Thio-, seleno-and telluro-ether complexes of aluminium( III ) halides: synthesis, structures and properties †

The reaction of AlCl 3 with Me 2 E (E = S, Se or Te) or n Bu 2 E (E = Se or Te) in CH 2 Cl 2 under rigorously anhy-drous conditions gave the pseudo -tetrahedral complexes [AlCl 3 (R 2 E)]. The [AlX 3 (Me 2 E)] (X = Br or I, E = S; X = Br, E = Te) were made from toluene solution since attempted syntheses in CH 2 Cl 2 resulted in substantial chloride incorporation. The synthesis of [(AlCl 3 ) 2 { o -C 6 H 4 (CH 2 SEt) 2 }], in which the ligand bridges two tetrahedral aluminium centres, and of the six-coordinate trans -[AlX 2 {MeE(CH 2 ) 2 EMe} 2 ][AlX 4 ] (X = Cl or Br, E = S, and X = Cl, E = Se) and cis -[AlI 2 {MeS(CH 2 ) 2 SMe} 2 ][AlI 4 ] are reported. The tripodal thioether forms [AlCl 3 {MeC(CH 2 SMe) 3 }], which is a chain polymer with κ 2 -coordinated ligand and a tbp arrangement at Al( III ). Chalcogenoether macrocycle complexes [AlCl 3 ([9]aneS 3 )], [AlCl 2 ([14]aneS 4 )][AlCl 4 ] and [AlCl 2 ([16]aneSe 4 )] [AlCl 4 ] are also described. All complexes were characterised by microanalysis, IR and multinuclear NMR ( 1 H, 27 Al, 77 Se or 125 Te) spectroscopy as appropriate. In CH 2 Cl 2 solution [AlCl 3 (Me 2 S)] with added Me 2 S forms [AlCl 3 (Me 2 S) 2 ], and the [AlX 2 {MeS(CH 2 ) 2 SMe} 2 ][AlX 4 ] exist as mixtures of cis and trans isomers which undergo rapid exchange at ambient temperatures. X-Ray crystal structures are reported for [AlCl 3 (Me 2 Se], [AlX 3 (Me 2 Te)] (X = Cl or Br), trans -[AlCl 2 {MeE(CH 2 ) 2 EMe} 2 ][AlCl 4 ] (E = S or Se), cis -[AlI 2 {MeS(CH 2 ) 2 SMe} 2 ][AlI 4 ], [AlCl 3 {MeC(CH 2 SMe) 3 }], and for the sulfonium salt [Me 2 SH][AlCl 4 ]. The aluminium halide chalcogenoether chemistry is compared with the corresponding gallium and indium systems, and the relative Lewis acidities of the metals discussed. Attempts to use [AlCl 3 ( n Bu 2 E)] (E = Se or Te) as LPCVD reagents to form aluminium chalcogenide ﬁ lms were unsuccessful.


Introduction
Aluminium chloride is widely used both in the laboratory and industry as a Friedel-Crafts catalyst for alkylation and acylation reactions, and also catalyses condensation, isomerisation and polymerisation reactions. 1 The reactions depend upon the strong Lewis acidity which produces incipient carbocations in combination with [AlCl 4 ] − anions. Aluminium bromide and iodide promote similar chemistry and have some niche applications. 1 The strong Lewis acidity of these three halides has resulted in the characterisation of very many adducts, mostly with N-or O-centres in solvents or donor ligands. 2 In contrast, AlF 3 is an inert polymer which forms few complexes. 3 Surprisingly little is known about AlX 3 adducts of chalcogenoethers, which are limited to early studies of adducts with Me 2 S or Et 2 S that reported phase diagrams, IR and 1 H NMR spectra, [4][5][6] and several studies of solution enthalpies. 7 The only crystallographically characterised example of a neutral thioether coordinated to AlX 3 is the recently reported 8 [AlCl 3 (thianthrene)], although there are a few AlMe 3 complexes with thiamacrocycles. 9 There are no reports of studies with telluroethers, and the single complex of a selenium ligand is [AlCl 3 (selenoxan)], characterised only by microanalysis. 10 Recent studies of gallium(III) halides GaX 3 (X = Cl, Br or I) with chalcogenoethers have established that most contain pseudo-tetrahedral gallium centres [GaX 3 L] (L = Me 2 S, Me 2 Se, Me 2 Te, etc.) or dinuclear [X 3 Ga(μ-L-L)GaX 3 ] (L-L = RS-(CH 2 ) 2 SR, MeSe(CH 2 ) 2 SeMe, MeTe(CH 2 ) 3 TeMe etc.). [11][12][13][14] Higher coordination numbers are rare in the gallium systems, but found with the thia-macrocycle [14]aneS 4 , ‡ which binds exodentate via two sulfur centres, affording the chain polymer [GaCl 3 ([14]aneS 4 )] with a trigonal bipyramidal geometry. 15 The larger rings [16]aneS 4

