₆])5 and platinum ([PtF₆]).6 The isolation of [AuF₇], claimed almost 20 years ago,7 has recently been shown by high-level quantum chemical calculations to be highly improbable.8 As the existence of [AuF₆] is also unlikely,8 oxidation state +V remains the highest oxidation state of gold that is known beyond doubt.9 Quantum-chemical calculations have furthermore strongly supported the thermochemical stability of mercury(+IV) as gaseous [HgF₄],4, 10–14 but no experimental confirmation has been obtained so far. The correctness of an early report of an electrochemically generated, spectroscopically characterized short-lived [Hg^III(cyclam)][BF₄]₃ species15 appears unclear from today's perspective.

Combining the results of the most accurate quantum-chemical predictions and of reliable experimental studies, a revised trend of the highest oxidation states of the 5d transition-metal row is obtained. Apart from the lack of iridium(+VII), there is a linear descent after osmium (Figure 1).

To establish the highest achievable iridium oxidation states, we report herein structure optimizations by density functional theory (DFT) methods,16 and of high-level coupled-cluster calculations16 of the stabilities of iridium fluoride complexes up to [IrF₉]. Figure 2 shows the DFT-optimized16 structures of [IrF₉], [IrF₇], and [IrF₅]. Whereas [IrF₉] exhibits a minimum with D_3h symmetry on the potential-energy surface (singlet state), the gas-phase elimination of F₂ for this complex is computed to be highly exothermic (see Table 1). Calculations on oxo-fluoro complexes of Ir^IX (e.g. on [IrO₃F₃]) indicate also very exothermic decomposition pathways with low barriers (details of these studies will be provided elsewhere17). This situation makes the existence of the highest theoretically possible oxidation state of iridium highly unlikely.

Figure 2. B3LYP-optimized minimum structures of [IrF₅] (C_2v), [IrF₇](D_5h), [IrF₉] (D_3h), and [IrOF₅] (C_4v). Bond lengths [pm], angles [°].

Table 1. Computed reaction energies [kJ mol⁻¹] for iridium fluoride complexes.^[a]

Reaction	CCSD	CCSD(T)	B3LYP^[b]
[a] Reaction energies given at scalar relativistic levels, for singlet [IrF9], doublet [IrF8], triplet [IrF7], quartet [IrF6], triplet [IrF5], quartet [IrF4], triplet [IrOF5], doublet [IrOF4], and singlet [IrOF3]. Spin-orbit corrections at B3LYP level to reactions c, d, and e amount to −36.9 kJ mol^−1, −7.6 kJ mol^−1, and −29.3 kJ mol^−1, respectively. [b] Values in parentheses are counterpoise and zero-point vibration corrected. [c] The experimental value is −60.2 kJ mol^−1. [d] The experimental value is +159.7 kJ mol^−1.
a) [IrF9]→[IrF7]+F2	−401.9	−319.6	−375.5 (−385.9)
b) [IrF8]→[IrF6]+F2	−329.6	−249.6	−246.7 (−257.2)
c) [IrF7]→[IrF5]+F2	32.8	102.6	130.3(119.6)
d) [IrF7]→[IrF6]+F	−100.1	−44.3	−32.9 (−40.3)
e) [IrF6]→[IrF5]+F	257.3	299.7	318.5 (308.2)
f) [IrF5]+[KrF2]→[IrF7]+Kr	−117.8	−163.5	−166.0
g) [IrOF5]→[IrOF3]+F2			261.6
h) [IrOF5]→[IrF4]+OF			176.9
i) [IrOF5]→[IrOF4]+F			172.8
j) [IrOF5]→[IrF5]+O			256.1
k) [IrF6]^+→[IrF4]^+ +F2			216.7
l) [IrF6]^+→[IrF5]^+ +F			154.5
m) [KrF2]→Kr+F2^[c]	−85.0	−60.9	−35.7
n) F2 → 2F^[d]	124.4	152.7	155.3

What about Ir^VIII? Synthesis of the most likely Ir^VIII complex, [IrO₄], has recently been attempted by matrix isolation but resulted in formation of the peroxide species [(O₂)IrO₂].19 This result reflects the energy gain from formation of an OO bond. The homoleptic fluoride complex [IrF₈] has a square antiprismatic (D_4d) minimum on the potential-energy surface (not shown). However, concerted F₂ elimination is computed to be highly exothermic (reaction (b) in Table 1). This elimination is so favorable because of the steric crowding in the Ir coordination sphere, which places this complex at very high energies (and probably leads to low barriers). Most mixed fluoro-oxo complexes of Ir^VIII are computed to be similarly unstable; only [IrOF₆] exhibits somewhat less exothermic decomposition pathways.17

Things look rather different for Ir^VII (d² configuration): [IrF₇] is computed to exhibit a pentagonal-bipyramidal triplet ground-state minimum (D_5h symmetry, see Figure 1).18 In contrast to [IrF₉], F₂ elimination from triplet [IrF₇] is appreciably endothermic (reaction (c) in Table 1), in fact much more so than the best calculations suggest for the long-sought [HgF₄].4, 10–14 Our coupled-cluster calculations predict an energy of +102.6 kJ mol⁻¹ (Table 1). As for related cases triple excitations contribute substantially to this positive value,4, 8, 12 and B3LYP DFT calculations compare reasonably well with the CCSD(T) results8, 12 (this makes B3LYP energies a good choice for larger systems where coupled-cluster calculations are not feasible).

A second potential channel for decomposition of [IrF₇] involves the homolytic dissociation of an IrF bond to give [IrF₆]. Although this reaction is calculated to be slightly exothermic (reaction (d) in Table 1; larger basis sets are expected to render this value less negative4, 12), the structural rearrangement required for this bond breaking is substantial and creates an appreciable barrier of +100.4 kJ mol⁻¹ (scalar relativistic DFT results). The transition state is a singly capped octahedron with C_3v symmetry, where the cap represents the IrF bond to be broken.

We note in passing that nonrelativistic pseudopotential calculations provide approximately 100 kJ mol⁻¹ less positive energies for F₂ elimination. Thus, as in all other cases studied to date,4, 8, 20, 21 the stability of the highest oxidation states of the 5d elements is largely due to relativistic effects.

Given the open-shell nature of several of the relevant species involved, we have also evaluated the influence of spin-orbit (SO) effects on stabilities, using single-point calculations with a recently implemented22 two-component non-collinear spin-density functional approach.16 Most of the relevant results are in footnote [a] of Table 1. While the SO stabilization increases from [IrF₇] to [IrF₆] to [IrF₅], its influence on the decomposition reactions of [IrF₇] is moderate and does not change the thermochemistry dramatically (the same holds for activation barriers; computed SO corrections to the barrier for homolytic IrF bond breakage amount to only −5.7 kJ mol⁻¹). Counterpoise corrections for basis-set superposition errors and zero-point vibrational energy corrections are also of no appreciable consequence for the relevant reaction energies (Table 1).

Thus it is likely that [IrF₇] is a viable target for gas-phase synthesis (e.g. in molecular-beam experiments) or for matrix-isolation studies. Characterization of [IrF₇] by vibrational spectroscopy may be aided by the harmonic vibrational frequency analysis provided in the Supporting Information. The electronic structure and oxidation state of Ir^VII species should be determinable by Mössbauer spectroscopy.

Oxidation of [IrF₅] by the endothermic fluorine compound KrF₂ is substantially exothermic (reaction (f) in Table 1). This situation is even more so with the strongest presently known oxidative fluorinating agent the [KrF]⁺ ion.23 Formation of the [KrF][IrF₆] ion-pair complex from (gas-phase) [KrF]⁺ and [IrF₆]⁻ is highly exothermic (by −491.6 kJ mol⁻¹ at B3LYP level) and provides a local minimum on the potential-energy surface (analogous to the known complex [XeF][IrF₆]24). However, the complex is calculated to decompose exothermically (−118.5 kJ mol⁻¹) into [IrF₇] and Kr. Figure 3 shows the reaction energies for [NgF][MF₆]→MF₇+Ng for a range of complexes (Ng=Kr, Xe and M=Ir, Pt, Au) at the corresponding computational level.

Figure 3. Computed energies (B3LYP) for the (gas-phase) reactions [NgF][MF₆]→[MF₇]+Ng (Ng=Kr, Xe; M=Ir, Pt, Au): □ [XeF]⁺ complexes, ○ [KrF]⁺ complexes.

While the formation of the ion-pair complexes from the separated ions is in all cases strongly exothermic (data not shown), only [KrF][IrF₆] decomposes exothermically to give the heptafluoride (note that these energies will be generally somewhat more positive in the condensed phase owing to electrostatic stabilization of the ion-pair complexes). These computational results suggest a possible pathway to obtain Ir^VII. Interestingly, in contrast to several known [KrF][MF₆] complexes of platinum and gold,24–26 and in spite of the existence of [XeF][IrF₆],24 observation of [KrF][IrF₆] has never been reported.

Initial data for an alternative Ir^VII target, the C_4v-symmetrical [IrOF₅], have also been obtained (data for the triplet state are provided in Table 1 and Figure 1 d; the singlet is only 2.4 kJ mol⁻¹ higher at the scalar relativistic level, but this difference is enhanced by SO effects17). [IrOF₅] has the advantage of a lower coordination number. Indeed, in this case elimination of F₂ is even more endothermic than for [IrF₇], and elimination of OF is also still appreciably endothermic (reactions (g) and (h) in Table 1). Even the homolytic splitting of an IrF bond to give [IrOF₄] is endothermic by 172.8 kJ mol⁻¹ (reaction (i) in Table 1). The harmonic vibrational frequencies of [IrOF₅] are provided in the Supporting Information.

Another Ir^VII species that comes to mind is the [IrF₆]⁺ ion. It is computed to prefer a triplet ground state with a slight Jahn–Teller distortion (D_4h symmetry, IrF_ax 179.3 pm, IrF_eq 184.7 pm). The adiabatic ionization potential IrF₆→[IrF₆]⁺ is calculated to be very large (13.5 eV at B3LYP level). The energies for concerted F₂ elimination and homolytic IrF dissociation are calculated to be both appreciably endothermic (reactions (k) and (l) in Table 1). As [IrF₆] is a volatile complex, it is unclear at the moment why its molecular ion has apparently never been observed in a mass spectrometry experiment.

Our present state-of-the-art quantum chemical calculations suggest thus that the highest iridium oxidation state that has a realistic chance of experimental observation is Ir^VII. The experimentally known highest 5d oxidation states for Groups 8, 10, and 11 are Os^VIII, Pt^VI, and Au^V, respectively (see ref. 27 for the computationally verified instability of Pt^VIII and ref.8 for exclusion of Au^VII and Au^VI). Adding the computationally predicted Hg^IV and Ir^VII states suggests that the trend for the later 5d metals should become a linear decrease once all computationally suggested possibilities have been verified experimentally (see solid line and open squares in Figure 1).

Dedicated to Professor Pekka Pyykkö on the occasion of his 65th birthday