# Fileset

[s41699-022-00356-y.pdf](https://mdr.nims.go.jp/filesets/f3590cc3-53f2-41bf-b715-60424ad5389d/download)

## Creator

Muhammad Awais Aslam, Tuan Hoang Tran, Antonio Supina, Olivier Siri, Vincent Meunier, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Marko Kralj, Christian Teichert, Evgeniya Sheremet, Raul D. Rodriguez, Aleksandar Matković

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Single-crystalline nanoribbon network field effect transistors from arbitrary two-dimensional materials](https://mdr.nims.go.jp/datasets/23d5a1ba-6968-41fc-b77c-9b21305d3556)

## Fulltext

Single-crystalline nanoribbon network field effect transistors from arbitrary two-dimensional materialsARTICLE OPENSingle-crystalline nanoribbon network field effect transistorsfrom arbitrary two-dimensional materialsMuhammad Awais Aslam 1✉, Tuan Hoang Tran 2, Antonio Supina3, Olivier Siri4, Vincent Meunier5, Kenji Watanabe 6,Takashi Taniguchi 7, Marko Kralj 3, Christian Teichert1, Evgeniya Sheremet2, Raul D. Rodriguez2 and Aleksandar Matković 1✉The last decade has seen a flurry of studies related to graphene nanoribbons owing to their potential applications in the quantumrealm. However, little experimental work has been reported towards nanoribbons of other 2D materials. Here, we propose auniversal approach to synthesize high-quality networks of nanoribbons from arbitrary 2D materials while maintaining highcrystallinity, narrow size distribution, and straightforward device integrability. The wide applicability of this technique isdemonstrated by fabricating molybednum disulphide, tungsten disulphide, tungsten diselenide, and graphene nanoribbon fieldeffect transistors that inherently do not suffer from interconnection resistance. By relying on self-aligning organic nanostructures asmasks, we demonstrate the possibility of controlling the predominant crystallographic direction of the nanoribbon’s edges.Electrical characterization shows record mobilities and very high ON currents despite extreme width scaling. Lastly, we exploredecoration of nanoribbon edges with plasmonic particles paving the way for nanoribbon-based opto-electronic devices.npj 2D Materials and Applications            (2022) 6:76 ; https://doi.org/10.1038/s41699-022-00356-yINTRODUCTIONThe successful synthesis of graphene nanoribbons (NRs)1,2 andtheir implementation in devices3,4 has brought them at theforefront as building blocks for information processing inquantum and classical electronics5. Graphene NRs enable variousfunctionalities including tunable band gap, high current carryingcapability, long mean free path, localized spin and topologicaledge states5. Similarly, other two-dimensional (2D) material NRscan display edge specific properties such as ferromagnetism6,7,efficient catalysis8,9, and enhanced sensing abilities3,10. Moreover,a recent study about MoS2 NRs demonstrated their potential forspintronics and quantum transport11. The development of 2Dmaterial NRs is largely driven by the needs in nanoelectronics,where three-dimensional (3D) gate-all-around architectures thatemploy nanotubes, nanorods, or NRs are considered as the likelysolution to the arising scaling challenges12–14.Despite all the possibilities that 2D material-based NRs hold,their sufficient quality, narrow widths, density, controlled edges,and high yield remain as technological challenges for realisticapplications. The most widely used preparation methods arebottom-up chemical synthesis15–18 and top-down lithogra-phy19–21. Chemical synthesis of NRs offers precise edge controland even allows synthesis of nanoporous graphene (NPG) ribbonswith widths down to 1 nm. Such NPG systems are very appealingdue to their ability to sieve as well as induce semiconductingbehaviour in graphene22,23. However, bottom-up synthesis routesfocus almost exclusively on graphene NRs, facing significantobstacles to develop more complex 2D materials or NR hetero-structures. Moreover, the device channels suffer from electricalpercolation issues and high junction (node) resistance3. Whereas,top-down lithography-based approaches do not offer a straightforward control over NR’s alignment with respect to highsymmetry directions of the 2D material. They also cause interfacecontamination thereby degrading device performance andoperation24.Recently, Aljarb et al.25 