Benchmarking break-junction techniques: electric and thermoelectric characterization of naphthalenophanes

Break-junction techniques provide the possibility to study electric and thermoelectric properties of single-molecule junctions in great detail. These techniques rely on the same principle of controllably breaking metallic contacts in order to create single-molecule junctions, whilst keeping track of the junction's conductance. Here, we compare results from mechanically controllable break junction (MCBJ) and scanning tunneling microscope (STM) methods, while characterizing conductance properties of the same novel mechanosensitive para- and meta-connected naphtalenophane compounds. In addition, thermopower measurements are carried out for both compounds using the STM break junction (STM-BJ) technique. For the conductance experiments, the same data processing using a clustering analysis is performed. We obtain to a large extent similar results for both methods, although values of conductance and stretching lengths for the STM-BJ technique are slightly larger in comparison with the MCBJ. STM-BJ thermopower experiments show similar Seebeck coefficients for both compounds. An increase in the Seebeck coefficient is revealed, whilst the conductance decreases, after which it saturates at around 10 μV K−1. This phenomenon is studied theoretically using a tight binding model. It shows that changes of molecule-electrode electronic couplings combined with shifts of the resonance energies explain the correlated behavior of conductance and Seebeck coefficient.

to the used solvent and coupling constants (J) are given in Hertz (Hz).The multiplicities are written as: s = singlet, d = doublet, t = triplet, dd = doublet of doublet, m = multiplet.Gas chromatography (GC-MS) was performed on a Shimadzu GC-MS-QP2010 SE gas chromatograph system, with a ZB-5HT inferno column (30 m × 0.25 mm × 0.25 mm), at 1 mL/min He-flow rate (split = 20:1) with a Shimadzu mass detector (EI 70 eV).Flash column chromatography (FCC) was performed with SiliaFlash® P60 from SILICYCLE with a particle size of 40-63 µm (230-400 mesh), and for TLC silica gel 60 F254 glass plates with a thickness of 0.25 mm from Merck were used.The detection was carried out with a UV-lamp at 254 or 366 nm.Gel permeation chromatography (GPC) was performed on a Shimadzu Prominence System with PSS SDV preparative columns from PSS (2 columns in series: 600 mm × 20.0 mm, 5 µm particles, linear porosity "S", operating ranges: 100-100.000g mol -1 ) using chloroform as solvent.For HPLC a Shimadzu LC-20AD and a LC-20AT HPLC were used, equipped with a diode array UV/Vis detector (SPD-M10A VP from Shimadzu, λ = 200-600 nm) and a column oven Shimadzu CTO-20AC for analytical measurements.The used column was a Reprosil 100 C18, 5 µm, 250 × 16 mm; Dr. Maisch GmbH.
For preparative HPLC a Shimadzu LC-20Ap was used, equipped with a diode array UV/Vis detector (SPD-20A from Shimadzu, λ = 200-600 nm).The used column was a Reprosil 100 C18, 10 µm, 250 × 30 mm; Dr. Maisch GmbH.High-resolution mass spectra (HRMS) were measured as HR-ESI-ToF-MS with a Maxis 4G instrument from Bruker or as HR-EI-MS spectrometry with a DFS double-focusing (BE geometry) magnetic sector mass spectrometer (ThermoFisher Scientific, Bremen, Germany).Mass spectra were measured with electron ionization (EI) at 70 eV, solid probe inlet, a source temperature of 200 °C, an acceleration voltage of 5 kV, and a resolution of 10'000.The instrument was scanned between e.g.m/z 300 und 350 at scan rate of 100-200 s/decade in the electric scan mode.Perfluorokerosene (PFK, Fluorochem, Derbyshire, UK) served for calibration.
Synthetic steps to the target structures: a) Overview: Synthesis of the anti- [2.2](1,4)naphthalenophanes, exposing a pair of acetyl protected thiol anchor groups, para-NP (1) and meta-NP (2).Reagents and conditions: a) 3 eq.NBS, 10 mol%  Analytic data for 4: Compound 5 was synthesized following a modification of a literature known procedure 2 : The dibromide compound 3 (1520 mg, 4.84 mmol, 1.0 eq) in THF (40 mL) was added dropwise slowly over 30 minutes to a 0.1 M solution of SmI 2 (5 g in 120 mL, 12.4 mmol, 2.5 eq.) stirred at room temperature under argon.The mixture was then stirred at that temperature for 16 hours, during which time the initial blue color became tinged slightly with green.Subsequently the reaction was slowly quenched with ice, extracted with DCM and the solvent was removed under vacuum.After flash column chromatography (SiO 2 , cyclohexane:ethylacetate 3:1) compound 5 (664 mg, 2.153 mmol, 89%) was yielded as a white solid.Note: Small amounts of impurities with similar polarity as the target structure were removed by washing with few mL of chloroform.
The crude sample of substituted arylboronic acids and K 2 CO 3 was added to a 20 mL roundbottomed flask equipped with a magnetic stirring bar.The tube was evacuated twice and backfilled with nitrogen.Acetonitrile (2 mL) and iodine (43 mg, 171 mmol) were added to the tube at room temperature under a stream of nitrogen, and the tube was sealed and put into a pre-heated oil bath at 80 °C for 4 h under nitrogen atmosphere.After the resulting solution was cooled to room temperature, Na 2 S 2 O 3 aq.(10 mL) was added to the resulting mixture, and then the aqueous layer was extracted with EtOAc (3 × 5 mL).The combined organic phase was dried over anhydrous Na 2 SO 4 , filtered and concentrated by rotary evaporation.After flash column chromatography (SiO 2 , cyclohexane) the diiodinated compounds (8: 74 mg, 0.111 mmol, 65%; 9: 88 mg, 0.131 mmol, 77%) were isolated as a white solid.

