Atom Transfer Radical Addition of Activated Primary Alkyl Chlorides Using In Situ Generated [Cp*RuII(Cl)(PR3)] Catalysts

Atom transfer radical addition (ATRA) of halogenated compounds with alkenes is well established but primary alkyl chlorides are understudied because of the difficult C−Cl bond activation. In this paper, we show that TONs of 61 can be achieved in the ATRA of ethyl chloroacetate onto styrene with [Cp*Ru(Cl)2(PPh3)] and 1,1′‐azobis(cyclohexanecarbonitrile) (ACHN) as a radical initiator, representing a three‐fold improvement compared to previous reports. New catalyst precursors of the type [Cp*Ru(Cl)2(PR3)] were synthesized and tested (R=Me, Et, Cy, Ph, p‐CF3C6H4 and p‐MeOC6H4). The kinetic reaction profiles were studied using in situ ATR−FTIR spectroscopy. Among these complexes, [Cp*Ru(Cl)2(PPh3)] gave the best yields while [Cp*Ru(Cl)2(PMe3)] showed the highest rate. While rates correlate with redox potentials (electronics), our investigation reveals that substrate sterics are important for the overall yield. Density functional theory calculations suggest an open‐shell singlet pathway, where polymerization is kinetically disfavored, explaining the selectivity towards ATRA products.


Introduction
[3] Minisci and colleagues reported that metal complexes can steer the selectivity towards either monoadduct formation or polymerization, creating the foundation for the field of atom transfer radical polymerization (ATRP). [4][7] Chlorinated substrates are notoriously more challenging, being a direct result of the higher bond dissociation energy (BDE) of the CÀ Cl bond, [8] and only the activated substrates have been extensively studied.
25][26][27][28] Monochlorinated substrates represent a class of affordable, stable and broadly available chemicals that could be used to introduce a new CÀ C bond and a single new CÀ Cl bond with 100 % atom efficiency.Dilman and colleagues have achieved impressive yields by combining a photocatalyst with a copper catalyst, albeit at relatively high additive loadings. [26]Recently, Bernhardt and coworkers obtained γ-halonitriles in high yields for the ATRA of chloroacetonitrile using an organocopper(II) catalyst that is regenerated electrochemically, but they did not investigate other primary alkyl halides. [28]Besides these two examples, the turnover numbers (TONs) remain low (< 30 TON). [23,24]In this article, we investigate whether the highly efficient Ru III system can be used for ATRA of primary alkyl chlorides.New precatalysts of the type [Cp*Ru(Cl) 2 (PR 3 )] with various phosphines are studied, in addition to system optimization.Reaction monitoring and density functional theory (DFT) calculations are used to gain insights into the reaction mechanism.

