Synthesis and structure of annulated dithieno[2,3- b ;3 ʹ ,2 ʹ - d ]thienyl-and ring-opened 3,3 ʹ -bithienyl Fischer carbene complexes

Nucleophilic attack on the central sulphur of dithieno[2,3- b ;3 ʹ ,2 ʹ - d ]thiophene (DTT) by n -BuLi opens the central thiophene ring and afforded, after subsequent reaction with Cr(CO) 6 and alkylation with [Et 3 O][BF 4 ], a series of mono- and biscarbene complexes containing a 3,3 ʹ -dithienyl backbone with a SBu substituent. Repeating the reaction with diisopropylamine as the nucleophile, leads to a dihydrodesulphurization reaction with ring-opening of the central thiophene ring of DTT and elimination of the sulphur atom. Subsequent reaction with n -BuLi or LDA, Cr(CO) 6 and [Et 3 O][BF 4 ] afforded 3,3 ʹ -dithienyl mono- and biscarbene complexes. In both instances α , α ʹ -dithienothiophene biscarbene complex was ob- served spectroscopically but not isolated. By using α , α ʹ -dibromodithieno[2,3- b ;3 ʹ ,2 ʹ ]thiophene as substrate, improved yields of the targeted mono- and biscarbene complexes of [2,3- b; 3 ʹ ,2 ʹ - d ]-DTT (M = Cr, W) could be obtained. The biscarbene complexes were unstable in the reaction mixture but in the case of tungsten could be isolated after in situ aminolysis with dimethylamine. The use of KHMDS as base con- verted Cr(CO) 6 to K[Cr(CO) 5 (CN)] and after reaction with DTT and subsequent alkylation with [Et 3 O][BF 4 ], afforded the chromium tetracarbonyl carbene complex of DTT.

able yields ( > 30%). The complexes are stable in the solid state, but the biscarbene complexes to a lesser extent in solution. The latter were readily oxidized to the carbene-ester complex by trace amounts of oxygen, and at elevated temperature a carbene-carbene coupled biscarbene complex with an extended conjugated linker formed. Notably, in all instances the [3,2-b ;2 ʹ,3 ʹd ]-DTT rings remained intact (inert) during carbene formation reactions and subsequent reactions occurred only at the carbene sites.
In this study, [2,3-b ;3 ʹ,2 ʹd ]-DTT is chosen as precursor for the preparation of Fischer carbene complexes (FCCs) and is compared with results obtained for the [3,2- The role of different bases as well as the reactivity patterns during lithiation and carbene formation are studied to expand the scope of possible reactions with DTT in the context of Fischer carbene chemistry.