demonstrated a technique based onvicinal growth to fabricate NRs of arbitrary 2D materials e.g.,(TMDCs). Although, the NRs produced by this approach are singlecrystals, the ribbon widths are rather large and non-uniform. Thegrowth of these ribbons is also dependent on specific substrate,requiring an additional transfer step for their integration25. VapourLiquid Solid (VLS) growth is another interesting method utilisedfor the NR synthesis11,26. Despite the fact that VLS uses silicondioxide (SiO2) as a substrate, it employs salt and metal precursorsthat can be detrimental for device integration.In this work we tackle the outlined challenges and demonstratea universal method to fabricate NRs of arbitrary 2D materials,including graphene, hexagonal boron nitride (hBN), transitionmetal dichalcogenides TMDCs, and nanoribbon heterostructureswith a width ranging from 6 to 100 nm. Our approach is based onepitaxially-grown organic needle-like nanostructures which selfassemble along high-symmetry directions of 2D materials. Weexploited these organic nano-needles as a mask through which 2Dmaterials could be etched by oxygen plasma. This results incrystalline nanoribbon-networks (NRNs) with high edge-to-surfaceratio and controlled predominant crystallographic edge-directions.To investigate the electrical performance of our NRNs and to showa challenging technological application, field effect transistors(FETs) were directly fabricated on Si/SiO2 and Si/SiO2/hBNsubstrates. Besides their inherent single-crystalline nature, NRN-FETs were obtained without any additional transfer steps. TMDCNRN-based devices show outstanding electrical properties,including WS2 and WSe2 nanoribbons. We also observed ferro-electric switching for graphene NR devices due to water1Institute of Physics, Montanuniversität Leoben, Franz Josef Strasse 18, 8700 Leoben, Austria. 2Tomsk Polytechnic University, Lenina ave. 30, 634034 Tomsk, Russia. 3Center forAdvanced Laser Techniques, Institute of Physics, Bijenička cesta 46, 10000 Zagreb, Croatia. 4Aix Marseille University CNRS CINaM UMR 7325, Campus de Luminy 13288, Marseillecedex 09, France. 5Department of Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic Institute, Troy, NY 12180, USA. 6Research Center for Functional Materials,National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan. 7International Center for Materials Nanoarchitectonics, National Institute for Materials Science, 1-1Namiki, Tsukuba 305-0044, Japan. ✉email: muhammad.aslam@unileoben.ac.at; aleksandar.matkovic@unileoben.ac.atwww.nature.com/npj2dmaterialsPublished in partnership with FCT NOVA with the support of E-MRS1234567890():,;http://crossmark.crossref.org/dialog/?doi=10.1038/s41699-022-00356-y&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41699-022-00356-y&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41699-022-00356-y&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41699-022-00356-y&domain=pdfhttp://orcid.org/0000-0002-1178-0722http://orcid.org/0000-0002-1178-0722http://orcid.org/0000-0002-1178-0722http://orcid.org/0000-0002-1178-0722http://orcid.org/0000-0002-1178-0722http://orcid.org/0000-0003-2116-8390http://orcid.org/0000-0003-2116-8390http://orcid.org/0000-0003-2116-8390http://orcid.org/0000-0003-2116-8390http://orcid.org/0000-0003-2116-8390http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-9786-3130http://orcid.org/0000-0002-9786-3130http://orcid.org/0000-0002-9786-3130http://orcid.org/0000-0002-9786-3130http://orcid.org/0000-0002-9786-3130http://orcid.org/0000-0001-8072-6220http://orcid.org/0000-0001-8072-6220http://orcid.org/0000-0001-8072-6220http://orcid.org/0000-0001-8072-6220http://orcid.org/0000-0001-8072-6220https://doi.org/10.1038/s41699-022-00356-ymailto:muhammad.aslam@unileoben.ac.atmailto:aleksandar.matkovic@unileoben.ac.atwww.nature.com/npj2dmaterialsadsorption at the ribbon edges27. Our proposed method allowsribbons which do not suffer from high node resistance betweeninterconnecting NRs, as the networks are ‘carved out’ from singlecrystals. To confirm this, we employ in operando Kelvin probeforce microscopy (KPFM). To demonstrate our methods scalabilityand ultimate control over the NR edge-direction, we havefabricated predominant armchair and zigzag NRNs from a large-area MoS2 ML obtained by chemical vapor deposited (CVD). Suchhigh edge-density in nanoribbon network field effect transistors(NRN-FETs) has potential benefits in sensing applications and intunable catalytic devices, especially when considering catalyticedge reactivity of MoS228,29. To go one step further, we illustrateedge-specific decoration of NRNs with silver nanoparticles,creating mixed-dimensional plasmonic heterostructures.RESULTSFabrication of NRNsFigure 1a–c depict the proposed NRN fabrication pathway.Detailed steps for MoS2 and graphene are given in SupplementaryFig. 