Stuck in Our Teeth? Crystal Structure of a New Copper Amalgam, Cu₃Hg

Abstract

We have synthesized a new Cu amalgam, the Cu-rich phase Cu${}_3$Hg. It crystallizes with the Ni${}_3$Sn structure type with a hexagonal unit cell (space group P6${}_3$/mmc , a = 5.408(4) Å, c = 4.390(3) Å) and shows some mixed occupancy of Cu on the Hg site, resulting in a refined composition of Cu${}_{3.11}$Hg${}_{0.89}$. This is the first example of an amalgam with the Ni${}_3$Sn structure type where Hg is located mainly on the Sn site. Cu${}_3$Hg might be one of the phases constituting dental amalgams and therefore has major relevance, as well as the only Cu amalgam phase described so far, Cu${}_7$Hg${}_6$ with the $\gamma$-brass structure. It occurs as a biphase in our samples. Thermal decomposition of Cu amalgam samples in a dynamic vacuum yields nanostructured copper networks, possibly suitable for catalytic applications.

1. Introduction

Modern dental amalgams consist of Hg, Ag, Sn, Cu and minor metallic additives. In order to prepare a typical amalgam filling, liquid mercury is mixed with powders or shavings of an alloy containing Ag, Sn and Cu in the suitable composition: 40–70 weight % Ag, 12–30% Sn and 12–24% Cu. Additives can be Zn, In or Pd up to ca. 4% [1,2,3]. The alloy is mixed with an equal weight amount of Hg to form the plastic amalgam, solidifying within minutes to hours. The solidification goes along with the formation of a number of intermetallic phases. The most important in the resulting metal matrix composite are Ag${}_2$Hg${}_3$ [4] as the major matrix phase, Ag${}_3$Sn [4,5] as the mechanically strongest phase, Sn${}_8$Hg [6] as the most corrosive phase, Ag${}_5$Sn [7], Cu${}_6$Sn${}_5$ and Cu${}_3$Sn [8,9,10], amongst others [11,12]. The $\gamma$2-phase Sn${}_8$Hg is posing a major problem as it is relatively corrosive and leads to degradation of the filling over the years, especially at the tooth-amalgam interface. Today, so-called $\gamma$2-free amalgams are in use, which are higher in Cu content [13,14]. As the reaction takes place at physiological temperatures, the resulting alloy may not be a thermodynamically stable system. Other combinations of the metal atoms may also be considered: Cu amalgams and Ag-Cu solid solutions, as well as ternary and quaternary phases may also come into play at equilibrium conditions.

First reports on the employment of amalgams in dental fillings date back to 659 during the Tang dynasty in ancient China [15]; in European medical history, the first evidence of amalgam fillings adopting the Chinese recipe dates back to the end of the 16th century [16]. Despite the slow, but constant replacement of amalgams by other dental filling materials [17], many people still rely on amalgam fillings as they are resistive for a very long time, and the mechanical removal of existing amalgam fillings harbors the risk of unnecessary Hg release. With respect to the vast employment of dental amalgams, it is very surprising that new binary Cu-Hg phases can be found. It becomes even close to improbable when taking into account that both mercury and copper have been important materials since prehistoric times: the combination of both is employed, e.g., in the centuries-old fire gilding process where a copper surface is amalgamated by submerging in a Hg nitrate solution prior to applying gold amalgam and subsequent distillation of the Hg. Despite the widespread and longtime preparation of Cu-Hg phases, knowledge hereon is scarce. The published phase diagram [18] shows only one phase, Cu${}_7$Hg${}_6$, and even for this phase, there exist concurring structure descriptions, some of them expressing reasonable doubt about the others [19,20,21,22]. Details on the behavior of the liquidus curve or thermodynamic stability ranges also remain unclear.

The problem with identifying amalgam phases often lies in their low thermal stability. As Hg has a very low melting point, the liquidus curve in most phase diagrams shows a steep run in the Hg-rich region, and the Hg-richest phases normally show low peritectic decomposition temperatures (see

Table 1. Decomposition temperatures or melting points of Hg-rich amalgams AHg${}_x$ with x > 3 and with A = non-Hg metal. Values are taken from [18] if not given otherwise. Congruent melting temperatures are given in italic numbers.

