Thi Mien Trung
Huynh‡
*ab,
Thanh Hai
Phan‡
ac,
Oleksandr
Ivasenko
a,
Stijn F. L.
Mertens
*ad and
Steven
De Feyter
*a
aKU Leuven-University of Leuven, Department of Chemistry, Division of Molecular Imaging and Photonics, Celestijnenlaan 200F, B-3001 Leuven, Belgium. E-mail: steven.defeyter@kuleuven.be; huynhthimientrung@qnu.edu.vn; stmerten@gmail.com
bQuy Nhon University, Department of Chemistry, 170 An Duong Vuong, Quy Nhon, Vietnam
cQuy Nhon University, Department of Physics, 170 An Duong Vuong, Quy Nhon, Vietnam
dInstitute of Applied Physics, Vienna University of Technology, Wiedner Hauptstraße 8-10/E134, A-1040 Vienna, Austria
First published on 7th December 2016
Highly oriented pyrolytic graphite (HOPG) can be covalently grafted with aryl radicals generated via the electrochemical reduction of 3,5-bis-tert-butyl-diazonium cations (3,5-TBD). The structure of the grafted layer and its stability under electrochemical conditions were assessed with electrochemical scanning tunneling microscopy (EC-STM) and cyclic voltammetry (CV). Stable within a wide (>2.5 V) electrochemical window, the grafted species can be locally removed using EC-STM-tip nanolithography. Using dibenzyl viologen as an example, we show that the generated nanocorrals of bare graphitic surface can be used to study nucleation and growth of self-assembled structures under conditions of nanoconfinement and electrochemical potential control.
Molecular functionalization of graphene addresses these challenges, and is a promising approach to widen the scope of its applications, such as in electronic devices, biosensors and composite materials.4 Both non-covalent and covalent functionalization protocols of graphitic surfaces have been explored extensively.5–17 The non-covalent approach relates to the physisorption of molecules, and in specific cases, organic molecules self-assemble into 2D crystalline films on top of the graphitic surface. Physisorption may alter the carrier concentration of graphene without affecting its materials properties.12,13,18 In contrast, covalent grafting results in band-gap opening near the Fermi level of graphene, turning pristine “metallic” graphene into a “semiconductor”.5,7–9,11,16
The most widely used procedure for covalent modification of graphitic surfaces involves the electroreduction of aryldiazonium salts.16,17,19–28 Electrons are transferred from the carbon surface to the diazonium cations at the solid–liquid interface. The release of a dinitrogen molecule results in a radical species that can attack the sp2 carbon lattice and be anchored there via covalent bond formation. Two major shortcomings of diazonium chemistry are the limited grafting density in combination with dendritic multilayer formation.22,23,26 As we recently demonstrated,16 both problems can be solved by grafting the sterically hindered 3,5-TBD, yielding high-density yet strict monolayer growth.16,19,20,24,26,27
For many applications, the stability of the grafted film is an important issue. The thermal and solvent stability of various grafted films on carbon-based and metallic surfaces has been reported.29–31 However, to the best of our knowledge, the in situ structural characterization at the nanoscale of grafted films and their stability under electrochemical control have not yet been reported.
We have recently developed a protocol for the creation of nanoconfined areas (nanocorrals) based on STM tip manipulation of these covalently modified graphitic surfaces, and provided proof-of-concept evidence for the self-assembly of a long-chain molecule, pentacontane, in those nanocorrals, at the interface between an organic solvent and the substrate.16 While in-depth studies are still in progress, we believe that confining the self-assembly of functional organic molecules within the corrals may lead to new structures and properties. It is, therefore, useful to explore and apply this concept to the self-assembly of electroactive organic molecules under electrochemical control.
As a representative example thereof, viologens have attracted much attention in the field of surface electrochemistry as they can be applied as chromophores, electron-transfer mediators and gating molecules.32–34 The reduced, uncharged species (V0), has been recently recognized as n-dopant for various carbon nanostructures including nanotubes35 and graphene11,36 as well as for other 2D materials such as MoS2.37 However, while reduced viologen has been synthesized for doping purposes, in situ electrochemical generation of viologen-doped 2D materials via the self-assembly approach has not been reported so far.
