NaGe6As6 is a ternary arsenide prepared by reaction of the elements at 650 °C. It crystallizes in a new monoclinic structure type [space group C2/m, Z = 2, a = 22.063(2), b = 3.8032(4), c = 7.2020(8) Å, β = 92.7437(15)°] that can be considered to be derived by inserting guest Na atoms between [Ge6As6] layers identical to those found in the layered binary arsenide GeAs. An unusual feature in both structures is the presence of ethane-like Ge2As6 units in staggered conformation, with Ge–Ge dumbbells oriented either parallel or perpendicular to the layers. Electronic band structure calculations have shown that the electron excess in NaGe6As6 is accommodated by raising the Fermi level across a 0.6 eV band gap in semiconducting GeAs so that it cuts the bottom of the conduction band, resulting in an n-doped semiconductor.
A large number of ternary pnictides A–M–Pn (A = electropositive metal; M = metal or metalloid from group 13 or 14; Pn = P, As, Sb, Bi) exhibit structures in which A cations are found in conjunction with heteronuclear anionic [MxPny]n– units . Most of these compounds satisfy the Zintl concept so that there is a definite relationship between the electron count and the crystal structure. In cases where the Pn atoms experience a deficiency of electrons with respect to the 8–N rule, polyanionic networks containing Pn–Pn bonds must necessarily develop. Less commonly, there is a surplus of electrons, which are assumed to reside on the M atoms, leading to formation of polycationic networks containing M–M bonds. Within the A–Ge–As (A = alkali or alkaline–earth metal) systems, the only known phases reported so far are Na5GeAs3 , K2GeAs2 , and BaGe2As2 . The latter is a polycationic compound with Ge atoms in homoatomic bonding arrangements of infinite cis–trans chains, consistent with the presence of Ge2+ species if oxidation states are assigned. A gas-phase molecular species NaGeAs3 has also been proposed .
As part of a broader investigation of other A-Ge-As phases, we report here the preparation and structure determination of the new compound NaGe6As6. Remarkably, it is related to the binary arsenide GeAs, a charge-balanced semiconducting compound that has so far been relatively little studied [6–14], through intercalation of Na atoms between layers. We examine the implications of this relationship for the electronic structure.
Na pieces (99.99 %), Ge ingots (99.999 %), and As lumps (99.999 %), all obtained from Alfa-Aesar and handled within an argon-filled glovebox, were combined in a 1:3:3 molar ratio (in an attempt to prepare “NaGe3As3”), loaded into an alumina crucible, and placed within a fused-silica tube which was evacuated and sealed. The tube was heated to 650 °C, held at this temperature for 10 days, and cooled to room temperature over 2 days. The product contained small needle-shaped crystals (typically 0.2 mm in length) which were selected under paraffin oil. Energy-dispersive X-ray (EDX) analysis of these crystals on a JEOL JSM-6010LA scanning electron microscope indicated a composition of Na0.8(1)Ge6.7(5)As5.5(5), which is close to the composition NaGe6As6 obtained from the structure determination. After this composition was established, the compound could also be prepared through reaction of a stoichiometric ratio (1:6:6) of the elements under the same heating conditions as above.
2.2 Structure determination
A crystal of NaGe6As6 was mounted within a small droplet of paraffin oil and placed under a cold nitrogen gas stream on a Bruker PLATFORM diffractometer equipped with a SMART APEX II CCD area detector and a graphite-monochromated MoKα radiation source. Intensity data were collected at −80 °C using ω scans with a width of 0.3° and an exposure time of 15 s per frame at six different ϕ angles. Face-indexed numerical absorption corrections were applied. Structure solution and refinement were carried out with use of the Shelxtl (version 6.12) program package . The centrosymmetric monoclinic space group C2/m was chosen on the basis of Laue symmetry, systematic absences, and intensity statistics. The initial positions for seven sites were located by Direct Methods. Of these sites, one (at 2a) was assigned as Na, three (at 4i) as Ge, and three (at 4i) as As atoms on the basis of their coordination geometries; this model is the most chemically reasonable because it places the more electronegative As atoms as the coordinating atoms around the Na atoms. Because the scattering factors of Ge and As are similar, refinements of occupancies were generally unstable when sites were allowed to be mixed with both Ge and As. However, these refinements tended towards all Ge in the first set of three sites and all As in the second set of three sites. All sites were found to be fully occupied and had reasonable displacement parameters. Atomic positions were standardized with the program Structure Tidy . The final refinement led to good agreeement factors and a featureless difference electron density map. Crystal data and experimental details are listed in Table 1, positional and displacement parameters in Table 2, and interatomic distances in Table 3.
