Key words
graphene nanoribbons - surface chemistry - edge extension - methyl–methyl coupling
Introduction
On-surface synthesis is an emerging approach to fabricate one-dimensional polymers
and two-dimensional graphenic nanostructures with atomic precision.[1]
[2]
[3]
[4] With the assistance of high-resolution surface-sensitive techniques and theoretical
simulations, several classical organic reactions have been successfully realized via
on-surface synthesis under ultrahigh vacuum (UHV) conditions.[5]
[6]
[7] Among these are coupling protocols yielding products similar to Ullmann,[8]
[9] Glaser,[10]
[11] and Sonogashira reactions,[12]
[13] as well as intramolecular processes like Bergman cyclization.[14]
[15] Moreover, on-surface synthesis enables the construction of molecules that are challenging
or not accessible via conventional solution chemistry.[16]
[17]
[18]
[19] Therefore, the exploration of new on-surface chemistry is highly desirable to complement
the limited tool kits and realize increasingly complex architectures.
Surface-assisted electrocyclic ring closure followed by loss of hydrogens is the final
and critical step to construct fully conjugated carbon materials, such as nanographenes
(NGs) and graphene nanoribbons (GNRs).[20]
[21] Although the common aryl–aryl coupling alone does not allow the formation of zigzag
edges, the simultaneous use of an on-surface methyl–aryl coupling enabled the formation
of a zigzag GNR (6-ZGNR) in 2016.[19] After polymerizing a U-shaped monomer 1 on a Au(111) surface, the obtained polymer was further annealed to 352 °C to initiate
the desired planarization process. Besides the cyclodehydrogenation between benzene
rings, the methyl groups were also involved to achieve fully conjugated zigzag edges
([Scheme 1A]). Thereafter, the methyl–aryl oxidative ring closure has also been applied to produce
an armchair GNR (7-AGNR) with extended edges ([Scheme 1B]).[22]
[23]
[24] Besides hexagonal structures, methyl groups could be fused with the adjacent benzenes
to also form five-[25]
[26]
[27]
[28] or seven-membered rings[29] through different precursor designs.
Scheme 1 Synthetic routes to (A) 6-ZGNR and (B) edge-extended 7-AGNR via on-surface methyl–aryl
couplings. (C) The on-surface synthesis involving methyl–methyl coupling in this work
toward edge-extended 9-AGNR (9-eGNR) with both sp3- and sp2-hybridized carbons on the edge. The methyl groups and the bonds formed through oxidative
ring closure of methyl groups are highlighted in red.
On the other hand, methyl–methyl coupling has been investigated on surfaces and furnished
success in the intermolecular coupling between alkane chains or preactivated bromomethyl
groups.[30]
[31]
[32] Recently, intramolecular methyl–methyl coupling has been achieved to construct circumcoronene,
a hexagonal NG with six zigzag edges.[33] However, the intramolecular coupling between two benzylic methyl groups has never
been explored in GNRs, although it could potentially be developed as a powerful edge
functionalization approach for structure engineering.
Therefore, in this work, we explored the on-surface methyl–methyl coupling using dimethyl
substituted o-terphenyl 3 as the monomer towards the synthesis of edge-extended 9-AGNR (9-eGNR) ([Scheme 1C]). We expected that this approach would potentially provide fully conjugated 9-eGNR that is predicted to have electronic bands of topological origin.[22] We found that the methyl–methyl coupling was indeed achieved along the ribbon at
350 °C, furnishing the edge structures as characterized by high-resolution scanning
tunneling microscopy (STM) and noncontact atomic force microscopy (nc-AFM). However,
not all of the ethanediyl bridges (CH2–CH2) could undergo complete dehydrogenation towards conjugated alkenes even under further
annealing at 440 °C. The loss of aryl units was also observed, similar to the previously
reported synthesis of pristine 9-AGNRs.[34]
[35] These results shed light on the scope and limitation of intramolecular methyl–methyl
coupling for future GNR synthesis.