Composition	Hg/A Ratio x	Decomposition Temperature (${}^{\circ }$C)
Cs${}_2$Hg${}_{27}$	13.5	12 [23]
KHg${}_{11}$	11.0	70 [24]
RbHg${}_{11}$	11.0	70 [24]
CaHg${}_{11}$	11.0	84
SrHg${}_{11}$	11.0	63 [24]
BaHg${}_{11}$	11.0	162 [24]
(N(CH${}_3$)${}_4$)Hg${}_8$	8.0	9 [25]
Rb${}_3$Hg${}_{20}$	6.67	132
Cs${}_3$Hg${}_{20}$	6.67	158
KHg${}_6$	6.0	170 [26]
BaHg${}_6$	6.0	410
Ba${}_{20}$Hg${}_{103}$	5.15	505
K${}_7$Hg${}_{31}$	4.43	187
Rb${}_7$Hg${}_{31}$	4.43	162
A${}_{11}$Hg${}_{45}$ ${}^1$	4.09	145–720
PtHg${}_4$	4.0	≥200 [27]
Rb${}_5$Hg${}_{19}$	3.8	193
Cs${}_5$Hg${}_{19}$	3.8	164
K${}_3$Hg${}_{11}$	3.67	195
A${}_{14}$Hg${}_{51}$ ${}^2$	3.64	157–480
K${}_2$Hg${}_7$	3.5	202 [28]
Rb${}_2$Hg${}_7$	3.5	197 [28]

${}^1$ Amalgams with the Sm${}_{11+x}$Cd${}_{45-x}$ structure type show compositional ranges between x = 3.79 (La, T = 720 ${}^{\circ }$C) and 4.09 (Gd, T = 145 ${}^{\circ }$C) and have been reported for A = La, Ce, Pr, Nd, Sm, Gd and U [29,30]. ${}^2$ Amalgams with variants of the Gd${}_{14+x}$Ag${}_{51-x}$ structure type show compositional ranges between x = 4.63 (Na, T = 157 ${}^{\circ }$C) and 5.44 (Sr, T = 480 ${}^{\circ }$C) and have been described for A = Na, Ca, Sr, Eu and Yb [29,31,32,33].

). Standard solid state preparation techniques usually are not suitable to prepare those delicate phases. Thermoanalytical methods can show the presence of low-decomposing phases; however, if accompanying structural elucidations are impeded, their exact compositions remain unclear. In addition, thermoanalytical studies can be hampered by kinetic effects, which are much more pronounced at low than at high temperatures (seed formation, phase transformations and the like). Problems with structural analyses of crystalline amalgams with high Hg content are caused by their high X-ray absorption coefficients. A special case is given for the widely-employed Au amalgams, which show an especially inconvenient combination of high absorption and especially low X-ray contrast, leading to only very few structurally-sufficiently described phases and a rudimentary phase diagram [18].

It now becomes clear that with all the named complications, a number of new amalgams can be expected to be found even in systems that have been employed technically for a very long time. Both synthesis and structural analysis require non-standard techniques and diligence.

2. Results

Single crystals of Cu${}_3$Hg were found together with crystals of Cu${}_7$Hg${}_6$ on the surface of an amalgamated copper spoon after 12 days at 90 ${}^{\circ }$C. Both phases form crystals of bronze color when Hg-free and of bright silver luster when a thin film of Hg covers the surface. Cu${}_3$Hg forms prismatic crystals, whereas Cu${}_7$Hg${}_6$ forms platelets. Both amalgams are air-stable at room temperature at least over several weeks. A rod-shaped crystal of Cu${}_3$Hg was selected under a binocular and glued on top of a glass fiber. Data were carefully corrected for considerable absorption effects on the basis of indexed crystal faces. Metric and extinction conditions pointed towards the Ni${}_3$Sn structure type (space group P6${}_3$/mmc), and structure solution and refinement showed this to be correct. Crystallographic details and results of the single-crystal structure refinement are compiled in

Table 2. Crystallographic data and details on single crystal data collection, structure solution and refinement for Cu${}_{3.11\left(4\right)}$Hg${}_{0.89\left(4\right)}$. Data collection was performed at room temperature. All standard deviations are given in parentheses in units of the last digit.

