Efficient Bottom-Up Preparation of Graphene Nanoribbons by Mild Suzuki–Miyaura Polymerization of Simple Triaryl Monomers

Article abstract of DOI:10.1002/chem.201602007

Herein an efficient bottom-up solution-phase synthesis of N=9 armchair graphene nanoribbons (GNRs) is described. Catalyzed by Pd(PtBu₃)₂, Suzuki–Miyaura polymerization of a simple and readily available triaryl monomer provides a nove

Full text of DOI:10.1002/chem.201602007

DOI: 10.1002/chem.201602007
Communication
&
Nanotechnology
Efficient Bottom-Up Preparation of Graphene Nanoribbons by
Mild Suzuki–Miyaura Polymerization of Simple Triaryl Monomers
Gang Li,[a] Ki-Young Yoon,[a] Xinjue Zhong,[b] Xiaoyang Zhu,*[b] and Guangbin Dong*[a]
tween the two monomers, which generally leads to lower mo-
Abstract: Herein an efficient bottom-up solution-phase
lecular weights.
synthesis of N=9 armchair graphene nanoribbons (GNRs)
To date, two indirect strategies have been employed to ad-
is described. Catalyzed by Pd(PtBu3)2, Suzuki–Miyaura poly-
dress the aforementioned issue. One is to generate a polymer
merization of a simple and readily available triaryl mono-
precursor for GNR with a more flexible backbone (non-PPP
mer provides a novel GNR precursor with a high molecular
type).[5–7] Despite the successful synthesis of GNRs with high
weight and excellent solubility. Upon cyclodehydrogena-
molecular weights in certain cases,[6a] this strategy generally re-
tion, the resulting GNR exhibits semiconducting properties
quires monomers with a relatively complex structure and often
with an approximately 1.1 eV band gap (LUMO: À3.52 eV;
HOMO: À4.66 eV) as characterized by UV/Vis-NIR spectros-
copy and cyclic voltammetry.
a high polymerization temperature.[6,7] The other strategy is to
produce GNRs by a surface-assisted protocol with Au(111) or
Ag(111) crystals.[8] While this non-solution-based process can
synthesize long GNRs (>200 nm), only a limited amount of
GNRs bound on a metal surface were produced at each batch
which also restrains the processing method to physical techni-
ques.[1d] In addition, neither approach has been used to pre-
pare N=9 armchair GNRs to date. Herein, to enable a practical
and efficient synthesis of GNRs with a large molecular weight,
we describe a new strategy for bottom-up solution-phase syn-
thesis of N=9 armchair GNRs through mild Suzuki–Miyaura
polymerization of a simple triaryl monomer.
Nanometer-wide strips of graphene, namely graphene nanorib-
bons (GNRs), are considered to be promising semiconducting
materials for nanoscale electronic devices because they pos-
sess nonzero bandgaps.[1] Initially, more intuitive “top-down”
approaches such as lithographic pattering of graphene,[2a] un-
zipping of carbon nanotubes,[2b] and sonochemical extraction
from graphite[2c] have been pursued to yield GNRs; yet, these
approaches are not ideal to produce GNRs with atomically pre-
cise edges and widths that are <10 nm.[1,2] Recent advances in
the “bottom-up” synthesis of GNRs from molecular precursors
through polymerization methods have allowed the precise ma-
nipulation of the edges and widths, which critically define the
optoelectronic properties of GNRs.[1,3] In 2008, Müllen and co-
workers reported a pioneering method for the solution-phase
synthesis of a structurally well-defined N=9 armchair GNR
through A2B2-type Suzuki–Miyaura polymerization, followed by
We hypothesized that the key to access N=9 armchair GNRs
with high molecular weights would be to employ a sterically
less-hindered AB-type monomer, because it should enhance
both the polymerization efficiency and the rotational flexibility
of the resulting PPP precursor. From a retrosynthetic view-
point, the design of a 2,3-bisarylated 4-bromophenylboric acid
ester (1) was conceived, which is expected to provide a unique
PPP GNR precursor (P1) (Scheme 1B). However, practical prepa-
ration of such a 1,2,3,4-tetrasubstitutedbenzene is not a trivial
issue. Thus, we first sought to develop an efficient route to
access monomer 1.
