17-AGNRs: Narrow-Bandgap Graphene Nanoribbons

• 17-AGNRs are atomically precise armchair graphene nanoribbons with 17 carbon dimer lines, classified in the 3p+2 family and exhibiting a narrow bandgap.
• Bottom-up on-surface synthesis on Au(111) combined with STM, nc-AFM, and STS confirms their structure, quantum confinement, and tunable electronic properties.
• They serve as a benchmark system for exploring width-controlled quantum confinement, substrate-induced gap renormalization, and challenges in device integration.

Seventeen-carbon-atom-wide armchair graphene nanoribbons (17-AGNRs) are atomically precise armchair graphene nanoribbons with $N=17$ carbon dimer lines across the width. In the standard AGNR family classification they belong to the $3p+2$ family, the branch associated with the smallest bandgaps at comparable width scale. This places 17-AGNRs at the narrow-bandgap end of bottom-up armchair nanoribbon research and makes them a reference system for studying width-controlled quantum confinement, substrate-renormalized electronic structure, Raman length metrology, and device integration (Wang et al., 2016).

1. Width notation, family assignment, and structural definition

The width of an armchair graphene nanoribbon is indexed by the number of carbon dimer lines across the ribbon. The literature uses equivalent symbols such as $N$, $N_a$, or $W$; for 17-AGNRs these all denote a ribbon that is 17 carbon atoms, or 17 dimer lines, wide across the armchair direction. Because

$17 = 3\times 5 + 2,$

17-AGNRs are members of the $3p+2$ family rather than the $3p$ or $3p+1$ branches (Wang et al., 2016).

Bottom-up Au(111) synthesis established the geometric scale of 17-AGNRs directly. Scanning tunnelling microscopy and bond-resolved nc-AFM identified planar armchair ribbons with width $2.5 \pm 0.1$ nm, an axial periodicity of $3p+2$0 nm, and an armchair edge morphology in which the filtered AFM images explicitly show 17 carbon atoms across the width. The DFT lattice constant along the ribbon axis is $3p+2$1 nm, consistent with the STM periodicity (Yamaguchi et al., 2019).

A complementary structural viewpoint comes from first-principles studies of H-terminated AGNRs, which showed that width quantization is reflected not only in the band structure but also in the relaxed geometry. For $3p+2$2, the longitudinal unit-cell elongation relative to bulk graphene was estimated as

$3p+2$3

that is, about $3p+2$4, with the threefold width periodicity encoded in the factor $3p+2$5. In that framework 17-AGNR sits in a definite phase of the oscillatory geometry–electronic-structure correspondence characteristic of armchair ribbons (Dasgupta et al., 2011).

2. Quantum confinement and electronic structure

The electronic structure of AGNRs is governed by transverse momentum quantization. In STM-based studies of atomically precise armchair ribbons, the relevant family classification is

$3p+2$6

with the $3p+2$7 family exhibiting systematically smaller gaps than the other two branches. For the specific trio $3p+2$8, first-principles calculations gave $3p+2$9 meV for 15-AGNR, $N$0 meV for 16-AGNR, and $N$1 meV for 17-AGNR, establishing 17-AGNR as the small-gap member between two large-gap neighbors (Wang et al., 2016).

The same conclusion emerges from the graphene-zone-folding picture. In the selection-rule formulation for AGNRs,

$N$2

For $N$3, the $N$4 line yields

$N$5

which passes through the graphene $N$6 point in the chosen parameterization. In a simple nearest-neighbor picture this would generate a zero-gap crossing; in realistic DFT with hydrogen termination and finite-width effects, the crossing is lifted and a small but finite gap opens (Dasgupta et al., 2011).

Different theoretical levels and dielectric environments produce different quantitative gaps for 17-AGNRs, and the reported values must therefore be read in context.

Context	Reported gap	Source
Free-standing H-terminated DFT-LDA	$N$7 meV	(Wang et al., 2016)
Gas-phase PBE, infinitely long ribbon	$N$8 eV	(Hwang et al., 15 Jul 2025)
Freestanding GW quasiparticle gap	$N$9–$N_a$0 eV	(Yamaguchi et al., 2019)
STS on Au(111)	$N_a$1 eV	(Yamaguchi et al., 2019)
STS on Au(111), apparent gap	$N_a$2 eV	(Hwang et al., 15 Jul 2025)

The Au-supported STS literature contains two distinct reported values. One study assigned a valence-band maximum at $N_a$3 V and a conduction-band minimum at $N_a$4 V, giving $N_a$5 eV, and showed that an image-charge-corrected quasiparticle gap $N_a$6 eV reproduces that measurement closely. Another study reported valence and conduction onsets near $N_a$7 V and $N_a$8 V and interpreted them as an apparent gap of about $N_a$9 eV (Yamaguchi et al., 2019). This suggests that the extracted band edges are sensitive to sample state and measurement protocol, even on the same nominal substrate.