Syntheses of precursors and complexes
The precursor [2,3-b ;3 ʹ,2 ʹd ]-DTT ( P1 ) was prepared as described by Nenajdenko [3] , with amendments as suggested by Allared [16] . The authenticity and purity of P1 were determined by spectroscopic characterization (Fig. S1, Supplementary data ). Carbene complexes were prepared by making use of P1 in three different methods by using a range of strong bases: n -BuLi ( nbutyllithium), a mixture of nucleophilic diisopropylamine (HDA) and n -BuLi (with subsequent formation of lithium diisopropylamide (LDA)), and potassium bis(trimethylsilyl)amide (KHMDS) (non-nucleophilic base). Reactions were performed at low temperatures (-78 °C) to create the opportunity to discriminate between reaction sites. As a measure of favoured reaction sites and stabilities of reaction intermediates, the yields and compositions of the final isolated carbene complexes serve as indicators. To further activate P1 and direct reactions to the α-positions of the DTT substrate the α-protons were exchanged by bromo substituents to facilitate lithium-halogen exchange reactions. Brominating the αpositions using NBS produces 5,5 ʹ-dibromo-[2,3b ;3 ʹ,2 ʹd ]-DTT ( P2 ), as an excellent building block to be used in Fischer carbene syn-thesis [3] . Two further experiments were executed with the aim to protect the DTT ring and to stabilize final reaction products ( Scheme 1 ).
Method 1 comprises of the classical Fischer method using n -BuLi, in THF at low temperatures ( −78 ˚C), ( Scheme 1 , method 1) [17] . The main reaction pathway involves a ring opening reaction of the central thiophene ring in P1 leading to the formation of a 3,3 ʹ-bithienyl-2-SBu backbone [18,19] . In a proposed sequence of reactions, the base (Bu − ) participates in nucleophilic attack on the central sulphur atom of the three fused thiophene rings, breaking open the central thiophene ring of P1 ( Scheme 2 ) . After opening the central ring, an anion resides on an α-position of the adjacent ring (2 ʹ position). Subsequent reaction with the metal carbonyl and alkylation will afford the monocarbene complex with both substituents on inside ring positions on separate rings, [2 ʹ-SBu-C 4 H 2 S-3,3 ʹ-{2-C(OEt)Cr(CO) 5 }C 4 H 2 S] ( 2 ). The anion is transferable to other active α-positions. The most active one is the remaining α-position (5) of the thiophene ring with the SBu substituent. We anticipate that conversion of the anion site to the remaining α-position on the thiophene bearing the SBu substituent is favoured. The activation is enhanced at this site by the presence of the SBu substituent in the other α-position of the same ring. After alkylation the product isolated is [{2-SBu,5-C(OEt)Cr(CO) 5 }C 4 HS-3,3 ʹ-C 4 H 3 S] ( 1 ) which represents the major product of the reaction. The next most active site for biscarbene complex formation will be at an inside αposition (2 ʹ) of the second thiophene ring. The biscarbene complex that formed after alkylation has the second highest yield, and is [{2-SBu,5-C(OEt)Cr(CO) 5 The second biscarbene complex of 3,3 ʹ-bithienyl-2-SBu is isolated in a very low yield and displays a second carbene substituent in the outside position (5 ʹ) of the second thiophene ring, [{2-SBu,5-C(OEt)(Cr(CO) 5 }C 4 HS-3,3 ʹ-{5 ʹ-C(OEt)Cr(CO) 5 }C 4 H 2 S] ( 4 ). The formation of the second carbene group in the electronically favoured inside position ( 3 ), is in competition with formation at the second outside α-position ( 4 ), which is sterically the less crowded position. The byproduct butylcarbene complex make up the remainder of complexes obtained by method 1. The targeted biscarbene complex [{Cr(CO) 5 C(OEt)} 2 -5,5 ʹ-C 8 H 2 S 3 ] ( 5 ) was spectroscopically identified as present in the reaction mixture in trace amounts, but could not be isolated.
Mono-and dinuclear complexes are obtained by using 1.8 equivalents of the lithiating agent to one equivalent of P1 . The stoichiometric ratio employed was rationalized as 1.5 molar equivalents are required for the formation of a 1:1 ratio of mono-:biscarbene products, with 1.2 equivalents excess. Nucleophilic attack of the mono-and dianions on carbonyl carbon atoms of chromium hexacarbonyl occur upon their addition. Quenching the resulting acylmetallates with [Et 3 O][BF 4 ] yielded the neutral carbene compounds. Monocarbene complexes 1 (yield 32%) and 2 (yield 8%) along with biscarbene complexes 3 (yield 24%), 4 (yield < 3%) and 5 (yield < 3%) were obtained. As major products, 1 and 3 prove intra-ring electron delocalization from C2 (SBu) to C5 stabilizes the carbene fragment on C5. In the case of 3 , additional interring stabilization from C2 to C2 ʹ allows a second carbene fragment to coordinate to C2 ʹ. [12][13][14][15] . Method 2 ( Scheme 1 ) involves the deprotonation of P1 using 2.4 equivalents KHMDS (2 equivalents including 1.2 times excess, allowing for dinuclear biscarbene formation), with the rest of the reaction following the classical Fischer method. The reaction proceeded poorly as 47% of the starting material was recovered and only little evidence of deprotonation was observed with the formation of [Cr(CO) 4 {C(OEt)-5-C 8 H 3 S 3 }CNEt] ( 7 , yield 12%). Hence the amide base also attacked carbonyl groups as is evident from the formation of 7 and [Cr(CO) 5  nism shows the base acting as a nucleophile and nucleophilic addition occurs to the carbonyl carbon atom of chromium hexacarbonyl producing K[Cr(CO) 5 CN] ( Scheme 3 ). The latter reacts with deprotonated P1 to afford 7 , while unreacted cyanide of [Cr(CO) 5 CN]is alkylated to give 6 ( Scheme 3 ).
