Synthesis, spectroscopic and crystallographic characterization of various cymantrenyl thioethers [Mn{C5HxBry(SMe)z}(PPh3)(CO)2]

Several manganese complexes of the type [MnCp(PPh3)(CO)2] containing a cyclopentadienyl (Cp) ligand with one to five SMe substituents were prepared and studied by X-ray crystallography. In these structures, a few S⋯S and/or S⋯Br interactions occur, and these are sometimes of significant importance for the arrangement of the molecules in the crystal.


Introduction
Aromatic thioethers, a long-known substance class, have attracted substantially increased interest over the last 30 years.A quick search in Scifinder (accessed on February 15, 2024) showed that while the annual number of publications stayed around 25 until 1996, this number then started to increase exponentially and reached a maximum of 205 in 2019 and was still at 174 in 2023.The main reason for this development can be attributed to the vast number of applications aromatic thioethers have found in agricultural chemistry (Li et al., 2021) and medicinal chemistry (Feng et al., 2016), and their importance in natural product biosynthesis (Dunbar et al., 2017).A special subgroup, bis(aryl) thioethers, has also found increased interest due to their photochemical properties (Riebe et al., 2017).While there are thousands of 'purely organic' aryl thioethers, the number of organometallic derivatives, particularly of the metallocene type, where a transition metal is �-coordinated to the aromatic part of the aryl thioether, is rather small.When it comes to persulfurated cyclopentadienyl complexes, there are only four compounds known.Three contain the pentakis(methylsulfanyl)cyclopentadienyl ligand, [{C 5 (SMe) 5 }ML n ] [ML n = Mn(CO) 3 (Su ¨nkel & Motz, 1988), RuCp* (Seneviratne & Winter, 1997) and FeCp (Blockhaus et al., 2019)] and one contains the pentakis(phenylsulfanyl) cyclopentadienyl ligand, [{C 5 (SPh) 5 }FeCp] (Blockhaus et al., 2019).Similarly, while there are ca 250 entries in Scifinder for the search mask 'Cr(� 6 -C 6 R 5 S-C)', there are only four entries for the benzene tris(thioether) and none for any higher thiolated benzene derivatives.When looking for crystal structure determinations of �-coordinated cyclopentadienyl thioethers in the Cambridge Structural Database (CSD; Groom et al., 2016;accessed on February 15, 2024), one finds 301 entries for the search mask with one thioether function.Of these, 233 are ferrocene-, 14 ruthenocene and 13 cymantrene derivatives.We therefore decided to look at the viability of a synthesis of a pentasulfurated cymantrene derivative where triphenylphosphane has substituted one carbonyl ligand and determine the crystal structures of these compounds.