1. Typically, NRs of about 10–30 nm width distribution wererealized as shown in Fig. 1d. By further optimization of the growthtime, etching time, or selection of organic nano-structures basedon shorter back bone (e.g., bi-phenylenes), the mean NRN widthcan be changed. The tunability of the mean width has beendemonstrated by changing the growth time and is presented inSupplementary Fig. 2.Organic molecules grow on 2D materials via van der Waals(vdW) epitaxy30. The molecules at the interface with the 2Dmaterial substrate adopt a flat-lying orientation and align theirπ-networks to optimize vdW interaction with the substrate31.Consequently, the molecules at the interface are ‘locked’ intopreferential adsorption sites on the substrate and the growingcrystallites adopt rotational commensuration with their 2Dmaterial support. This provides an inherent self-alignment withthe substrates’ high symmetry directions, i.e., armchair or zig-zag32–34.After organic nanostructure growth, the hybrid organic/2Dmaterial stacks were precisely etched to form NRNs via exposureto oxygen plasma, i.e. reactive ion etching (RIE). An etch rate of~1 layer in 3 s was established for graphene and TMDCs. Uponetching, the remaining organic molecules can either be left asan encapsulation layer35 or removed by rinsing in chloroform oralso by vacuum annealing. Our method allows for fabrication ofmonolithic NRNs of different exfoliated and CVD 2D materials, asdemonstrated for graphene, hBN, MoS2, WS2 and WSe2Fig.1(e–j). Apart from individual 2D materials, in Fig. 1i we showNRN-heterostructures consisting of vertically stacked monolayerWS2 (n-type) and bilayer WSe2 (p-type), thus enabling atomicallythin p-n junctions.The structural integrity of NRs was probed by Raman andphotoluminescence (PL) spectroscopies. Results for graphene NRsshow extremely low values of (Intensity of the D peak/Intensity ofthe G peak) ID/IG peak ratios in the Raman spectra after etching ofthe flake into NRN (Supplementary Fig. 3 and SupplementaryTable 1). Similarly, for ML MoS2 (Supplementary Fig. 4) there are nodefect activated peaks, which indicates high crystallinity of thefabricated nanoribbons. Figure 2a, b show results for the NR pairs(considering 3 layer MoS2) with ~70∘ relative inclination. Theyexhibited a prominent difference in the intensities of E12g and A1gFig. 1 Fabrication of 2D material NRNs. a–c Schematic representation of the key fabrication steps. a Organic nanostructures self-assemblyand self-alignment. b After reactive ion etching. c NRN after removal of the sacrificial organic layer. d Histogram of NR widths for agraphene NRN shown in e. e–i Optical micrographs of various 2D material NRNs, presenting respectively NRNs of graphene, hBN, MoS2,WS2, and WS2/WSe2 heterostructure. j AFM topography image of a NRN from chemical vaopour deposition (CVD) Monolayer(ML) MoS2(22 × 22 μm2, z-scale 15 nm).M.A. Aslam et al.2npj 2D Materials and Applications (2022)    76 Published in partnership with FCT NOVA with the support of E-MRS1234567890():,;Raman active modes. Such anisotropy for Raman modes of MoS2NRs36 and MoS2 flakes37 was previously observed by changing thepolarization configuration. The intensity of a Raman mode isproportional to the dot product of the Raman tensor with the lightpolarization. Since the NRs of both directions are etched out of thesame flake, their crystallographic orientation, and consequentlythe dot products are the same. Thus, this anisotropy cannot beexplained by the selection rules per se. Raman spectra for rotatednanoribbons and flake are shown in (Supplementary Fig. 5).Anisotropy has an apparent effect on the relative A1g modeintensity that warrants further investigation into its variations innarrow NRs. Furthermore, WS2 NRs exhibited a dominant excitonpeak and a suppressed