In this study, the diazonium grafting process on graphite and the stability of the grafted films as a function of electrode potential are investigated using cyclic voltammetry and EC-STM. Further, we show that it is possible, under electrochemical control, to locally degraft molecules via EC-STM tip based nanomanipulation. In the process, nanocorrals of pristine graphite surface are generated, in which reduced viologen molecules can self-assemble under electrochemical control. This experimental finding opens new avenues to investigate supramolecular self-assembly of n-/p-doping molecules in nanoconfined spaces under electrochemical control towards bottom-up creation of nanoconfined n-/p-doped 2D materials for nanoscale electronics applications.
All CV measurements were performed employing an Autolab PGSTAT101 potentiostat (Metrohm-Autolab BV, the Netherlands). Prior to each experiment, the HOPG electrode (ZYB grade, Momentive Performance Materials) was freshly cleaved using scotch tape. The electrochemical grafting of 3,5-TBD on HOPG was carried out in a lab-built single-compartment three-electrode cell, exposing a geometric working electrode area of 38.5 cm2. Pt wire and Ag/AgCl/3 M NaCl served as counter and reference electrodes, respectively. All potentials are reported versus the reversible hydrogen electrode (RHE). During the measurements, the electrolyte was kept under Ar. After grafting, the 3,5-TBD modified HOPG sample was rinsed with hot toluene and Milli-Q-water to remove physisorbed material from the surface, and finally dried with a stream of nitrogen before being transferred to the EC-STM cell.
All EC-STM experiments were carried out with an apparatus designed at the University of Bonn as described elsewhere;38 relevant potentials are defined in Fig. 1a. In order to eliminate the influence of oxygen as well as acoustic and electromagnetic interference, the entire EC-STM system is housed in a sealed aluminum chamber with electrical and liquid feedthroughs and filled with Ar. The STM tips were electrochemically etched from 0.25 mm tungsten wire in 2 M KOH solution, rinsed with water, dried and subsequently coated by passing the tip through a lamella of hot-melt glue.
The efficiency of the grafting process is characterized by EC-STM measurements. Fig. 1 shows large scale and high resolution EC-STM images of bare HOPG (Fig. 1b and c) and 3,5-TBD grafted HOPG (Fig. 1d and e). All images were recorded in 50 mM HCl. Since chloride anions hardly adsorb on HOPG,39 high-resolution STM images reveal the typical hexagonal structure of bare HOPG (Fig. 1c), whereas a high-density monolayer is observed after the electrochemical reduction and grafting of the 3,5-TBD species. This result is in line with STM observations under ambient conditions, as reported earlier.16 The presence of the tert-butyl groups at the meta-positions of the phenyl ring prevents the growth of multilayers, which otherwise readily form by the attachment of generated radicals onto those positions.16,27
Fig. 2 shows the CVs of bare HOPG (red curve) and 3,5-TBD grafted HOPG (black curve) in 50 mM HCl. In the double layer region (Fig. 2, inset), grafting decreases further the low interfacial capacitance of HOPG. The potential window of bare HOPG in HCl solution is limited by the oxygen evolution reaction (OER) at the anodic limit and the hydrogen evolution reaction (HER) at the cathodic limit. The presence of the grafted layer on the graphitic surface is seen to shift the onset of both OER and HER. Compared to bare HOPG, the HER on 3,5-TBD grafted HOPG surface starts 190 mV more negative than on bare HOPG whereas the OER onset is shifted positively by 240 mV. The change in onset of both OER and HER is constant upon multiple cycling (Fig. S2†). This observation may indicate the blocking effect of active sites on HOPG by 3,5-TBD for both reactions, or may find its origin in an increase of ohmic potential drop across the grafted layer. This effect was also reported for HOPG and metal electrode surfaces that were non-covalently functionalized by organic thin films.40,41
EC-STM measurements were used to verify the stability of the 3,5-TBD grafted layer on HOPG as a function of substrate potential. Fig. 3 shows a series of STM images that was recorded at the same grafted surface area while the applied potential was altered within the full potential window. The grafted layer of 3,5-TBD on HOPG remains intact at all potentials including OER and HER potential regimes, and no signs of 3,5-TBD desorption were observed. The strong covalent bond between grafted 3,5-TBD and the carbon lattice is thus impervious to the effects of cycling in the entire available potential window.