2.1.1.Synthesis of [Mn(C 5 H 4 SMe)(PPh 3 )(CO) 2 ], 1b.A solution of 1a (0.050 g, 0.097 mmol) in tetrahydrofuran (THF, 8 ml) was treated at 195 K with 2.5 M n-BuLi solution (0.040 ml, 0.10 mmol) with stirring for 30 min.MeSSMe (0.010 g, 0.10 mmol) was then added and the mixture was warmed gradually to room temperature within 16 h.The reaction mixture was filtered through a plug of silica gel and then evaporated in vacuo.The residue was taken up in the minimum amount of petroleum ether (PE) and placed on top of a silica-gel column.PE/CH 2 Cl 2 (85:15 v/v) eluted a yellow band.Evaporation of the solvent in vacuo left 1b as a yellow powder (yield: 0.040 g, 0.0820 mmol, 85%).For spectra, see Figs.S1-S3 and S27 in the supporting information.
problems.For spectra, see Figs.S4-S6 in the supporting information.
2.1.3.Synthesis of [Mn{C 5 H 2 Br(SMe) 2 }(PPh 3 )(CO) 2 ], 3. A solution of 2 (0.43 g, 0.76 mmol) in THF (15 ml) was treated at 195 K with 1.0 M LDA solution (0.92 ml, 0.92 mmol) with stirring for 1 h.MeSSMe (0.080 ml, 1.00 mmol) was then added and the mixture was warmed gradually to room temperature within 16 h.The mixture was filtered through a plug of silica gel and then evaporated in vacuo.The residue was taken up in the minimum amount of PE and placed on top of a silica-gel column.PE/Et 2 O (85:15 v/v) eluted a yellow band.Evaporation of the solvent in vacuo left 3 as a yellow powder (yield: 0.29 g, 0.48 mmol, 63%).Recrystallization from PE yielded yellow crystals, which were suitable for X-ray diffraction and full structure refinement.For spectra, see Figs.S7-S9 in the supporting information.A solution of 3 (0.29 g, 0.48 mmol) in THF (10 ml) was treated at 195 K with a freshly prepared lithium tetramethylpiperidide (LiTMP) solution (1.19 mmol in 2 ml THF) with stirring for 1 h.MeSSMe (0.110 ml, 1.19 mmol) was then added and the mixture was warmed gradually to room temperature within 16 h.The mixture was filtered through a plug of silica gel and then evaporated in vacuo.NMR and mass spectra (see Figs. S10,S11 and S28) showed this product to be a mixture of at least four compounds.The MS showed only 3, 4 and 5, but of course without any indication of stereochemistry; the 31 P NMR spectrum showed five signals of relevant intensity, assignable to compounds 2, 3, 4 and X, and one further unknown, possibly 5.In the 1 H NMR spectrum, there are several signals in the Cp region (5.0-3.7 ppm), that have apparently no counterpart in the SMe region of the spectrum.The residue was taken up in the minimum amount of PE and placed on top of a silica-gel column.PE/Et 2 O (85:15 v/v) eluted two yellow bands.The first (F1) gave apparently unreacted starting material 3 (yield: 0.060 g, 21%; Fig. S12).The second still yielded a mixture and was therefore rechromatographed, using PE/Et 2 O (1:1 v/v) as eluent.The first fraction (F2.1) left, after full evaporation of the solvent, 4 as a yellow powder (yield: 0.10 g, 0.15 mmol, 31%; Figs.S13 and  S14).The second fraction (F2.2) left, after evaporation of the solvent, a yellow product, which, according to its NMR spectra (Figs.S15 and S16), was still a mixture of two main products, 4 and 'X', together with small amounts of unidentified by-products.Although compound X could not be isolated in a pure form, the appearance of its 1 H NMR spectrum suggests that it is a stereoisomer of 4, like [{C 5 H(Br-1)[(SMe) 3 -2,3,4]}-Mn(PPh 3 )(CO) 2 ].