cyclodehydrogenation, also known as
a Scholl oxidation
(Scheme 1A).[4] While ground-breaking, this strategy was chal-
lenging to prepare GNRs with high molecular weights due to
the use of sterically congested penta- and hexaaryl monomers
that hampered the efficacy of the polymerization and the solu-
bility[5] of the poly(p-phenylene) (PPP) precursor. In addition,
compared to AA- or AB-type polymerizations, using A2B2-type
polymerization is hard to avoid stoichiometric imbalance be-
Starting from inexpensive 1,2-dibromobenzene, a bromine-
directed ortho lithiation/silylation was adopted through modifi-
cation of a known procedure[9] to access 1,4-disilane
2
(Scheme 2). Subsequent Suzuki–Miyaura coupling followed by
bromination with Br2 afforded 1,4-dibromo triaryl 5, which can
potentially serve as a monomer for Yamamoto polymeri-
zation.[10] Finally, selective monolithiation/borylation provided
monomer 1 in good yield.
With monomer 1 in hand, various Suzuki–Miyaura polymeri-
zation conditions were examined.[11] Applying the commonly
used [Pd(PPh3)4]/K2CO3 conditions already led to unfractionated
polymer P1 with a relatively high molecular weight (Mn =
14.5 kDa), determined by size-exclusion chromatography (SEC)
analysis (Table 1, entry 4). After careful optimization (see
Table S1, Supporting Information), Pd(PtBu3)2 and K3PO4 were
found to be a superior catalyst-base combination,[12] allowing
[a] Dr. G. Li, K.-Y. Yoon, Prof. Dr. G. Dong
Department of Chemistry, University of Texas at Austin
100 East 24th Street, Austin, TX 78712 (USA)
E-mail: gbdong@cm.utexas.edu
[b] X. Zhong, Prof. Dr. X.-Y. Zhu
Department of Chemistry, Columbia University
3000 Broadway, New York, NY 10027 (USA)
E-mail: xyzhu@columbia.edu
Supporting information for this article can be found under
http://dx.doi.org/10.1002/chem.201602007.
Chem. Eur. J. 2016, 22, 9116 – 9120
9116
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
Scheme 1. Bottom-up solution synthesis of N=9 armchair graphene nanoribbons.
lated yield (entry 1). This high molecular weight material was
successfully characterized by SEC analysis due to its excellent
solubility. The molecular weight was also confirmed up to
30.0 kDa (degree of polymerization:~60) by MALDI-TOF MS,
which also indicated that the end groups of P1S were HÀ/HÀ
or HÀ/PhÀ (Figure S1, Supporting Information).
To explore the graphitization step and assist our understand-
ing and characterization of the final GNR product, dimer 6 and
trimer 7 were rapidly synthesized as model substrates (see the
Supporting Information for the synthetic routes). After survey-
ing a number of Scholl oxidation methods,[13] the use of DDQ
with triflic acid (TfOH) in CH2Cl2 was found to offer the cleanest
reaction (Scheme 3).[14] The corresponding nanographenes, G2
and G3, were obtained in 60 and 70% yields, respectively[15] as
yellow solids. Their structures were unambiguously confirmed
Scheme 2. Monomer synthesis. LDA=lithium diisopropylamide.
Table 1. Selected optimization for the polymerization of 1.
1
by H/13C NMR, IR, MALDI-TOF MS, and X-ray crystallography. It
is worth noting that nanographene G3 does not exhibit a per-
fect planar surface around the edge, with steep out-of-plane
side chains (Scheme 3), which might provide an important im-
plication to the structure of the GNR. Finally, GNR G1S was pre-
pared in 56% yield (after Soxhlet extraction) as a black powder
from polymer P1S by using the same oxidation protocol with
the exception of employing a longer reaction time (Scheme 3).