|Formula mass, amu||908.05|
|Space group||C2/m (No. 12)|
|ρcalcd, g cm−3||5.00|
|Crystal dimensions, mm3||0.22 × 0.04 × 0.03|
|Radiation||Graphite-monochromated MoKα, λ = 0.71073 Å|
|μ(Mo Kα), mm−1||31.1|
|2θ limits, deg||3.70–66.44|
|Data collected||−33 ≤ h ≤ 32, −5 ≤ k ≤ 5, −10 ≤ l ≤ 10|
|No. of data collected||4333|
|No. of unique data, including Fo2 < 0/Rint||1270/0.034|
|No. of unique data, with Fo2 > 2 σ(Fo2)||1039|
|No. of variables||42|
|R(F) for Fo2 > 2 σ(Fo2)a||0.026|
|Goodness of fitc||1.040|
|Δρmax/Δρmin, e Å−3||1.91/ −2.35|
aR(F) = Σ||Fo| − |Fc||/Σ|Fo|; bRw(F2) = [Σw(Fo2 − Fc2)2/Σw(Fo2)2]1/2, w = [σ2(Fo2) + (AP)2 + BP]−1, where P = (Max(Fo2, 0) + 2Fc2)/3 and A and B are constants adjusted by the program; cGoF = S = [Σw(Fo2 − Fc2)2/(nobs − nparam)]1/2, where nobs is the number of data and nparam the number of refined parameters.
|Atom||Wyckoff position||x||y||z||Ueq (Å2)a|
aUeq is defined as one-third of the trace of the orthogonalized Uij tensor.
Further details of the crystal structure investigation may be obtained from Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (fax: +49-7247-808-666; e-mail: firstname.lastname@example.org, http://www.fiz-karlsruhe.de/request_for_deposited_data.html) on quoting the deposition number CSD-430531.
2.3 Band structure calculations
Tight-binding linear muffin tin orbital band structure calculations were performed on GeAs and NaGe6As6 within the local density and atomic spheres approximation with use of the Stuttgart Tb-Lmto-Asa program (version 4.7) . The basis set consisted of Na 3s/3p/3d, Ge 4s/4p/4d, and As 4s/4p/4d orbitals, with the Na 3p/3d, Ge 4d, and As 4d orbitals being downfolded. Integrations in reciprocal space were carried out with an improved tetrahedron method over 164 (for GeAs) or 242 irreducible k points (for NaGe6As6) within the first Brillouin zone.
3 Results and discussion
NaGe6As6 is a new phase in the Na-Ge-As system, in which Na5GeAs3 and NaGeAs3 were the only previously reported representatives. Na5GeAs3 is a Zintl phase that contains Na+ cations and [Ge2As6]10− anions in the form of discrete edge-sharing tetrahedral units . NaGeAs3 was prepared in a Knudsen cell and is proposed to consist of gaseous “Zintl molecules” that are aggregates of Na+ cations and [GeAs3]− anions in the form of tetrahedral clusters isoelectronic to As4 .