Transport-relevant band curvature is also unusually favorable. GW analysis of freestanding 17-AGNRs yielded $W$0 and $W$1, markedly smaller than the corresponding values reported for 13-AGNRs, 9-AGNRs, and 7-AGNRs in the same comparison (Yamaguchi et al., 2019).

3. Bottom-up synthesis on Au(111)

Controlled synthesis of 17-AGNRs has been demonstrated by on-surface synthesis on Au(111) from the BADBB precursor. In that route the surface acts as both template and catalyst for a two-stage transformation: Ullmann-type polymerization after dehalogenation, followed by cyclodehydrogenation into a fully planar ribbon. Stepwise annealing studies reported bromine detachment and formation of extended one-dimensional “17-polymers” around $W$2C, and complete planarization into 17-AGNRs after annealing to $W$3C for 2 h (Yamaguchi et al., 2019).

The polymer stage is structurally distinct from the final ribbon. The 17-polymers show an axial periodicity of $W$4 nm, compared with $W$5 nm from DFT, and the anthracene side units remain non-planar because of steric hindrance. STM heights evolve from about $W$6 nm for intact BADBB islands, to about $W$7 nm for the polymer, and then to about $W$8 nm for the planarized 17-AGNR (Yamaguchi et al., 2019).

A later optimization study shifted the emphasis from proof-of-principle synthesis to length control. There, BADBB was sublimated onto room-temperature Au(111), dehalogenation was described as starting around $W$9C, and cyclodehydrogenation as occurring at $17 = 3\times 5 + 2,$0C. The critical process variable was a slow linear ramp from $17 = 3\times 5 + 2,$1C to $17 = 3\times 5 + 2,$2C at about $17 = 3\times 5 + 2,$3C/min, designed to separate polymerization from cyclodehydrogenation and to suppress premature end-capping by hydrogen released during planarization (Hwang et al., 15 Jul 2025).

That study also identified a specific termination mechanism. Although the BADBB design suggests staggered termini, most synthesized ribbons ended in a motif with two protrusions at about $17 = 3\times 5 + 2,$4 to the ribbon axis. DFT modeling attributed this to flipping of the terminal anthracene unit in the last precursor, followed by end-group cyclization that creates pentagons and blocks further growth. In that picture, minimizing premature terminal flipping is essential for extending the ribbon length (Hwang et al., 15 Jul 2025).

4. Microscopy, tunnelling spectroscopy, and Raman fingerprints

The structural identity of 17-AGNRs has been established by a combined STM/nc-AFM/STS workflow. Bond-resolved constant-height nc-AFM with CO-functionalized tips on Au(111)/mica resolved the internal carbon skeleton and directly confirmed both the armchair edge topology and the 17-atom width. STM topography of the final ribbons reproduced the expected interior ring-like contrast and matched Tersoff–Hamann simulations (Yamaguchi et al., 2019).

STS provided local band-edge assignments. In one study, edge spectra showed an occupied peak at $17 = 3\times 5 + 2,$5 V and an unoccupied peak at $17 = 3\times 5 + 2,$6 V, interpreted as the valence-band maximum and conduction-band minimum. Fourier-transformed STS on a single 17-AGNR of length $17 = 3\times 5 + 2,$7, sampled with $17 = 3\times 5 + 2,$8 nm along the edge, produced dispersive features whose shapes and curvatures agreed with rigidly shifted GW bands (Yamaguchi et al., 2019). Another study reported four prominent peaks at VB-1 $17 = 3\times 5 + 2,$9 V, VB $3p+2$0 V, CB $3p+2$1 V, and CB+1 $3p+2$2 V, with differential-conductance maps showing states spread along the ribbon and enhanced near the edges (Hwang et al., 15 Jul 2025).