Optimization of catalytic conditions
We started our investigations by applying the Ru III system reported by Severin and coworkers [21,22] to the ATRA of ethyl chloroacetate 2 on styrene 1 (Table 1).The primary acetonyl radical is partially stabilized by the neighboring ester group, and this was anticipated to facilitate halogen atom abstraction.At room temperature, the reaction did not proceed with [Cp*Ru(Cl) 2 (PPh 3 )] complex 4 Ph and magnesium as reducing agent (entry 1), while it gives quantitative yields for CCl 4 , [21] illustrating well the enormous difference in reactivity between these substrates.When using radical initiator azobisisobutyronitrile (AIBN) at 65 °C as the reducing agent only traces of product 3 a were obtained (entry 2), but increasing the temperature to 85 °C led to 39 % yield (entry 3).This is already an improvement compared to the direct use of analogous Ru II systems (Table S1, entry 8). [23,24]In the absence of activator or complex 4 Ph (Table 1 entry 6 & 7), when using [Ru(Cl) 3 ] or [Cp*Ru(Cl) 2 ] 2 as metal source, or when performing the reaction under air, product 3 a was not formed (Table 1 and S1).
Using a radical initiator to reduce and activate precatalyst 4 Ph has two major consequences on the system: (1) Undesired free radical polymerization of styrene occurs as side reaction, and (2) the concentration of radicals in solution is highly dependent on the decomposition rate of the initiator, and therefore on the reaction temperature.At 85 °C, AIBN has a (calculated) half-life of 39 min (Table S4), meaning that the radical concentration is high at the start, promoting styrene polymerization.In addition, AIBN is quickly depleted and active catalyst cannot be regenerated later on, resulting in earlier catalyst deactivation.To circumvent both problems, initiator 1,1'-azobis(cyclohexanecarbonitrile) (ACHN) was tested because of its higher 10 h half-life temperature of 85 °C.Using ACHN gave a substantial increase in yield of 3 a from 39 % (entry 3) to 61 % (entry 4).This corresponds to a TON of 61, which is the highest value reported for ruthenium-catalyzed ATRA of ethyl chloroacetate 2 to styrene 1 (a three-fold improvement compared to literature). [23,24]Upon increasing the temperature to 100 °C, the yield of product 3 a decreases to 45 %, due to  11 [c] 1 mol % ACHN (5 mol %) 85 37 12 [d] 1 mol % ACHN (5 mol %) 85 16 [a] Standard conditions: [styrene] = 1.38 M, [chloroacetate] = 2.76 M. [b] Average of 2 experiments.Yields were determined by 1 H NMR integration with an internal standard (1,3,5-trimethoxybenzene). [c] [chloroacetate] = 6.9 M. [d] [chloroacetate] = 1.38 M.
increased styrene oligomerization/polymerization (entry 5).A similar effect occurs when increasing only the ACHN loading (entry 9).Other reducing agents or radical initiators did not yield the desired product (Table S2).Doubling both the loading of precatalyst 4 Ph and ACHN resulted in a negligible increase in yield to 63 % (entry 10), while using half the amount resulted in less than 10 % yield (entry 8).Changing the equivalents of ethyl chloroacetate 2 resulted in lower yields (entry 11 & 12), due to competition with polymerization or decreased catalyst solubility in the reaction medium.The amount of ethyl acetate byproduct (i.e., the dehalogenated substrate) is negligible, suggesting that there is no considerable radical quenching via hydrogen atom abstraction from the solvent (i.e., reactions in benzene and chlorobenzene as the solvent afforded a similar or lower yield as reactions in toluene, Table S3).No byproduct resulting from double styrene insertion was observed either.

Catalyst screening
To improve catalytic activity, we turned our attention towards the synthesis of analogues of paramagnetic Ru III complex 4 Ph .Increasing the electron donating character of the phosphine was expected to facilitate substrate activation and increase catalyst activity, in analogy to the improvement observed for more electron rich η 5 ligands such as Cp*. [16]Similar studies have been published for [Cp X Ru II (X)(PR 3 ) 2 ] analogues, but, in the case of Ru II , phosphine decoordination is essential to create the active catalyst.For those catalysts the electronic effect of the phosphine on the intrinsic reactivity of the catalyst is (largely) masked by the change in RuÀ P bond strength. [14,16,29]When using [Cp*Ru III (Cl) 2 (PR 3 )] as precatalyst, the vacant site is created by halide abstraction instead and hence the direct electronic effect of the phosphine ligand on catalysis should be more pronounced.We therefore decided to prepare a series of paramagnetic [Cp*Ru III (Cl) 2 (PR 3 )] complexes with varying phosphines PR 3 and test them as ATRA catalyst.