The electrochemical stability of the grafted layer is further demonstrated by the response of a hexacyanoferrate redox probe, which on freshly exfoliated HOPG shows reversible behaviour.42 Accordingly, CVs of 3,5-TBD/HOPG were measured in 1 mM K3Fe(CN)6 + 0.2 M Na2SO4 after polarization of the electrode for 3 min at −800 mV, 0 mV, +800 mV and +1400 mV in 50 mM HCl, Fig. 4. The presence of the non-conductive grafted layer blocks the electron transfer between HOPG electrode and the Fe(CN)63− species at the solid/liquid interface.43 No matter how far the potential excursion is extended, the electron transfer blocking behavior of the 3,5-TBD-modified HOPG remains intact.29
The combined EC-STM and CV results indicate that the covalently grafted 3,5-TBD layer on HOPG is highly stable against changes in the electrochemical potential. This finding opens the possibility of using the grafted layer as a robust mask to drive the selective self-assembly within nanocorrals of electroactive organic molecules under electrochemical control.
Fig. 5 illustrates the degrafting process at an HOPG–electrolyte interface by employing the EC-STM tip. The grafted layer is almost unaffected by the EC-STM tip under mild tunneling imaging conditions, i.e. high bias voltage and low tunneling current. However, grafted molecules are removed under more drastic tunneling conditions, i.e. low bias voltage and high tunneling current, leaving a corral free from grafted molecules behind as marked by the dashed square in Fig. 5a. The surface morphology (Fig. 5a) and high-resolution image (Fig. 5b) of the degrafted corral are similar to that of bare HOPG shown in Fig. 1b and c. This finding suggests that also EC-STM tip based degrafting converts the sp3 hybridization back to original sp2 pristine graphite lattice, which concurs with our previous study.16
Our experimental results demonstrate that also the EC-STM tip can act as a “nanoshaver” to locally remove covalently grafted molecules from HOPG. The as-formed degrafted areas are prone to functionalization under electrochemical control such as supramolecular self-assembly. The advantage over both UHV and ambient conditions is that electrochemical conditions allow strong control over adsorption/desorption processes and phase transition of molecules.
To illustrate this concept, we explored the self-assembly of dibenzyl viologen (DBV) in the nanocorrals. This molecule (Fig. 6a) is a redoxactive compound possessing three redox states, viz. dication (DBV2+), monocation (DBV+) and uncharged form (DBV0). The redox states of DBV can be reversibly converted by proper selection of the electrochemical potential (Fig. S3†).51,52
Fig. 6 shows STM images recorded in 0.1 mM DBV + 50 mM HCl, immediately following tip-assisted degrafting. Note that degrafting itself took place in the presence of the same electrolyte. During degrafting, the substrate potential was kept at E1 = −300 mV (Fig. 6b–d) or E2 = −510 mV (Fig. 6e–g), where DBV is expected to occur in its radical monocationic (DBV˙+) and uncharged form (DBV0), respectively (see Fig. S3†). As seen from Fig. 6, immediately following the nanocorral formation, the reduced viologen species selectively self-assemble within the nanocorrals and form different ordered structures at different substrate potentials. The size of the self-assembled monolayer is determined by the dimensions of the degrafted nanocorrals. We note that the defect density in the self-assembled layers inside the nanocorrals is higher than on standard HOPG surfaces (see Fig. S7a and b†). We propose that this difference is a direct expression of the nanoconfinement itself: the very small domains that can form in the corrals do so under very different conditions from the large domains on unrestricted HOPG surfaces, where Ostwald-like ripening takes place and large domains grow at the expense of smaller domains, leading to longer-range order. Also, nucleation is expected to take place primarily at the edges of the nanocorral, whose length/area ratio is much higher than on an open surface. The difference in behaviour in our opinion underscores the added value of our approach, as much may be learned about nucleation and growth of self-assembled structures by providing restricted substrate areas.