trion peak in the PL measurements. This isdue to a reduction in the free electron density over small widths(<30 nm) caused by the predominant oxygen terminatededges38,39. An example of the PL spectrum of WS2 is presentedin Fig. 2(c), comparing the initial ML flake and the resulting NR.The Raman spectra for the nanoribbon WS2/WSe2 heterostructureis provided in the (Supplementary Fig. 6).Predominant crystallographic orientations of nanoribbonsTo demonstrate that our proposed method offers control of NRspredominant crystallographic orientation, two different organicmolecules—parahexaphenyl (6P) and dihydrotetraazaheptacene(DHTA7)—were grown epitaxially on ML MoS2 obtained by CVD.As their phenylene (6P) and acene (DHTA7) backbones could beseen as armchair and zig-zag motifs, respectively, and uponadsorption, molecular backbones will align with the correspond-ing high-symmetry directions of the 2D material substrate34,40,41.The control over the orientation of the predominant NR directioncan be verified by using triangular CVD MoS2 flakes that terminatewith zig-zag edges due to the growth kinetics42–44. Figure 3a, bcompares the NR directions and the triangular MoS2 flake-edgedirections, presenting 2D Fast Fourier Transform (2D-FFT) analysisof the atomic force microscopy (AFM) topography images(corresponding insets). In the case of 6P masks (Fig. 3a),predominant NR directions are tilted by (8.5 ± 0.4)∘ from the edgedirections, i.e., NR edges are close to parallel with the zigzagcrystallographic direction. By altering the backbone of themolecular mask (the case of DHTA7—Fig. 3b) the NR edgeschange predominantly following the armchair crystallographicdirection. Moreover, employing other molecular species couldallow controlling this angle for a particular 2D material of interest,and to exploit orientation specific properties for 2D materials45,46in the one-dimensional NR-regime.TMDC-NRN field effect transistorsField-modulation of the NRNs was tested by fabricatingtwo-terminal field-effect transistors (FETs) and investigating theirtransfer characteristics source-drain current vs source gate voltageID(VSG) at 77 K and 300 K. Van der Waals graphite electrodes wereemployed to probe electrical response of the devices betweeneach step of the fabrication. Our proposed NRN fabricationmethod is compatible with conventional two-dimensional fieldeffect transistor (2D-FET) fabrication schemes (as mask-lithographyor e-beam lithography) since the 2D material films can bepatterned into NRNs prior to the electrode fabrication. Figure 4apresents the scheme of the device geometry, and an opticalmicrograph for one of the WS2-NRN-FETs in the inset.Figure 4b–d provide typical semi-logarithmic transfer curves at300 K for flakes with organic nanostructures for WS2, MoS2, andWSe2, respectively and after patterning them into NRs. For WSe210–15 nm hBN was used for bottom capping and NRN waspatterned on top of it. MoS2 and WS2 devices exhibit an n-typebehaviour whereas WSe2 exhibited an ambipolar behaviour, bothbefore and after patterning of the flakes into NRNs. A shift in thepositive VSG direction of the ID(VSG) curves was observed after NRNformation, indicating p-type doping by the RIE process. Onaverage, ID(VSG) curves for MoS2 devices showed a positive shift of40 V after the NRN formation. This large positive shift can beattributed to electron depletion by the oxygen terminatingNR-edges47.The devices exhibited exceptional transfer characteristics evenwhen SiO2 was used as the gate dielectric. The performance ofNRN-FETs could be further enhanced by employing high-Kdielectric materials such as HfO248. An increase in the hysteresiswas observed in the NR devices (except WSe2) which can be dueto two possible mechanisms including high density of edgesfacilitating charge trapping/de-trapping mechanisms from adosr-bates20 or increased capacitive gating effect-traps due to SiO249.To investigate the origin of hysteresis the devices were annealedin vacuum at 400K which lead to reduction of hysteresis due toremoval of adsorbates on the edges50. However, a full closure ofthe hysteresis was noted by performing low temperaturemeasurements (77 K) as shown in (Supplementary Fig. 7) whichpoints towards capacitive gating as the possible reason49.Compared to the unetched flakes, the NRN-devices experienceda decrease of the ID, as a consequence of the severely reducedchannel widths when compared to the original 2D material-FET47.An increase in ID and mobilities was observed when NRN-FETs areFig. 