Single-Molecule Conductance Studies of Organometallic Complexes Bearing 3-Thienyl Contacting Groups

Article abstract of DOI:10.1002/chem.201604565

The compounds and complexes 1,4-C₆H₄(C≡C-cyclo-3-C₄H₃S)₂(2), trans-[Pt(C≡C-cyclo-3-C₄H₃S)₂(PEt₃)₂] (3), trans-[Ru(C≡C-cyclo-3-C₄H₃S)₂(dppe)₂] (4; dppe=1,2-bis(diphenylphosphino)ethane) and trans-[Ru(C≡C-cyclo-3-C₄H₃S)₂{P(OEt)₃}₄] (5) featuring the 3-thienyl moiety as a surface contacting group for gold electrodes have been prepared, crystallographically characterised in the case of 3–5 and studied in metal|molecule|metal junctions by using both scanning tunnelling microscope break-junction (STM-BJ) and STM-I(s) methods (measuring the tunnelling current (I) as a function of distance (s)). The compounds exhibit similar conductance profiles, with a low conductance feature being more readily identified by STM-I(s) methods, and a higher feature by the STM-BJ method. The lower conductance feature was further characterised by analysis using an unsupervised, automated multi-parameter vector classification (MPVC) of the conductance traces. The combination of similarly structured HOMOs and non-resonant tunnelling mechanism accounts for the remarkably similar conductance values across the chemically distinct members of the family 2–5.

Full text of DOI:10.1002/chem.201604565

A Journal of
Accepted Article
Title: Single-Molecule Conductance Studies of Organometallic
Complexes Bearing 3-Thienyl Contacting Groups
Authors: Sören Bock, Oday Al-Owaedi, Samantha Eaves, David
Milan, Mario Lemmer, Brian Skelton, Henrry Osorio, Richard
Nichols, Simon Higgins, Pilar Cea, Nicholas Long, Tim
Albrecht, Santiago Martín, Colin Lambert, and Paul J. Low
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201604565
Link to VoR: http://dx.doi.org/10.1002/chem.201604565
Supported by

10.1002/chem.201604565
Chemistry - A European Journal
Single-Molecule Conductance Studies of Organometallic
Complexes Bearing 3-Thienyl Contacting Groups
Sören Bock,a Oday A. Al-Owaedi,b,c Samantha G. Eaves,a,d David C. Milan,e
Mario Lemmer,f Brian W. Skelton,a,g Henrry M. Osorio,h,i,j Richard J. Nichols,e
Simon J. Higgins,e Pilar Cea,h,i Nicholas J. Long,f Tim Albrecht,f Santiago Martính,k*
Colin J. Lambert,b* Paul J. Lowa*
a School of Chemistry and Biochemistry, University of Western Australia, 35 Stirling
Highway, Crawley, 6009, WA, Australia
b Department of Physics, Lancaster University, Lancaster, LA1 4YB, UK
c Department of Laser Physics, Women faculty of science, Babylon University, Iraq.
d Department of Chemistry, Durham University, South Rd, Durham, DH1 3LE, UK
e Department of Chemistry, University of Liverpool, Crown St, Liverpool, L69 7ZD,
UK
f Department of Chemistry, Imperial College London, London SW7 2AZ, UK
g Centre for Microscopy, Characterisation and Analysis, University of Western
Australia, Crawley, Western Australia 6009, Australia
h Departamento de Química Física, Facultad de Ciencias, Universidad de Zaragoza,
50009, Spain
i Instituto de Nanociencia de Aragón (INA) y Laboratorio de Microscopias Avanzadas
(LMA), Edificio I+D Campus Rio Ebro, Universidad de Zaragoza, C/Mariano
Esquillor, s/n, 50018 Zaragoza, Spain
j Departamento de Física, Escuela Politécnica Nacional, Av. Ladrón de Guevara, E11-
253, 170525 Quito, Ecuador.
k Instituto de Ciencias de Materiales de Aragón (ICMA), Universidad de Zaragoza-
CSIC, 50009 Zaragoza, Spain.
Email: S.M. smartins@unizar.es, C.J.L. c.lambert@lancaster.ac.uk, P.J.L.
paul.low@uwa.edu.au
1
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
ABSTRACT
The compounds and complexes 1,4-C6H4(C≡C-cyclo-3-C4H3S)2 (2), trans-Pt(C≡C-
cyclo-3-C4H3S)2(PEt3)2 (3), trans-Ru(C≡C-cyclo-3-C4H3S)2(dppe)2 (4) and trans-
Ru(C≡C-cyclo-3-C4H3S)2{P(OEt)3}4 (5) featuring the 3-thienyl moiety as a surface
contacting group for gold electrodes have been prepared, crystallographically
characterised in the case of 3 – 5, and studied in metal|molecule|metal junctions using
both STM-I(s) and STM-BJ methods. The compounds exhibit similar conductance
profiles, with a low conductance feature being more readily identified by STM-I(s)
methods, and a higher feature by the STM-BJ method. The lower conductance feature
was further characterised by analysis using an unsupervised, automated multi-
parameter vector classification (MPVC) of the conductance traces. The combination
of similarly structured HOMOs and non-resonant tunnelling mechanism accounts for
the remarkably similar conductance values across the chemically distinct members of
the family 2- 5.