Results and Discussion
The synthesis of the new monomer 3 was carried out as displayed in [Scheme 2], adopting the procedure in previous reports.[36]
[37] Starting from commercially available 1,2-dibromobenzene (4), lithiation/silylation gave 1,4-disilyl intermediate 5 ([Scheme 2]). Subsequent Suzuki–Miyaura coupling of 5 with p-tolylboronic acid followed by bromination with Br2 at room temperature (RT) afforded monomer 3. To guarantee the high purity required by the on-surface polymerization, monomer
3 was recrystallized several times from methanol to completely remove the mono-brominated
side-product, which could terminate the on-surface polymerization and limit the lengths
of obtained GNRs. The contents of C and H atoms in elementary analysis were measured
to be 57.8% and 3.8%, respectively, which well matched with the calculated values
for C20H16Br2 (C: 57.72%, H: 3.88%). Matrix-assisted laser desorption/ionization-time of flight
(MALDI-TOF) mass spectrometry (MS) analysis of 3 was done using trans-2-[3-(4-tert-butylphenyl)-2-methyl-2-propenylidene]malononitrile as the matrix and silver trifluoroacetate
as the cationizing salt, thus leading to pseudo molecular ions where one silver cation was noncovalently attached to each molecule
of 3. An intense signal at m/z = 520.8669 was observed with isotopic distribution patterns well-matched by the calculated
spectrum ([Figure S5], calculated value for C20H16Br2Ag+: 520.8664).
Scheme 2 Synthetic route to monomer 3. LDA = lithium diisopropylamide. TMSCl = trimethylsilyl chloride. dppf = 1,1'-bis(diphenylphosphino)ferrocene.
For investigating the on-surface synthesis, monomer 3 was first sublimed onto the Au (111) surface at RT under UHV conditions. Densely
packed molecular islands of 3 were imaged by STM as displayed in [Figure 1A]. Due to the nonplanar geometry of 3, it is nontrivial to identify the individual molecules even from the high-resolution
STM images. By gradually increasing the temperature to 200 °C, as proposed in our
earlier work,[1] thermally activated debromination furnished biradical intermediates, which further
polymerized to yield linear polyphenylene chains. At this stage, a different phase
of the sample could be clearly observed in the STM image ([Figure 1B]). Because of the significant steric hindrance within the polymer, the central polyphenylene
backbone tends to be flat and lie closer to the substrate than the branched methyl–phenyl
groups. The brighter spots with the apparent height of ~5 Å occasionally appearing
in the STM image can be assigned to the tilting of some methyl groups out of plane,
or the bromine atoms adsorbing on top of the molecule.
Figure 1 Large-scale (left) and zoom-in (right) STM images, as well as chemical structures
of (A) monomer 3 sublimed on the Au (111) surface at RT, (B) the polyphenylene chain after annealing
at 200 °C, and (C) the fully cyclodehydrogenated GNR after annealing at 350 °C. Scanning
parameters: (A) left: V
s = −1 V, I
t = 100 pA; right: V
s = −1 V, I
t = 100 pA; (B) left: V
s = −1 V, I
t = 10 pA; right: V
s = −1 V, I
t = 10 pA; (C) left: V
s = −0.5 V, I
t = 30 pA; right: V
s = −0.5 V, I
t = 30 pA.
By further increasing the temperature to 350 °C, the cyclodehydrogenation was triggered,
fusing the benzene rings as well as inducing the coupling between the peripheral methyl
groups ([Figure 1C]). A closer investigation using the bond-resolved nc-AFM imaging technique with a
CO-functionalized tip[38] was performed to reveal the fine structures at the atomic level. As clearly resolved
in [Figure 2C], three distinguishable edge structures, indicated by the coloured arrows, were formed
after the reaction. The edge structures with bright features in the nc-AFM image marked
by the red arrows are attributed to the ethanediyl bridges with doubly hydrogenated
sp3-hybridized carbons (–CH2–CH2–).[39] On the other hand, the formation of conjugated six-membered rings was also observed,
leading to the π-extension of the armchair edge, as highlighted by the blue arrows.