Nucleophilic attack on (and ring-opening of) P1 with the use of n -BuLi was unexpected and is not observed for the reaction of n -BuLi on the isomer of P1 , [3,2- , or other annulated thiophenes [12][13][14][15]21] . The reaction is unique for P1 (three sulphurs on the same side of DTT) and might be ascribed to the lack of conjugated pathways from the two ends of P1 to the central sulphur atom (now more electrophilic). Hence an experiment was planned to compare the rates of nucleophilic attack and deprotonation by using a stronger nucleophile ( e.g. amine, in this case diisopropylamine HDA) along with n -BuLi ( Scheme 1 , method 3). In the reaction, 2.4 equivalents of HDA and n -BuLi (2 equivalents to allow for dinuclear complex formation with 1.2 times excess) were used. At very low temperatures ( ca. −80 °C), HDA remains protonated, and n -BuLi unreacted. The occurrence of nucleophilic attack on the central sulphur atom of P1 by HDA can therefore be compared to the probability of nucleophilic attack by the stronger base/weaker nucleophile, n-BuLi. As the temperature of the reaction mixture is allowed to increase gradually, LDA is generated in situ (deprotonation of HDA by n -BuLi), and either LDA or excess remaining n-BuLi can deprotonate P1 . It was determined that the first reaction was the nucleophilic attack of HDA on the central sulphur atom of P1 , leading first to ring-opening and after a second attack, to dihydrodesulphurization of P1 . Thus a route that yields S(DA) 2 (DA = diisopropylamide) along with 3,3 ʹ-bithiophene (3,3 ʹ-BT) is proposed ( Scheme 4 ). A more reactive and electrophilic central sulphur initialises the first addition of HDA, resulting in an intermediate with favourable electronic properties facilitated by the external thiophene rings, allowing for a second attack with elimination of the sulphur atom.
Bromination of the α-positions of DTT produces 5,5 ʹ-dibromo- Fig. S1 for NMR characterization), which is an excellent building block to be used in Fischer carbene synthesis, as lithium-halide exchange reactions show many advantages compared to deprotonation reactions in annulated thiophene substrates. The double lithium-halogen exchange reaction affords the debromination of P2 using 2.4 equivalents n -BuLi ( Scheme 1 , method 4) and subsequent reaction with the metal carbonyl and alkylating agent lead to the targeted compounds [Cr(CO) 5 {C(OEt)-5-C 8 H 3 S 3 }] ( 10 , yield 44%) and 5 (trace amount). Deprotonation ( 10 ) is in competition with ring-opening ( 1 ). Compound 1 (5% yield) was observed as by-product of the reaction. Repeating the reaction with tungsten carbonyl gave no ring-opening products but [W(CO) 5 {C(OEt)-5-C 8 H 3 S 3 }] ( 11 , yield 36%), and the corresponding biscarbene complex [{W(CO) 5 C(OEt)} 2 -5,5 ʹ-C 8 H 2 S 3 ] in very low yield (not purified/isolated), as well as 20% P1 (debrominated P2 ). During work-up it was observed that P2 readily converts to P1 . The biscarbene complex 5 could be isolated ( Scheme 1 ), but the low yield allowed only for characterization by 1 H NMR and FT-IR spectroscopy, as well as single crystal analysis. The tungsten analogue (method 4), could not be isolated, leading to the conclusion that the bisethoxycarbene complexes are highly reactive in the reaction mixtures in solution. Method 5 (Scheme 1) was therefore introduced to stabilize the biscarbene complexes before purification and isolation, and applied to the more reactive tungsten complexes. The mono-and bisethoxycarbene tungsten complexes are first synthesized using method 4, followed by in situ aminolysis of the reaction mixture. Excess dimethylamine hydrochloride and sodium hydroxide are added to the reaction mixture. Thereafter, distilled H 2 O was added dropwise until all the suspended salts dissolved and reacted [23] . [W(CO) 5 {C(NMe 2 )-5-C 8 H 3 S 3 }] ( 12 , yield 69%) and [{W(CO) 5 C(NMe 2 )} 2 -5,5 ʹ-C 8 H 2 S 3 ] ( 13 , yield 21%) were isolated in moderate to good yields along with 8% of P1 .  a Proton chemical shifts for the SCH 2 CH 2 CH 2 CH 3 fragment are reported with the first value being the chemical shift of the first methylene group after the sulphur atom, followed by the values of the second and third methylene groups. The last chemical shift is that of the methyl group. b Proton chemical shifts for the ethoxy fragment are reported with the first value being the chemical shift of the methylene group, and the second the chemical shift of the methyl group. c Proton chemical shifts for the NEt fragment ( 7 ) are reported with the first value being the chemical shift of the methylene group, and the second the chemical shift of the methyl group. In the case of 12 and 13 the chemical shifts represent the methyl groups of NMe 2 .