INTRODUCTION
The development of a range of complementary and relatively facile methods for the
measurement of the electrical properties of single molecules has seen a renaissance in
the field of molecular electronics.1-4 The continued progress of the area from
fundamental science towards technology now rests on a number of key issues,5 among
which are the reliable contacting of molecules within a junction,6,7 the reduction in
electronic variation between individual junctions,1,8-12 and the optimisation of the
transport properties of these junctions.13-16 To these ends, considerable effort is being
made to explore the effects of the electrode-molecule contact groups and structure of
2
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
the contact,17-20 the potential applications of non-metallic electrodes to create an ‘all-
carbon’ molecular electronic device platform,21 as well as the backbone structure of
the molecular component on the electrical properties of the junction.22-24
Although the majority of single-molecule and thin-film junctions studied to date have
been based on organic molecules, such as alkanes,25,26 oligo(arylene)ethynylenes27-29
and polyynes,30-33 metal complexes have also been recognised as potential
components in a future molecular electronics technology.34-38 Metal complexes offer
a range of potential advantages over structurally and electronically simpler organic
molecules, including redox activity and a wider range of readily accessible and
systematically variable spin-states and magnetic properties,39-42 diversity of molecular
structure and potential for modular construction through in situ or ‘on surface’
coordination chemistry,43-46 better alignment of the frontier molecular orbitals with
the Fermi level of the (usually metallic) junction electrodes,8,47-50 as well as high
thermoelectric efficiency.51
In some earlier studies of organometallic complexes in molecular electronics, the
complex trans-Pt(C≡CC6H4SAc)2(PPh3)2 was assembled within a mechanically
controlled break-junction (MCBJ) based molecular junction. From the resulting I/V
curves, collected over a bias range of ±5 V, a resistance of 5 – 50 GΩ (i.e. G = 0.2 –
0.02 nS; 20 - 2 x 10-6 G0) was estimated at the extremes of the bias range, some three
orders of magnitude less conductive than similarly–contacted organic oligoarylene
systems.52 This ‘insulating’ behavior, even under such a high applied bias, was
ascribed to the largely σ-type Pt-C(sp) bonds in the C≡C-Pt-C≡C backbone, although
it is clear that at this bias voltage, the conductance mechanism is likely to be field
3
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
emission rather than tunneling.53 In contrast, a later study with a family of complexes
trans-Pt(C≡CC6H4SAc)2(L)2 (L = PCy3, PPh3, P(OEt)3) in crossed-wire junctions at
more modest bias (up to 1 V) revealed a 2 – 3 fold higher conductance than 1,4-(4-
AcSC6H4C≡C)2C6H4, which was ascribed to the shorter sulfur-sulfur distance in the
metal complexes.54
In seeking to enhance the wire-like response, significant attention was turned to
ruthenium bis(alkynyl) complexes, which are generally thought to offer more
significant d-π orbital mixing in the occupied frontier molecular orbitals.9,34,50,55-57
The thioacetate complex trans-Ru(C≡CC6H4SAc-4)2(dppm)2 has been assembled into
monolayers and studied within an STM-break junction, with comparison made to the
oligo(phenyleneethynylene) (OPE) compound 1,4-(4-AcSC6H4C≡C)2C6H4 as a
benchmark. Extrapolation to single-molecule conductances gave values of 19 ± 7 nS
(ca. 2.5 x 10-4 G0) for the ruthenium complex and 3.6 ± 2.0 nS (ca. 4.6 x 10-5 G0) for
the OPE. The higher conductance has been attributed to both the shorter molecular
length and the extensive Ru(d)-C≡C(π) mixing in the metal complex.56 These
concepts have been extended to other examples of organometallic wires based on
group 8 metal centers and the trans-bis(alkynyl) motif, with topics of interest
including the exploration of surface contacting groups,9,50 the inclusion of multiple
metal centers along the molecular back-bone,35,47,57 and electronic function beyond
that of a simple wire, such as charge storage and gated transistor-like response.41,58
One particular advantage of organometallic complexes within the field of molecular
electronics lies in the ability to systematically alter the molecular structures of these
systems with a fair degree of synthetic ease, which permits a modular approach to
4
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
molecular designs and a systematic search for structure-property relationships. In
seeking to further explore the electrical properties of oligophenyleneethynylene
(OPE), and trans-bis(alkynyl) complexes of platinum and ruthenium, we have turned
to such a systematic study here. Here, the 3-thienyl moiety,59-61 which is readily
introduced into both organic and organometallic structures, is used as a contacting
group for the ready attachment of organic, ruthenium and platinum based
organometallic complexes within Au|molecule|Au junctions and electrical
characterisation using both the I(s)62 and STM-BJ63 methods. The conductance results
are interpreted with the aid of DFT level calculations and junction simulations.