The structure of NaGe6As6 is of a new monoclinic type (space group C2/m) containing [Ge6As6] layers alternately stacked with Na atoms along the c direction (Fig. 1a). Each layer is built by condensing chains of edge-sharing octahedra, each centered by Ge–Ge dumbbells, extending along the b direction (Fig. 1b). There are two types of Ge2As6 units, each of which resembles ethane in staggered conformation. These units are oriented so that Ge1–Ge1 pairs lie roughly parallel to the layers whereas Ge2–Ge3 pairs lie perpendicular to the layers. The sequence of these pairs (one horizontal Ge1–Ge1 pair followed by two vertical Ge2–Ge3 pairs) as they propagate along the a direction leads to a characteristic motif of two five-membered rings fused to opposite sides of a central six-membered ring. The Na atoms lying between the layers are surrounded by six As atoms in octahedral coordination. Other structures containing ethane-like Ge2As6 units have been previously reported in only a few examples: GeAs , A3Ge2As4 (A = Ca, Sr) , ACdGeAs2 (A = K, Rb) , K5In5Ge5As14 , and K8In8Ge5As17  (the latter two exhibit disorder of In and Ge atoms). Similar motifs of fused 5–6–5 rings are found in K5In5Ge5As14  and Ga3Te3I .
The [Ge6As6] layers in NaGe6As6 are the same as those found in the binary arsenide GeAs (Fig. 2). As the layers are stacked, they are in perfect registry in NaGe6As6 but they are displaced with respect to each other in GeAs. Both structures can also be regarded as being built from rumpled close-packed nets of As atoms stacked in an hcp (AB) arrangement, with Ge–Ge dumbbells centered in the octahedral sites between alternate pairs of layers. This viewpoint draws an analogy to the CdI2-type structure: when half of all available octahedral sites are filled with Ge–Ge dumbbells within a close-packed array of As atoms, the resulting formula is □1/2(Ge2)1/2(As), or GeAs. When a third of the remaining vacant octahedral sites between the GeAs layers is occupied by Na atoms, the resulting formula is □2/6(Na)1/6(Ge2)1/2(As), or NaGe6As6. This relationship is similar to the intercalation of guest atoms within layered transition-metal dichalcogenides .
The possibility that the Ge and As atoms are not ordered, as was assumed in the structure refinement, must be considered given their similar X-ray scattering factors. Evidence to support the ordered model comes from an analysis of the bond valence sums , which are 1.3 for the Na atom, 3.5–3.6 for the Ge atoms, and 2.6–3.0 for the As atoms, in agreement with the expected valences. If the assignments of Ge and As atoms are reversed, the bond valence sums would show worse agreement.
The derivation of NaGe6As6 by insertion of Na atoms into GeAs poses interesting questions about how the electronic structure is affected. GeAs itself can be considered to be a polycationic Zintl phase, attaining the charge-balanced formulation (Ge3+)(As3−) consistent with the presence of homoatomic Ge–Ge bonds within the dumbbells. (SiAs [10, 24, 25], GeP [10, 26], and GaTe  are the only other known compounds isostructural to GeAs.) On proceeding from GeAs  to NaGe6As6, there is virtually no change in the Ge–As (2.44(1)–2.47(1) vs. 2.4349(8)–2.4755(9) Å) and Ge–Ge distances (2.43(1)–2.46(1) vs. 2.4376(11)–2.4572(9) Å). It is reasonable to propose that electron transfer will take place from the Na atoms to the rest of the framework to give the formulation Na+[Ge6As6]−. If this increase in electron count is assumed to cause local chemical reduction of the Ge atoms, corresponding to the formulation (Na+)(Ge2.83+)6(As3−)6, one expects a strengthening of the Ge–Ge bonds, but this is not reflected in the observed distances. Interestingly, although insertion of Na atoms pries the layers further apart (the center-to-center distance between the layers expands from 6.4 to 7.2 Å), the closest As–As separation in the gap between the layers actually decreases slightly from 3.34(1) Å in GeAs to 3.270(1) Å in NaGe6As6. These As–As distances are certainly too long to be considered as fully covalent bonds, but it is notable that they are quite a bit shorter than twice the van der Waals radius of As (4.0 Å) .