1 H NMR spectroscopy
NMR spectra, recorded in CDCl 3, support the assigned molecular structures of the compounds. 1 H NMR data are summarized in Table 1 . Two classes of monocarbene complexes ( 1 (C5-) and 2 (C2-)) were isolated from method 1 ( Scheme 1 ) containing a 3,3 ʹ-BT-2-SBu-and 3,3 ʹ-BT-2 ʹ-SBu-thienyl substituent, respectively. Both SBu ( inside ) and the carbene substituent ( outside ) are on the same thiophene ring in 1 , while both substituents are on the 'inside' position on different thiophene rings in 2 ( Fig. 3 ). Resonance structures predict carbene carbon stabilization of C5-and C2-monocarbene complexes to reflect in the downfield shift of the H4 and H5 resonance, respectively. This is confirmed by experimental data, as these resonances are the most downfield ( Table 1 ). The favoured formation of 1 is due to the carbene fragment attaching to the least steric congested C5-position; representing a linear (open) arranged molecule, with stabilizing intra-ring electron delocalization between C5 (C carb -acceptor) and C2-SBu (S-donor), Fig. 3

(a).
In contrast the main delocalization pathway between the C2 ʹ-SBu substituent and the C2-carbene substituent ( inside to inside ) is between the two thiophene rings in 2 , Fig. 3 (b). Therefore ring, SBu and OEt proton resonances for 1 are significantly more downfield compared to 2 , indicating more stabilization/electron density delocalization towards the carbene carbon. Compared to C5-and C2-3,3 ʹ-BT monocarbene complexes, 1 and 2 respectively show no significant difference in proton resonances caused by the additional C2-SBu and C2 ʹ-SBu fragment, respectively [22] . Two classes of 3,3 ʹ-BT-2-SBu biscarbene complexes ( 3 (C5,C2 ʹ-) and 4 (C5,C5 ʹ-)) were isolated from method 1 ( Scheme 1 ), with 3 as the major product. Compared to the analogous Fischer carbene reaction with 3,3 ʹ-BT, the C5,C2 ʹ-biscarbene complex formed only in trace amounts and the major product was the C5,C5 ʹbiscarbene complex ( 9 ) [22] . This leads to the supposition that the SBu fragment stabilizes the two carbene carbons in 3 through intra-and inter-ring electron delocalization, although sterically more crowded, where only intra-ring electron delocalization is possible in 4 . Again no significant difference in proton resonances caused by the additional SBu fragment is seen when comparing 4 with 9 ( Table 1 ) [22] .
The 1 H NMR spectrum of 7 is contaminated with 6 ( Fig. 4 ). Compound 6 is known and the chemical shifts for the protons in the NEt fragment are found at 3.63 ppm (methylene group) and 1.44 ppm (methyl group) [20] . Replacing a carbonyl group with a CNEt ligand ( 10 vs. 7 ), which have very similar electronic bonding properties, has very little effect on the proton resonances. Protons H4 ʹ and H5 ʹ of 7 now resonate at 7.45 ppm as a broadened singlet and no longer appear as separate doublets at 7.47 and 7.50 ppm as for 10 ( Fig. 4 ). The targeted [2,3-b ;3 ʹ,2 ʹd ]-DTT biscarbene complex ( 5 ) display H4 and ethoxy chemical shifts slightly but not significantly more downfield compared to its monocarbene analogue, 10 .
Replacing an ethoxy with a stronger π -donor amino substituent has a significant effect on H4, shifting the resonance upfield from 8.38 ( 11 ) to 6.86 ppm ( 12 ), due to the superior stabilization of Fischer carbene complexes by N-donor heteroatoms compared to oxygen. This results in a marked quenching of the electron withdrawing effect of the carbene metal carbonyl moiety, whereby the DTT-carbene substituent chemical resonances are not significantly shifted downfield from the precursor [2,3- resonances ( δ 7.40 ppm and 7.38 ppm). Manifestation of the bonding properties for the two methyl substituents (NMe 2 ) of 12 and 13 is found in two separate resonances, respectively, indicating that the methyl substituents are in different electronic environments due to restricted rotation around the C carb −N bond.