RESULTS AND DISCUSSION
Synthesis and characterisation. The parent alkyne 3-ethynyl thiophene (HC≡C-cyclo-
3-C4H3S, 1) was obtained from the Sonogashira cross-coupling of 3-bromothiophene
with trimethylsilylacetylene, and subsequent deprotection.59 Further cross-coupling of
1 with 1,4-diiodobenzene gave 1,4-C6H4(C≡C-cyclo-3-C4H3S)2 (2) (Scheme 1).59 The
metal complex trans-Pt(C≡C-cyclo-3-C4H3S)2(PEt3)2 (3) was prepared from the
Cu(I)-catalysed reaction of 1 and PtCl2(PEt3)2 in NEt3,64 whilst trans-Ru(C≡C-cyclo-
3-C4H3S)2(dppe)2 (4) was obtained from the one-pot reaction of [RuCl(dppe)2]OTf
with 1, in the presence of either KOBut or DBU and TlBF4.65 Complexes of general
form trans-Ru(C≡CR)2{P(OEt3)3}4 have been prepared in moderate yield from
reaction of trans-RuCl2{P(OEt)3}4 with an excess of LiC≡CR.66 The complex trans-
Ru(C≡C-cyclo-3-C4H3S)2{P(OEt)3}4 (5) was obtained here simply by allowing the
reaction of trans-RuCl2{P(OEt)3}4 67 with 1, KPF6 and NHPri2 in ethanol to proceed
for 12 days at r.t., with isolation of the desired compound being achieved by
precipitation from methanol. The long reaction time was compensated by the simple
5
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
reaction conditions and work-up, compatibility with alkynes substituted with sensitive
functional groups and improved yields. The compounds were each characterised by
the usual array of 1H, 13C{1H} and, in the case of 2 - 5, 31P{1H} NMR spectroscopies,
mass spectrometry and elemental analysis.
Scheme 1. The preparation of compounds 2 - 5.
Molecular structures
Single crystals of 3, 4 and 5 suitable for X-ray diffraction were obtained by
recrystallisation by slow diffusion of hexanes (3, 4) or EtOH (5) into CH2Cl2
solutions of the complexes. Plots of the molecules showing the atom labelling
schemes are given in Figures 1 - 3 and important bond lengths and angles are
summarised in the relevant Figure caption.
6
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Figure 1. A plot of a molecule of trans-Pt(C≡C-cyclo-3-C4H3S)2(PEt3)2(3) (ellipsoids
drawn at 50% probability level) showing the atom labelling scheme. Bond lengths
(Å): Pt(1)-P(1) 2.3022(5); Pt(1)-C(1) 2.007(2); C(1)-C(2) 1.185(3); C(2)-C(21)
1.452(3); C(21)-C(22) 1.373(3); C(22)-S(23) 1.705(3); S(23)-C(24) 1.682(3); C(24)-
C(25) 1.393(3); C(21)-C(25) 1.428(3). Bond angles (°): P(1)-Pt(1)-C(1) 88.32(6),
91.68(6); Pt(1)-C(1)-C(2) 177.83(19); C(1)-C(2)-C(21) 177.3(2).
In the crystal, 3 is situated on a crystallographic inversion center. The platinum atom
shows the expected square-planar geometry with trans-disposed alkynyl and
phosphine ligands. The Pt(1)-P(1) (2.3022(5) Å), Pt(1)-C(1) (2.007(2) Å) and C(1)-
C(2) (1.185(3) Å) are all within the usual range of values for complexes of this type,68
and display little variation of significance with the electronic character of the alkynyl
substitutent. In the crystal, the P(1)-Pt(1)-C(21)-C(22) torsion angle (136.6, –43.4 °)
prevents any significant extended conjugation through the molecule.
7
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Figure 2. A plot of one molecule of trans-Ru(C≡C-cyclo-3-C4H3S)2(dppe)2(4)
(ellipsoids drawn at 50% probability level) showing the atom labelling scheme. Bond
lengths (molecule 1, Å): Ru(1)-P(1) 2.3539(5); Ru(1)-P(2) 2.3602(5); Ru(1)-C(11)
2.0611(19); C(11)-C(12) 1.218(3); C(12)-C(13) 1.433(3); C(13)-C(14) 1.373(3);
C(14)-S(15) 1.709(3); S(15)-C(16) 1.709(3); C(16)-C(17) 1.364(3); C(13)-C(17)
1.448(3). Bond angles (molecule 1, °): P(1)-Ru(1)-P(2) 82.90(2), 97.10(2); P(1)-
Ru(1)-C(11) 94.38(5), 85.62; P(2)-Ru(1)-C(11) 97.28(5), 82.72(5); Ru(1)-C(11)-
C(12) 176.2(2); C(11)-C(12)-C(13) 176.1(2). Bond lengths (molecule 2, Å): Ru(2)-
P(3) 2.3464(4); Ru(2)-P(4) 2.3586(4); Ru(2)-C(21) 2.0626(18); C(21)-C(22)
1.217(3); C(22)-C(23) 1.431(3); C(23)-C(24) 1.389(10); C(24)-S(25) 1.715(10);
S(25)-C(26) 1.673(8); C(26)-C(27) 1.352(12); C(23)-C(27) 1.411(10). Bond angles
(molecule 2, °): P(3)-Ru(2)-P(4) 97.86(2), 82.14(2); P(3)-Ru(2)-C(21) 96.61(5),
83.39(5); P(4)-Ru(1)-C(21) 84.93(5), 95.07(5); Ru(2)-C(21)-C(22) 177.3(2); C(21)-
C(22)-C(23) 176.0(2).
8
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
The general structural features of trans-Ru(C≡CR)2(dppe)2 complexes have been
summarised recently,65 and compound 4 offers some points worthy of brief comment.
