Letter
https://doi.org/10.1038/s41586-019-1331-z
Living annulative π-extension polymerization for
graphene nanoribbon synthesis
Yuuta Yano^1 , Nobuhiko Mitoma1,2, Kaho Matsushima^1 , Feijiu Wang1,2, Keisuke Matsui1,2, Akira takakura1,2, Yuhei Miyauchi1,2,3,
Hideto Ito1,2 & Kenichiro Itami1,2,4
The properties of graphene nanoribbons (GNRs)^1 –^5 —such as
conductivity or semiconductivity, charge mobility and on/off ratio—
depend greatly on their width, length and edge structure. Existing
bottom–up methods used to synthesize GNRs cannot achieve
control over all three of these parameters simultaneously, and
length control is particularly challenging because of the nature of
step-growth polymerization^6 –^18. Here we describe a living annulative
π-extension (APEX)^19 polymerization technique that enables rapid
and modular synthesis of GNRs, as well as control over their width,
edge structure and length. In the presence of palladium/silver salts,
o-chloranil and an initiator (phenanthrene or diphenylacetylene),
the benzonaphthosilole monomer polymerizes in an annulative
manner to furnish fjord-type GNRs. The length of these GNRs can
be controlled by simply changing the initiator-to-monomer ratio,
achieving the synthesis of GNR block copolymers. This method
represents a type of direct C–H arylation polymerization^20 and
ladder polymerization^21 , activating two C–H bonds of polycyclic
aromatic hydrocarbons and constructing one fused aromatic ring
per chain propagation step.
To achieve full synthetic control over the structures of GNRs,
we must draw inspiration from organic synthesis, where a target
molecular entity is built up with atom-by-atom precision^7 –^18.
Müllen^10 –^14 , Fasel^11 ,^13 ,^14 and others^11 ,^13 –^18 have reported bottom–up
synthetic approaches, such as the solution-phase Suzuki–Miyaura
coupling polymerization^10 ,^18 , the Diels–Alder polymerization^12 and
on-surface polymerization^11 ,^13 –^15 ,^17 of small aromatic components to
form polyarylene precursors, which are stitched into GNRs by cyclo-
dehydrogenation (Scholl reaction). These methods have received
considerable attention owing to their high potential of controlling the
width (~1 nm) and edge structures of GNRs, especially with respect to
top–down synthesis by lithographic cutting of graphene or unzipping
of carbon nanotubes^6 –^9. Whereas the width and edge structures can be
controlled precisely in solution-phase syntheses, precise length control
of GNRs is yet to be achieved. The same is true for on-surface GNR
synthesis, where length control remains an unresolved issue^11 ,^13 –^15.
Recently, we established APEX methodologies^19 for the rapid synthe-
sis of nanographenes from simple aromatic templates^22 ,^23. For exam-
ple, unfunctionalized polycyclic aromatic hydrocarbons can be directly
coupled at the K regions (convex armchair edge) with dibenzosilole
derivatives in the presence of a palladium catalyst^22 ,^23. We envisaged
that the use of phenanthrene (I) as an initiator and benzonaphthosilole
(M) as a monomer would result in a living^24 and controllable APEX
polymerization (Fig. 1a). A crucial feature of this approach is the use
of M as a monomer substrate. M contains a masked K region^22 , which
is only exposed upon carrying out the Pd-mediated APEX reaction.
Thus, the initial APEX reaction occurs selectively between the K region
of I and M to provide tribenzochrysene (1a) as the initial intermediate.
This results in the formation of a new fused aromatic ring and unmasks
a new K region, primed for subsequent APEX reactions. Because of
the steric congestion around the K region of 1a at the propagation
terminus, it is expected that the reaction rate of propagation between
1a and M is slower than the initiation reaction. Propagation would
then continue in a regioselective manner to afford fjord-type GNR 2.
After extensive optimization studies, we established the conditions
for the living APEX polymerization. The benzonaphthosilole monomer
M polymerized smoothly in the presence of I, Pd(OCOCF 3 ) 2 , AgSbF 6
and o-chloranil in 1,2-dichloroethane at 80 °C to afford fjord-type
GNR 2 (Fig. 1a). Analysis of the crude product by size-exclusion chro-
matography (SEC) with polystyrene (PS) standards indicated that 2 was
monomodal with narrow dispersity under various M/I ratios (Fig. 1b, c).
The estimated number-average molecular weight (Mn), dispersity
(Đ = Mw/Mn; where Mw is weight-average molecular weight), degree
of polymerization (DP) and estimated polymer length of the crude
GNRs, as well as the yields of purified GNRs, are shown in Fig. 1b.
This APEX polymerization enabled the length-controlled synthesis of
fjord-type GNR 2 merely by rational changes to the M/I ratio. In the
reaction with M/I = 10, the relatively short GNR 2 (Mn = 2.9 × 103 Da,
DP = 7) was obtained in a narrow range of Đ values. Increasing the
M/I ratio resulted in the formation of a longer GNR 2 : the reaction with
a M/I ratio of 500 furnished 2 in 82% yield with Mn = 1.5 × 105 Da,
Đ = 1.22 and DP = 391. We also conducted matrix-assisted laser
desorption/ionization time-of-flight (MALDI–TOF) mass analysis of a
representative sample (GNR 2 with Mn = 1.3 × 104 Da and Đ = 1.23)
to confirm the validity of the PS-based Mn, as conventional PS stand-
ards may be conformationally too flexible to reflect the real molecular
weight of rigid GNRs^12. MALDI–TOF mass analysis revealed eleven
distinct mass peaks, each observed with gaps of m/z = 382 (m, mass; z,
charge; Extended Data Fig. 1a, b), supporting the specific formation of
the fjord-type GNR structure without any defects such as non-cyclized
or non-aromatized products. Furthermore, the dispersity of the pol-
ymer ranged from DP(n) = 21–31 (m/z = 8,000–12,000), implying a
narrow dispersity range in the current APEX polymerization. The m/z
value of the peak top (m/z = 10,116.9099) was considered to reflect
the real Mn of GNR 2 , which is slightly smaller than the PS-based Mn.
Therefore, we used the PS-based Mn in the current study to estimate
the molecular weight of GNRs.
The APEX polymerization was found to be highly regulated. The Mn
value changed predictably with the M/I ratio, whereas Đ was virtually
constant (~1.23) and independent of the M/I ratio (Fig. 1d). The Mn
value was also completely proportional to conversions of M without any
change of Đ (Fig. 1e). In the reaction with M/I = 10, the chain propa-
gation stopped within 1 h, but the propagation resumed to produce the
longer GNR 2 (Mn = 3.3 × 104 Da) when additional M (90 equivalents
relative to I) was added (Extended Data Fig. 1c, d). Notably, the Mn and
Đ values of 2 were almost identical to those obtained in the reaction with
M/I = 100 (Fig. 1b). Thus, the present APEX polymerization satisfies the
requirements for living polymerization. Although there have been some
reports of GNRs with low Đ values (1.06–1.12) after purification and
fractionation by SEC^25 , this work demonstrates the first living polym-
erization for GNRs with low Đ even in the crude mixture.(^1) Graduate School of Science, Nagoya University, Nagoya, Japan. (^2) JST-ERATO, Itami Molecular Nanocarbon Project, Nagoya, Japan. (^3) Institute of Advanced Energy, Kyoto University, Kyoto, Japan.
(^4) Institute of Transformative Bio-Molecules (WPI-ITbM), Nagoya University, Nagoya, Japan. *e-mail: [email protected]; [email protected]
18 JULY 2019 | VOL 571 | NAtUre | 387