13 C NMR spectroscopy
The 13 C NMR data of the carbene complexes, recorded in CDCl 3 , are summarized in Table 2 . The electrophilic carbene carbon lacks electron density; hence its chemical shift appears in the most 157.0 In the case of 7 , two trans signals are reported first (one trans to the carbene carbon and one trans to CNEt) followed by one cis signal ( trans to CO) that is higher in intensity. f Assignments could not be made unambiguously.  Table 2 . This is also the trend in analogous 3,3 ʹ-BT carbene complexes (330.1 and 316.3 ppm, respectively) [22] . The carbene carbon chemical shifts of 3,3 ʹ-BT C2-and C5-monocarbene complexes are at ca. 7 ppm lower field compared to their analogous 3,3 ʹ-BT-2 ʹ/2-SBu carbene complexes, 2 and 1 resepctively [22] . In the case of 4 , the carbene carbon resonance of its C5 ʹ-carbene fragment (on the C3 ʹ-ring without a C2-SBu fragment) is ca. 4 ppm more downfield compared to the carbene carbon signal of its C5-carbene fragment (on the C3-ring containing C2-SBu). This we ascribe to the donating property of the SBu-substituent into the carbene carrying thiophene ring, once connected via a conjugated pathway with a strong withdrawing carbene carbon on the same thiophene ring ( Fig. 3 ). The carbene carbon signal of the C5 ʹ-carbene fragment of 4 (155.1 ppm, Table 2 ) is similar to the carbene carbon resonance of 9 (155.8 ppm), the analogous 3,3 ʹ-BT biscarbene complex [22] .
The observation of π -donation is in contrast with the σinductive electron withdrawing effect of the SBu-fragment on the thienyl substituent. This is a consequence of replacing H by a more electronegative S atom. In 1 and 2 , the presence of the SBugroup results in a ca. 20 ppm downfield shift of C2 and C2 ʹ respectively, when compared to their 3,3 ʹ-BT equivalents (129.1 (C2) and 122.4 ppm (C2 ʹ), respectively) [22] . The same principle is seen when comparing the C3-ring (containing a SBu fragment) of 4 with its own C3 ʹ-ring (without a SBu fragment) and again with 9 (SBu fragment is absent). Hence carbene complexes with 3,3 ʹ-BT-2/2 ʹ-SBu substituents have less electron density in the thienylene substituent to stabilize the carbene carbons compared to 3,3 ʹ-BT carbene complexes.
In the 1 H NMR spectrum of 7 , the presence of 6 can be observed. However, because of the long data acquisition time needed to acquire the 13 C NMR spectrum of 7 , decomposition of 6 is observed, accompanied with the absence of signals belonging to 6 ( Fig. 5 ) [20] . When one carbonyl group of 10 is replaced with CNEt to give 7 , downfield shifts are seen in the carbene carbon and the carbonyl groups' resonances ( Fig. 5 ). The carbonyl signals change from a characteristic pentacarbonyl fingerprint (one trans and one cis ) to a tetracarbonyl fingerprint (two trans and one cis ).
Characteristic values for different transition metals are seen in the resonances of carbene carbons and carbonyl groups. These resonances are ca. 23 ppm downfield when comparing 10 (Cr) with 11 (W) ( Fig. 5 ), and similar to their respective analogous [2,3-b ]-TT carbene complexes [21] . The aromatic carbon most affected by a carbene fragment is the carbon ipso to the carbene carbon. Comparing C3 ʹ-ring resonances of 3 and 4 , the ipso carbons (C2 ʹ and C5 ʹ respectively) are ca. 20 ppm downfield compared to their carbene free analogue C2 ʹ and C5 ʹ positions in 4 and 3 respectively, Table 2 . The chemical shifts for the carbene carbon atom and C4 of aminocarbene complex 12 is lowered by ca. 37 and 23 ppm, respectively, compared to ethoxycarbene complex 11 , with the same metal and thienylene substituent. The two methyl substituents of NMe 2 in 12 and 13 resonate as separate peaks in their 13 C NMR spectra, ca. 8 ppm apart. Due to the reactivity of 5 , no 13 C NMR spectrum could be obtained for this compound, as decomposition occurred during the long data acquisition time needed to observe the carbene carbon resonance.