The structure is composed of two independent molecules, both of which are situated
on crystallographic inversion centres and which differ in the orientation of the
thiophene group. The dihedral angles between the thiophene plane and the plane
containing the Ru and C(n1) atoms and the midpoints of the two ligand P atoms are
3.4° for molecule 1, and 96.3 and 77.8° for the two components of the disordered
thiophene of molecule 2. Therefore, at least in the crystal, molecules 1 and 2 of 4 are
rare examples of such bis(arylacetylide) complexes trans-Ru(C≡CR)2(dppe)2 in which
the aromatic rings sit close to the ‘privileged’ orientations that allow maximum d-π
conjugation along the molecular backbone, and the first in which both conformational
isomers are observed for the same chemical compound. The comparable bond lengths
within the two molecules are essentially indistinguishable, with the possible exception
of the most precisely determined Ru-P bond lengths, which appear to be marginally
shorter in molecule 2.
9
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Figure 3. A plot of a molecule (ellipsoids drawn at 30% probability level) of trans-
Ru(C≡C-cyclo-3-C4H3S)2{P(OEt)3}4 (5) showing the atom labelling scheme. Bond
lengths (Å): P(1)-Ru(1) 2.3149(3); Ru(1)-C(1) 2.0592(15); C(1)-C(2) 1.221(2); C(2)-
C(21) 1.421(2); C(21)-C(22) 1.36(2); C(22)-S(23) 1.599(12); S(23)-C(24) 1.741(7);
C(24)-C(25) 1.480(11); C(25)-C(21) 1.45(2). Bond angles (°) P(1)-Ru(1)-P(1)
90.338(2), 171.19(2); P(1)-Ru(1)-C(1) 94.40(1), 85.60(1); Ru(1)-C(1)-C(2) 180;
C(1)-C(2)-C(21) 180.
Although
the
mixed
ligand
vinylidene-acetylide
has been
complex
[Ru(C≡CPh){C=C(Me)Ph}{P(OEt)3}4][CF3SO3]
structurally
characterised,66 compound 5 appears to be the first structurally characterised
bis(acetylide) derivative Ru(C≡CR)2{P(OR)3}4. In the crystal, the
tetrakis(triethylphosphite) derivative 5 is situated on a crystallographic 4 axis so that
there is only one unique phosphite group. The dihedral angle between the two
thiophene groups (which are disordered about the crystallographic 2- fold axis) is
therefore 90°. The Ru-P (2.3149(3) Å) and Ru-C(1) (2.0592(15) Å) distances in 5 are
10
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
shorter than in the mixed vinylidene-acetylide derivative (Ru-P 2.341(3) - 2.350(2) Å;
Ru-C 2.114(8) Å) reflecting the increased electron density at Ru in 5 and increased
Ru-P and Ru-C back bonding. Although back-bonding plays only a modest role in the
bonding of metal-alkynyl complexes,69 the notion is also supported by the trends in
C≡C
bonding
lengths
in
5
(1.221(2)
Å)
and
the
cation
[Ru(C≡CPh){C=C(Me)Ph}{P(OEt)3}4]+ (1.209(11) Å). Overall, with the obvious
exception of the phosphine and phosphite ligands, there are few if any substantive
differences in the structures of 4 and 5.
Single Molecule conductance - STM-BJ and I(s). Single molecule conductance
measurements were carried out using substrates with a low surface coverage of
molecules of interest on gold substrates. Low surface coverage was chosen to
minimise the formation of multi-molecule junctions and promote formation of single
molecule events. Adsorption of 2 - 5 at low surface coverage was achieved by
immersion of a gold-on-glass substrate in CHCl3 solutions of the analyte (1 mM) for
about 80 s. After adsorption, the samples were washed in ethanol and then blown dry
in a stream of nitrogen. All in situ I(s) and STM-BJ measurements were conducted in
mesitylene, a non-polar solvent commonly used in STM-based single molecule
electrical measurements because of its high boiling point and relatively low vapor
pressure. For a given set-point current and bias voltage, typically 6,000–7,000 events
were observed in both STM-BJ and I(s) experiments.
Taking compound 2 as a representative example, without any data selection, it is
rather difficult to assign a conductance value to the data (Figure 4). Possible reasons
for this include a low junction formation probability and short plateau features. The
11
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
one dimensional (1D) conductance histogram of the whole data set shows only a faint
shoulder, i.e. a conductance peak partially obscured by the exponential background
(Figure 4, left). Matching the peak, a faint plateau feature can be seen in the 2D
conductance histogram (Figure 4, right), but there is a need for data selection for this
system to increase signal to noise ratio of the data.
Figure 4: All data 1D (left) and log 2D (right) conductance histograms composed
from I(s) data. Note the 2D histogram is plotted against units of 10-4 G0.
Data selection can be made manually, using a rational criterion, for example by
selecting traces with a current plateau that exceeds 0.1 nm in length, and disregarding
those without. However, since manual data selection can never be fully objective, it is
of interest to compare the results against an automated data selection approach. Here,
the unsupervised, automated multi-parameter vector classification, MPVC, has been
adopted to verify the conclusions reached from the manually sorted data.70 By way of
example, the data for molecule 2 are analysed in more detail in the following
paragraphs, which compare the results of manual and automated data selection
methods.
12
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
For the MPVC, an exponentially-decaying current-distance trace was created as
reference vector, R (I0 = 30 nA, β = 0.5 Å-1). Three vector properties (classifiers) were
then calculated for each I(s) trace with respect to the reference:
1) ΔX - The length of the distance vector, Y, between reference and I(s) trace;
2) θ - The angle, between R and –Y;
3) hr, the reduced Hamming distance, with hr being the number of component changes
to transform the reduced distance vector, Y, into the reduced reference vector, R. In
this context, ‘reduced’ means that every vector element is divided by its absolute
value, so that the resulting vector consists only of 1, 0 and -1.
The whole data set, consisting of 3838 I(s) traces is thus transformed into 3838
vectors in 3-dimensional space (see cylinder plots in the SI). Here, similar traces, e.g.
traces with plateaus, are in close proximity to each other. Plain exponential and
plateau-containing traces are expected to form distinct clusters in this representation
and fuzzy c-means clustering (FCM) was then used to assign the cluster
membership.71,72 Note that the total number of clusters k was selected to be two in this
case, to account for plain exponential decays and molecular events, but can be chosen
to be a higher number if any cluster consists of sub-clusters (e.g. to account for a
variety of different junction geometries).
13
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Figure 5: 1D (left) and log 2D (right) conductance histograms of cluster 1, containing
predominantly I(s) traces with plateaus as found by MPVC. These are assigned to
‘molecular events’ (see main text).
Figure 6: 1D (left) and log 2D (right) conductance histograms of cluster 2, containing
predominantly plain exponential traces without plateaus as found by MPVC. These
are assigned to I(s) traces without molecular junction formation.
During FCM, 217 (5.65 %) I(s) traces were assigned to cluster 1, containing
predominantly exponential decays with plateaus ('molecular events', Figure 5).
Cluster 2 contains the remaining, predominantly plain exponential traces (3621, 94.35
14
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
%) (Figure 6). By manual data selection 459 (11.96 %) were selected as plateau
containing (Figure 7).
Figure 7: 1D (left) and log 2D (right) conductance histograms generated by manual
selection of data. Note that the 2D histogram is plotted against units of 10-4 G0.
The results of the unsupervised algorithm approach shows excellent agreement with
the data selected on the basis of the 0.1 nm plateau length criterion described above in
terms of conductances, but with understandable differences in terms of number of
selected traces. With respect to the latter, some 81 (37.3 %) of the plateau traces in
cluster 1 were also marked as plateau-containing during the hand-sorting process.
Also, 17.6 % of the manually-selected traces were found by the clustering algorithm.
The traces found both by MPVC and hand selection are predominantly long plateaus
around the most probable conductance value. In addition, 135 traces were included by
the MPVC algorithm but not by hand sorting. These traces contained plateaus at
various conductance values or unconventional features, meaning deviations from the
plain exponential decay other than plateaus. These can possibly originate from
different molecular processes during junction formation (or rupture) or noise features.
In contrast, some 377 traces were only marked as plateau-containing during hand
15
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
selection and not by MPVC. Mostly, those were traces with very short plateau
features, or longer plateaus in exponential traces with large decay coefficients. Such
features can arise from changes in the molecular junction geometry during the tip
retraction process.
After MPVC analysis cluster 1 exhibits a conductance peak around 0.41×10-4 G0
(Figure 5). Hand sorting gives a most probable conductance of 0.42×10-4 G0 (Figure
7). This indicates that although there are differences in the curve selection between
hand sorting and automated sorting, the most probable conductance of both data
selection methods shows excellent agreement.
While the non-contact I(s) technique favors low conductance groups,1 the STM-BJ
method generally leads to a greater propensity of higher conductance values. These
differences can be explained in terms of the way in which the junctions are formed in
both methods. In the I(s) method, the (typically gold) STM tip is brought into close
proximity of the surface to encourage molecular junction formation, but without any
initial contact between the STM tip and substrate. In contrast, in the STM-BJ
technique, the STM tip is fused (or crashed) into the substrate and withdrawn to give
a metallic filament between the tip and the substrate. Molecular junctions form
immediately after the Au-Au point contact breaks.1 The rough or fractal nature of
these cleaved gold contact junctions often leads to a variety of conductance features in
STM-BJ-based metal|molecule|metal junctions formed from common anchoring
groups such as thiol,1,73 carboxylic acid,74 or pyridine75 where in each case more than
one single molecule conductance value has been reported, and attributed to differing
contact morphologies between the contacting groups and the gold electrode(s).
16
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Data recorded using the STM-BJ technique for compound 2 at Utip = 0.6 V are
summarized in Figure 8. As shown, the conductance profile from these STM-BJ data
for 2 shows only one conductance group (labeled H). Its most probable conductance
((2.83±0.65) × 10-4 G0, Table 1) is in good agreement to that reported by van der Zant
et al. using the mechanically controlled break-junction (MCBJ) technique (4x10-4
G0).59
17
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Figure 8. Conductance histograms built by adding all conductance traces (ca. 550)
that showed discernible plateaus (with a current plateau which exceeds 0.1 nm in
length) as those displayed in the inset of the figures using either the STM-BJ (left
side) or the I(s) method (right side). H = high conductance group. L = low
conductance group. (a) compound 2, (b) compound 3, (c) compound 4 and (d)
18
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
compound 5. Conductance data are referenced to the conductance quantum G0 = 2e2/h
= 77.5 µS. Utip = 0.6 V.
Table 1. Single molecule conductance data for compounds 2-5.