2 Evidence of structural integrity of the NRs. a Raman spectraof the characteristic vibrational modes of MoS2 showing the tri-layerflake before patterning, and two ribbons with ≈ 70∘ of relativeinclination. The spectra are normalized with respect to the E12g-modeintensity, and relative changes of the A1g mode are indicated byhorizontal dashed lines and red arrows. b Raman intensity maps of E12g and A1g modes of two intersecting NRs (scale-bar: 0.5 μm),highlighting a distinct variation of A1g mode as the ribbon directionis changed. The direction of linearly polarized light is indicated byblack arrows. This is in agreement with the published result that therelative intensity of the A1g to E12g peaks is strongly dependent onthe orientation angle of the material’s crystallographic axes, for afixed in-plane polarization60. c PL spectra of a ML WS2 flake andcorresponding NR, highlighting exciton and trion components.M.A. Aslam et al.3Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2022)    76 measured at low temperature (77 K). These observations confirmthe presence of band transport in NRNs, as at low temperaturesthe contribution from phonons are minimized (Supplementary Fig.8).Figure 4e summarizes the apparent linear electron mobilities (μ)obtained from the ID(VSG) curves measured at 77 K and calculatedboth by parallel and fringing capacitance models. Mobility plotsare shown in Supplementary Fig. 8. The commonly used parallelcapacitance model overestimates the mobilities when applied toNRs as their widths are much smaller compared to the oxidethickness11. Therefore, taking into account the capacitance perunit area for the fringing capacitance model51, a more realisticarea-specific gate capacitance can be expressed as:Cox � εoxε0πln 6 toxW þ 1� �� �Wþ 1tox( )(1)Fig. 4 TMDC-NRN-FETs. a Schematic diagram of a NRN-based FET utilizing vdW graphite electrodes. Inset of a shows an optical micrograph ofa 3L WS2 FET (scale-bar 5 μm). b–d Semi-logarithmic transfer curves of WS2, MoS2, and hBN-WSe2 FETs before and after patterning the flakesinto NRNs. e Parallel and fringing capacitance apparent linear electron mobilities at 77 K, for the devices presented in b–d.Fig. 3 Orientation control of nanoribbons. a (scale-bar: 5 μm) CVD MoS2 NRN with ribbons nearly parallel to the flake edges (zigzagdirection), using 6P molecules for the self-assembled mask. Inset shows 2D-FFT analysis of the NR directions with respect to the triangularflake edges. 2D Fast Fourier Transform (2D-FFT) image is rotated by 90∘ to represent real-space directions. b (scale-bar: 1 μm) Similar to a onlyusing DHTA7 molecules for the self-assembled mask, and resulting in NRs in armchair direction (perpendicular to flake edges).M.A. Aslam et al.4npj 2D Materials and Applications (2022)    76 Published in partnership with FCT NOVA with the support of E-MRSwhere tox is the oxide thickness and W is channel width. Effectivechannel length and width were estimated considering parallel andserial connections of NRs for each particular NRN-FET.Graphene NRN-FETs and edge-induced ferroelectric effectBesides very high mobilities ≈ 1000–1200 cm2V−1s−1 (using thefringing capacitance model) graphene-based NRN devices exhibitpronounced hysteresis in the ID(VSG), see Fig. 5(a). The observedhysteresis ΔCNP (difference between two CNPs) was not presentin the original flakes, nor is introduced by the deposited organicnano-structures. The effect appears after the RIE in oxygen plasmaonce the samples are exposed to the ambient environment. Thisexposure allows the attachment of water molecules from the air tothe oxygenated edges of the nanoribbons. The hysteresis remainsin high-vacuum, at low temperatures Fig. 5b, and is practicallyindependent of the VSG sweep-rates Fig. 5c. Similar effect waspredicted for edge-adsorbed water molecules and was observedfor graphene-FETs with oxygen-plasma etched edges27. Theorientation of water molecules can be changed due to the torqueinduced by the external electric field. The total field experiencedby the graphene NRN is a sum of the gate-bias induced field andthe net field produced by the edge-adsorbed water