FT-IR spectroscopy
All the compounds represent a M(CO) 5 L system [24] , with the exception of 7 , and their IR data are summarized in Table 3 . Compound 4 was isolated in trace amounts as an inseparable mixture along with 3 and due to their similar chemical properties, they could not be purified further. The A 1 (2) and E bands of 5 , 12 and 13 overlap to give one strong broad band in each case ( Table 3 ).
The cis -M(CO) 4 L 2 system ( 7 ) has C s symmetry and display four IR and Raman active carbonyl stretching frequency modes (2A 1 , B 1   (1)

Molecular structures
The carbene complexes were crystallized by dissolving samples in DCM, layering the saturated solution with hexane, and allowing the crystals to form at low temperatures by slow diffusion and evaporation.
Single crystal X-ray diffraction studies confirmed the molecular structures of complexes 1 , 5 , 7 and 10 -13 ( Fig. 6 ). Selected bond lengths, angles and torsion angles are given in Table 4 [26][27][28] . All the ethoxycarbene complexes reported have the preferred Z -isomer arrangement, where the stereoelectronic interaction of the alkyl substituent and a cis-CO ligand prevails against the steric repulsion in alkoxy Fischer carbene complexes [29] .
To determine if the thienylene spacer, carbene carbon, metal and ethoxy/amino fragment are all in the same plane, for each structure a mean plane was drawn through C2, C3, C4 and C5, and a second through M, C carb and O/N. The angle between the two planes was measured, revealing that most of the crystal structures have a planar arrangement. Steric congestion in 5 prevents the molecule from being planar and the angles between the two mean planes, on each side of the molecule, are 7.39 and 8.87 ˚. These angles are significantly greater in 12 and 13 ( Table 4 ), due to the rigid, curved and condensed DTT backbone and the bulky dimethylamine fragment/s resulting in a more crowded molecule. This result is supported by the torsion angles M −C carb −C5 −C4 and O/N −C carb −C5 −C4 reported in Table 4 .
From inspection of the C carb −C5 bond lengths, it is evident that 12 and 13 (aminocarbene complexes) are less dependent on electron density from the thienylene substituent to stabilize the carbene carbon compared to the crystal structures of the ethoxycarbene complexes, hence their significantly longer bond lengths, as expected from the conclusions drawn from the NMR data. The M −CO trans to Ccarb bond lengths of the compounds are shorter than the M −CO trans to CO bond lengths and the respective averaged M −CO bond length of Cr(CO) 6 and W(CO) 6 [31,32] . This is a result of poorer π -acceptor properties of the carbene carbon compared to a carbonyl ligand and the trans to C carb carbonyl ligand compensates for this by increasing metal-to-carbonyl-carbon backbonding. The same principle in seen in 7 , where the longer M −CNEt bond length is compensated for by the shorter M −CO bond length trans to CNEt ( Table 4 ) ( Table 4 and Figure 6 ). In the case of aminocarbene complexes ( 12 and 13 ) the M −C carb −C5 bond angles are significantly decreased and the N −C carb −C5 bond angles increased as a result of the sterically crowded carbene substituents ( Figure 7 ). The aminocarbene substituent, because of electronic involvement with the carbene carbon, places the methyl groups of the nitrogen in unfavourable planar orientations with respect to the bulky metal carbonyl fragment. The thiophene substituent rotates out of the plane to alleviate congestion. As a result, the DTT substituent now has an unfavourable orientation of the ring pi-cloud to effectively overlap with the carbene p π -orbital.
The two thiophene rings in 1 are non-planar as indicated by the inter-ring torsion angles C2 −C3 −C3 ʹ−C2 ʹ of 145.3(2) ˚and C4 −C3 −C3 ʹ−C4 ʹ of 146.4(2) ˚, striking a balance between electronic and steric factors. Little inter-ring conjugation/electron delocalization between the thiophene rings occur, as concluded from the 1 H NMR spectral data. Rotation around the inter-ring bond of the thiophene rings generates two positions for the sulphur atom of the second thiophene ring. Two molecules are observed in the unit cell; one molecule with the sulphur atom of the second thiophene ring rotated to the same side as the first sulphur atom in the spacer and in the second molecule the second thiophene ring is rotated away from the other sulphur atom in the spacer. The one molecule showed distortion of the SBu fragment. Compounds 10 and 11 packed in a similar fashion, with ππ interaction distances measured at 3.250 and 3.940 Å , respectively (Fig. S6, Supplementary data ), while the solid state structure of 13 packs such that both stacking and columnar packing is observed when viewed down two different axes (Fig. S7, Supplementary data ).