Composition	Cu${}_{3.11\left(4\right)}$Hg${}_{0.89\left(4\right)}$
Crystal system	hexagonal
Space group	P$6_3$/mmc, (No. 194)
Lattice parameters, a, c (Å)	5.408(4), 4.390(3)
Unit cell volume, V (Å${}^3$)	111.18(13)
No. of formula units, Z	2
Density (X-ray), $\rho$ (g cm${}^{-3}$)	11.235
Diffractometer	STOE IPDS 1, Ag$K_{\alpha }$ radiation
Data collection temperature, T (K)	295(2)
Absorption coefficient, $\mu$ (mm${}^{-1}$)	48.62
Diffraction angle range, $\vartheta$ (${}^{\circ }$)	5.02–23.53
Index range	−7 ≤ h, k ≤ 7, −6 ≤ l ≤ 6
No. of collected reflections	2154
No. of independent reflections	72
No. of independent reflections (I ≥ 2$\sigma$(I))	71
$R_{int}$	0.2198
$R_{\sigma }$	0.0485
Structure factor, F(000)	322.8
Corrections	Lorentz, polarization, absorption effects
Absorption correction	numerical [34,35]
Structure solution	direct methods [36]
Structure refinement	full-matrix least-squares on $F^2$ [36]
No. of lest-squares parameters	9
GooF	1.180
R values (I ≥ 2$\sigma$(I))	R1 = 0.0345, wR2 = 0.0757
R values (all data)	R1 = 0.0368, wR2 = 0.0773
Residual $\rho$(e${}^{-}$) max/min (e${}^{-}$Å${}^{-3}$)	+1.698/−1.419
Extinction coefficient	0.047(4)

,

Table 3. Standardized fractional atomic coordinates [37] and equivalent isotropic displacement parameters (Å${}^2$) for Cu${}_{3.11\left(4\right)}$Hg${}_{0.89\left(4\right)}$. The equivalent isotropic displacement parameter is defined as 1/3 of the trace of the anisotropic displacement tensor. All standard deviations are given in parentheses in units of the last digit.

Atom	Occupation Factor	Wyckoff Letter	x	y	z	${\mathit{U}}_{\mathit{equiv}}$
Hg1	0.89(5)	2d	$\frac{1}{3}$	$\frac{2}{3}$	$\frac{3}{4}$	0.0160(7)
Cu1	0.11(5)	2d	$\frac{1}{3}$	$\frac{2}{3}$	$\frac{3}{4}$	0.0160(7)
Cu2	1	6h	0.1589(3)	2x	$\frac{1}{4}$	0.0201(18)

,

Table 4. Coefficients U${}_{ij}$ of the anisotropic displacement tensor (Å${}^2$) for Cu${}_{3.11\left(4\right)}$Hg${}_{0.89\left(4\right)}$. U${}_{ij}$ is defined as U${}_{ij}$ = exp{$-2{\pi }^2[U_{11}{\left(h{a}^{\ast }\right)}^2+\dots +2U_{21}hk{a}^{\ast }{b}^{\ast }]$}. All standard deviations are given in parentheses in units of the last digit.

Atom	${\mathit{U}}_{11}$	${\mathit{U}}_{12}$	${\mathit{U}}_{33}$	${\mathit{U}}_{23}$	${\mathit{U}}_{13}$	${\mathit{U}}_{12}$
Hg1	0.0156(7)	=U${}_{11}$	0.0168(8)	0	0	0.0078(4)
Cu1	0.0156(7)	=U${}_{11}$	0.0168(8)	0	0	0.0078(4)
Cu2	0.0195(19)	0.024(2)	0.019(2)	0	0	0.0118(10)

and

Table 5. Selected interatomic distances and their frequencies in Cu${}_{3.11\left(4\right)}$Hg${}_{0.89\left(4\right)}$ in Å. All standard deviations are given in parentheses in units of the last digit.

Atom 1	Atom 2	Distance
Hg1	Cu2	2.7048(19) (6×)
	Cu2	2.736(2) (6×)
Cu2	Cu2	2.578(5) (2×)
	Cu2	2.6521(19) (4×)

.

Cu${}_3$Hg crystallizes with the Ni${}_3$Sn structure type. The lattice parameter a of the Ni${}_3$Sn structure type is doubled with respect to simple hcp by creating two crystallographic sites with non-equal Wyckoff numbers for Sn (Hg) on 2d and Ni (Cu) on 6h while retaining the original space group type P6${}_3$/mmc. This shows the Ni${}_3$Sn structure type to be a coloring variant of hcp as the hcp topology is retained in an ordered packing of two different spheres. By the coloring, Hg atoms are only coordinated by 12 Cu atoms forming [HgCu${}_{12}$] anticuboctahedra. Cu atoms form streaks of trans-face-sharing [Cu${}_6$] octahedra (see Figure 1). In an alternative picture, the crystal structure can be visualized with plane hexagonal nets at heights z = 1/4 and 3/4, which are stacked along c with ...ABAB... periodicity, where A and B are shifted by (1/3, 2/3, 1/2) with respect to each other.