The resulting GNR G1S was characterized by solid-state tech-
niques. Analogous to the previously reported data on GNRs
prepared from solution-phase synthesis,[4–7] the FTIR spectrum
of G1S showed the absence of signals derived from free rota-
tion of the phenyl rings (4054 cmÀ1), and significantly dimin-
ished signal triad from the aryl CÀH stretching vibrations
(3024, 3050, 3082 cmÀ1) compared to the polymer precursor
P1S (Figure 1a and Figure S2, Supporting Information). Raman
spectroscopy of the G1S powder revealed two intense peaks
at 1340 and 1621 cmÀ1, typically assigned to the first-order D
and G bands of graphitic materials, respectively, as well as less
intense second-order bands (Figure 1b). MALDI-TOF MS was
used for quantitative analysis of the cyclodehydrogenation by
comparing the corresponding mass between the precursor
(P1) and the GNR (G1) (Figure S1, Supporting Information). As-
suming that the mass difference was only caused by the loss
of hydrogen, the efficiency of the cyclodehydrogenation step
Yield[a]
Mn
Mw
“
[b]
[b]
[b]
Entry Variation from
the “standard conditions” [%]
[kDa]
[kDa]
1
none
89
(59)[c]
92
94
92
19.5
(30.6)
18.0
16.0
14.5
33.7
(42.9)
31.1
27.4
18.7
1.73
(1.40)
1.73
1.71
1.29
2
3
4
258C
758C
3 mol% [Pd(PPh3)4],
K2CO3,
toluene, 3 d
[a] Isolated yield. [b] Determined by THF SEC calibrated using polystyrene
standards. [c] After Soxhlet extraction under reflux of acetone.
the polymerization to proceed smoothly at 258C in THF to
give P1 with Mn =18.0 kDa (Entry 2). Changing the reaction
temperature from 25 to 508C slightly increased the Mn to
19.5 kDa with a moderate molar mass dispersity (“) (entry 1);
however, further increasing the reaction temperature to 758C
decreased the Mn, likely due to partial decomposition of the
catalyst at this temperature (entry 3). Notably, Soxhlet extrac-
tion of the P1 synthesized at 508C provided the GNR precursor
(P1S) with an enhanced Mn (30.6 kDa) and a low “ in 59% iso-
Chem. Eur. J. 2016, 22, 9116 – 9120
www.chemeurj.org
9117
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
Scheme 3. Graphitization to access nanographenes and GNR G1S.
DDQ=2,3-dichloro-5,6-dicyano-1,4-benzoquinone.
Figure 1. (a) IR spectra of P1S (black) and G1S (red). (b) Raman spectrum of
G1S.
was 99%. Altogether, these observations support a successful
graphitization step.
visible, and even near IR (NIR) regions (Figure 3),[16] clearly
showing three excitonic states in the Vis-NIR region as expect-
ed by calculations in the previous literature.[3f] The lonset of G1
was observed in the NIR region at 1110 nm, corresponding to
the optical Egap of 1.12 eV. The lonset of G1S was at 1170 nm,
similar but slightly bathochromically shifted, corresponding to
an optical Egap of 1.06 eV. Furthermore, the cyclic voltammo-
gram of G1S revealed that the HOMO and lowest unoccupied
molecular orbital LUMO were À4.66 and À3.52 eV, respectively,
from the vacuum level (Figure 3 and Figure S5, Supporting In-
formation). The observed electrochemical Egap (1.14 eV) is mar-
ginally larger than the optical Egap, as typically seen in organic
semiconductors. The Egap of G1S was, to some extent, larger
than the theoretically predicted bandgap of N=9 armchair
GNR (0.86 eV).[3d,e] This discrepancy is presumably due to the
finite-length effect[17] and/or imperfect planarity of GNR G1S as
suggested by the structure of G3.