This result indicates that the methyl–methyl coupling can indeed be achieved on Au(111)
surfaces under UHV conditions to afford the ethanediyl bridges, but the further dehydrogenation
towards the fully conjugated structure is not efficient enough for the clean conversion.
An approximate statistical analysis of the occurrence of the –CH = CH– and –CH2–CH2– motifs yields a ratio of 1:4 (another example of a typical 9-eGNR formed in our experiments is shown in [Figure S6]). Further annealing at 440 °C did not lead to the dehydrogenation of the ethanediyl
bridges, suggesting a higher kinetic energy barrier and that different reaction mechanisms
could be involved in the simultaneous formation of saturated and unsaturated C2 units.
Figure 2 (A) High-resolution STM (V
s = 0.01 V, I
t = 10 pA) and (B) constant-height STM current image and the simultaneously acquired
(C) nc-AFM image of the formed GNR (V
s = 10 mV). (D) Proposed chemical structure of the defective 9-eGNR. Blue arrows: sp2 hybridized carbon; red arrows: sp3 hybridized carbon; gray arrows: defects caused by the loss of aryl units.
Besides the coupling of methyl groups, the loss of aryl units was also observed during
the cyclodehydrogenation process, which appeared as the “bite defects” (marked with
gray arrows in [Figure 2C]). We note that these defects are not due to possible impurities in the precursor
compound, but occur intrinsically during the on-surface cyclodehydration of polyphenylene.
Similar defects were also observed during the previous synthesis of pristine 9-AGNRs
using methyl-free o-terphenyl-based monomers.[34]
[35]
Raman spectroscopy was applied for further characterization of the obtained 9-eGNR on Au(111). The radial-breathing-like mode (RBLM, ~300 cm−1) displays very low intensity embedded in the background noise ([Figure 3], marked by a dashed line),[34] indicating that the width of the ribbons is not uniform. The CH/D region of the
Raman spectrum is a signature of the GNR's edge structure. The Raman spectrum of pristine
9-AGNRs exhibits two distinct narrow peaks at 1232 cm−1 (C–H bending mode) and 1332 cm−1 (D mode, peak width ~15 cm−1), as displayed in [Figure 3].[40] However, a single broad peak at 1327 cm−1 was observed in the current 9-eGNR (peak width ~100 cm−1), which is a clear indication of structural diversity. In addition, the appearance
of a small peak around 1657 cm−1 (marked with *) seems similar to the D' mode associated with defects on GNR edges
made by top-down approaches, as well as on graphenes and carbon nanotubes.[41]
[42]
[43] More detailed investigations, for example by using tip-enhanced Raman spectroscopy
methods,[44] may thus provide further insights into the detailed chemical structures in the defective
graphene materials.
Figure 3 Raman spectra of the obtained 9-eGNR (green) and the pristine 9-AGNR (red) measured on Au(111) surfaces in air using 532 nm
and 785 nm lasers, respectively.
Conclusions
In summary, methyl–methyl coupling was explored on a Au(111) surface under UHV conditions
as a new synthetic approach for edge extension of AGNRs. As visualized by STM and
nc-AFM, the coupling of methyl groups proceeded when heated to 350 °C, mainly forming
the ethanediyl bridges alongside fully conjugated six-membered rings. Considering
the success of intramolecular methyl–methyl coupling in circumcoronene synthesis from
dodecamethyl hexa-peri-hexabenzocoronene,[33] we envision that the preplanarization might allow more efficient aromatization.
Besides, a more reactive surface like Cu(111) is expected to achieve more sp2-hybridized carbons from the intramolecular methyl–methyl coupling. New monomer designs
and further optimizations of dehydrogenation conditions will be conducted in the future
to employ the methyl–methyl coupling for GNRs with complex architectures.