dipoles,yielding a robust bi-modal—ferroelectric—behavior of grapheneNRN-FETs.To rule out any causes of hysteresis due to trap-states thedevices were measured in high vacuum (10−7 mbar), after vacuumannealing (at 410K for over 90 min), and were subjected to lowtemperature (77 K) measurements. In all cases, the observedhysteresis was preserved. As the trapping is sensitive totemperature a significant quenching would occur at lowtemperature52,53. This was observed for TMDC-based NRN-FETs,where the bi-stable states of the adsorbed water molecules at theribbon edges are not expected. For graphene NRN-FETs at 77 Kthe hysteresis of the transfer curves is only slightly reduced asshown in Fig. 5b. In addition, VSG sweep-rate dependentmeasurements were carried out both at 300 K and at 77 K. Figure5c presents the VSG sweep-rate dependence of the ΔCNP,representing the negligible difference in the CNP position forthe forward and the backward sweeps. This further helps usexcluding any contributions from capacitive gating which acts onseconds time scale52. Lastly, to identify the temperature requiredfor water dissociation from graphene NR edges the devices wereannealed for various temperatures (373 K, 473 K and 573 K) undervacuum conditions. A large reduction of the hysteresis wasobserved after annealing the devices at 573 K (Supplementary Fig.9) whereas no significant changes to the hysteresis were observedfor lower temperature. A shift of the CNP towards negative VSGwas also noted. This is a direct indication of water removal whichotherwise causes a p-type doping of graphene54. Our results pointto the induced ferroelectric effect in oxygen-terminated graphenenanoribbon-FETs, which is very similar and more robust thanobserved previously for the oxygen-terminated flake-edges27.While ferroelectric-graphene nanoribbons and their integrationinto heterostructures are very interesting and promising pathwaysfor future nanoelectronics, optoelectronics, neuromorphic electro-nics, and sensing applications, such research is beyond the scopeof this study. To directly probe the resistivity of the nodesbetween the adjacent NRs and the potential drops across theNRN-FET channel in operando frequency modulated (FM) KPFMwas performed on graphene NRN-FETs. Figure 5d presents aFig. 5 Graphene NRN-FETs. a Length and width scaled transfer curves of the bi-layer graphene flake after organic nanostructure growth, andof the corresponding NRN-FET after annealing. Forward and reverse sweep direction is indicated by the arrows. b Length and width scaledtransfer curves for graphene NRN-FET at 77 K, demonstrating that bi-modal switching persists at low temperature. c Difference in the forwardand reverse bias charge neutrality point (CNP) positions as a function of the VSG sweeping rate. d in operando FM-KPFM image of a grapheneNRN-FET (scale-bar: 1 μm). Potential profile lines from the ribbons indicated by (1) and (2a–c) in sub-panel d are presented in e andf, respectively.M.A. Aslam et al.5Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2022)    76 contact potential difference (CPD) map across a grapheneNRN-FET during operation.To highlight the relevant potential drops across the channel,cross-sections marked in Fig. 5d are provided in Fig. 5(e, f). Thetransitions from the electrodes to the channel do not introduceany significant potential drops, as seen earlier for gold electrodeson both graphene and MoS255,56. Graphene NR labeled (1)interconnects between the source and drain electrodes, it exhibitsan almost perfectly linear potential drop across the 10 μm longchannel Fig. 5e. NRs labeled (2a–c) form a parallel connection toNR (1). No potential drop was observed at the nodes between (2a)-(2b) and (2b)–(2c), as indicated in Fig. 5f. Further, many moreribbons that do not bridge the source and drain electrodes,maintain a constant potential as these are not part of the currentflow across the device. Above all, the consistent potentialobserved at all nodes is due to the translation of single-crystallinity of the original 2D material into the nanoribbonnetwork. By not suffering from high node resistance, most of thepotential drop of the provided VSD bias is utilized for the transportalong the NRs in the channel (Fig. 5d–f). As a consequence, ourNRN FETs exhibit high on state currents and allow for high