advances.sciencemag.org/cgi/content/full/3/6/e1602297/DC1

Supplementary Materials for

Stable anchoring chemistry for room temperature charge transport

through graphite-molecule contacts

Alexander V. Rudnev, Veerabhadrarao Kaliginedi, Andrea Droghetti, Hiroaki Ozawa, Akiyoshi Kuzume,

Masa-aki Haga, Peter Broekmann, Ivan Rungger

Published 9 June 2017, Sci. Adv. 3, e1602297 (2017)

DOI: 10.1126/sciadv.1602297

This PDF file includes:

section S1. Synthesis

section S2. Grafting and characterization of modified HOPG

section S3. Conductance measurements—Additional data

section S4. Conductance calculations—Additional data and discussion

scheme S1. Synthetic route of 3,5-dimethyl-4-nitrobenzenediazonium

tetrafluoroborate salt.

fig. S1. 1H NMR spectrum of 1-amino-3,5-dimethyl-4-nitrobenzene (500 MHz,

CDCl3).

fig. S2. 1H NMR spectrum of 3,5-dimethyl-4-nitrobenzenediazonium

tetrafluoroborate salt (500 MHz, DMSO-d6).

fig. S3. Grafting on HOPG.

fig. S4. Characterization of modified HOPG surface.

fig. S5. Dendrimeric structures.

fig. S6. Blank experiment in argon.

fig. S7. Blank experiment in 1,2,4-trichlorobenzene.

fig. S8. Individual conductance traces.

fig. S9. Geometries of grafted DMAB molecules.

fig. S10. Transmission of different DMAB structures.

fig. S11. Transmission of AB, DMAB, and PPD.

fig. S12. Structures of PPD junctions.

References (48–55)

section S1. Synthesis

Synthesis of 3,5-dimethyl-4-nitrobenzenediazonium tetrafluoroborate salt was performed

according to Scheme S1.

scheme S1. Synthetic route of 3,5-dimethyl-4-nitrobenzenediazonium tetrafluoroborate

salt.

1-iode-3,5-dimethyl-4-nitrobenzene was synthesized from 1-iode-3,5-dimethylbenzene

according to previous reports (48).

1-amino-3,5-dimethyl-4-nitrobenzene:

1-iode-3,5-dimethyl-4-nitrobenzene (1 g, 3.6 mmol), CuI (120 mg, 0.53 mmol), K2CO3 (1.66 g,

12 mmol), trans-4-hydroxy-L-proline (200 mg, 1.5 mmol) were added in 10 mL of DMSO. 4 ml

of aqueous ammonia (28%) was added in the mixture solution and refluxed for 12 h. The

solution color changed to green during the addition of aqueous ammonia. CH2Cl2 was added to

the reaction solution and the mixture was washed with water twice. The organic layer was dried

over Na2SO4 and the solvent removed under reduced pressure. The crude product was purified

by silica-gel column chromatography using 100% CH2Cl2 as eluent, affording 1-amino-3,5-

dimethyl-4-nitrobenzene as yellow solid (206 mg, 34 %).

1H NMR (500 MHz, CDCl3) = 6.34 (s, 2H, benzene ring), 2.29 (s, 6H, CH3). FT-IR (ATR)

3422, 3354, 3331, 3217, 2963, 2924, 2851, 1641, 1593, 1477, 1466, 1429, 1379, 1288, 1030,

853, 841, 760 cm-1.

3,5-dimethyl-4-nitrobenzenediazonium tetrafluoroborate salt:

To a solution of the 1-amino-3,5-dimethyl-4-nitrobenzene (81 mg, 0.49 mmol) in 5 mL of

CH2Cl2 at 0 C was added BF3∙OEt (100 mg, 0.7 mmol) followed by t-butyl nitrite (100 l, 0.84

mmol). The resulting reaction mixture was allowed to warm to room temperature and the

precipitate appeared in the solution and was filtered. The crude product was washed with CH2Cl2

and dried under vacuum to give pale yellow solid (78 mg, 60%).