Detailed information on the nanostructure of G1S was ob-
tained by AFM. Figure 2a shows an AFM image of GNR G1S
deposited on highly oriented pyrolytic graphite (HOPG) (also
see Figure S3, Supporting Information). The GNR molecules
self-assemble into 2D islands consisting of highly ordered
stripes. Each 2D island is characterized by a height of 0.40Æ
0.05 nm and an interstripe separation of 4.8Æ0.2 nm, as
shown by a typical line profile across the boundary (Figure 2b)
and Fourier transform of the image (Figure 2a, inset). The
height of each island is close to the interlayer distance in
graphite (0.34 nm) and the stripe separation is in excellent
agreement with the expected intermolecular spacing between
parallel GNR molecules adopting a flat-lying geometry, as de-
picted by the molecular model in Figure 2c. Similar self-assem-
bled 2D-striped structures have also been observed in other
types of GNRs.[4,7b,c]
To gain a better understanding of GNR G1 as a semiconduct-
ing material, UV/Vis-NIR spectra of two GNRs, G1S and short
G1 (made from a short P1, Mn =11.0 kDa, “=1.56, see
Table S1, entry 7, Supporting Information) as thin films, and
nanographenes, G2 and G3, were obtained at room tempera-
ture and compared. G2 and G3 only absorbed light in the UV
and visible regions showing a bandgap (Egap) of 2.82 and
2.64 eV, respectively, as determined by their absorption onset
wavelengths (lonset) (Figure S4, Supporting Information). In con-
trast, G1 and G1S exhibit a strong, broad absorption in the UV,
In summary, a unique strategy for the efficient solution-
phase synthesis of N=9 armchair GNRs was developed, which
features a simple triaryl monomer and mild Suzuki–Miyaura
polymerization conditions. Benefited by a highly soluble PPP
precursor, this approach allows for a rapid entry to longer
GNRs. Given the conciseness and modular character of the mo-
nomer preparation, this strategy holds great potential to be
generalized for accessing other functionalized GNRs. In-depth
studies of the electronic structure and charge carrier dynamics
in these GNRs are ongoing in our laboratories.
Chem. Eur. J. 2016, 22, 9116 – 9120
www.chemeurj.org
9118
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
viersity. G.L. thanks EFRC-CST (DOE) for a postdoc fellowship.
Dr. H. S. Ahn in the Bard group (UT Austin), Dr. V. Lynch (UT
Austin), and Dr. M. Tuan Trinh (Columbia) are acknowledged
for the CV, X-ray, and Raman experiments, respectively. We
thank Dr. M. C. Young for proofreading the manuscript. John-
son Matthey is thanked for a donation of Pd salts.
Keywords: conjugated polymer
·
graphene nanoribbon
·
nanographene
polymerization
·
palladium catalysis Suzuki–Miyaura
·
[1] a) A. Narita, X.-Y. Wang, X. Feng, K. Müllen, Chem. Soc. Rev. 2015, 44,
6616; b) K. Müllen, ACS Nano 2014, 8, 6531; c) A. Narita, X. Feng, K.
Müllen, Chem. Rec. 2015, 15, 295; d) Y. Segawa, H. Ito, K. Itami. Nat. Rev.
Mater. doi:10.1038/natrevmats.2015.2.
[2] a) Z. Chen, Y.-M. Lin, M. J. Rooks, P. Avouris, Physica E 2007, 40, 228;
b) L. Jiao, L. Zhang, X. Wang, G. Diankov, H. Dai, Nature 2009, 458, 877;
c) X. Li, X. Wang, L. Zhang, S. Lee, H. Dai, Science 2008, 319, 1229.
[3] a) B. Obradovic, R. Kotlyar, F. Heinz, P. Matagne, T. Rakshit, M. D. Giles,
M. A. Stettler, D. E. Nikonov, Appl. Phys. Lett. 2006, 88, 142102; b) J.
Wang, R. Zhao, M. Yang, Z. Liu, J. Chem. Phys. 2013, 138, 084701; c) S.