currentmodulation by the gate.Decoration of nanoribbons with plasmonic particlesTo demonstrate high edge-to-surface ratio of our 2D materialnanoribbon networks, we investigated edge-specific decoration ofthe NRs by metallic nanoparticles (NPs). Figure 6a schematicallypresents the decoration process. The details are provided in themethods section. The NR edges induce selective nucleation of Agnanoparticles via the photo-activated reduction of Ag ions at theedges via electron transfer from graphene. Figure 6b, c present agraphene NRN before and after decoration with Ag NPs.Edge-specific decoration of 2D materials with metallic NPs wasalready demonstrated57, and utilizing NRN enhances the benefitsof these hybrid systems.To investigate the photocatalytic activity of the edge-decoratedNPs, we use 4-nitrobenzenethiol (4-NBT) as a model for photo-catalysis experiment. This is shown in Fig. 6d. The photocatalyticconversion of 4-NBT to p,p'-dimercaptoazobenzene (DMAB) hasbeen intensively investigated58. Both Ag NP-decorated and baregraphene NRs were exposed to the solution of 4-NBT, and theresulting Raman spectra are shown in Fig. 5e. Without the NPsonly the G-mode of the graphene NRs can be observed. However,edge decorated NPs not only enable surface-enhanced ramanspectroscopy (SERS) signal, but also induce the desired photo-catalytic reaction of 4-NBT into DMAB, as evident from theappearance of the DMAB characteristic Raman mode at~ 1440 cm−1 and ~ 1390 cm−1.These Raman modes are relatedto ag16 and ag17 vibrations of N=N of DMAB58. Such aphotocatalytic reaction on nanoribbons decorated with plasmonicnanoparticles shows its future potential towards photocatalyticapplications. In further studies, we will focus on employing 2Dmaterial-based NRN-FETs combined with edge-specific decoratedplasmonic NPs, gaining an additional ‘knob’ via gate biasing. Suchcoupled mixed-dimensional plasmonic systems can be utilised ingate-controlled photocatalytic reactions, tunable SERS sensors,and high-sensitivity optoelectronic devices, however such experi-ments would go beyond the scope of this study.DISCUSSIONWe propose a method to fabricate nanoribbon networks startingfrom arbitrary 2D materials, including WS2 and WSe2 NRs, andtheir heterostructures, which were not demonstrated until now.The method allows achieving NR widths below 20 nm while alsoenabling a straight-forward integration of the 2D material basedNRNs into high-performance FETs with high yields of nanoribbon(Supplementary Table 2 and Supplementary Fig. 10). Further, withthe appropriate choice of the self-aligned molecular masks controlof the NR direction with respect to the crystallographic high-symmetry directions is achieved. Examined TMDC nanoribbonFig. 6 Decoration of nanoribbon edges. a Illustration of the process for the decoration of the NRNs with metallic nanoparticles, and (b, c)graphene NRN before and after edge-decoration with Ag NPs (scale bars 500 nm, z-scales 25 nm). d Dimerization of 4-NBT to DMAB and (e)demonstration of the photocatalytic activity and (surface-enhanced raman spectroscopy) SERS capability of the hybrid graphene-NRN+ Ag-NPsystem.M.A. Aslam et al.6npj 2D Materials and Applications (2022)    76 Published in partnership with FCT NOVA with the support of E-MRSnetwork FETs exhibit band transport, maintain high carriermobility values, clear off-states, high ON-state currents, andmaintain stable operation over a large number of sweeping cycles(Supplementary Fig. 11).By bridging between top-down and bottom-up approaches, ourmethod provides high-quality NR connections (nodes) that do notact as scattering centers (high resistivity points), as proven by inoperando KPFM of graphene-NRN FETs. Further, using graphene-NRN FETs we show bi-modal switching of the transfer curveswhich has been theoretically predicted, and thus far demonstratedonly for graphene edges27. By Raman spectroscopy we haveobserved that MoS2 ribbons with the different growth directionsexhibit Raman anisotropy36. Our method facilitates both highcrystallinity and large-area coverage without the high resistanceissues of the adjacent nanoribbon nodes. In comparison, thebottom up approaches usually suffer from percolation and noderesistance issues3, while the top-down approaches tend tointroduce defects24 in the ribbons yielding lower