1H NMR (500 MHz, DMSO-d6) = 8.07 (s, 2H, benzene ring), 2.43 (s, 6H, CH3). FT-IR (ATR)

3100, 2307, 2293, 1537, 1458, 1371, 1306, 1109, 1084, 1015, 885, 824, 795 cm-1.

fig. S1. 1H NMR spectrum of 1-amino-3,5-dimethyl-4-nitrobenzene (500 MHz, CDCl3).

fig. S2. 1H NMR spectrum of 3,5-dimethyl-4-nitrobenzenediazonium tetrafluoroborate salt

(500 MHz, DMSO-d6).

section S2. Grafting and characterization of modified HOPG

Grafting of the molecules onto HOPG

The 4-aminobenzene (AB) and 3,5-dimethyl-4-aminobenzene (DMAB) molecules were

covalently attached to the HOPG surface by electrochemical reduction. In particular, the graphite

surface modification was performed in two steps (fig. S3) with the experimental protocol that

was identical for AB and DMAB. First, the molecules were grafted onto HOPG by applying the

potential sweep from positive to negative potentials in a cyclic voltammetry experiment (49).

The electrochemical reduction led to the cleavage of diazonium group as molecular nitrogen and

to the formation of an aryl-radical. The aryl-radical is active enough to break the sp2-

hybridization of the HOPG top layer and form a covalent carbon-carbon bond (50, 51). Figure

S3a shows the cyclic voltammograms (CVs) of HOPG in 30 mM of respective diazonium salts in

acetonitrile solution. The initial potential was 0.41 V and no electrochemical reaction occurred.

Then the potential was swept to -0.05 V and back. The first cycles showed irreversible peaks

around 0.1 V. This feature was previously assigned to the reductive cleavage of diazonium-

groups and the formation of the corresponding aryl radicals (49). The consecutive cycles were

nearly featureless, which means that after the first cycles the HOPG surface became blocked by

grafted and physisorbed reduction products. After 5 cycles in the assembly solution, the cell with

the HOPG electrode was thoroughly rinsed with acetonitrile, tetrahydrofuran, dichloromethane,

again acetonitrile, and finally with water to remove any physisorbed material. Subsequently, the

nitro group of the grafted molecules was electrochemically reduced to amine in water-ethanol

electrolyte solution according to the reported protocol (38). CVs are shown in fig. S3b. The

cathodic peaks at

~-1.0 V correspond to the reduction of nitro-group to amine (38). The reduction was completed

in the first cycle.

fig. S3. Grafting on HOPG. The schematics of HOPG surface modification by AB and DMAB

in the consecutive electrochemical reactions. Corresponding CVs for AB and DMAB: (a)

grafting by reductive cleavage of diazonium group in 30 mM 4-nitrobenzenediazonium

tetrafluoroborate (or 3,5-dimethyl-4-nitrobenzenediazonium tetrafluoroborate) + 0.1 M TBA-PF6

in AcN and (b) reduction of nitro-group to amine in 0.1 M KCl in H2O + EtOH (9:1 v/v). Scan

rate is 50 mVs-1.

Characterization of the modified HOPG surface

In order to confirm the covalent grafting of the AB and DMAB molecules to the HOPG surface

and to detect the formation of sp3 nodes on the perfect sp2 HOPG lattice, we employed ex-situ

Raman spectroscopy (fig. S4a). In fact, the presence of sp3-defects in sp2-hybridized carbon

systems manifests itself through the characteristic D-mode at ~1350 cm-1 (51, 52). Consistently,

for HOPG modified with both AB and DMAB we observed the D band at 1354 cm-1

accompanied by the typical bands for a smooth HOPG surface (G at 1582 and 2D at 2717 cm-1)

(fig. S4a). In contrast, the Raman spectra obtained for a bare HOPG substrate displayed no D

band. This observation confirms the presence of sp3-defects caused by the covalent attachment of

the molecules.

fig. S4. Characterization of modified HOPG surface. (a) Representative Raman spectra of

(a0) bare, (a1) AB- and (a2) DMAB-modified HOPG electrodes. (b) Representative STM

images, (c) cross-sections along the white line in the image and (d) cluster height distributions of

HOPG modified with (b1-d1) AB and (b2-d2) DMAB. Imaging conditions: Vbias = 0.1 V, ISP =

25 pA. The histograms in d were obtained from height analysis of different 120 bright spots.