Osella, A. Narita, M. G. Schwab, Y. Hernandez, X. Feng, K. Müllen, D. Bel-
jonne, ACS Nano 2012, 6, 5539; d) H. Raza, E. C. Kan, Phys. Rev. B 2008,
77, 245434; e) Y.-W. Son, M. L. Cohen, S. G. Louis, Phys. Rev. Lett. 2006,
97, 216803; f) D. Prezzi, D. Varsano, A. Ruini, A. Marini, E. Molinari, Phys.
Rev. B 2008, 77, 041404.
[4] X. Yang, X. Dou, A. Rouhanipour, L. Zhi, H. J. Räder, K. Müllen, J. Am.
Chem. Soc. 2008, 130, 4216.
[5] L. Dçssel, L. Gherghel, X. Feng, K. Müllen, Angew. Chem. Int. Ed. 2011,
50, 2540; Angew. Chem. 2011, 123, 2588.
[6] For using Diels–Alder reactions to prepare GNR precursors, see: a) A.
Narita, X. Feng, Y. Hernandez, S. A. Jensen, M. Bonn, H. Yang, I. A. Verzh-
bitskiy, C. Casiraghi, M. R. Hansen, A. H. R. Koch, G. Fytas, O. Ivasenko, B.
Li, K. S. Mali, T. Balandina, S. Mahesh, S. De Feyter, K. Müllen, Nat. Chem.
2014, 6, 126; b) A. Narita, I. A. Verzhbitskiy, W. Frederickx, K. S. Mali, S. A.
Jensen, M. R. Hansen, M. Bonn, S. De Feyter, C. Casiraghi, X. Feng, K.
Müllen, ACS Nano 2014, 8, 11622; c) J. Wu, L. Gherghel, M. D. Watson, J.
Li, Z. Wang, C. D. Simpson, U. Kolb, K. Müllen, Macromolecules 2003, 36,
7082.
Figure 2. (a) AFM image of G1S self-assembled on HOPG. The inset shows
a Fourier transform of the image with the two diffraction spots correspond-
ing to an interstripe distance of 4.8Æ0.2 nm. (b) A representative cross-sec-
tional profile of the image (red line in panel a). (c) Molecular model of flat-
lying GNR molecules.
[7] For using Yamamoto polymerization methods to prepare GNR precur-
sors, see: a) M. G. Schwab, A. Narita, Y. Hernandez, T. Balandina, K. S.
Mali, S. De Feyter, X. Feng, K. Müllen, J. Am. Chem. Soc. 2012, 134,
18169; b) T. H. Vo, M. Shekjirev, D. A. Kunkel, F. Orange, M. J.-F. Guinel, A.
Enders, A. Sinitskii, Chem. Commun. 2014, 50, 4172; c) T. H. Vo, M. Sheki-
hirev, D. A. Kunkel, M. D. Morton, E. Berglund, L. Kong, P. M. Wilson, P. A.
Dowben, A. Enders, A. Sinitskii, Nat. Commun. 2014, 5, 3189; d) M. E.
Gemayel, A. Narita, L. F. Dçssel, R. S. Sundaram, A. Kiersnowski, W.
Pisula, M. R. Hansen, A. C. Ferrari, E. Orgiu, X. Feng, K. Müllen, P. Samori,
Nanoscale 2014, 6, 6301.
[8] a) J. Cai, P. Ruffieux, R. Jaafar, M. Bieri, T. Braun, S. Blankenburg, M.
Muoth, A. P. Seitsonen, M. Saleh, X. Feng, K. Müllen, Nature 2010, 466,
470; b) J. Cai, C. A. Pignedoli, L. Talirz, P. Ruffieux, H. Sçde, L. Liang, V.
Meunier, R. Berger, R. Li, X. Feng, K. Müllen, R. Fasel, Nat. Nanotechnol.
2014, 9, 896; c) R. R. Cloke, T. Marangoni, G. D. Nguyen, T. Joshi, D. J.