Results and discussion
Aluminium trihalides (AlX 3 (X = Cl, Br or I), are strong hard Lewis acids with a very high affinity for water. 2 Successful synthesis of their complexes with soft donor ligands such as thio-, seleno-or telluro-ethers requires anhydrous AlX 3 , rigorously anhydrous solvents and ligands and exclusion of water at all stages of the manipulations. Trace water displaces the neutral ligands and also generates [AlX 4 ] − which are readily identified in the solids by their characteristic IR spectra ([AlCl 4 ] − t 2 = 498, [AlBr 4 ] − 394, [AlI 4 ] − 336 cm −1 ) 18 and in solution by 27 Al NMR spectroscopy (Table 1), where they have sharp characteristic resonances. The moisture sensitivity of the halides and the complexes is much greater than observed in the corresponding gallium(III) systems. 12,13 The higher reactivity of the aluminium halides also affects the choice of solvent for the synthesis. Whilst complexes of AlCl 3 are readily made in anhydrous CH 2 Cl 2 , use of this solvent for the AlBr 3 or AlI 3 reactions results in incorporation of substantial amounts of chloride into the products, and the heavier halides are best made from anhydrous toluene solution. Similar observations were made by Burford et al. 19 in AlX 3 /R 3 PO/CH 2 Cl 2 systems. However, the pre-isolated pure [AlX 3 L] (X = Br or I) react only slowly with CH 2 Cl 2 (or CD 2 Cl 2 ), which remains the NMR solvent of choice, as non-coordinating and useable down to 180 K. The reactivity of AlCl 3 in CH 2 Cl 2 has been ascribed to the formation of the intermediate carbenium ion [CH 2 Cl][AlCl 4 ], 8,20 and in previous studies of GaCl 3 complexes of thioethers we observed the formation of [o-C 6 H 4 (SMeCH 2 Cl) 2 ][GaCl 4 ] 2 when [(GaCl 3 ) 2 {μ-o-C 6 H 4 (SMe) 2 }] is allowed to stand in CH 2 Cl 2 solution for several days. 13 Cleavage of C-Se or C-Te bonds is also observed, e.g. the formation of the selenonium cation in [o-C 6 H 4 (CH 2 ) 2 SeMe]-[GaCl 4 ] from o-C 6 H 4 (CH 2 SeMe) 2 . 12 In the aluminium systems many of the complexes degrade on standing in solution at ambient temperatures (below), hence rapid isolation of the complexes from solution is advisable. Decomposition in solution is much slower at low temperatures, permitting growth of X-ray quality crystals overnight at −18°C.

[AlX 3 (R 2 E)] complexes
The reaction of AlCl 3 in CH 2 Cl 2 or AlX 3 (X = Br or I) in toluene with Me 2 S affords [AlX 3 (Me 2 S)] in high yield. The chlorocomplex is an oil at room temperature, the others crystalline solids. The complexes can also be made by condensing excess Me 2 S onto the appropriate AlX 3 at −196°C, allowing the mixture to thaw, when the halide dissolves to give a clear solution, and then removing excess Me 2 S in vacuo. The phase diagram of the AlCl 3 /Me 2 S system 4 shows both [AlCl 3 (Me 2 S)] and [AlCl 3 (Me 2 S) 2 ], but on removing the volatiles in vacuo from a mixture of AlCl 3 and 3Me 2 S, only the 1 : 1 complex was isolated.