During structure refinement, unusual residual electron densities and weight factors pointed towards possible mixed occupation on the two crystallographic sites. This was tested by independently refining mixed occupations with retaining an overall full occupation of the two sites. The Cu2 site on 6h shows no sign of mixed occupation by Hg, whereas there is significant Cu content on the Hg/Cu1 site on 2d (see Table 3). Refinement of the mixed occupation according to the denotation Cu${}_3$(Hg${}_{1-x}$Cu${}_x$) leads to x = 0.11(4) and the overall refined composition Cu${}_{3.11\left(4\right)}$Hg${}_{0.89\left(4\right)}$.

Figure 2 shows a crystal of Cu${}_3$Hg under an optical microscope (left) and the specimen used for X-ray data collection after EDX measurements (right). The severe degradation of Cu${}_3$Hg in an electron beam under high vacuum becomes evident. The sensitivity of the amalgam towards decomposition in a vacuum hampers detailed EDX analyses; still, we managed to record spectra at three different points of the crystal. The recorded spectra show the absence of further metal atoms and a composition similar to the single-crystal refinement results, but slightly lower in Cu content. The mean measured composition is Cu${}_{2.79}$Hg (see

Table 6. Results of EDX spectroscopic measurements on the Cu${}_3$Hg crystal shown in Figure 2 (right). The values are averaged over three measurements on different spots. The largest deviations from the mean value are given in parentheses. Expected values are calculated for the composition Cu${}_{3.49}$Hg.

Atom	Atom-% Detected	Atom-% Calculated
Cu	73.6(2)	77.7
Hg	26.4(2)	22.3

), calculated: Cu${}_{3.49}$Hg. This is somehow surprising as decomposition in a high vacuum of the electron microscope can clearly be seen in the crystal picture, which would mean evaporation of Hg and therefore a higher Cu content with respect to the results from single-crystal refinement. A reason for this deviation could be the formation of a Hg surface layer after decomposition and prior to evaporation of the Hg in the vacuum. As EDX primarily is a surface-sensitive method, Hg may thus be overrepresented in the results.

In order to further explore the phase width of Cu${}_{3+x}$Hg${}_{1-x}$ and to gather knowledge about thermal stability and phase relations of both Cu amalgams Cu${}_3$Hg and Cu${}_7$Hg${}_6$, thermochemical investigations are in progress.

3. Discussion

All amalgams adopting the Ni${}_3$Sn structure type [38,39,40,41] are compiled in

Table 7. Amalgams AHg${}_3$ (A = non-Hg metal) adopting the Ni${}_3$Hg structure type. Hg has an atomic radius of 151 pm [42]. The ideal c/a ratio for hcp is 1.633. Due to a doubling of a, the ideal c/a ratio in the Ni${}_3$Sn structure type is 0.8167. In Ni${}_3$Sn (r${}_{Ni}$ = 124 pm, r${}_{Sn}$ = 132 pm, ratio r${}_{Sn}$/r${}_{Ni}$ = 1.06), the c/a ratio is 0.802 [40].

Phase AHg${}_3$	r${}_{\mathit{A}}$ [pm]	r${}_{\mathit{A}}$/r${}_{\mathit{Hg}}$	c/a
LiHg${}_3$ [29,47]	152	1.01	0.768
CaHg${}_3$ [31]	197	1.30	0.757
SrHg${}_3$ [31,48]	215	1.42	0.741
ScHg${}_3$ [29,49]	160	1.06	0.748
YHg${}_3$ [29,49,50]	181	1.20	0.744
LaHg${}_3$ [29,51]	187	1.24	0.729
CeHg${}_3$ [52]	182	1.21	0.731
EuHg${}_3$ [31]	180	1.19	0.747
GdHg${}_3$ [29,53]	180	1.19	0.742
TbHg${}_3$ [53]	177	1.17	0.748
DyHg${}_3$ [29,53]	178	1.18	0.746
HoHg${}_3$ [29,53]	176	1.17	0.747
ErHg${}_3$ [29,53]	176	1.17	0.748
TmHg${}_3$ [29,53]	176	1.17	0.748
YbHg${}_3$ [53]	176	1.17	0.761
LuHg${}_3$ [29,53]	174	1.15	0.750
ThHg${}_3$ [54]	179	1.19	0.730
UHg${}_3$ [55]	156	1.03	0.735

. Always, the minority component A is coordinated only by Hg atoms in an anticuboctahedral environment, and all known examples have the Hg-rich composition MHg${}_3$. This can be seen as an expression of some Coulombic contributions A${}^{\delta +}$[Hg${}_3$]${}^{\delta -}$ and the formation of coordination spheres of negatively polarized Hg atoms around the cationic species A. This is corroborated by the fact that this class of amalgams is only formed by explicitly electropositive A metals.