Low
High
Break-off
calculated
conductance
(L)a / G0
conductance
(H)b / G0
distance / nm (crystallographic) S...S
distance / nm
2
3
4
5
(0.42 ±
(2.83 ±
1.60
1.70 ± 0.28
1.85 ± 0.24
1.90 ± 0.24
1.90 ± 0.24
0.10)×10-4
0.65)×10-4
(0.77 ±
(2.70 ±
1.43 (1.42)
1.45 (1.44)
1.45 (1.43)
0.14)×10-4
0.66)×10-4
(1.03 ±
(3.18 ±
0.28)×10-4
1.04)×10-4
(0.98 ±
(3.12 ±
0.14)×10-4
0.58)×10-4
a I(s) method. b STM-BJ method
As shown in Figure 8 and Table 1, distinct conductance groups were also obtained for
the metal complexes 3 - 5 using the I(s) (L-group) and the STM-BJ (H-group)
method. Interestingly, the compounds 2 – 5 conductance values differ by a factor of
about two for the L group, while this factor is lower for the H group (Table 1). This
demonstrates that the central moiety [C6H4 vs Pt(PEt3)2 vs Ru(dppe)2 vs
Ru{P(OEt)3}4] does not exert a great influence on the conductance of these
organometallic complexes within these 3-thienyl contacted Au|molecule|Au junctions.
19
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Quantum Chemical Modeling
In the quest to better understand the conductance behavior, the electronic properties of
the molecules and electrical behavior of the junctions have been investigated using
DFT-based methods. Initial studies of the electronic structures of 2 – 5 were carried
out at B3LYP level of theory76 with LANL2DZ basis set used for metal atoms (Ru,
Pt)77 and 6-31G** 78 (all other atoms) basis set, to explore the influence of the central
fragment (C6H4 (2), Pt(PEt3)2 (3), Ru(dppe)2 (4), Ru{P(OEt)3}4 (5)) on the
distribution and composition of the frontier molecular orbitals. Plots of the HOMOs
are given in Figure 9, and plots of the LUMOs are given in the Supporting
Information.
The organic compound 2 again provides a convenient point to commence discussion
and a basis for comparison of the metal complexes 3 – 5. Unsurprisingly, the lowest
energy structure features a co-planar arrangement of the thienyl and 1,4-phenylene
rings, with the frontier orbitals distributed almost evenly across the molecular
backbone making a linear, π-type conjugated pathway between the two sulfur atoms.
For the platinum complex 3, the lowest energy identified minimum featured the
thienyl moieties lying perpendicular to the square plane defining the coordination
geometry at the metal centre (perp-3). However, a second minimum, barely 0.8 kcal /
mol higher in energy, in which the thienyl moieties lie in the same plane as the metal
coordination sphere (planar-3) was also identified. Again, the HOMOs of these
complexes are π-type and delocalized over the molecular backbone, and feature a
small but important metal contribution (perp-3, 10% Pt; planar-3, 19% Pt). The
20
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
LUMOs are more metal in character (perp-3, 42%; planar-3, 28%) and rather
delocalized in the case of planar-3.
The ruthenium complexes 4 and 5 offer HOMOs that are similarly structured to those
described for 2 and offer only marginally more metal character than planar-3 (4, 33%
Ru; 5, 24% Ru). The LUMO of 4 is largely of metal/dppe character, in the case of the
phosphite analogue 5 the LUMO is thienyl-π* in character, with the unoccupied metal
orbital lying slightly (ca. 0.04 eV) higher in energy.
(b)$
(a)$
(c)$
(d)$
(e)$
Figure 9. The iso-surfaces (±0.04 (e / bohr3)1/2) of the HOMOs for (a) 2 (b) perp-3 (c)
planar-3 (d) 4 (e) 5.
To provide further insight into the experimentally observed trends, and to better
evaluate the properties and behavior of these molecular junctions, calculations using a
combination of DFT (the SIESTA code)79 and a non-equilibrium Green’s function
formalism were also carried out. For the transport calculations, eight layers of (111)-
oriented bulk gold with each layer consisting of 6×6 atoms and a layer spacing of
21
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
0.235 nm were used to create the molecular junctions as shown in Figure 6, and
described in detail elsewhere.80 These layers were then further repeated to yield
infinitely-long current-carrying gold electrodes. Each molecule was attached to two
(111) directed pyramidal gold electrodes. The molecules and first layers of gold atoms
within each electrode were then allowed to relax again, to yield the optimal junction
geometries shown in Figure 10. From these model junctions the transmission
coefficient, T(E), was calculated using the GOLLUM code.80
Figure 10. Relaxed geometries of molecular junctions of 2-5.
22
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
A key factor governing the conductance of a molecular junction is the position of the
Fermi level of a metal electrode with respect to the molecular HOMO and LUMO
levels. In turn this energy alignment is sensitive to not only the chemical nature of the
contacting groups which binds the molecule to the electrode, but also the precise
configuration of the metal electrode-molecule contact.15,81 However, it is well-known
that the Fermi energy predicted by DFT (EFDFT) is often not reliable,33 and as such the
room temperature electrical conductance G was computed for a range of Fermi
DFT
energies EF. The calculated conductances G are plotted as functions of EF – EF
in
Figure 11, which reveal similar conductance values over a wide range of Fermi
energies, between -0.4 eV to +0.4 eV relative to the DFT-predicted value. The
predicted conductance values of all molecules were compared with the experimental
values and a single common value of EF was chosen, which gave the closest overall
agreement. This yielded a small correction of EF – EFDFT = - 0.075 eV, which has been
used in all of the theoretical results described below. Thygesen and colleagues have
discussed similar situations for C60-contacted molecular wires, and have shown that
critical molecular orbitals can become pinned close to the Fermi level due to partial
charge transfer, leading to good quantitative agreement between calculated and
experimentally determined conductance.82
The experimental data, now interpreted with the aid of Figure 11, indicates that in all
cases the Fermi level lies close to the centre of the HOMO-LUMO gap, but shifted
slightly towards the HOMO resonance, and therefore HOMO-mediated hole
tunnelling mechanism is anticipated in each case. 47-50,83,84 However, in contrast to the
studies of Wang,56 Rigaut57 and Mayor52 with organic, ruthenium and platinum
bis(alkynyl) compounds and complexes contacted into molecular junctions by thiolate
23
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
groups, the conductance values only differing by of factor of ~2 are obtained across
the thienyl-contacted series 2 – 5. This lack of variation occurs because although
HOMO and LUMO transport resonances differ significantly between molecules 2 - 5,
transport in the vicinity of the middle of the HOMO-LUMO gap is similar for all
molecules (Figure 11).