S3. STM conductance measurements
Molecular compounds were deposited onto Au(111) samples using the drop casting technique.
Au samples were annealed at approximately 900 K for 1-2 minutes, allowed to cool down to room Breaking traces were acquired by recording the current while retracting the tip of the tip-sample junction.2,200 traces are obtained for each compound.They are aligned by setting them to zero displacement, when the monoatomic Au junction between the tip and the sample is broken at a conductance of .In this way, we ensure a normalized analysis of all traces.The same procedure 1 0 was followed for the MCBJ measurements.

S4. Apparent stretching length
For both break-junction techniques the apparent stretching length ( ) of the molecular plateaus was obtained in the same way by fitting a Gaussian distribution to each conductance peak and determining the length differences of every trace between the Au-Au monoatomic breakage point ( ) and , where and are the most probable conductance value and the standard deviation of the Gaussian fitting curves, respectively.These end points are represented in Fig. S3a as crosses on top of the individual breaking traces.To take into account multiple junction configurations, Gaussian distributions were fitted to one-dimensional displacement histograms, constructed from the trace length ( ) of individual breaking traces.The most probable value,

𝐿 𝑡𝑟𝑎𝑐𝑒
, was then determined as the maximum of the distribution (as shown in Fig. S3b).

S7. Low-conductance plateau
One-dimensional conductance histograms of both compounds reveal a small conductance peak at lower values in addition to the main conductance maximum, indicating another possible stable configuration of the molecules inside the junctions.To learn about the origin, we applied the clustering technique of section S5 in order to create an ensemble that includes the traces with this low-conductance plateau.To achieve this, we changed the parameters for the traces included in   We find that 11%-15% of the molecular traces show a lower conductance plateau for the MCBJ and 20%-31% for the STM measurements.The lower conductance plateau is always accompanied by that preceding at a higher (main) conductance, as can be seen in the single trace examples of Fig. S5. Figure S5e shows an example of the two obtained classes from sub-clustering the molecular class, displayed in Fig. 3c and measured with the STM.