Rizzo, C. Bronner, T. Cao, S. G. Louie, M. F. Crommie, F. R. Fischer, J. Am.
Chem. Soc. 2015, 137, 8872; d) J. Liu, B.-W. Li, Y.-Z. Tan, A. Giannakopou-
los, C. Sanchez-Sanchez, D. Beljonne, P. Ruffieux, R. Fasel, X. Feng, K.
Müllen, J. Am. Chem. Soc. 2015, 137, 6097; e) Y.-C. Chen, T. Cao, C. Chen,
Z. Pedramrazi, D. Haberer, D. G. de Oteyza, F. R. Fischer, S. G. Louie, M. F.
Crommie, Nat. Nanotechnol. 2015, 10, 156; f) C. Rogers, C. Chen, Z. Pe-
dramrazi, A. A. Omrani, H.-Z. Tsai, H. S. Jung, S. Lin, M. F. Crommie, F. R.
Fischer, Angew. Chem. Int. Ed. 2015, 54, 15143; Angew. Chem. 2015, 127,
15358.
Figure 3. UV/Vis-NIR spectra of short G1 (black) and G1S (red), and HOMO/
LUMO levels obtained by cyclic voltammogram (inset table).
Acknowledgements
G.D. thanks the Welch Foundation (F 1781) for research grants.
G.D. is a Sloan fellow. X.Y.Z. acknowledges support by NSF
DMR 1420634 (Materials Research Science and Engineering
Center). The AFM imaging experiment utilizes facilities of the
Share Materials Characterization Laboratory at Columbia Un-
[9] V. Diemer, F. R. Leroux, F. Colobert, Eur. J. Org. Chem. 2011, 327.
[10] Attempts of using 5 in a Yamamoto polymerization resulted in P1 with
a low molecular weight (<5.2 kDa).
Chem. Eur. J. 2016, 22, 9116 – 9120
www.chemeurj.org
9119
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
[11] For using Suzuki–Miyaura reactions to prepare PPPs, see: a) T. I. Wallow,
B. M. Novak, J. Am. Chem. Soc. 1991, 113, 7411; b) T. Yokozawa, H.
Kohno, Y. Ohta, A. Yokoyama, Macromolecules 2010, 43, 7095.
[12] For seminal work on using Pd(PtBu3)2 in Suzuki–Miyaura couplings, see:
a) A. F. Littke, G. C. Fu, Angew. Chem. Int. Ed. 1998, 37, 3387; Angew.
Chem. 1998, 110, 3586; b) A. F. Littke, C. Dai, G. C. Fu, J. Am. Chem. Soc.
2000, 122, 4020.
[13] For a recent review, see: M. Grzybowski, K. Skonieczny, H. Butenschçn,
D. T. Gryko, Angew. Chem. Int. Ed. 2013, 52, 9900; Angew. Chem. 2013,
125, 10084.
[15] The moderate yields were likely caused by the intensive washing
during the purification.
[16] The difference in their absorption maxima is likely due to their dissimi-
lar aggregation states caused by different lengths. For a similar obser-
vation, see ref. [6b].
[17] O. Hod, J. E. Peralta, G. E. Scuseria, Phys. Rev. B 2007, 76, 233401.
[18] CCDC 1454122 (5), 1454123 (G2) and 1454124 (G3) contain the supple-
mentary crystallographic data for this paper. These data can be ob-
tained free of charge from The Cambridge Crystallographic Data
Centre..
[14] a) L. Zhai, R. Shukla, R. Rathore, Org. Lett. 2009, 11, 3474; b) B. T. King, J.
Kroulík, C. R. Robertson, P. Rempala, C. L. Hilton, J. D. Korinek, L. M. Gor-
tari, J. Org. Chem. 2007, 72, 2279.
Received: April 29, 2016
Published online on May 31, 2016
Chem. Eur. J. 2016, 22, 9116 – 9120
www.chemeurj.org
9120
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim