Although elemental gold has long been recognized for its noble character and chemical inertness, ionic gold hosts a rich variety of chemistry. Its inertness is partially a result of relativistic effects that are especially pronounced for heavy atoms. Relativistic effects result in a contraction and stabilization of the gold 6s atomic orbital and make gold the most electronegative metal on the periodic table. Gold is so electronegative, in fact, that it can act as an anion when paired with electropositive elements, such as cesium and rubidium.¹ Relativistic effects also explain why the 2+ oxidation state of gold is unstable and rarely observed: the 5d atomic orbitals of gold are well-shielded from the nucleus and therefore destabilized, which makes the 5d⁹ electronic configuration of gold energetically unfavorable.² This is in stark contrast to copper, a member of the same family as gold, which has a ubiquitous 2+ oxidation state. To overcome the energetic penalty associated with the 5d⁹ electronic configuration, gold atoms can form bonds with one another to pair and stabilize the high-energy 5d electrons, such as in Au₂(SO₄)₂.³ Alternatively, mononuclear gold(II) can be stabilized by using redox non-innocent ligands, such as dithiolenes or cyclic thioethers,^4-6 or through exotic superacid chemistry, such as with Au₃F₈ • 2(SbF₅) and AuXe₄²⁺ (Sb₂F₁₁⁻)₂.^7-8

Generally, gold prefers to adopt the 1+ or 3+ oxidation states, sometimes at the same time within a single material, such as in the double perovskite Cs₂Au^IAu^IIIX₆ (X = Cl, Br, I).^9-10 This perovskite has been studied for over a century, including high pressure studies attempting to access the 2+ oxidation state of gold through charge comproportionation.^11-12 While searching the veritable trove of literature on Cs₂Au^IAu^IIIX₆ as I was working on a prior study of the “gold-cage” perovskites,¹³ I noticed that the original 1922 report from Wells additionally reported a composition with the formula Cs₄Cu^IIAu^III₂Cl₁₂.⁹ Given the recent interest in metal halides, I was surprised to find only a handful of subsequent reports on this material and its palladium analog, Cs₄Pd^IIAu^III₂Cl₁₂, with the most recent report being a PhD thesis from 2004.^{9, 14-18} This was despite an exhaustive literature search that included texts that were unavailable online and were only available at the library on microfilm. We therefore decided to synthesize these materials and study their structures and optoelectronic properties. In the process of developing their syntheses, I noticed that omitting palladium and copper from the synthesis yielded a few tiny black crystals mixed in among the large yellow needles of the expected product, CsAu^IIICl₄. Collecting a single-crystal X-ray diffraction pattern of one of these black crystals gave confusing results—the material had cubic structure with a formula of Cs₄Au₃Cl₁₂ and a single unique AuCl₄ square-planar complex having Au–Cl bond lengths of 2.32 Å, slightly larger than the well-precedented value of 2.29 Å for Au³⁺–Cl. We couldn’t figure out how this formula would balance in charge having only gold(III), and we figured that we must have been missing something.

 After dragging my feet for some time, I finally decided to spend a few hours hunched over the microscope mechanically separating enough black crystals to get a powder X-ray diffraction pattern. This pattern revealed a pronounced peak splitting that did not match the cubic single-crystal structure, which prompted a reexamination of the single-crystal diffraction. Taking a page from our prior work on the “gold-cage” perovskites, which had crystals that twinned on the micron length scale and required fragmentation into tiny pieces to find single domains, I crushed one of the black crystals into tiny pieces of about 15 microns. When I solved the crystal structure from one of these tiny fragments, it all came together. 

I still remember the moment when I first solved the crystal structure of Cs₄Au^IIAu^III₂Cl₁₂ in the correct, tetragonal space group, which revealed three distinct AuCl₄ complexes, one of which had longer Au–Cl bonds than the other two. I knew what this meant: it was gold(II). I was elated; I went out and bought a double cheeseburger, fries, and a large milkshake. However, no reasonable scientist would be convinced of the presence of mononuclear gold(II) from slightly longer bond lengths in a single-crystal structure. But how would we collect the data we needed to prove my hypothesis when we could only produce a few milligrams of tiny crystals at below 1% yield through painstaking mechanical separation?

As it turns out, solving this problem of obtaining a phase pure synthesis was as easy as adding a mild reductant to the crystallization solution for CsAu^IIICl₄. As luck would have it, ascorbic acid was the first reductant I tried, and it worked like a charm. The next challenge was to convince ourselves that we had the material we thought we had. Thankfully, our neighbors in the Solomon lab were excited to use their expertise in electron paramagnetic resonance spectroscopy to characterize an element which seldom has an unpaired spin. These EPR results were the first convincing evidence we had of a 2+ oxidation state for gold. We followed up with further structural characterization, X-ray absorption spectroscopy, and magnetometry, all of which agreed with the Cs₄Au^IIAu^III₂Cl₁₂ formulation. But we wanted more proof, so we enlisted Prof. Dominic Ryan at McGill University to carry out ¹⁹⁷Au Mössbauer spectroscopy. This unconventional experiment is challenging due to the ~24 h half-life of the ¹⁹⁷Pt source, which meant that the source needed to be shipped immediately after activation at the McMaster Reactor, followed by an experimental marathon to collect the data before the activity of the source was too low. After the results of this experiment, too, concurred with the other evidence supporting Cs₄Au^IIAu^III₂Cl₁₂ as the formulation, we breathed a collective sigh of relief after holding our breaths for so long hoping that we weren’t mistaken about having gold(II). Around the same time, our computational collaborators in the Neaton group at UC Berkeley began to see results consistent with our experiments, allowing us to conclude that there is an intervalence charge-transfer interaction between gold(II) and gold(III) in the material and again providing further support of our formulation. 