The IR spectrum of [AlCl 3 (Me 2 S)] shows features at 541 and 410 cm −1 assigned to the E and A 1 modes of the pyramidal AlCl 3 unit; both bands are quite broad and the E mode shows some evidence of further splitting. Similar observations were made in the spectra of many of the aluminium complexes in this work, and are probably due to solid state effects, such as lower site symmetry or cation-anion interactions. The 1 H NMR spectrum of [AlCl 3 (Me 2 S)] in CD 2 Cl 2 (295 K) shows a singlet at δ = 2.53 which does not change significantly on cooling the solution to 190 K. Addition of aliquots of Me 2 S to the solution produces progressive shifts in the single resonance to low frequency, and on cooling to 185 K, two broad resonances are resolved at δ = 2.24 and 2.20, suggesting ligand exchange is slowing, but that the low temperature limit has not been reached. The latter shift is similar to that of free Me 2 S (δ = 2.15), whilst the former is assigned to a new aluminium complex. The 27 Al NMR spectra are more informative (Table 1). At ambient temperatures a broad singlet at δ = 111.6 is present in the spectrum of [AlCl 3 (Me 2 S)], which is little changed on cooling the solution to 190 K. However, addition of 1 mol. equivalent of Me 2 S to the solution generates a new resonance at δ = 73.5 (W 1/2 = 650 Hz), and this resonance is unchanged upon addition of more Me 2 S and shows only a small low frequency drift on cooling the solution to 190 K. The new 27 Al chemical shift is in the range expected for five-coordinate Al species. 21 The combination of the 1 H and 27 Al NMR results show that in the presence of excess Me 2 S in CH 2 Cl 2 solution the 2 : 1 complex [AlCl 3 (Me 2 S) 2 ] forms; further addition of Me 2 S does not produce any evidence for a 3 : 1 complex. As noted above, work-up of the solution results in decomposition to reform [AlCl 3 (Me 2 S)]. Five-coordination is established in the solid state with aluminium-phosphine complexes, e.g.
[AlI 3 (PEt 3 ) 2 ], which has a tbp geometry with axial phosphines. 22 The complexes [AlX 3 (Me 2 S)] (X = Br or I) are generally similar to the chloride complex, and exhibit progressively lower frequency shifts in the 27 Al NMR spectra as the halogen becomes heavier (Table 1). However, although the 1 H NMR spectra show fast exchange with added Me 2 S in CH 2 Cl 2 solution, no new resonances were evident in the 27 Al NMR spectra in the presence of a large excess of Me 2 S, indicating that in these cases 2 : 1 complexes do not form.
The [AlX 3 (Me 2 E)] (X = Cl, Br, E = Se or Te) were obtained in high yields and their 1 H NMR and IR spectroscopic properties are similar to those of the thioether analogues. The 27 Al NMR spectra (Table 1) show only small low frequency shifts along the series E = S > Se > Te, and no new complexes are formed by adding excess Me 2 E to CH 2 Cl 2 solutions of the appropriate [AlX 3 (Me 2 E)]. [AlCl 3 (Me 2 Se)] shows a 77 Se NMR chemical shift of δ = −11.3, which corresponds to a small low frequency coordination shift (Δ = −11.3); this can be compared with small high frequency coordination shifts observed in [GaX 3 (Me 2 Se)]. 12 Although high frequency coordination shifts are seen in most transition metal selenoether (and telluroether) complexes, in p-block complexes both high and low frequency shifts are seen in different systems, and the causes are not understood. 23 We were unable to observe a 77 Se NMR resonance from [AlBr 3 (Me 2 Se)] or 125 Te resonances from [AlX 3 (Me 2 Te)] over the temperature range 295-190 K, presumably due to fast exchange. The solutions of the selenoether and telluroether complexes develop new resonances on standing, some of which may be due to Me 2 E 2 , Me 3 E + or Me 2 EX 2 from their chemical shifts, but given the sensitivity of 77 Se and 125 Te chemical shifts to concentration, solvent etc. 21,24 their identification was not pursued. They do, however, provide evidence of the fragility in solution of the AlX 3 complexes with the heavier chalcogenoethers.