Figure 2
One-pot lithiation of [Mn(C 5 Br 5 )(PPh 3 )(CO) 2 ] (6) followed by electrophilic quenching with Me 2 S 2 .Compounds 8 and 10 possess planar chirality and only one enantiomer is shown.evaporation of the solvent in vacuo, 7 (still impure) was left as a yellow powder (yield: 0.16 g, <0.21 mmol, <87%).Part of this product (0.060 g, <0.08 mmol) was dissolved in THF (10 ml) and treated at 183 K with BuLi solution (0.030 ml, 0.075 mmol) with stirring for 30 min.MeSSMe (0.010 ml, 0.12 mmol) was then added and the mixture was warmed to room temperature within 16 h.The mixture was filtered through a plug of silica gel.Evaporation of the solvent left a yellow powder (0.020 g).This crude product was redissolved in THF (8 ml) and treated at 183 K with BuLi solution (0.010 ml, 0.025 mmol) with stirring for 60 min.Then, still at 183 K, MeSSMe (0.003 ml, 0.04 mmol) was added and the temperature was raised to ambient temperature within 16 h.After complete evaporation of the solvent, the residue was taken up in the minimum amount of PE and placed on top of a silica-gel chromatography column.Elution with PE/Et 2 O (9:1 v/v) produced two fractions.Evaporation of the second fraction (F2) left a yellow powder (0.010 g).NMR spectroscopy Experiments were carried out with Mo K� radiation using a Bruker D8 VENTURE diffractometer.Absorption was corrected for by multi-scan methods (SADABS2016; Krause et al., 2015).H-atom parameters were constrained.Computer programs: APEX2 (Bruker, 2011), SAINT (Bruker, 2011), SHELXT2014 (Sheldrick, 2015a), SHELXL2018 (Sheldrick, 2015b) and Mercury (Macrae et al., 2020).
(see Figs. S22 and S23) showed this product to be nearly pure 11 contaminated with an unknown product 'Y'.Although the latter could not be isolated in a pure form, its 1 H NMR data suggest its formulation as [Mn{C 5 H(SMe) 4 }(PPh 3 )(CO) 2 ].Method (b).The conditions of method (a) were slightly changed, using 0.19 ml n-BuLi solution and stirring for only 30 min.The NMR spectra of the crude product showed the presence of only two compounds, 7 and 8 (Figs.S19-S21).The residue was redissolved in THF (10 ml) and treated at 183 K with 2.5 M n-BuLi solution (0.19 ml, 0.48 mmol) with stirring for 30 min.MeSSMe (0.050 ml, 0.60 mmol) was then added at this temperature.The mixture was warmed gradually to room temperature within 16 h with continuous stirring and was then filtered through a plug of silica gel and evaporated in vacuo.For 10, MS (EI, 70 eV): m/z = 631.8(M + -2CO), 616.8 (M + -Me), 601.8 (M + -2CO -2Me), 370.0 (M + -2CO -PPh 3 ).