The Ni${}_3$Sn structure type is a favored sphere packing in amalgams where the atomic radii of A do not differ much from the one of Hg (151 pm, [42]). Differences according to Table 7 seem only favorable if the A metal is larger than Hg. The largest difference can be found for Sr with an atomic radius 42% larger than the one of Hg. No amalgams are known with A atoms smaller than Hg so far. In addition, it can be stated that all known amalgams of this type show a ratio of the lattice parameters c/a smaller than its ideal value for a packing of incompressible spheres of equal size with all spheres in contact. The ab plane is widened with respect to the stacking of the planes. This distortion cannot be due to only geometrical reasons because the larger A atoms should distort the packing equally in all directions, not only the ab plane. The anticuboctahedral coordination polyhedra [AHg${}_{12}$] thus are oblate with the six equatorial Hg atoms having larger distances than the other six. The variation of the c/a ratio does not follow the quotient of the atomic radii: c/a is largest (and therefore closest to the ideal geometric ratio) for LiHg${}_3$ (0.768) and for YbHg${}_3$ (0.761), while Li and Hg have almost identical atomic radii, and Yb is 17% larger than Hg. The smallest c/a ratios are found for LaHg${}_3$ (0.729, La being 24% larger than Hg), CeHg${}_3$ (0.731, 21%), Th (0.730, 19%) and U (0.735, 3%). Geometric, as well as electronic reasons may contribute to the structural distortion, and heteroatomic interactions seem to have higher bonding contributions than homoatomic ones, as suggested by studies on the electronic structure of Ni${}_3$Sn [39]. In the amalgam cases, Coulombic contributions may be the underlying reason; however, the degree of interplay of different contributions still remains speculative. One could assume bonding Hg${}^{\delta -}$-Hg${}^{\delta -}$ interactions to cause a contraction of the [AHg${}_{12}$] anticuboctahedra along the [001] direction, as Hg${}^{\delta -}$ formally behaves like an early p group metal [43] and tends to form Hg-Hg bonding.

A similar case of deviation from ideal sphere packing geometry is known for elemental mercury. Rhombohedral $\alpha$-Hg crystallizes in a distortion variant of fcc where the individual hexagonal sphere planes are widened with respect to their stacking along the {111} directions of the cubic fcc cell, leading to a rhombohedral metric with $\alpha$ = 70.44(6)${}^{\circ }$ [44] instead of 60${}^{\circ }$ for an ideal fcc packing. Band structure calculations show differences in bond strengths for $\sigma$- and $\pi$-type interactions between the Hg(p${}_z$) orbitals within and between the hexagonal layers, resulting in the structural distortion of the sphere packing [45,46].

Most interesting within this context is Cu${}_3$Hg as the first amalgam in this structure type with a reversed composition. In comparison to the other amalgams, Cu${}_3$Hg may be seen as the anti-Ni${}_3$Sn-type, as it is the only amalgam where the A element constitutes the majority component, and Hg is coordinated in an anticuboctahedron of Cu atoms. The structural deviation from ideal geometry leads to a c/a ratio of 0.812, which is much closer to the ideal value of 0.8167 than for all amalgams with AHg${}_3$ composition. Here, the bonding situation is obviously quite different. Obviously, Coulombic interactions cannot play a major role here, as Cu is not a very electropositive metal. This would explain the absence of bonding Hg-Hg contributions and therefore no contraction along c.

While the details of chemical bonding still remain unclear, the inversed composition may also be explained by geometrical reasons. If the Ni${}_3$Hg structure type demands a structural necessity for the minority component to be the bigger atom, then Cu with atomic radius of 128 pm [42] favors the formation of Cu${}_3$Hg with a quotient r${}_{Hg}$/r${}_{Cu}$ = 1.18 over the formation of CuHg${}_3$ with a quotient of r${}_{Cu}$/r${}_{Hg}$ = 0.878. If the atomic radii really play a decisive role in the formation of amalgams of the Ni${}_3$Sn structure type with a composition A${}_3$Hg, then the respective amalgams should also be found for A = Cr (r = 128 pm), Mn (r = 127 pm), Fe (r = 126 pm), Co (r = 125 pm) and Ni (r = 124 pm). As from these elements only Mn and Ni are known to form amalgams and as from these no amalgam with composition A${}_3$Hg is known so far, we consider it worthwhile to have a closer look at the respective binary phase diagrams in the near future.