Synthesis and characterization of [Cp*Ru(Cl) 2 (PR 3 )]:
A series of phosphine ligands with varying electronic and steric parameters (PMe 3 , PEt 3 , PCy 3 , P(p-CF 3 C 6 H 4 ) 3 and P(p-MeOC 6 H 4 ) 3 were selected.Coordination of the ligand to commercial [Cp*Ru(Cl) 2 ] 2 proceeded smoothly in DCM, yielding complexes 4 as paramagnetic red/orange solids.The complexes are relatively air stable as solids, but oxygen sensitive in solution.
Single crystal X-ray diffraction confirms the piano stool structure in all cases (Figure 2, Table 2).Solvent molecules did not cocrystallize, in contrast to the reported structure of 4 Ph , and complexes 4 Me and 4 Et crystallized with two molecules per unit cell. [21]All structures show differences in the two RuÀ Cl bond lengths.On average, RuÀ Cl bond is the longest for complex 4 Ph .The RuÀ P bond is the longest for 4 Cy reflecting the bulky nature of the PCy 3 ligand.The electronic variation in the aryl phosphine ligands does not translate to the RuÀ P bond length.Low temperature X-band EPR spectra showed rhombic spectra without resolved hyperfine couplings for each of the Ru III complexes 4 (Figures S4-S9, g values in Table 2), characteristic of low spin d 5 ruthenium complexes with a (distorted) octahedral geometry.
We employed cyclic voltammetry to determine the reducing power of the various Ru III/II couples in DCM (Table 2).Complexes 4 display a reversible oxidation to cationic Ru IV between 0.2-À 0.54 V vs. Fc + /Fc in accordance with literature, [30] with the exception of complex 4 Et , for which this oxidation is irreversible (Figure S10).The Ru III/II redox event is found between À 0.8 and À 1.0 V vs. Fc + /Fc and reduction to Ru II is irreversible, except for complex 4 PhCF3 .We infer that the negative charge on the formed [Cp*Ru(Cl) 2 (PR 3 )] À anion (except for 4 PhCF3 ) cannot achieve stabilization due to the electron rich nature of the complexes, consequently leading to rapid loss of a chloride anion on the CV time scale, accompanied by structural rearrangement. [30]nterestingly, complex 4 PhCF3 affords a quasi-reversible Ru III/II wave, indicating that redox reversibility can be improved by employing electron withdrawing ligands that provide better anion stabilization.Comparison of the complexes based on their Ru III/II reduction peak should be performed cautiously as different degrees of electrochemical reversibility are apparent throughout the series.Nevertheless, for the Ru III/II couple, the following series is obtained ordering the Ru II complexes from the least to most reducing: 4 PhCF3 < 4 Ph < 4 PhOMe < 4 Cy < 4 Et < 4 Me .Although the Tolman electronic parameter suggests that PCy 3 is the most electron-rich phosphine, this does not translate to the measured electrochemical potential of complex 4 Cy . [31]his deviation can be rationalized by the bulky nature of PCy 3 ligand that limits the extent of σ-donation to the metal center.
Catalytic activity of [Cp*Ru(Cl) 2 (PR 3 )]: The various ruthenium catalysts were tested using the optimized conditions described in Table 1, entry 4 (1 mol % catalyst, 5 mol % ACHN, 85 °C), with all complexes demonstrating activity except 4 Cy .The high steric demand of the PCy 3 ligand likely hampers substrate access to the metal center, consistent with its electrochemical behavior (vide supra).Surprisingly, the complexes 4 Me , 4 Et and 4 PhOMe did not give higher yields compared to complex 4 Ph , despite their greater reductive power (E 1/2 values in Table 2).Complexes 4 Et and 4 Ph performed similarly (59 % and 61 % respectively) despite the former having a much lower reduction potential (À 1.02 V vs. À 0.87 V respectively).Complexes 4 Me and 4 Et have similar reduction potentials, but 4 Me yields only 40 % of product.Similarly, higher yields were obtained with complex 4 PhCF3 over 4 PhOMe , despite the former being the least reducing.Overall, no clear correlation can be found between the Ru III/II redox potential and overall product yield.RuÀ Cl metrics or computed BDEs (Table S7) did not correlate to yields either.We hypothesize that, while the redox potential likely influences the activity (rate) of the reaction, it doesn't translate to overall yields because additional factors, such as catalyst activation and deactivation rates, are important.To assess this hypothesis, the reaction was monitored over time to gain insights into the reaction profile and relative reaction rates.