We further monitored the surface modification by using atomic-force (AFM) and scanning

tunnelling (STM) microscopies. According to our experience and the published reports (50, 53),

AFM imaging does not allow to distinguish physisorbed and covalently-attached molecules on

the HOPG substrate. In contrast, during STM imaging, physisorbed molecules are easily dragged

across the HOPG by an STM tip, while covalently-attached ones remain unaffected (50, 53).

Figure S4b shows representative images of AB/HOPG and DMAB/HOPG, which display HOPG

terraces covered with bright spots. These spots were stable during the multiple scanning of the

same area and have rather uniform width of a few nanometres (2-4 nm, fig. S4c). These spots are

attributed to clusters of a few molecules (in agreement with Ref.(50)) covalently attached at

adjacent sites of the topmost HOPG layer. On the one hand, for DMAB/HOPG, the apparent

height of such clusters is 0.65 ± 0.08 nm (fig. S4d), which correlates with the length of the

grafted DMAB molecules (~0.8 nm as calculated in Ref. (54)). On the other hand, the apparent

height of AB/HOPG clusters is 0.94 ± 0.24 nm, which is significantly larger and much less

uniform than for AB/HOPG. Gauss fitting gives a 3-fold larger standard deviation in case of AB.

These findings indicate the formation of dendrimers during AB assembly. It is well-known that

the aryl-radical can react with another molecules in meta-positions (respective diazonium group),

thus leading to a dendrimeric oligomer growth (39). The larger apparent height and the wider

distribution of the cluster heights indicate the formation of different dendrimeric structures. A

pair of examples are presented in fig. S5. In contrast, the narrow height distribution and the mean

apparent height of 0.65 nm indicate that the dendrimeric growth is absent for DMAB. These

results are in agreement with previous works (50).

fig. S5. Dendrimeric structures. Possible dendrimeric oligomer structures obtained during AB

assembly.

section S3. Conductance measurements—Additional data

fig. S6. Blank experiment in argon. (a) 1D and (b) 2D conductance histograms constructed

from ≈1000 individual STM-BJ traces recorded at 100 mV with an Au tip and an HOPG

substrate in an Ar atmosphere.

fig. S7. Blank experiment in 1,2,4-trichlorobenzene. (a) 1D and (b) 2D conductance

histograms constructed from ≈1000 individual STM-BJ traces recorded at 100 mV with a Au tip

and an HOPG substrate in 1,2,4-trichlorobenzene.

fig. S8. Individual conductance traces. Withdrawing individual conductance traces for (a) a

junction with a probable dendrimeric structure, which can be obtained upon the AB grafting; (b)

a junction with DMAB; (c) a non-covalent junction with p-phenylenediamine (PPD).

section S4. Conductance calculations—Additional data and discussion

Geometries of the 2-molecule system for all inter-molecule distances

For the total energy calculations as function of distance between two DMAB molecules we used

a single layer of HOPG (a graphene sheet) in the supercell shown in fig. S9. Adding more layers

is not expected to affect the rather large changes of binding energy as function of distance. For

two molecules attached to graphene atoms in the same sublattice we found a supercell magnetic

moment equal to 2 owing to the presence of two unsaturated dangling bonds (see the main

text). In contrast, when two molecules are attached to graphene atoms belonging to different

sublattices (e.g. nearest neighbour atomic sites), the electrons from the dangling bonds pair up so

that the supercell magnetic moment vanishes.