4. Materials and Methods

4.1. Synthesis of Copper Amalgams

We did not intend to prepare copper amalgams. Crystals of Cu${}_3$Hg and Cu${}_7$Hg${}_6$ were the serendipitous results of the attempt to synthesize Hg-rich uranium amalgams by electro-crystallization, starting from a solution of UI${}_4$ in N,N-dimethylformamide (DMF) and a reactive cathode consisting of a Hg drop suspended in an amalgamated Cu spoon. The preparative approach to temperature-sensitive Hg-rich amalgams via electro-crystallization had proven very convenient in previous cases [23,32,33,56,57]. Electro-crystallization was performed at 90 ${}^{\circ }$C for two weeks and resulted in the formation of nanocrystalline UO${}_2$ (see Figure 3, left), while the Hg drop formed the copper amalgams Cu${}_3$Hg and Cu${}_6$Hg${}_7$ over the reaction time of two weeks. The reason for the unsuccessful U amalgam synthesis is that UI${}_4$ dissolves in molecular form in DMF and does not form ionic complexes [U(DMF)${}_x$]${}^{4+}$ as other metal iodides do. This has been shown by recording UV-Vis spectra of the solutions and comparison with the literature.

The reaction of a copper foil with mercury at 105 ${}^{\circ }$C was repeated outside of the electro-crystallization chamber and also resulted in the formation of the Cu amalgams Cu${}_3$Hg and Cu${}_7$Hg${}_6$ directly from the elements; see Figure 3. Both amalgams decompose at temperatures between 110 and 200 ${}^{\circ }$C. By heating the amalgamated copper foil in a dynamic vacuum to 200 ${}^{\circ }$C, the released mercury is distilled off, and a homogeneous copper network with interpenetrating nano-scaled copper rod structures is formed (see Figure 4). This copper network with high porosity might be interesting for utilization in heterogeneous catalysis reactions due to its simple and straightforward preparation.

4.2. Single Crystal Investigations

Single crystals of the copper amalgams were selected under binoculars, and specimens with well-shaped faces (see Figure 2, left) were glued on top of a glass fiber with cyanoacrylate glue. They were centered on the one-circle goniometer of a diffractometer system IPDS 1 (Stoe & Cie., Darmstadt, Germany) equipped with graphite-monochromatized Ag-K$\alpha$ radiation (fine-focus X-ray tube) and an imaging plate detector After checking crystal quality and crystal metrics, data collection was performed in $\phi$ scans; data of the accessible part of an entire Ewald sphere were collected. Data were corrected for Lorentz and polarization effects with the diffractometer software package [58]. Absorption corrections were performed carefully on the basis of optimized indexed crystal faces [34,35,58]. Structure solution for Cu${}_3$Hg was performed with direct methods [36] in space group P6${}_3$/mmc as indicated by extinction conditions together with statistics on E${}^2$−1 (E being the normalized structure factor). Structure refinement was performed with full-matrix least-squares cycles [36] on F${}^2$ (F being the structure factor). Both atoms were treated with anisotropic displacement tensors. Further information can be obtained from Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (fax +49 (0)7247 808 666; e-mail: [email protected]; http://www.fiz-informationsdienste.de/en/DB/icsd/depot_anforderung.html), on quoting the deposition number CSD-433425.

4.3. Powder Diffractometry

For powder diffraction, the amalgam samples were prepared by grinding together with diamond powder for both optical dilution due to high absorption coefficients and mechanical reasons as the amalgams are rather ductile. The powders were sealed in glass capillaries (⌀ = 0.3 mm), and data collection was performed on a diffractometer system STADI P (Stoe & Cie., Darmstadt, Germany) equipped with Ge-monochromatized MoK$\alpha$1 radiation and a Mythen 2K detector in parafocusing Debye–Scherrer geometry. For better comparison of the recorded diffraction patterns with calculated patterns based on single-crystal data from literature, they were converted to Cu-K$\alpha$1 wavelength.

4.4. EDX Spectroscopy

EDX spectra were collected on an electron microscope system JSM-6500F (Jeol, Freising, Germany) with a field emission source and EDX detector at 25 kV.