In Storing a solution of [AlCl 3 (Me 2 S)] in CH 2 Cl 2 in the refrigerator, produced a few small crystals which were identified as the sulfonium salt [Me 2 SH][AlCl 4 ] (Fig. 4), by the X-ray structure solution, and probably formed by adventitious hydrolysis. Solid sulfonium salts are rare, but we have previously obtained examples from serendipitous hydrolysis of some niobium(V) fluoride-thioether complexes, 25 the formation being promoted by "anhydrous" conditions and a large weakly coordinating Ellipsoids are drawn at the 50% probability level and H atoms are omitted for clarity. Symmetry operation: a = x, y, 3/2 − z. Selected bond lengths (Å) and angles (°) Se1-Al1 = 2.486(2), Cl1-Al1 = 2.121(2), Cl2-Al1 = 2.1130 (14), Cl2-Al1-Cl2a = 115.30(9), Cl2-Al1-Cl1 = 111.32(5), Cl2-Al1-Se1 = 106.47(5), Cl1-Al1-Se1 = 105.20 (6). with HCl gas. The n-butyl derivatives [AlCl 3 ( n Bu 2 E)] (E = Se, Te) were also synthesised as potential single source precursors for LPCVD as discussed below. Their characterisation data matched well with that of the equivalent methyl substituted complexes. The telluroether complexes are not stable when stored at room temperature over a period of days to weeks, even in an N 2 filled glove box. They gradually become darker and a quantity of black solid (elemental Te) forms. Samples can be stored at −18°C for a few weeks with minimal decomposition. A sample of [AlCl 3 ( n Bu 2 Te)] that had been stored in the glove box for several weeks and had visibly decomposed was analysed by ESI + mass spectrometry, which showed the presence of [ n Bu 3 Te] + , and this ion was also present in the 125 Te NMR spectrum (δ = 491). 13

Complexes with bidentate ligands
By analogy with the corresponding GaX 3 adducts, 12 and considering the preference for four-coordination in [AlX 3 (R 2 E)] adducts described above, it was expected that flexible dithioethers or diselenoethers (L-L) would produce complexes of the type [X 3 Al(μ-L-L)AlX 3 ]. In fact this was only true for the bulky o-xylyl-backboned dithioether, o-C 6 H 4 (CH 2 SEt) 2 , which gave yellow crystals of [(AlCl 3 ) 2 {μ-o-C 6 H 4 (CH 2 SEt) 2 }]. The crystals were of modest quality and had the characteristics of a modulated structure (see Experimental), and thus comparison of the detailed bond lengths and angles is not warranted. However, they are isomorphous with the gallium(III) analogue, 13 and serve as an example of this structure type (Fig. 5).
The spectroscopic data are consistent with four-coordinate aluminium in solution. The corresponding diselenoether o-C 6 H 4 (CH 2 SeMe) 2 was completely converted to the selenonium cation [o-C 6 H 4 (CH 2 ) 2 SeMe][AlCl 4 ] upon reaction with AlCl 3 in CH 2 Cl 2 (see ESI †). The same selenonium cation is formed upon reaction of this ligand with GaCl 3 or InCl 3 . 12,15 Unexpectedly, reaction of AlX 3 with MeE(CH 2 ) 2 EMe (E = S or Se) failed to give [X 3 Al{μ-MeE(CH 2 ) 2 EMe}AlX 3 ], and the products had an  (Fig. 6 and 7) reveal centrosymmetric cations (trans isomer) with identical d(Al-Cl), which are as expected longer than in the four-coordinate complexes. The d(Al-S) and d(Al-Se) differ by ∼0.14 Å, which approximates to the difference in covalent radii of the chalcogens. 26 The bond angles around the aluminium show only small deviations from 90°.
Although both of the structures contain the chalcogenoether in the DL conformation, the two crystals are not isomorphous and in the selenoether complex there are short contacts Se2⋯Se2′ (3.436 Å) and Se3⋯Se3″ (3.542 Å). These link Al2centred cations into chains through Se2 (along the a direction)   and similarly, the Al3-centred cations are linked into chains through Se3 (again along the a direction) (Fig. 8).
The Nujol mull IR spectra for these salts all confirm the presence of [AlX 4 ] − , but the Al-X stretches of the cations could not be identified with certainty. The solution speciation is less clear, and all the complexes are extremely moisture sensitive in solution. All four complexes show 27 27 Al = 37.4) which is assigned to the six-coordinate cation. For the other complexes it is likely that the increasing electric field gradients promote fast quadrupolar relaxation, resulting in the loss of the cation resonance.