Reaction monitoring
The addition of ethyl chloroacetate 2 to styrene 1 was monitored using an in situ ATRÀ FTIR probe to obtain insights into reaction profiles and to understand to what extent electronics influence the reaction rate.The consumption of styrene 1 (909 cm À 1 ), chloroacetate 2 (1764 cm À 1 ), and the formation of the addition product 3 a (1734 cm À 1 ), were monitored by following the differential absorption spectrum (Figure S1).In the absence of ruthenium catalyst, styrene 1 conversion occurs immediately due to free radical polymerization, while in the presence of complex 4 Ph a long induction period (10 h) is observed (Figure 3, left).ACHN has a half-life temperature (t 1/2 ) of 85 °C, therefore, after 10 h, 50 % of the initiator has decomposed to free radicals, corresponding to approximately 5 eq. of free radicals per catalyst complex.Even though recombination between initiator radicals occurs due to their proximity upon formation -effectively reducing the free radical concentration available for precatalyst activation -it is plausible to consider 4 Ph as an efficient radical scavenger.b] Redox potentials for the Ru III/II couple were determined in a DCM solution with 100 mM [TBA][PF 6 ] as electrolyte, RE=Ag/AgCl, WE = glassy carbon, CE = platinum wire, scan rate = 0.1 mV s À 1 .Potentials given are vs.c] Average of 2 experiments.Yields are determined by 1 H NMR integration with an internal standard (1,3,5-trimethoxybenzene). [d] Crystallographic data reported in literature. [21]f active Ru II catalyst.A shorter induction period of 2 h is observed when using AIBN instead with complex 4 Ph at 85 °C, alongside a 7 times higher rate for product 3 a formation within the first 10 h, after which the catalyst suffers from deactivation.These observations are consistent with the lower 10 h half-life temperature of AIBN (65 °C), which results in a higher free radical concentration at earlier stages of the reaction (Figure S2).
When sufficient precatalyst has been activated (10 h for ACHN), the rate of product formation increases.Conversions of ethyl chloroacetate 2 and styrene 1 align well with formation of product 3 a up to 20 h (Figure 3, left), after which conversion of styrene 1 starts to exceed formation of product 3 a due to polymerization.The undesired styrene polymerization is most likely a result of free radical polymerization, but we cannot fully exclude ruthenium-catalyzed ATRP (albeit less likely under the current conditions, using excess of ethyl chloroacetate 2).
To compare the catalysts, we look at the maximum rates relative to 4 Ph .Aryl phosphine complexes 4 PhOMe and 4 PhCF3 demonstrate a similar induction period to 4 Ph , but are 0.17 and 0.57 times less active in product 3 a formation compared to 4 Ph (Figure 3, right).In contrast, alkyl phosphine complexes 4 Me and 4 Et feature a negligible induction period (< 1 h) and a higher rate of product 3 a formation at the beginning of the reaction, i. e. a 3.2 fold increase for 4 Me and 1.96 for 4 Et .This difference between alkyl and aryl phosphines indicates that the former are considerably more active catalysts.When comparing the maximum rates for product 3 a formation, the following reactivity order is found: 4 Me > 4 Et > 4 Ph > 4 PhCF3 > 4 PhOMe .This reactivity order correlates well with the reduction potential of the catalysts (Figure 3, right), except for 4 PhOMe , for which we hypothesize a higher sensitivity for catalyst deactivation (see also Table 2).

Substrate scope
Various substrates were tested to determine the applicability of the reaction (Figure 4).Complex 4 Ph was chosen as catalyst due to its easier synthesis and similar performance for overall yields to 4 Et .Moderate-to-good yields were obtained for chlorinated substrates that are able to provide sufficient stabilization to the primary radical (3 a, 3 b, 3 e, 3 f).Weakly stabilized or unactivated chloroalkanes gave poor or no yield (3 hÀ 3 k).In reactions where the yields of desired products are low, styrene oligomerization/polymerization is predominant.Interestingly, various chloroesters that can provide a similar radical stabilization independent of the ester group, afforded very different yields.The methyl and ethyl ester performed well (3 b 64 %; 3 a 61 %) but the tert-butyl ester gave a poor yield (3 c 30 %) and the benzyl ester was unreactive (3 d).This result suggests that steric hindrance of the substrate plays a crucial role for the overall yield.For 3 d a steric effect is less obvious.The absence of reactivity could be caused by a stronger interaction with the catalyst, leading to deactivation, or by πÀ π stacking with the solvent, which might hamper substrate activation.For tosyl chloride, a higher yield of 3 g was obtained in literature because polymerization is suppressed at room temperature. [21]Parasubstituted styrenes gave moderate-to-low yields (3 l 36 %; 3 m 19 %; 3 n 16 %), lower than styrene itself (3 a 61 %), poorly correlating to the electronic influence of the substituent.Of note was the low yield of 3 n (16 %) obtained from the reaction with 4-vinylbiphenyl -electronically similar to styrene but affording significantly lower yields -again suggesting the important influence of substrate sterics.Internal alkenes (3 o, 3 p) only yielded trace product formation, while electron-poor or unactivated alkenes (3 q-s) were completely unreactive under these conditions.We hypothesize that the observed steric influence arises during the radical recombination step, when the newly formed benzyl radical approaches the ruthenium complex for chloride atom abstraction (See Figure 5).