Raman spectroscopy supplies the ensemble-averaged width, edge, and length fingerprints. On Au(111), 17-AGNR spectra showed CH/D-region features around $3p+2$3, $3p+2$4, and $3p+2$5 cm$3p+2$6, a mode at $3p+2$7 cm$3p+2$8 under $3p+2$9 nm excitation, and a G peak around $3p$0–$3p$1 cm$3p$2. After transfer to Raman-optimized substrates, low-frequency modes became accessible and the observed RBLM and overtone features appeared at $3p$3, $3p$4, $3p$5, and $3p$6 cm$3p$7. Deconvolution of the G region yielded components at $3p$8, $3p$9, $3p+1$0, and $3p+1$1 cm$3p+1$2, while the CH/D region displayed peaks at $3p+1$3, $3p+1$4, $3p+1$5, $3p+1$6, $3p+1$7, and $3p+1$8 cm$3p+1$9 (Hwang et al., 15 Jul 2025).

For AGNRs more generally, width metrology in Raman is organized around the radial-breathing-like mode,

$2.5 \pm 0.1$0

with empirical parameters $2.5 \pm 0.1$1 and $2.5 \pm 0.1$2. That relation was established for narrow confined AGNRs rather than for 17-AGNRs specifically, but it captures the same inverse-width trend that underlies the 17-AGNR RBLM assignment (Zhang et al., 2023).

Length metrology is provided by the universal longitudinal compressive mode (LCM),

$2.5 \pm 0.1$3

which was shown to be present in 5-, 7-, and 9-AGNRs and to depend primarily on ribbon length rather than width. For 17-AGNRs, this was operationalized by combining REBOII, Phonopy, and bond-polarizability calculations with experiment: low-frequency peaks at $2.5 \pm 0.1$4, $2.5 \pm 0.1$5, $2.5 \pm 0.1$6, and $2.5 \pm 0.1$7 cm$2.5 \pm 0.1$8 were assigned to finite ribbons ranging from 5-mer to 11-mer, corresponding to lengths from about $2.5 \pm 0.1$9 nm to $3p+2$00 nm (Overbeck et al., 2019).

5. Length optimization, substrate transfer, and device integration

Length control is the key practical bottleneck for 17-AGNR device integration. Under optimized on-surface-synthesis conditions, ribbon length increased systematically with coverage and thermal protocol: moderate coverage produced an average length $3p+2$01 nm, higher coverage increased it to $3p+2$02 nm, and near-monolayer coverage combined with the slow thermal ramp yielded $3p+2$03 nm, with individual ribbons exceeding $3p+2$04 nm. The study attributed this to a template-like effect generated by dense monomer assembly and parallel ribbon growth within ordered islands (Hwang et al., 15 Jul 2025).

Transfer off the Au growth substrate was achieved by a polymer-free process. The GNR/Au(111)/mica stack was floated on aqueous HCl until the Au film delaminated from mica, transferred onto the target substrate, treated with ethanol to improve adhesion, annealed at $3p+2$05C for 10 min, and then stripped of Au with a KI/I$3p+2$06-based etchant. Raman before and after transfer showed that the width and edge fingerprints were retained, demonstrating stability under ambient conditions and in harsh chemical environments including acid vapors and etchants (Hwang et al., 15 Jul 2025).

Thermal robustness on oxide-based device substrates is more limited. Annealing at $3p+2$07C for 2 h broadened and merged CH/D-region Raman peaks, and strong local laser heating produced similar degradation. The available evidence therefore separates chemical robustness during wet transfer from post-transfer thermal stability on dielectrics (Hwang et al., 15 Jul 2025).

These materials have nevertheless been incorporated into functional field-effect transistor geometries. The demonstrated architecture used graphene source and drain electrodes separated by $3p+2$08–$3p+2$09 nm nanogaps on a local $3p+2$10 gate dielectric, with highly p-doped Si/$3p+2$11 as the supporting wafer. Because optimized 17-AGNRs have an average length close to the gap size, they can bridge the electrodes after transfer (Hwang et al., 15 Jul 2025).

Transport measurements showed strongly nonlinear, S-shaped $3p+2$12–$3p+2$13 curves at room temperature, with currents up to the nA range at $3p+2$14 V and resistances distributed around $3p+2$15–$3p+2$16 G$3p+2$17. Room-temperature gate modulation was weak. At low temperature, $3p+2$18 maps revealed gate-dependent resonances and regions of suppressed conductance consistent with tunnelling through discrete states and Coulomb interactions, although clear isolated Coulomb diamonds were not obtained. The device study interpreted the low-temperature response as evidence that transport occurs through the GNRs, with multiple channels in parallel or series likely contributing (Hwang et al., 15 Jul 2025).