Refinement
Crystal data, data collection and structure refinement details are summarized in Table 1.All structures were solved with SHELXT (Sheldrick, 2015a) and refined with SHELXL (Sheldrick, 2015b).In the refinement of all structures, several low-angle reflections had to be omitted due to beam-stop interferences (five in 1b, five in 3, six in 4 and five in 11).For compound 4, the SHELXT solution suggested one Mn, one Br and four S atoms.One of the S atoms turned out to actually be a P atom.Another S atom had significantly higher electron density than the remaining two.It was concluded that this was due to positional disorder with a Br atom.A first difference Fourier synthesis, which showed a residual electron-density peak at a distance of 1.800 A ˚from the 'bromine' atom, confirmed this assumption.No significant electron density was found near the Cp-ring 'CH carbon'.Therefore, a model was refined where the two 'inner' ring substituent atoms were both in part S and in part Br atoms.Refinement gave a 76:24 disorder in favour of the 'original' structure solution.In order to stabilize the refinement, some SADI restraints had to be employed (SADI restrains particular distances to be identical within certain standard deviations).
The data for compound 2 are included for comparison.While there was no problem obtaining the solution, refinement proceeded only with difficulty.First, three mediumintensity peaks close to an inversion centre turned up in a difference Fourier synthesis, which when connected had the appearance of a cyclohexane ring.Although cyclohexane was not explicitly used during the synthesis, it may have been part of the petroleum ether that was used for recrystallization.Apparently, 'petroleum ether' is not a particular compound, but a mixture of hydrocarbons within a specific range of boiling points.Therefore, the three atoms were refined isotropically with a common displacement parameter, and the site occupancy factor (s.o.f.) was refined as well.Refinement gave a value for the s.o.f. of ca 0.75, with a reasonable displacement parameter.Then the s.o.f. was fixed at 0.75 and the displacement parameters were allowed to refine freely, first isotropically and then anisotropically.Finally, methylene H atoms were added according to the standard riding model of SHELXL.While the s.o.f. and displacement parameters appear reasonable, some of the bond lengths appear unreasonably short.One possible explanation might be conformational disorder within the cyclohexane ring, which is quite usual for this molecule.The quality of the data set, however, did not allow for a proper resolution of this disorder.At the same time as the appearance of the solvent molecule, a medium-intensity peak was also localized close to atom H3 of the cyclopentadienyl ligand, with a distance of approximately 1.85 A ˚to atom C3.As compound 2 possesses planar chirality, we assumed this electron density derived from an alternative bromine position, corresponding to a rotational isomer of the enantiomer of the major orientation (of course, in a centrosymmetric space group, both enantiomers are present anyway).Again, both bromine positions were given a common displacement parameter, and their s.o.f.values were refined.This resulted in an s.o.f.value of only 0.045 for the minor orientation.Then the s.o.f.values were fixed at 0.955 and 0.045, respectively, and the displacement parameters were refined, first isotropically and then anisotropically.The corresponding H atoms were positioned according to the riding model.After this 'problem' was solved, another one appeared.A rather large residual electron-density peak turned up only 0.82 A ˚from the S atom, 1.73 A ˚from atom C2 and 0.71 A ˚on the distal side of the Cp ring.We could not find any explanation for this observation.The distance from sulfur is too small to be a methyl group or O atom, and the distance from the Cp plane is too large for a ring substituent.We also tried a refinement without the cyclohexane, using the SQUEEZE (Spek, 2015) model available in PLATON (Spek, 2020).However, this refinement neither provided better statistics nor change anything about the residual high electron density next to the S atom.With this unexplained residual electron density in mind, we refrained from any further structure discussion.However, displacement ellipsoid plots of the current 'best' solution are displayed in the supporting information.
Treatment of 1a with n-BuLi in THF followed by addition of Me 2 S 2 led to the product of Br-Li exchange, i.e. compound 1b, in 85% yield.This compound has been known for over 50 years and had originally been prepared by photolysis of [Mn(C 5 H 4 SMe)(CO) 3 ] in the presence of PPh 3 (Kursanov et al., 1970).When LDA was used as the base instead of n-BuLi, 1a was deprotonated in the �-position and, after electrophilic quenching with Me 2 S 2 , the disubstituted complex 2 was obtained in 79% yield, as an enantiomeric pair due to the planar chirality.Renewed treatment with LDA and Me 2 S 2 gave the trisubstituted compound 3 in 63% yield, apparently exclusively as the 1-bromo-2,5-bis(methylsulfanyl)-isomer.When 3 was treated with LDA/Me 2 S 2 , apparently only unreacted 3 could be recovered.Therefore, we decided to use the stronger base LiTMP, in 2.5 equivalents before adding Me 2 S 2 .NMR and mass spectrometric examination of the crude reaction product showed the presence of compound 4, together with unreacted starting material 3, presumably 5 and other unknown compounds.Chromatography allowed the isolation of rather pure compound 4, albeit in rather low yield (31%).The parallel formation of 5 hints at the occurrence of 'halogen dance' reactions (Blockhaus et al., 2020).
We then turned to an alternative approach, similar to that described by us for the synthesis of [Mn{C 5 (SMe) 5 }(CO) 3 ].We