Conclusions
The use of a more reactive isomer [2,3-b ;3 ʹ,2 ʹd ]-DTT, as an annulated thiophene substrate for the preparation of Fischer carbene complexes with different bases, revealed unexpected ring-opened products. The central sulphur atom of [2,3-b ;3 ʹ,2 ʹd ]-DTT was attacked by organolithium reagents ( e.g. n -BuLi) or amine nucleophiles (HDA). While using n -BuLi the majority of carbene complexes formed resulted from a ring-opened DTT spacer. Nucleophilic attack on the central thiophene sulphur by n -BuLi afforded a SBu substituent on the inside position of a 3,3 ʹ-BT-2-SBu backbone. Employment of nucleophilic HDA/ n -BuLi again resulted in attack by HDA on the central sulphur atom, but this time with excision of the central sulphur atom, producing mono-and biscarbene complexes of 3,3 ʹ-BT. In both instances carbene complexes from deprotonated DTT were spectroscopically observed, but could not be isolated. Although deprotonation was observed, KHMDS also facilitated unprecedented nucleophilic addition to the carbonyl carbon atom of chromium hexacarbonyl to afford K[Cr(CO) 5

Methods and materials
All operations were carried out using standard Schlenk techniques or vacuum line techniques under an inert atmosphere of nitrogen or argon, using oven-dried glassware. Silica gel 60 (parti-cle size 0.063-0.20 mm) was used as resin (stationary phase) for all column chromatography separations.
Triethyloxonium tetrafluoroborate was prepared according to literature procedure and stored in diethyl ether under Ar (g) [33] . Boron trifluoride etherate was distilled before use. Under N 2 atmosphere THF and diethyl ether were distilled over sodium wire and benzophenone, hexane and benzene over sodium wire and DCM and acetonitrile over CaH 2 . CDCl 3 was dried over CaH 2 . Other chemicals commercially supplied by Sigma Aldrich and Strem Chemicals were used as received. The n -BuLi used in syntheses, was from a stock 1.6 M solution in hexane.
P1 was prepared from 3-bromothiophene that involved coupling of two thiophene rings using a lithium bromide exchange and copper chloride mediated coupling, followed by bromination with N-bromosuccinimide (NBS). Another lithium bromide exchange and thio-ring closure with S(PhSO 2 ) 2 yields pure P1 after standard reaction work-up and purification [3,16] . P2 is prepared by bromination of P1 with NBS. [3] Assignment of the NMR resonances of these precursors is given in the Supplementary data (Fig.  S1).
Compounds 1 , 2 , 7 and 10 could not be characterized using mass spectrometric (MS) analysis as the compounds did not ionize during the technique. In the case of 4 , only NMR analysis was carried out as the compound was obtained as an inseparable mixture along with 3 . The low yield of 5 resulted in too small a sample size to collect 13  Preparation of the samples was carried out under Ar (g) and the NMR tubes were sealed before data collection. The 1 H NMR data are reported in the format: chemical shift (integration, multiplicity, coupling constant, assignment) and the 13 C NMR data in the format: chemical shift (assignment), in the order of assignments. The spectral coupling patterns are: s -singlet, d -doublet, t -triplet, q -quartet and m -multiplet. First-order analysis is carried out to assign signals of the 1 H NMR spectra. Additional 2D [ 1 H, 1 H] COSY NMR experiments were done where confirmation of the proton assignments were required. Assigning the carbon chemical shifts, obtained from proton-decoupled 13 C NMR spectra, was possible with the assistance of 2D [ 1 H, 13 C] HSQC and 2D [ 1 H, 13 C] HMBC NMR experiments. Standard Bruker pulse programs were used in the experiments.

FT-IR spectroscopy
Infrared spectroscopy was performed on a Bruker ALPHA FT-IR spectrophotometer with a NaCl cell, using dried hexane as solvent. Insoluble samples were measured using dried DCM as solvent. The absorptions were measured from 400 to 40 0 0 cm −1 . The IR data are reported in the format: absorption intensity (assignment) in the order of highest to lowest wavenumber. The wave intensities are: vw -very weak, w -weak, m -medium, s -strong, vs -very strong and sh -shoulder.