Figure 11. The calculated conductance as a function of the Fermi energy for 2-5.
Black dashed line shows the chosen Fermi energy (EF = - 0.075 eV).
24
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
Table 2. The HOMO and LUMO energies of the isolated molecules (eV, c.f. Figure
5), the total number of electrons of the isolated molecule (QI), the total number of
electrons of the molecule attached to the gold electrodes (QMG), the total number of
electrons transferred from the molecule (Г= QI – QMG) and the calculated and
experimental G/G0.
QMG
Molecule
EHOMO
ELUMO
QI
Г
Th. G/G0
Ex. G/G0
(L group)
0.45×10-4
0.44±0.1×10-4
2
-5.67
-4.78
-4.34
-4.50
-1.98
-0.71
-0.87
-0.08
94
93.83
0.17
0.225
0.304
0.328
163.775
planar-3a
164
350
322
0.77×10-4 0.77±0.14×10-4
1.06×10-4 1.03±0.28×10-4
0.99×10-4 0.98±0.14×10-4
4
5
349.696
321.672
a perp-[3] HOMO -4.96 eV, LUMO -0.39 eV
To further explore the reasons for small differences in conductance across the series,
the nature of the molecule-gold contact was also examined. Table 2 summarises the
molecule-gold interaction in terms of the number of valence electrons (QI) associated
with the molecule, the number calculated on the molecule in the junction (QMG) and
hence the number of electrons associated with the thienyl S!Au contacts (Γ) (or
‘bonds’) (based on calculated Mulliken charges). Mulliken charges are basis-set-
dependent mathematical constructions and therefore only approximately coincide
with the physical charge on a molecule. However, it is clear from the data in Table 2
25
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
that the value of Г, and the nature of the contact, is only weakly dependent on the
nature of the backbone and auxillary ligands in 2 – 5. Overall, the molecular
conductance of these molecules are similar, with minor variation arising through a
convolution of the strength of the S!Au bond, and the position of the tail of the
HOMO resonance relative to the Fermi level of the electrodes (Table 2).
CONCLUSION
The family of 3-thienylethynyl contacted compounds 1,4-C6H4(C≡C-cyclo-3-C4H3S)2
(2), trans-Pt(C≡C-cyclo-3-C4H3S)2(PEt3)2 (3), trans-Ru(C≡C-cyclo-3-C4H3S)2(dppe)2
(4) and trans-Ru(C≡C-cyclo-3-C4H3S)2{P(OEt)3}4 (5) have been prepared and studied
in metal|molecule|metal junctions using both STM-I(s) and STM-BJ methods. The
compounds 2 – 5 each display two conductance values, which differ by a factor of ~2
within the following range of conductance values: (0.44±0.10 – 1.03±0.28) × 10-4 G0
(low conductance group) and (2.70±0.66 – 3.18±1.04) × 10-4 G0 (high conductance
group). The MPVC method has been applied to verify the lowest conductance group
in an algorithmically definable fashion. For the 3-thienyl contact employed here, the
conductance values obtained using MPVC and manual data selection were very
similar, although there were some differences between the current-distance data sets
assigned by each method. The MPVC method, which allows reproducible and
objective analysis of conductance features close to the limit of the current amplifier, is
therefore a promising avenue for the further exploration of low conductance feature.
In addition, with an increase in the number of sub-clusters the method should also
prove useful in the analysis of a wider array of junction configurations or in cases
where the junction evolves over time or with distance. Further efforts to develop and
exploit the MPVC tool are now underway. A quantum chemical analysis of the
26
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
electronic structures of the isolated molecules reveals a similarly structured HOMO in
each case. Within model junctions, the Fermi level lies slightly towards the HOMO
resonance in each case, and a non-resonant hole tunnelling mechanism mediated by
the similarly structured HOMOs is proposed. The positioning of the Fermi level well
within the HOMO-LUMO gap is proposed to account for the similar conductance
behaviour across the series. Our study demonstrates that while for some systems,
platinum complexes may well be less conductive than purely organic analogues or
similarly–structured complexes of the group 8 metals, this is not a universal situation,
and by appropriate use of contacts and ancillary ligands to position key molecular
orbitals with respect to the Fermi levels of the electrodes, rather efficient molecular
wires may be engineered. For the future, it will be of interest to study thermal
transport through such wires, since although they have similar electrical properties,
their vibrational properties and phonon thermal conductances are likely to differ
significantly. This ability to tune the latter, while preserving electronic conductance is
an attractive proposition for the design of thermoelectric thin films.85
EXPERIMENTAL
General conditions All reactions were performed under an N2 atmosphere using
standard Schlenk techniques. Reaction solvents were purified and dried by
appropriate means prior to distillation and storage under nitrogen. No special
precautions were taken to exclude air or moisture during work-up. The compounds
HC≡C-cyclo-3-C4H3S (1),86 1,4-C6H4(C≡C-cyclo-3-C4H3S)2 (2),59 PtCl2(PEt3)2 (from
a 10 wt % solution of triethylphosphine in hexane),87 Pd(PPh4)4,88 [RuCl(dppe)2]OTf89
67
and trans-RuCl2{P(OEt)3}4 were prepared by literature methods, or minor
27
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
variations. All other materials were obtained from commercial suppliers and used as
received.