Figure S9a shows the relative positions for 1st to 5th nearest neighbour distances. The magenta

circle indicates the position of the first molecule, while the green circles (black squares) indicate

the positions of the 2nd molecule for binding on the different (same) sublattice to the one of the

first molecule. The numbers inside the circles and squares correspond to the number of C-C

bonds between the two molecules. The considered supercell is rectangular and extends over 16

(20) C-C bonds along the zigzag (armchair) HOPG edge.

fig. S9. Geometries of grafted DMAB molecules. Unit cells used in the total energy

calculations as function of distance of two DMAB molecules. (a) top view with the molecules at

5h nearest neighbour distance, (b) side view for 1st nearest neighbour separation, (c) side view

for 5th nearest neighbour separation.

Scissor operator

We calculated the PPD LDA HOMO Kohn-Sham state to be at about -3.8 eV below the vacuum

energy, compared to an experimental ionization potential of about 6.8 eV (32). Therefore, a gas

phase correction of -3 eV is required for the LDA HOMO energy. However, the proximity of the

HOPG and Au electrodes to the molecule induces an increase of the HOMO energy due to the

so-called image-charge effect, which reduces the applied correction compared to the gas phase

(32, 40-42). This can be captured in the calculations by applying an image charge correction (32,

40-42), which is inversely proportional to the distance between the centre of the molecule and

the image plane of the electrodes. For both Au and graphene this image plane is set to be 1 Å

above the surface (32, 41, 42, 55). We note that small shifts of the image plane only lead to

minor quantitative differences. Similar arguments apply to the LUMO state, so that we apply a

correction of equal magnitude but opposite sign to the LUMO. For DMAB we use the same

parameters as for the PPD.

Zero bias transmission

Transmission functions for DMAB and PPD are presented in fig. S10 and S11. In fig. S11 the

transmission is plotted over a larger energy range. The shift of the energy of the HOMO peak to

lower values for increasing angles in the PPD molecule is due to the increase in the scissor

operator correction to the HOMO state, which itself is caused by the increasing effective

molecule-electrode separation for larger angles (40, 42). For DMAB the molecule has a nearly

vertical alignment and therefore a largest shift. The additional features in the transmission for

DMAB when compared to the one of PPD are due to the modification of the DOS of the surface

HOPG layer caused by the covalently bonded molecule.

fig. S10. Transmission of different DMAB structures. Transmission as function of energy for

the five shown structures, which have different angles between the DMAB and the HOPG (for

S1-MH the angle from the plane is 89°, while for the remaining structures it is about 75°), and

different angles between Au and N, as well as passivation with a second molecule and a H atom.

fig. S11. Transmission of AB, DMAB, and PPD. Transmission, T, as function of energy for

DMAB and an AB molecule with a second benzene attached (longer dendrimer) (a), and PPD at

three different angles (b). The considered geometries are the same as the ones used for Fig. 2 in

the main manuscript.

We evaluate the changes in transmission when replacing the passivating H-atom with a second

DMAB molecule. Here we first relaxed the unit cell for the H-passivated system for a fixed

position of the Au tip and substrate (fig. S10a), which corresponds to the structure shown in the

main text in Fig. 1a. The 2-molecule system is then first relaxed without Au, and the Au tip is

subsequently added separately to each of the molecules (fig. S10e-f), keeping the vector

separating the N atom and the closest Au tip atom the same as for the H-passivated system (fig.

S10a). The Au-N distance is therefore kept fixed, while the angle between the Au-N bond and

the axis of the molecule, as well as the angle between the axis of the molecule and the HOPG

substrate changes significantly for the three considered structures. Furthermore, we compare the

effect of passivation with either a second molecule or an H atom by replacing the molecule not

attached to the tip with an H atom and without performing any further geometry relaxation (fig.

S10b-c). The resulting transmission curves (fig. S10d) are very similar for all 5 structures,

showing that the results are robust for both changes in the contact angles, as well as for changes

in the nature of the passivating molecule and atom.

Figure S12 shows the scattering region for the three considered different angles of the PPD

molecule.

fig. S12. Structures of PPD junctions. Structures used for the transmission calculations for PPD

presented in Fig. 2g of the main manuscript, for the PPD molecule at different angles from the

plane of the HOPG: (a) 3°, (b) 12°, (c) 21°.