At room temperature a CD 2 Cl 2 solution of [AlCl 2 {MeS-(CH 2 ) 2 SMe} 2 ][AlCl 4 ] shows singlet CH 3 (δ = 2.26) and CH 2 (δ = 3.08) resonances. On cooling to 223 K the spectrum shows three CH 3 resonances (δ = 2.26, 2.50, 2.66) and overlapping CH 2 resonances (δ = 3.08-3.31), which we tentatively assign to a mixture of cis and trans isomers of [AlCl 2 {MeS(CH 2 ) 2 SMe} 2 ] + , the changes reversing on warming the solution. The bromoand iodo-complexes behave similarly. Notably, none of the complexes show resonances due to free dithioether, which would seem to rule out significant amounts of [AlCl 2 {MeS-(CH 2 ) 2 SMe}] + , being present. At 185 K further splitting of the resonances is evident, which is probably due to slowing of the pyramidal inversion at S, leading to separate resonances for the individual invertomers. The solutions decompose slowly on standing.
The trans-[AlCl 2 {MeSe(CH 2 ) 2 SeMe} 2 ][AlCl 4 ] exhibits a singlet 77 Se NMR resonance at room temperature (δ = 95.5) which is a low frequency coordination shift (Δ = −25.5) and singlet CH 3 (δ = 2.36) and CH 2 (δ = 3.21) resonances in the 1 H NMR spectrum. Cooling the solution to 190 K produces little change in the 1 H NMR spectrum, although the 77 Se resonance is lost below ∼240 K. In the selenoether complex only the trans isomer appears to be present in significant amounts. On standing, new resonances grow in due to decomposition. Attempts to record spectra in CD 3 CN solution resulted in displacement of the thio-or seleno-ethers by the nitrile.
Reaction of AlCl 3 with the ditelluroether t BuTe(CH 2 ) 3 Te t Bu produced a mixture of species resulting from ligand fragmentation. The 1 H NMR spectrum of the product showed multiple resonances for the t-butyl groups and the CH 2 units. Multiple signals were also observed in the 125 Te NMR spectrum, whilst the 27 Al NMR spectrum indicated the presence of [AlCl 4 ] − . Very air sensitive, yellow crystals were isolated of one of the decomposition products, which proved to be [ t BuTe(CH 2 ) 3 Te-( t Bu)Te(CH 2 ) 3 Te t Bu][AlCl 4 ], derived from fragmentation of the ditelluroether (see ESI †).

Polydentates and macrocycles
The reaction of the tripodal trithioether MeC(CH 2 SMe) 3 with AlCl 3 in a 1 : 1 molar ratio in anhydrous CH 2 Cl 2 gave colourless crystals whose structure (Fig. 10) showed a chain polymer with the ligand binding as a bridging bidentate with one uncoordinated -CH 2 SMe arm. The structure forms a chain in the a direction The geometry at aluminium is a distorted trigonal bipyramid with equatorial chlorines and there are two slightly different aluminium environments in the unit cell.

LPCVD investigation
Following previous success in using neutral chalcogenoether adducts of GaCl 3 to deposit thin films of crystalline Ga 2 Se 3 and Ga 2 Te 3 , 17 several coordination complexes of AlCl 3 were synthesised as potential single source precursors to Al 2 E 3 films. Ligands with n Bu substituents were selected as these had previously been shown to be more effective than ligands with Me substituents, probably because they have the β-hydride elimination route available. 17 [AlCl 3 ( n Bu 2 E)] (E = Se, Te) were synthesised as yellow oils as described above. LPCVD using both complexes was attempted at temperatures between 723 and 873 K, using the CVD equipment described previously. 13 In all cases the precursor evaporated cleanly, having changed colour to dark brown during the evaporation. There was no deposition on the substrates and some elemental selenium or tellurium was deposited on the tube at the exit of the furnace. We conclude that these reagents are unsuitable for LPCVD under these conditions.