[Mn(C 5 H 4 SMe)(PPh 3 )(CO) 2 ], 1b.
Compound 1b crystallizes in the monoclinic space group P2 1 /c, with one molecule in the asymmetric unit (Fig. 3).The bond parameters of 1b, together with those of the other structures described here, can be found in Table 2.There are no unusual features.In comparison with the ring-unsubstituted parent compound (Su ¨nkel & Klein-Hessling, 2021), the Mn-CO and Mn-Ct Cp (Ct Cp is the centroid of the cyclopentadienyl ring) distances are nearly identical, and the Mn-P bond is slightly elongated.In both compounds, one Mn-CO bond bisects one cyclopentadienyl C-C bond, while the other Mn-CO and the Mn-P bond nearly eclipse a C-H bond of the ring.The SMe group in 1b is in a relative trans position with respect to the PPh 3 ligand.The methyl group at sulfur is in an axial position.

Figure 4
Top and side views of the molecular structure of compound 3, with displacement ellipsoids drawn at the 50% probability level.3.2.2.[Mn{C 5 H 2 Br(SMe) 2 }(PPh 3 )(CO) 2 ], 3. Compound 3 crystallizes in the orthorhombic space group Pbca, with one molecule in the asymmetric unit (Fig. 4).The bond parameters can be found in Table 2. Again, there are no unusual features.This time, the C-Br bond is in a relative trans position with respect to the Mn-P bond.One methyl group is in an axial position at atom S5, while the other at S2 is in an equatorial position.The Mn-P, Mn-CO and Mn-Ct Cp bonds are virtually identical with the corresponding bond lengths in 1b.The same is true for the C Cp -S and S-CH 3 bonds.The bond angle at the S atom with the equatorial methyl group is significantly larger than the corresponding angle with the axial methyl group, which in turn is identical to the corresponding angle in compound 1b.
In comparison with the structures of 1b and 3, the Mn-P, Mn-CO and Mn-Ct Cp bonds are slightly (but significantly, � > 5�) elongated (Table 2).The two 'outer' methylsulfanyl groups are equatorial, with both methyl groups directed towards the unsubstituted C-H bond, while the 'inner' methyl group is in an axial position, directed away from the Mn atom.A little bit surprising is the observation that the Mn-P bond nearly eclipses one C Cp -S bond, instead of the neighbouring C-H bond, as one might expect.Thus, a rather short S� � �P distance of 3.601 (1) A ˚results, which is identical to the sum of the van der Waals radii.In addition, all S atoms are significantly closer to the Mn atom than the sum of their van der Waals radii (3.80A ˚).This feature is more distinct for the S atoms with axial methyl groups (R ' 3.47 A ˚) than for those with equatorial methyl groups (R ' 3.57A ˚), and is also observed in the structures of 1b and 3.
3.2.4.[Mn{C 5 (SMe) 5 }(PPh 3 )(CO) 2 ], 11.Compound 11 crystallizes in the monoclinic space group P2 1 /c, with one molecule in the asymmetric unit (Fig. 6).The Mn-P bond length in 11 is longer than in the other three compounds, while the Mn-CO and Mn-Ct Cp distances are very similar to the values found in 4. All methylsulfanyl groups are significantly tilted away from an 'ideal' axial position (C-C Cp -S-C Me in the range between 50 and 68 � versus 92 � in 1b).Four methyl groups are directed away towards the distal side of the cyclopentadienyl ring, while one is on the proximal side.This

Figure 6
Top and side views of the molecular structure of compound 11, with displacement ellipsoids drawn at the 50% probability level., where all the SMe groups are in distal positions (Blockhaus et al., 2019).This orientation with four methyl groups on one side of the ring and one methyl group on the other resembles, however, the situation found in the structure of the uncomplexed 'free' anion (Wudl et al., 1981).Theoretical studies of the conformational preferences of poly(methylsulfanyl)benzenes, including hexakis(methylsulfanyl)benzene, have been reported (Lumbroso et al., 1986;Fleurat-Lessard & Volatron, 2009), but to our knowledge no such studies of polythiolated cyclopentaienyl rings exist.All the Mn� � �S distances are in the range 3.52-3.59A ˚and are thus significantly shorter than the sum of their van der Waals radii (3.80A ˚).For comparison, in [Mn{C 5 (SMe) 5 }(CO) 3 ], the Mn� � �S distances range from 3.39 to 3.59 A ˚. Still, it seems unlikely that there is explicit bonding between Mn and S, as such distances are also the simple geometrical result of �-bonding of the substituted Cp ring to the metal.