S8. STM-BJ Seebeck coefficient measurements
To perform Seebeck coefficient measurements a home-built STM was used, capable of measuring simultaneously the conductance (G) and the thermovoltage ( ) of the formed  ℎ molecular junctions.The tip was heated using a 1 kΩ surface resistor, creating a temperature difference ( ) between the tip and the sample, with the tip being at and the sample at . This temperature difference not only generates a thermovoltage, , in the molecular junction but also in the copper lead that connects the tip to the rest of the setup.Considering these factors the thermoelectric equation of the circuit can be expressed as , where and are the Seebeck coefficients of the molecular junction and the copper lead,    respectively.Figure S6a shows a scheme of the equivalent circuit of the STM.

II. Theoretical S9. Four-site model
In addition to the theoretical analysis presented in Fig. 6 of the main text, we discuss here the four-site tight-binding model for the metal-molecule-metal junction with asymmetric couplings to left and right electrodes.This allows us to inspect the robustness of the model.Furthermore, we study the influence of variations in interdeck hopping .

𝑑
The four-site tight-binding model is depicted in Fig. S8a.As in Fig. 6, there are on-site energies , and hopping terms and .However, distinct couplings to left and right electrodes and are taken into account this time, which translate to the following Hamilton operator and linewidth broadening matrices: Γ  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ) , ̂Γ = ( 0 0 0 0 0 0 0 0 0 0 0 0 We evaluate the transport properties using the Landauer-Büttiker approach 4 within the wideband-limit approximation 5 .Since we do not consider any level shift inside the molecule from the coupling to the electrodes but only level broadenings, the self-energy matrices and (5) 3 = 1/2 ( ( 2 + 4 2 ) - ) ,  4 = 1/2 ( ( 2 + 4 2 ) +  ) The parameter represents the intradeck hopping within naphthalene units, while parametrizes   the interdeck hopping, which is typically smaller 10 .The HOMO-LUMO gap is then given by , and the splitting of nearly degenerate HOMO and LUMO level pairs on the naphthalene units is determined by .We use these relations to fit appropriate values for and from electronic structure calculations.For this purpose we calculate the energies of the frontier orbitals of an isolated para-NP structure with SH anchors attached, using density functional theory (DFT) as implemented in the TURBOMOLE program suite 11 .In the DFT calculations we employ the def-SV(P) Gaussian basis set 12 for all atoms and the PBE exchange-correlation functional 13 .
Afterwards the energies of the frontier orbitals are corrected using a calculation 14 .Since the  0  0 HOMO levels for the equilibrium geometry are nearly degenerate whereas the LUMO levels exhibit a larger splitting, we fix the sulfur atoms and stretch the molecule along the sulfur-sulfur axis by .Subsequently, a constrained geometry optimization is performed.In this way the close and electrode 15-19 .In contrast to the analysis in Fig. 6 of the main text, we will thus keep one coupling fixed and lower the other coupling to simulate an asymmetric situation.

7
At low temperatures the conductance simplifies to , and the thermopower becomes . Accordingly, the expression for the Lorentzian model reads isolated yields: 69% of 1 and 84% of 2g, 12.8 mmol, 1.0 eq) was dissolved in dry dichloromethane (80 mL) and degassed with argon.Under an active argon stream, N-bromosuccinimide (6.904 g, 38.4 mmol, 3.0 eq) and benzoyl peroxide (413 mg, 1.28 mmol, 10 mol%) were added and the suspension was degassed to give a yellow suspension.The reaction mixture was heated under nitrogen at 55 °C for 18 hours.The reaction mixture was cooled to room temperature and then washed with 2 M HCl (2 x 15 mL), 2 M NaOH (2 x 20 mL), brine, and dried with MgSO 4 .The solvent was evaporated, and the crude was purified by flash column chromatography (SiO 2 , cyclohexane:ethyl acetate 6:1) to yield the desired product in 78% yield (3.124 g, 9.948 mmol).
Figure S2: a) Conductance measurements of para-NP.Left and central panels show the 2D conductance vs. electrode displacement histograms, containing all the raw data and molecular data, respectively, obtained from MCBJ experiments.The raw data consists of 10,000 consecutive traces.The right panel shows the corresponding 1D histograms: raw data (black) and molecular data (red).Dashed lines are the Gaussian fits of the respective conductance peaks for raw and molecular data.b) Same as a) but for meta-NP.