Lewis acidity falls AlCl 3 > AlBr 3 > AlI 3 , and AlCl 3 > AlBr 3 > GaCl 3 > GaBr 3 (ref. 27-31 and references therein)
. One should note in passing that the order with halogen is reversed for boron. The modelling work is based upon gas phase molecules and does not take into account solid state effects (lattice energies, intermolecular interactions and packing effects), or the effects of lattice solvent, which may complicate the interpretation of experimental data, and in some cases lead to apparently anomalous results. 30 The various contributions listed above, means that interpreting changes in metal-ligand bond lengths simply in terms of Lewis acidity must be done with care, and one might expect the occasional anomaly, but as a result of recent studies, there are sufficient data available to attempt some comparisons for Al/Ga/In-Group 16 donor complexes. Table 2 shows some illustrative data. The data show firstly, that if one compares complexes of the same element with the same coordination number, the M-X distances seem unaffected by the specific chalcogen donor type present, which is consistent with the metalhalogen being the dominant interaction. A similar comparison of the M-E bond lengths shows that these increase (sometimes only marginally) with halide, Cl < Br < I, consistent with the trends deduced for lighter donor atoms. The M-X, and M-E bond lengths in comparable complexes of Al and Ga are also nearly identical, consistent with their almost identical covalent radii, resulting from the "3d block contraction", i.e. the increased nuclear charge resulting from the 3d metals only partially screened by the d electron shell. 26 As expected, In-X and In-E bonds are typically ∼0.2 Å longer. A very recent dft study 31 suggested that whilst Ga and In halide complexes of Me 2 Se had a high degree of covalency in the M-Se bonds, those of aluminium had a markedly higher electrostatic component to the bonding. Our experimental data reported in the present paper, show no evidence for a significant change in the bonding type present along the series of group 16 donor complexes, the differences noted being due to the higher Lewis acidity of Al(III). We note that the dft calculations predict 31 an Al-Se bond length (for the gas phase molecule) of 2.53 Å compared to the X-ray crystallographic result (for the solid) of 2.486(2) Å.
Our data also show the ready formation of six-coordinate cations with aluminium, [AlX 2 (L-L) 2 ] + , contrasting with the reluctance of gallium to exceed four-coordination, except in macrocyclic compounds, cannot be due to steric effects, but must be a further consequence of the stronger Lewis acidity of aluminium. These differences must originate in the donor/ acceptor orbital energy match (or mis-match) rather than in charge/radius effects. The larger indium centre has a less clear preference, easily accommodating four-, five-or six-coordination depending upon the ligand and reaction conditions.

Conclusions
Chalcogenoether complexes of aluminium(III) halides with four-, six-and (rarely) five-coordinate metal centres have been prepared, and their structures and properties compared with those of the heavier analogues GaX 3 and InX 3 . The aluminium complexes are extremely moisture sensitive, and complexes with selenium or tellurium ligands are prone to slow E-C bond cleavage in solution. Nonetheless, the formation and structural characterisation of telluroether complexes of the hard AlX 3 acceptors is notable. In contrast to the gallium complexes, it does not appear that the aluminium systems are suitable for LPCVD applications. The detailed study of Al(III) complexes with soft, modest donor chalcogenoethers has confirmed the trends in Lewis acidity observed with hard O or N donor ligands and are broadly in line with expectations based upon the dft calculations. The work further demonstrates that a significant range of chalcogenoether complexes with hard p-block Lewis acids are obtainable despite the hard/soft-acceptor/donor mismatch.

X-Ray experimental
Details of the crystallographic data collection and refinement parameters are given in Table 3. Crystals suitable for single crystal X-ray analysis were obtained as described above. Data collections used a Rigaku AFC12 goniometer equipped with an enhanced sensitivity (HG) Saturn724+ detector mounted at the window of an FR-E+ SuperBright molybdenum (λ = 0.71073 Å) rotating anode generator with VHF Varimax optics (100 µm focus) with the crystal held at 100 K (N 2 cryostream). Structure solution and refinement were straightforward, 37,38 except as detailed below, with H atoms bonded to C being placed in calculated positions using the default C-H distance. For [AlCl 3 {MeC(CH 2 SMe) 3 }] the data were collected using the Rigaku automated routines which normally gives close to 100% of the data out to 2θ of 55°. For reasons that are not clear this did not happen in this case and it proved difficult to obtain more suitable crystals for a re-collection. Judged by the high R int value, the data are of modest quality although the intensities seems satisfactory (80% exceed the Shelxl test, I > 2σ(I)). The structure that emerges from the analysis appears sound, with no unusual adp values or other causes for concern. For [(AlCl 3 ) 2 {C 6 H 4 (CH 2 SEt) 2 }] the diffraction pattern exhibits many additional reflections in the 100 projection. These are likely due to a modulation. In fact, the Ga analogue 13 exhibits the same behaviour and in that case it was possible to index the modulated cell. As with the Ga analogue, the data were indexed on a strong sub-cell making refinement of the basic structure possible; however, the ignored reflections result in unrealistic thermal parameters for many of the atoms. For both of these structures therefore, detailed comparisons of the geometric parameters require caution.