This study really was a surprise for us; we did not expect to stabilize mononuclear gold(II) in an air-stable material with simple ligands and a simple synthesis. We hope that this report will spur further studies into the properties of gold(II), perhaps with the possibility of accessing exotic electronic transport properties by leveraging the d⁸/d⁹ metal combination, like what’s seen in some of the cuprate high-temperature superconductors. Stay tuned for our report on the bromide analog of Cs₄Au^IIAu^III₂Cl₁₂, which has a rich and complicated phase space and is surprisingly quite distinct from its chloride counterpart.

References

1. Norrby, L. J., Why is mercury liquid? Or, why do relativistic effects not get into chemistry textbooks? J. Chem. Educ. 1991, 68 (2), 110.
2. Heinze, K., The Quest for Mononuclear Gold(II) and Its Potential Role in Photocatalysis and Drug Action. Angew. Chem., Int. Ed. 2017, 56 (51), 16126-16134.
3. Wickleder, M. S., AuSO₄: A True Gold(II) Sulfate with an Au₂ ⁴⁺ Ion. Z. Anorg. Allg. Chem. 2001, 627 (9), 2112-2114.
4. Waters, J. H.; Gray, H. B., A Stable Paramagnetic Complex of Gold. J. Am. Chem. Soc. 1965, 87 (15), 3534-3535.
5. Huang, D.; Zhang, X.; McInnes, E. J. L.; McMaster, J.; Blake, A. J.; Davies, E. S.; Wolowska, J.; Wilson, C.; Schröder, M., Crystallographic, Electrochemical, and Electronic Structure Studies of the Mononuclear Complexes of Au(I)/(II)/(III) with [9]aneS₂O ([9]aneS₂O = 1-oxa-4,7-dithiacyclononane). Inorg. Chem. 2008, 47 (21), 9919-9929.
6. Blake, A. J.; Greig, J. A.; Holder, A. J.; Hyde, T. I.; Taylor, A.; Schröder, M., Bis(1,4,7-trithiacyclononane)gold Dication: A Paramagnetic, Mononuclear Au^II Complex. Angew. Chem. Int. Ed. 1990, 29 (2), 197-198.
7. Seidel, S.; Seppelt, K., Xenon as a Complex Ligand: The Tetra Xenono Gold(II) Cation in AuXe₄²⁺(Sb₂F₁₁^–)₂. Science 2000, 290 (5489), 117-118.
8. Hwang, I.-C.; Seppelt, K., The Reduction of AuF₃ in Super Acidic Solution. Z. Anorg. Allg. Chem. 2002, 628 (4), 765-769.
9. Wells, H. L., Some Complex Chlorides containing Gold. II. Cesium Triple Salts. Am. J. Sci. 1922, 3 (17), 315-326.
10. Elliott, N.; Pauling, L., The Crystal Structure of Cesium Aurous Auric Chloride, Cs₂AuAuCl₆, and Cesium Argentous Auric Chloride, Cs₂AgAuCl₆. J. Am. Chem. Soc. 1938, 60 (8), 1846-1851.
11. Liu, X. J.; Moritomo, Y.; Nakamura, A.; Kojima, N., Pressure-induced phase transition in mixed-valence gold complexes Cs₂Au₂X₆ (X =Cl and Br). J. Chem. Phys. 1999, 110 (18), 9174-9178.
12. Kojima, N.; Hasegawa, M.; Kitagawa, H.; Kikegawa, T.; Shimomura, O., P–T Phase Diagram and Gold Valence State of the Perovskite-Type Mixed-Valence Compounds Cs₂Au₂X₆ (X = Cl, Br, and I) under High Pressures. J. Am. Chem. Soc. 1994, 116 (25), 11368-11374.
13. Lindquist, K. P.; Boles, M. A.; Mack, S. A.; Neaton, J. B.; Karunadasa, H. I., Gold-Cage Perovskites: A Three-Dimensional Au^III–X Framework Encasing Isolated MX₆^3– Octahedra (M^III = In, Sb, Bi; X = Cl^–, Br^–, I^–). J. Am. Chem. Soc. 2021, 143 (19), 7440-7448.
14. Heines, P. Darstellung und Untersuchungen ternärer und quaternärer Halogenopalladate mit besonderer Beachtung von Redoxreaktionen: Pd(II) / Pd(IV) im Festkörper. Universität Dortmund, 2004.
15. Ferrari, A.; Cecconi, R.; Cavalca, L., Isomorfismo fra esacloroaurati di elementi a valenz diversa. – Nota IV sull'sistenza degli esacloroaurati. Gazz. Chim. Ital. 1943, 73, 23-29.
16. Linus Pauling Day-by-Day (September 6, 1937). http://scarc.library.oregonstate.edu/ (accessed January 31, 2022).
17. Ferrari, A.; Cavalca, L.; Nardelli, M., Su alcuni cloroaurati e su un caso di ampio isomorfismo di massa. Gazz. Chim. Ital. 1955, 85, 137-144.
18. Gomm, P. S.; Underhill, A. E., Electrical conduction studies on the mixed-valence triple salts of gold. Inorg. Chem. Nucl. Chem. Lett. 1974, 10 (4), 309-313.