Intermolecular contacts and Hirshfeld analysis
The importance of intermolecular contacts, also termed 'noncovalent interactions', for the build-up of crystal structures is undisputed.While the near omnipresence of hydrogen bonds has been known for a long time (mainly due to their structure-directing effects in biomolecules), in recent decades it was recognized that interactions involving halogens, chalcogens and even pnictogens and tetrel elements also have a great influence on the mutual arrangements of molecules in crystals and the terms 'halogen bond', 'chalcogen bond', 'pnictogen bond' and 'tetrel bond' were created (Brammer et al., 2023;Vogel et al., 2019;Scheiner, 2023;Scilabra et al., 2019;Mahmudov et al., 2022).For the present study, we looked first only at the interactions involving S and/or Br atoms, using the corresponding feature in Mercury (Macrae et al., 2020).In compound 1b, no such interactions are observed.However, in compound 3, double S� � �Br contacts of 3.414 A ˚, well below the sum of the van der Waals radii, lead to the formation of 'dimers' (Fig. 7).
The C-Br-S angle at Br1 is 153.6 (1) � , while the C-S-Br angles at S2 are 128.0(1) and 128.8 (1) � .Atom S5 is not involved in such interactions; however, it accepts a hydrogen bond from a phenyl C-H group.
Compound 4 was not included for this study, due to the presence of the S/Br disorder, which did not allow a proper resolution of the relative contributions of these elements.
Compound 11 displays a molecular chain in the crystallographic c direction, which is held together via weak S� � �S contacts (distance of 3.588 A ˚between atoms S1 and S3, just below the sum of the van der Waals radii).The C Cp -S-S angles at S1 and S3 are 139.8(1) and 168.4 (1) � , respectively.Atoms S2 and S5 serve as hydrogen-bond acceptors towards two arene C-H bonds (Fig. 8).For comparison, in closely related [Mn{C 5 (SMe) 5 }(CO) 3 ] (Su ¨nkel & Motz, 1988), only dimer formation via an S� � �S interaction between inversionrelated S atoms (3.510A ˚) is observed.On the other hand, pentakis(methylsulfanyl)ferrocene employs four S atoms for the formation of parallel undulating (wavy) chains along b using double S� � �S bridges on both sides (Blockhaus et al., 2019).
In order to get a better overview of the intermolecular interactions at work, we undertook a Hirshfeld analysis, using the program CrystalExplorer (Spackman et al., 2021).First, we determined the Hirshfeld surfaces (Fig. 9) and fingerprint plots (Fig. 10).
Evaluation of the fingerprint plots allowed the calculation of the relative contributions of interactions of elements inside and outside the Hirshfeld surface (Table 3).
As can be seen, interactions between hydrogen and other elements make up for more than 95% of all interactions and ca a 50% contribution comes from H� � �H interactions.This  Motz, 1988), interactions with H-atom contributions make up ca 88%, with 34.8% coming from H� � �H and 24.6% from H� � �S; S� � �S interactions contribute 2.3%.
We also performed an orbital calculation of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of compounds 1b, 3, 4 (major component) and 11, using the program TONTO (HF/STO-3G), as provided with CrystalExplorer.The results are shown in Fig. 11.
As can be seen, for 1b, the HOMO resides on the S atom, the CO ligands and one arene ring, while the LUMO is concentrated on the cyclopentadienyl ring.In 3, the HOMO is distributed over the CO ligands and the P atom and part of the arene rings, while the LUMO is mainly situated on the Br atom and the SMe groups.In compound 4 (major component), the HOMO resides mainly on two SMe groups, as well as on one arene ring.The LUMO is spread over the metal, the remaining SMe group and the Br atom, as well as on one other arene ring.For compound 11, the HOMO resides on the Cp ring, including two SMe groups and the P atom, while the LUMO is distributed over the Mn atom and the PPh 3 ligand.

Conclusions
The best approach for the synthesis of [Mn{C 5 (SMe) 5 }-(PPh 3 )(CO) 2 )] appears to be the one-pot reaction of [Mn(C 5 Br 5 )(PPh 3 )(CO) 2 ] first with 2 equivalents of n-BuLi/ MeSSMe and then with four equivalents of these reagents.Stepwise reactions, as well as the bottom-up approach starting with [Mn(C 5 H 4 Br)(PPh 3 )(CO) 2 ], lead only to complicated product mixtures, which need multiple purification steps with large losses in yield.The structures of 1b, 3, 4 and 11 with one, two, three or five SMe groups, respectively, show an unpredictable distribution of axial and equatorial methyl groups.S� � �Br contacts in 3 leads to the formation of 'dimers', while S� � �S contacts in 11 lead to polymeric one-dimensional strands.Besides these, no structure-directing effects of halogen or chalcogen (S) bonding can be observed.

Special details
Geometry.All esds (except the esd in the dihedral angle between two l.s.planes) are estimated using the full covariance matrix.The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry.An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s.planes.

Special details
Geometry.All esds (except the esd in the dihedral angle between two l.s.planes) are estimated using the full covariance matrix.The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry.An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s.planes.

Special details
Geometry.All esds (except the esd in the dihedral angle between two l.s.planes) are estimated using the full covariance matrix.The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry.An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s.planes.

Figure 5
Figure 5Top and side views of the molecular structure (major component) of compound 4, with displacement ellipsoids drawn at the 50% probability level.Only one enantiomer of this planar chiral compound is shown.

Figure 7
Figure 7Dimer formation in the crystal of 3,

Figure 8
Figure 8 Packing plot of 11, viewed along the a axis, showing the intermolecular S� � �S contacts.

Table 1
Experimental details.

Table 3
Relative contributions (%) of elements to the interactions of atoms inside and outside the Hirshfeld surface (mainly contributions > 2%).