Fig.
Fig. S2b a faint blue region is visible for displacement values longer than 1 nm and values temperature and then introduced into a 1 mM dichloromethane (DCM) solution of the corresponding molecule.After 20 minutes samples were dried off with nitrogen gas to eliminate possible molecular clusters on the surface.Mechanically cut Au wires (0.25 mm diameter, 99.99% purity, Goodfellow) were used as STM tips.A bias voltage was applied to the sample, using .The tunnelling current was amplified using a double-stage, home-made, linear   = 100  current-voltage (IV) converter with an overall gain of ( in the first stage, 2.5 × 10 10 / 5 × 10 8 / multiplied by a factor of in the second one).The Au tip is welded to a homemade printed circuit 50 board (PCB) tip holder and bolted under a piezoelectric tube.With this system we can move the STM tip over the substrate vertically and horizontally with a resolution of .All electronic 10 -20  signals are generated and read using a field programmable gate array (FPGA) with a range.± 10  In the case of the piezoelectric tube signals, the FPGA output signals are amplified by a factor of 14 using a high voltage amplifier.

Figure
Figure S3: a) Left panel: 1D conductance histogram of the para-NP compound measured with the MCBJ technique.Dashed lines are the Gaussian fits to each peak, black for the high-plateau   and green for the low-plateau.Right panel: individual breaking traces.Grey dashed lines on top  of both panels represent the value, where all the lengths are acquired.Black and green   + ∆ crosses on top of the breaking traces represent the apparent stretching length, , of high-and    low-plateaus.Note that the breaking traces are offset horizontally for better visibility.b)  Histograms compiling the lengths of traces ( ) for high-and low-plateaus from panel a)    (solid lines).Peaks are fitted with Gaussian functions (dashed lines and filled areas), and crosses on top of the Gaussian fits represent the value for each plateau.

1. 5
With this technique traces with or without a molecular plateau were separated, and traces with different conductance plateaus could be divided into different clusters for each set of measurements.To obtain the cluster with a plateau at low , limits for the input traces are changed  to[  ,  ]  for both the 2D and 1D histogram.10 -  0 10 -6.2  0S6.High-conductance plateauIndividual breaking traces, containing only the high-conductance plateau, are displayed in Fig.S4for both molecules.They stem from the molecular classes for both the para-and meta-NP, as presented in Fig.3c and 3d.Both the MCBJ and STM methods give rise to similarly shaped single traces.Comparing the plateaus amongst the molecules, the plateaus in the single breaking traces of the para-NP contain more conductance variation than the plateaus of the meta-NP.

Figure S4: a )
Figure S4: a) Examples of single breaking traces for para-NP, obtained from datasets as displayed in Fig. 3c and 3d, for both the MCBJ (red) and STM-BJ (blue) measurement methods.b) Same as a) but for meta-NP.

𝐺
the clustering analysis to[  ,  ]  for both histogram inputs.In Fig.S5we show the 2D 10 -5  0 10 -6.2  0 vs. displacement and 1D histograms containing all the traces with lower conductance plateaus   for both compounds and methods.The most probable conductances ( ) and apparent stretching   lengths ( ) of both plateaus are summarized in Table S2, representing the main conductance  plateau as 'High G' and the low conductance plateau as 'Low G'.The classes were obtained by sub-clustering the molecular class in two classes, with molecular classes as shown in Fig. S2a and S2b for MCBJ and Fig. 3c and 3d for STM.