The absence of significant structural changes within the layers implies that the major electronic effect of introducing Na atoms into GeAs is to essentially produce an n-doped semiconductor. This picture is confirmed by comparing the electronic band structures of GeAs and NaGe6As6. The density of states (DOS) curve for GeAs clearly reveals valence and conduction bands separated by a gap of 0.58 eV (Fig. 3a), which agrees fortuitously well with an optical band gap of 0.65 eV reported in early experimental measurements . The atomic projections of the DOS show that, as a result of the similar energies of Ge and As atomic orbitals, there is a strong hybridization of these states, which are dominated by 4s character at lowest energy (−14 to −6 eV) and by 4p character at higher energy (−5 to 0 eV). The crystal orbital Hamilton population (COHP) curves indicate that the filled valence bands correspond to Ge–As and Ge–Ge bonding levels. If a rigid band model is applied and the electron count is increased by one, corresponding to NaGe6As6, the Fermi level would be raised so that it lies at the bottom of the conduction band. Because antibonding levels now start to be occupied, the expectation is that Ge–As and Ge–Ge bonds will be weakened, or the structure may undergo distortion to avoid this situation. The DOS curve for the actual structure of NaGe6As6 (Fig. 3b) is very similar to that of GeAs. The calculated band gap is slightly larger (0.64 eV) and the Fermi level just cuts the bottom of the conduction band. The bonds do weaken, but only very slightly; the integrated COHP values (–ICOHP) change from 2.61 to 2.56 eV/bond for the Ge–As contacts and from 2.74 to 2.66 eV/bond for the Ge–Ge contacts on proceeding from GeAs to NaGe6As6. This small destabilization will, of course, be compensated by ionic bonding interactions between the Na atoms and the layers. The situation is reminiscent of Ba7Ga4Sb9, an “electron-rich metallic Zintl phase” in which the apparent violation of charge balance can be reconciled by recognizing the important stabilizing role of the cations [29, 30]. Except for the region around the Na atoms in NaGe6As6, the layers are non-interacting and held together only by van der Waals forces, as confirmed by the small and negative –ICOHP values for the interlayer As–As contacts (−0.03 eV/bond in GeAs and −0.10 eV/bond in NaGe6As6).
The successful preparation of NaGe6As6 suggests that the GeAs structure can tolerate modest changes of its electron count; it may thus be amenable to doping within layers or intercalation between layers by other elements. To establish if the proposed description of intercalation is accurate, it will be important to perform experiments to attempt deintercalation of NaGe6As6. Moreover, the physical properties will be expected to be modified; for example, metallic-like electrical conductivity can be predicted for NaGe6As6. It will be worthwhile to examine other compounds that are isostructural to GeAs, such as SiAs, or those with related layered structures, such as GaS and GaSe [31–35], to see if they can also accommodate guest atoms between the layers.
Dedicated to: Professor Wolfgang Jeitschko on the occasion of his 80th birthday.
This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC).
 B. Eisenmann, G. Cordier in Chemistry, Structure, and Bonding of Zintl Phases and Ions (Ed.: S. M. Kauzlarich), VCH, New York, 1996, chapter 2, p. 61.Search in Google Scholar
 B. Eisenmann, J. Klein, M. Somer, Z. Kristallogr.1991, 197, 265.Search in Google Scholar
 B. Eisenmann, J. Klein, J. Less-Common Met.1991, 175, 109.Search in Google Scholar
 B. F. Mentzen, R. Hillel, A. Michaelides, A. Tranquard, J. Bouix, C. R. Acad. Sci., Ser. II1981, 293, 965.Search in Google Scholar
 G. M. Sheldrick, Shelxtl (version 6.12), Bruker Analytical X-ray Instruments Inc., Madison, WI (USA) 2001.Search in Google Scholar
 O. Jepsen, O. K. Andersen, Tb-Lmto-Asa Program (version 47), Max Planck Institut für Festkörperforschung, Stuttgart (Germany) 2000. http://www2.fkf.mpg.de/andersen/LMTODOC/LMTODOC.html (accessed Febrary 2016).Search in Google Scholar
 R. Blachnik, E. Irle, J. Less-Common Met.1985, 113, L1.Search in Google Scholar
 L. Pauling, The Nature of the Chemical Bond, 3rd ed. Cornell University Press, Ithaca, NY, 1960.Search in Google Scholar
 H. d’Amour, W. B. Holzapfel, A. Polian, A. Chevy, Solid State Commun.1982, 44, 852.Search in Google Scholar
 S. Benazeth, N.-H. Dung, M. Guittard, P. Laruelle, Acta Crystallogr.1988, C44, 234.Search in Google Scholar
©2016 by De Gruyter