6. Edge chemistry, confined synthesis, and networked extensions

The intrinsic small-gap character of 17-AGNRs can be altered substantially by edge chemistry. First-principles studies of chemically modified AGNRs did not tabulate single-H 17-AGNR directly, but they established the family trend that $3p+2$19 ribbons remain the smallest-gap semiconductors under H, F, NH$3p+2$20, and NO$3p+2$21 edge functionalization. For 17-AGNRs specifically, one explicit result was given for double hydrogenation: a double-H-passivated $3p+2$22 AGNR, whose effective width is reduced to 16, was reported to have a $3p+2$23 eV bandgap. N passivation was qualitatively different: N-passivated AGNRs were metallic for all studied widths because N-derived states drive a semiconductor-to-metal transition (0711.1700).

A distinct synthetic extension is confined growth inside carbon nanotubes. Microwave-heating work established that AGNR formation inside $3p+2$24 nm single-walled carbon nanotubes is width-selected by steric fit and that microwave processing in seconds yields flatter, less twisted 6- and 7-AGNRs than furnace annealing. That study did not synthesize 17-AGNRs and did not treat any $3p+2$25 armchair ribbons experimentally or theoretically. It nevertheless stated that a 17-AGNR, with nominal width of about $3p+2$26 nm, would not fit inside the $3p+2$27 nm hosts used there and would require larger-diameter nanotubes of roughly $3p+2$28–$3p+2$29 nm inner diameter. A plausible implication is that the same confinement-plus-microwave strategy could be transferable to 17-AGNRs if an appropriate host diameter and precursor chemistry were identified (Zhang et al., 2023).

Seventeen-AGNRs also enter recent theory as building blocks of two-dimensional defective carbon networks. In periodic 4–8 defect-line sheets, the network can be interpreted as an array of covalently connected AGNRs. For $3p+2$30, the resulting 17-d$3p+2$31CN structure is explicitly classified as a class III network, with lattice parameters $3p+2$32 \AA\ and $3p+2$33 \AA. Class III networks inherit the $3p+2$34 family behavior of their constituent ribbons and are semimetallic: the valence band crosses the Fermi level near $3p+2$35, while an electron pocket forms near $3p+2$36 for odd $3p+2$37. In that construction, the small-gap one-dimensional 17-AGNR is transformed into a two-dimensional semimetal with closed electron- and hole-like Fermi surfaces (Gillen et al., 11 Apr 2025).

Several recurrent misconceptions are resolved by this broader literature. First, 17-AGNRs are not metallic in realistic isolated-ribbon treatments; rather, they are small-gap semiconductors whose gap depends strongly on the theoretical level and dielectric environment. Second, not every AGNR Raman or confinement study is a 17-AGNR study: the universal LCM work established length metrology for armchair ribbons in general, and the microwave confinement work established a synthetic framework on 6- and 7-AGNRs rather than on 17-AGNRs directly (Overbeck et al., 2019). Third, the most important unresolved practical issue is not structural identity, which is already secure, but simultaneous control of length, alignment, contacts, and thermal budget during device processing (Hwang et al., 15 Jul 2025).

7. Position within the AGNR field

Within the AGNR width series, 17-AGNR occupies a special position because it is both wide enough to reach the narrow-gap regime and still accessible by atomically precise bottom-up synthesis. Freestanding theory places it on the small-gap $3p+2$38 branch; Au-supported spectroscopy shows that substrate screening renormalizes the gap strongly; and GW analysis indicates exceptionally small, nearly symmetric electron and hole effective masses. These features explain why 17-AGNRs were described as having the smallest bandgap and the smallest electron/hole effective mass among bottom-up AGNRs reported at the time of their initial detailed characterization (Yamaguchi et al., 2019).

At the same time, the device literature shows that narrow gap alone does not guarantee transistor-like behavior. In the presently demonstrated short-channel geometries, 17-AGNR transport is contact-limited, room-temperature gate modulation is weak, and thermal post-processing is constrained by substrate-dependent degradation. This suggests that future progress will depend at least as much on interface engineering and alignment control as on the intrinsic band structure of the ribbon itself (Hwang et al., 15 Jul 2025).

Taken together, the literature defines 17-AGNRs as a benchmark narrow-bandgap armchair nanoribbon system: atomically precise in width and edge structure, experimentally accessible on Au(111), spectroscopically identifiable by STM/STS and Raman, transferable to device substrates, and extensible in theory toward chemically modified edges, confined-growth geometries, and semimetallic two-dimensional ribbon networks.