High resolution mass spectrometry
Mass spectral analyses were performed on a Waters® Synapt G2 high definition mass spectrometer (HDMS) that consists of a Waters Acquity Ultra Performance Liquid Chromatography (UPLC®) system hyphenated to a quadrupole-time-of-flight (QTOF) instrument. Data acquisition and processing was carried out with Mass-LynxTM (version 4.1) software. A leucine encephalin solution (2 pg/ μL, m/z 555.2693) was used as an internal lock mass control standard to compensate for instrumental drift and ensure good mass accuracy. The internal control was directly infused into the source through a secondary orthogonal electrospray ionization (ESI) probe to allowing intermittent sampling. Flow injection analysis (FIA, 0.4 mL/min flow rate) with the injection volume set at 5 μL. Samples were made up in ultra purity liquid chromatography methanol to an approximate concentration of 10 μg/mL. The methanol was spiked with 0.1% formic acid and used throughout the 1 min run. The capillary voltage for the ESI source was set at 2.6 kV for negative mode ionization. The source temperature was set at 110 °C, the sampling cone voltage at 25 V, extraction cone voltage at 4.0 V and cone gas (nitrogen) flow at 10.0 L/Hr. The desolvation temperature was set at 300 °C with a gas (nitrogen) flow of 600.0 L/Hr. The mass to charge ratios ( m/z ) were measured in the range of 50-1500 Da with the raw data presented in the form of a centroid profile (scans were collected every 0.3 s). Negative electron spray was chosen as the ionization technique. The MS data are reported in the format: calculated mass, found mass (percentage intensity, fragmentation) in the order of highest to lowest mass to charge ratio.

Single crystal X-ray diffraction
Single crystal diffraction data for 1, 5, 7, 10 and 13 were collected at 150 K on a Bruker D8 Venture diffractometer with a kappa geometry goniometer and a Photon 100 CMOS detector using a Mo-K α I μS micro focus source. Data were reduced and scaled using SAINT and absorption intensity corrections were performed using SADABS (APEX III control software) [34] . Single crystals of 11 and 12 were analysed on a Rigaku XtaLAB Synergy R diffractometer, with a rotating-anode X-ray source and a HyPix CCD detector. Data reduction and absorption were carried out using the CrysAl-isPro (version 1.171.40.23a) software package [35] . X-ray diffraction measurements were performed at 150 K, using an Oxford Cryogenics Cryostat. All structures were solved by an intrinsic phasing algorithm using SHELXTS [36] and were refined by full-matrix least-squares methods based on F 2 using SHELXL [37] . All nonhydrogen atoms were refined anisotropically. All hydrogen atoms were placed in idealized positions and refined using riding models.

UV-Vis spectroscopy
Measurements were performed on 10 mL DCM solutions of 0.01 mM analyte concentration at 25 °C. Absorptions were measured in the range 20 0-10 0 0 nm using a UV-Vis spectrophotometer Specord 200 plus. WinASPECT PLUS (version 4.2) software was used for data visualization.

Synthesis of complexes
Method 1: Lithiation of [2,3-b ;3 ʹ,2 ʹd ]-DTT using n -BuLi n -BuLi (1.6 M in hexane, 2.1 mL, 3.3 mmol) was added to [2,3b ;3 ˈ,2 ˈd ]-DTT ( P1 ) (0.36 g, 1.8 mmol) dissolved in 20 mL of THF at −78 °C. After 30 min, Cr(CO) 6 (0.59 g, 2.7 mmol) was added to the reaction mixture. The solution stirred for 15 min at −78 °C, followed by 40 min at room temperature. The solvent was reduced in vacuo . The residue was dissolved in 10 mL DCM and cooled to −30 °C before being treated with [Et 3 O][BF 4 ] in DCM (1.33 g, 7.0 mmol). The reaction mixture was allowed to reach room temperature after which the solvent was reduced in vacuo . The products were purified using column chromatography starting with eluent n -hexane and increasing the polarity with DCM. The products isolated, in sequence of elution, are listed in Table  S1, Supplementary data . 20 mL of THF at -78 °C. After 30 min at -78 °C the solution was allowed to reach room temperature for 1 hour. The solvents were reduced in vacuo . The residue was dissolved in 10 mL DCM and cooled to -30 °C, then treated with [Et 3 O][BF 4 ] in DCM (1.33 g, 7 mmol). The reaction mixture was allowed to reach room temperature after which the solvent was reduced in vacuo . Column chromatographic purification of the products was carried out using nhexane and DCM as eluents. The products isolated, in sequence of elution, are listed in Table S2, Supplementary data .