Computational study
We employed density functional theory (DFT, B3LYP/def2-TZVP/ disp3, 358 K) to gain further insight into the relative barriers for each step of the catalytic cycle (Figure 5).As homolytic bond breaking and formation are a key step in the cycle, the transition states were calculated both at the triplet (T) and the open-shell singlet (OSS) surface.A previously reported mechanism determined experimentally was taken as basis for our computational study. [32]As seen in Figure 5, after activation, the low-coordinate Ru II active species (A) abstracts a chloride atom from the ethyl chloroacetate starting material 2, generating a Ru III À Cl complex and the ethoxyacetonyl radical (B).This homolytic CÀ Cl bond splitting is best represented at the OSS surface, with a low barrier transition state of 11.7 kcal mol À 1 (Figure 5, TS1).In this transition state two electrons of opposite spin are spatially separated, one localized on the organic substrate and the other on the metal center (Figure 6).The   resulting ethoxyacetonyl radical adds onto the alkene to generate a benzyl radical (C) that recombines with the Ru III À Cl complex to regenerate the Ru II catalyst (A), with concomitant product formation (D).Again, this process has the lowest barriers at the OSS surface (Figure 5B).The overall energy gain in the reaction is relatively low (ΔG°2 98K < 20 kcal mol À 1 ), even for more reactive chlorinated substrates (Table S6).This can be rationalized by the overall decrease in entropy resulting from combining two reactants into one product.The reaction is enthalpically driven by the formation of a strong CÀ C bond.
The rate determining step (RDS) for the OSS energy surface is the addition of the ethoxyacetonyl radical onto styrene (ΔG � = 10.1 kcal mol À 1 , TS2), matching experimental observations for the ATRA of di-and trichloroacetate to styrene. [32]The relatively low computed energy barriers (ΔG � < 15 kcal mol À 1 ) suggest that the reaction should be fast at room temperature, but this is disproven experimentally.Firstly, elevated temperatures are required for catalyst activation with ACHN.In addition, the catalyst is only slowly activated over time, resulting in low active catalyst concentrations, which affects the efficiency of the reaction.Even though we could not identify off-cycle species experimentally, we cannot fully exclude their formation, for instance from recombination of a Ru II complex and a radical forming a Ru III À alkyl complex.Regeneration of active Ru II species from this Ru III -alkyl species likely requires heating.
Finally, the energy barrier for radical recombination of the benzylic radical with the Ru III À Cl bond at the OSS surface (TS3; ΔG � = 5.4 kcal mol À 1 ) is lower than the barrier for second styrene insertion (TS4; ΔG � = 17.9 kcal mol À 1 ), showing that polymerization (E) is kinetically disfavored over the desired ATRA reaction.This significant difference in energy barriers explains the high selectivity for monoadduct formation and the lack of double insertion products observed experimentally.

Summary & Conclusions
ATRA of primary alkyl chlorides to alkenes remains a difficult reaction, but by employing precatalysts of the type [Cp*Ru-(Cl) 2 (PR 3 )] and activating them in situ, decent yields and TONs were achieved.By employing ACHN as radical initiator, higher reaction temperatures could be used without altering the free radical concentration and a three-fold improvement in TON (61) using [Cp*Ru(Cl) 2 (PPh 3 )] 4 Ph was achieved for ethyl chloroacetate 2. Chloroalkanes with radical-stabilizing functional groups gave moderate to good yields, while unactivated chloroalkanes proved unreactive.Para-substituted styrenes gave moderate yields, albeit lower than styrene itself.The influence of the phosphine ligand was studied via reaction monitoring with in situ ATRÀ FTIR.A long induction period is present for the aryl phosphine complexes 4 Ph , 4 PhOMe and 4 PhCF3 but is absent for alkyl phosphine complexes 4 Me and 4 Et .The maximum rate of product formation correlates well with the catalyst redox series (electronic properties).Conversely, an important influence of steric effects on the reaction is observed in the substrate scope and likely stems from the radical recombination step.To the best of our knowledge, an important steric influence for this type of ATRA reaction has not been previously described.DFT indicates an open-shell singlet surface for the reaction.Our calculations also shows that polymerization is kinetically disfavored due to a higher energy barrier, explaining the good selectivity towards ATRA products over double insertion products.