Figure
Figure S5: a) Conductance measurements of para-NP.Left and central panels show the 2D conductance vs. displacement histograms of the molecular traces with two conductance plateaus, obtained using the non-supervised clustering technique applied to MCBJ and STM results, respectively.The right panel shows the corresponding 1D histograms of MCBJ (red) and STM (blue) measurements, displaying two peaks in the conductance counts which correspond to the respective plateaus in the 2D histograms.Dashed lines are Gaussian fits to the respective conductance peaks for MCBJ and STM.b) Single trace examples of MCBJ and STM techniques with the red and blue lines obtained from the shown data sets for the para-NP in a), respectively, which are offset for visibility.c) Same as a) but for meta-NP.d) Same as b) but for meta-NP.e) Example of the two sub-classes obtained from sub-clustering the STM molecular class in Fig. 3c.Note that the right panel is identical to the central panel shown in a).

Figure
Figure S6: a) Scheme of the thermoelectric circuit of the STM, where is the bias voltage   applied, and are the Seebeck coefficients of the molecular junction and the copper lead,    respectively, is the conductance of the molecular junction, and is the temperature difference  Δ between the tip (at ) and the sample ( ). b) (Top) Tip displacement and (bottom)  ℎ >     =    signals, respectively, during a thermovoltage measurement.While the molecular junction is   formed, the tip displacement is momentarily stopped and V bias is ramped between ±10 mV.c) Examples of current-voltage traces with temperature difference (red) and without temperature difference (blue), where is obtained from the zero current crossing point and from the slope  ℎ

2 Å
lying energies of the HOMO-1 and HOMO states are separated, and fitting yields the parameters and . = 2.7   = 0.6  Let us now analyze how the asymmetrically coupled four-site tight-binding model explains the experimentally observed behavior of decreased conductance and simultaneously growing thermopower during the stretching.As before, we study first a decrease in the coupling strength or due to the stretching, which can be justified by a reduced electronic overlap of molecule Γ  Γ Figs. 6b and S8b against the corresponding Fermi energy.Variations of are expected to be Fig. 5), indicating that a change in does not explain the global behavior of an increasing for a

Figure S9 :
Figure S9: Four-site tight-binding model with on-site energies and , intradeck hopping    = 1,…,4 terms and interdeck hopping , and a symmetric coupling to left and right electrodes.Model   Γ results are obtained for for all , , and a varying interdeck hopping .For   = 0   = 2.7  Γ = 0.75   evaluations of the conductance and thermopower we assume a temperature of .a)    = 300  Transmission plotted as a function of energy (lower x-axis) and thermopower as a function of () the respective Fermi energy (upper x-axis) for different parameters , indicated in the legend.b)  and plotted against interdeck hopping .The behavior is depicted for the four different Fermi    energies, which are marked by vertical dashed orange lines in panel a).c) plotted as a function

References………………………………………………...………………...….46 I. Experimental S1. Synthesis General Procedures:
All commercially available chemicals were used without further purification.Dry solvents were used as crown cap and purchased from Acros Organics and Sigma-Aldrich.NMR solvents were obtained from Cambridge Isotope Laboratories, Inc. (Andover, MA, USA) or Sigma-Aldrich.All NMR experiments were performed on Bruker Avance III or III HD, two-or four-channel NMR spectrometers operating at 400.13, 500.13 or 600.27MHz proton frequency.The instruments were equipped with direct observe BBFO, indirect BBI or cryogenic four-channel QCI (H/C/N/F) 5 mm probes, all with self-shielded z-gradient.The experiments were performed at 298 K or 295 K.All chemical shifts (δ) are reported in parts per million (ppm) relative

Table S1 :
Most probable conductance ( ) and apparent stretching length ( ), considering all the     molecular traces for all compounds in additional data sets (middle panels of Fig.S2a and S2b).Uncertainties, obtained from the standard deviations, are indicated for each value.

Table S2 :
Most probable conductance ( ) and apparent stretching length ( ) of the main high-High G") and low-conductance plateau ("Low G") for the molecular traces showing both conductance plateaus for para-NP and meta-NP, as measured with MCBJ and STM methods.Uncertainties, obtained from the standard deviation, are provided for each value.
(  )  =  0 (  )  the same time.This can clearly be seen from the relationship between and for a given , (  ) Γ