At E1 = −300 mV, each bright feature is believed to consist of two individual radical monocations (DBV˙+) forming a so-called “dimer phase” as shown in Fig. 6b–d. By contrast, the DBV0 uncharged species forms a stacking phase at E2 = −510 mV as shown in Fig. 6e–g. Different domains are rotated with respect to each other by 117 ± 2° (Fig. 6e). Individual DBV˙+ and DBV0 species appear as elongated features (blue lines), which align face-to-face due to π–π interactions between neighboring benzyl groups and spin pairing between neighboring reduced bipyridinium.52 Strikingly, in both the dimer and the stacking phases, individual molecules are rotated with respect to the long row axis (yellow lines) by 121 ± 2° (Fig. 6d and g) suggesting the influence of the graphite substrate in directing the self-assembly. The orientation of the molecular rows and of the individual species within the rows in both phases runs parallel to the symmetry axes of the underlying hexagonal HOPG lattice as observed with respect to bare HOPG (Fig. S4 and S5†). Tentative models for the respective phases are shown in Fig. 6h and i.
The dynamics of molecular phase transitions on changing the substrate potential was also experimentally observed. Fig. 7 shows the same surface area in response to shifting the substrate potential to more negative values. The dimer phase (Fig. 7a) is persistent in the potential regime −420 mV < E < −280 mV, and assigned to the formation of DBV˙+ (Fig. S3†). On passing the second reduction peak (DBV˙+ → DBV0) at E = −450 mV, this ordered phase is gradually broken down and the stacking phase of DBV0 partially appears. The phase transition starts at the borders of the nanocorral, which is therefore identified as a nucleation center (yellow arrow in Fig. 7b). A complete phase conversion is obtained when the substrate potential reaches E = −520 mV (Fig. 7c). Interestingly, a structural optimization of the stacking phase is also observed by ripening of domain size and orientation as indicated by yellow arrows in Fig. 7d. The dimer-to-stacking phase transition within the nanocorrals is a reversible process: sweeping the substrate potential back in positive direction leads to the reappearance of the dimer structure corresponding to the conversion of DBV0 to DBV˙+ (Fig. S6†). At potentials E > −280 mV, dicationic DBV2+ is formed, which binds only weakly to HOPG53 and results in a 2D adsorbate gas phase instead of an ordered structure (see Fig. S7c†).
Fig. 7 Dynamics of phase transition of the viologen molecule driven by change in substrate potential. Ub = +150 mV, It = 0.1 nA. |
The self-assembly of DBV on HOPG under electrochemical conditions has been studied previously, albeit in a different electrolyte.53 The voltammetric behavior in that case was limited to one pair of sharp peaks, indicative of a faradaic phase transition between the 2D adsorbate gas and the stacking phase, which was assigned to DBV˙+. In the present case, the voltammogram clearly shows a third stability region in the cathodic scan (see Fig. S3†), where the dimer phase is observed (DBV˙+), before the stacking phase (DBV0) is reached at the most negative potentials.
Nanocorrals of three different sizes were constructed (57 nm × 57 nm, 42 nm × 34 nm and 30 nm × 20 nm, Fig. 8) and the self-assembly of DBV0 (stacking phase) was monitored. In these proof-of-concept experiments, self-assembly is observed irrespective of the size of the nanocorrals. This level of control over the size and shape of nanocorrals will allow investigating nucleation and monolayer growth processes in nanoconfined spaces under electrochemical control.
Fig. 8 Self-assembly of DBV0 within nanocorrals of different size created by the EC-STM tip, E = −530 mV, Ub = +230 mV, It = 0.1 nA. |
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6nr07519c |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2017 |