General considerations
All reagents were purchased from commercial suppliers and used without further purification unless stated otherwise.All reactions were carried out under an inert argon or nitrogen atmosphere using standard Schlenk line techniques and glovebox apparatus.Additional details can be found in the supplementary information.

Crystallographic details
X-ray diffraction data was measured on a Bruker D8 Quest Eco diffractometer using graphite-monochromated (Triumph) Mo Ka radiation (λ = 0.71073 Å) and a CPAD Photon III C14 detector.The sample was cooled with N 2 to 100 K with a Cryostream 700 (Oxford Cryosystems).Intensity data were integrated using the SAINT software. [35]Absorption correction and scaling was executed with SADABS. [36,37]The structures were solved using intrinsic phasing with the program SHELXT 2018/2 [38] against F 2 of all reflections.
Least-squares refinement was performed with SHELXLÀ 2018/3. [38]ll non-hydrogen atoms were refined with anisotropic displacement parameters.The hydrogen atoms were introduced at calculated positions with a riding model.

General experiment for catalysis
In a flame dried 10 mL Schlenk flask were added [Cp*Ru(Cl) 2 (PPh 3 )] 4 Ph (7.85 mg, 0.0138 mmol, 0.01 eq.) and ACHN (16.85 mg, 0.069 mmol, 0.05 eq.).Dry toluene (0.55 mL) was added, in order to have a total reaction volume of 1 mL (taking into account the volume of the substrates needed).Subsequently, styrene 1 (159 μL, 1.38 mmol, 1 eq.) and ethyl chloroacetate 2 (295 μL, 2.76 mmol, 2 eq.) were added.The reaction mixture was stirred and degassed with argon for two minutes, the flask was closed and heated to 85 °C for 24 hours.After cooling down, the flask was opened to air and 1,3,5-trimethoxybenzene was added as internal standard (200 μL from a stock solution in CDCl 3 , approximately 0.45 M).The mixture was stirred, a sample was taken and diluted with CDCl 3 for 1 H NMR analysis.Yields were determined using the characteristic product doublet of doublet at 4.96 ppm (CHÀ Cl, CDCl 3 ) with respect to the internal standard singlet at 6.08 ppm.

Reaction monitoring using in situ ATRÀ IR probe
In a 2-necked 10 mL Schlenk flask, catalyst (0.0276 mmol, 0.01 eq.) was placed, followed by 0.89 mL dry toluene, styrene 1 (318 μL, 2.76 mmol, 1 eq.) and ethyl chloroacetate 2 (590 μL, 5.52 mmol, 2 eq.).The reaction mixture was stirred and degassed with argon for 5 minutes.The ATRÀ IR probe was inserted in the flask, the system was closed and heated to 85 °C.When the temperature was reached, ACHN was added (0.2 mL of 0.069 M stock solution, 0.138 mmol, 0.05 eq.).The reaction was stirred at 85 °C for about 40 hours, recording a spectrum every 15 minutes.After cooling down to room temperature, the flask was opened to air and 1,3,5trimethoxybenzene was added as internal standard (400 μL from a stock solution in CDCl 3 , approximately 0.45 M) and the probe was removed.A sample was taken for 1 H NMR analysis.The yield of the reaction was determined and used to interpret the ATRÀ IR data.

Figure 1 .
Figure 1.A) ATRA reaction equation.B) Literature examples of ruthenium catalysts used for ATRA of chlorinated substrates.

Figure 4 .
Figure 4. Substrate scope for the ATRA of various alkyl chlorides and alkenes.Yields are averages of 2 experiments and were determined by 1 H NMR integration with an internal standard (1,3,5-trimethoxybenzene).

Figure 6 .
Figure 6.Spin density plots for TS1 and TS3 in the open shell singlet spin state, showing spatial separation of the alpha and beta electron in the transition states.

Table 1 .
Optimization of the reaction conditions for the ATRA of ethyl chloroacetate 2 to styrene 1[a]

Table 2 .
Selected crystallographic metrics for complexes 4, EPR parameters, redox potentials and yields of product 3 a obtained in catalysis.
The resulting X-ray crystallographic data was deposited at the Cambridge Crystallographic Data Centre (CCDC), under the deposition numbers CCDC 2225048-2225052.contain(s) the supplementary crystallographic data for this paper.These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.