The NMR spectra were recorded on 400 MHz Bruker Avance, Bruker AV-500 or
1
13
Bruker AV-600 spectrometers. H and C{1H} spectra were referenced to residual
31
protic solvent signals, while P{1H} spectra were referenced to external phosphoric
acid. IR spectra were recorded using a Thermo Scientific Nicolet 6700 spectrometer
as CH2Cl2 solutions in a cell fitted with CaF2 windows. Electrospray mass spectra
were recorded on a Waters LCT Premier. Elemental analyses were performed by
London Metropolitan University, London, United Kingdom.
Synthesis of trans-Pt(C≡C-cyclo-3-C4H3S)2(PEt3)2 (3)
A Schlenk flask was charged with dry triethylamine (25 ml), PtCl2(PEt3)2 (150 mg;
0.30 mmol) 3-ethynylthiophene (1, 76 mg; 0.7 mmol) and CuI (14 mg; 0.08 mmol)
and the resulting mixture stirred at room temperature for 7 hours. After removing the
solvent under reduced pressure the residue was purified using column
chromatography (silica; CH2Cl2:hexanes 1:1). The first band was discarded, the
second band, containing some mono-substituted by-product was discarded as well and
the third yellow band, containing the desired product, was collected. After removing
the solvent the product was obtained as pale yellow powder (76 mg; 39 %). IR
(CH2Cl2) / cm–1: v(C≡C) 2107 (s). 1H-NMR (CDCl3, 500 MHz) δ / ppm: 1.18 – 1.26
(m, 18H); 2.12 – 2.21 (m,6H); 6.95 – 7.00 (m, 2H, H2); 7.03 – 7.10 (m, 2H, H4 or H5);
7.13 – 7.18 (m, 2H, H4 or H5). 13C-NMR (CDCl3, 126 MHz) δ / ppm: 8.52 (s, CH3);
16.55 (t, J = 17.6 Hz, CH2), 103.62 (br, Cα or Cβ), 106.32 (br, Cα or Cβ), 123.57 (s, C4
or C5), 124.07 (s, C4 or C5), 128.03 (s, C3), 130.30 (s, C2).31P-NMR (CDCl3,243
28
This article is protected by copyright. All rights reserved.

10.1002/chem.201604565
Chemistry - A European Journal
MHz) δ / ppm: 11.46 (s, JCPt = 1186 Hz). EI(+) –MS (m/z): 646 [M]+, 579
(Pt(PEt3)2(C≡C-thp) + MeCN]+. Anal. Found: C, 44.52; H, 5.71. C24H36P2PtS2 calc.:
C, 44.64; H, 5.62 %.
Synthesis of trans-Ru(C≡C-cyclo-3-C4H3S)2(dppe)2 (4)
Method A. An oven-dried Schlenk flask was charged with dry DCM (15 ml),
[RuCl(dppe)2]OTf (108 mg; 0.10 mmol), 3-ethynylthiophene (1, 80 mg; 0.74 mmol)
and potassium tert-butoxide (29 mg; 0.26 mmol). After stirring overnight at ambient
temperature the dark yellow mixture was filtered through an alumina plug (basic,
oven dried). After removing the solvent the crude product was obtained as dark
yellow solid, which was recrystallised by slow diffusion of hexane in a CH2Cl2-
solution (52 mg; 47 %).
Method B. An oven-dried Schlenk flask was charged with dry CH2Cl2 (5 ml), DBU (3
drops) and [RuCl(dppe)2]OTf (162 mg; 0.15 mmol). Then, 3-ethynylthiophene (1,
108 mg; 1.00 mmol) was added and the mixture was stirred for 5 minutes. To the
yellow solution TlBF4 (42 mg; 0.14 mmol) was added and the suspension was stirred
for 5 minutes. The yellow mixture was filtered through a alumina plug (basic, oven-
dried) and the solvent of the filtrate removed giving the crude product as yellow solid,
which was recrystallised by slow diffusion of hexane in a CH2Cl2-solution (68 mg; 41
%).
1
IR (CH2Cl2) / cm-1: v(C≡C) 2060 (s, sh), 2085 (m). H-NMR (CDCl3, 500 MHz) δ /
ppm: 2.55 – 2.65 (m, 8H, CH2), 6.50 (dd, J = 3.0, 0.8 Hz, 2H, H2 or H4); 6.61 (dd, J =
4.9; 0.8 Hz, 2 H, H2 or H4); 6.93 – 7.01 (m, 16H, Hmeta); 7.09 (dd, J = 4.9; 3.0 Hz, 2H,
29
This article is protected by copyright. All rights reserved.