crystallinity.Lastly, the high edge-to-surface ratios of our NRNs allowed us toselectively decorate the edges with plasmonic nanoparticles.These hybrid mixed-dimensional systems can provide a platformfor next generation optoelectronic and plamsonic sensing devicesdue to the flexibility provided by our method for size tuning of thenanoribbons and the applicability of the process to heterostruc-tures and vertical 2D material p-n junctions.METHODS2D materials, organic masks, and device fabricationFlakes of 2D materials were mechanically exfoliated from bulkcrystal and transferred onto a 300 nm SiO2/Si substrate usingcommercially available Nitto tape and polydimethylsiloxane (Gel-Pak-DLG-X4). Monolayer and few layer flake thicknesses wereidentified via optical contrast, PL, and Raman measurements.Graphite flakes 10–50 nm thick kish graphite) were thentransferred on 2D materials as electrodes to make devicechannels. 6P and DHTA7 nanostructures were grown on devices/flakes by hot wall epitaxy. The growth procedures were adoptedfrom refs. 34,41. MoS2 triangular flakes were grown from solution-based CVD at atmospheric pressure similarly to the procedure inref. 59. The liquid Mo precursors used were NaMo and AHM in 1:1ratio and dissolved in ultra-pure water at concentration of200 ppm.Reactive ion etchingThe reactive ion etching process was developed using an OxfordPlasma 80 plus RIE system. For all devices the forward power waskept at 80W with an oxygen flow of 50 sccm under a pressure of40mTorr. Etching time was optimized according to the thicknessof 2D materials.Electrical characterizationElectrical characterization of the flake- and NRN-FETs were doneusing Keithley 2636A Source-Meter attached to the Instec probestation. The samples were contacted with Au coated Ti electricalcantilever microprobes. Low temperature electrical measurementswere performed using liquid nitrogen on a silver plate for thermaluniformity. The temperatures were monitored via mK2000temperature controller connected to the probe station with atemperature resolution of 0.01 K.AFM and FM-KPFM MeasurementsAFM and FM-KPFM measurements were performed using Horiba/AIST-NT Omegascope AFM system. Aseylec probes were employed(spring constant ~ 42N/m, resonant frequency ~ 70 kHz, tip radiusbelow 30 nm). For width measurements ‘Nanosensors’ probes wereused (spring constant of 10–130N/m, resonant frequency ~ 300 kHzand tip radius of 2 nm). For in-operando FM-KPFM experiments, thegraphene-NRN-FETs were controlled by a Keithley 2636A sharing thesame ground with the KPFM-setup. FM-KPFM measurements werecarried out in a two-pass mode, with the probe lifted by 12 nm in thesecond pass. Topography and CPD images were processed in theopen-source software Gwyddion v2.56. For topography images zero-order line filtering was applied and leveling of the base plane. ForCPD images only zero order line filtering was applied.Micro-PL MeasurementsAll micro-PL and Raman measurements were performed using aHoriba LabRam HR Evolution confocal Raman spectrometer using600 lines/mm and 1800 lines/mm gratings. A 532 nm laser sourcewas used to excite the samples with an excitation power of0.1–3.2 mW. The laser spot was focused by a 100 × , 0.9 NAobjective.NP edge-decoration and photocatalysis experimentsAg deposition on graphene nanoribbons was carried out byphoto-deposition method. 10 μl of 1 mM AgNO3 was dropped onthe sample. Thin glass was placed on top of the sample for ease offinding the region of interest. Red laser (633 nm) and objective100x were used to irradiate the sample. Laser power, laserscanning speed and area were optimized to control the size of AgNPs. For photocatalytic experiment, 0,1 mM 4-NBT with water toethanol ratio 50:50 was prepared. Nanoribbon networks decoratedwith Ag NPs was immersed in this solutions overnight. After that,sample was washed and Raman spectra were recorded with NT-MDT Raman spectroscope.DATA AVAILABILITYCorrespondence and the requests for the data and/or materials should be addressedto Aleksandar Matković.Received: 11 June 2022; Accepted: 14 October 2022;REFERENCES1. Li, X., Wang, X., Zhang, L., Lee, S. & Dai, H. Chemically derived, ultrasmoothgraphene nanoribbon semiconductors. Science 319, 1229–1232 (2008).2. Cai, J. et al. Graphene nanoribbon heterojunctions. Nat. Nanotechnol. 9, 896–900(2014).3. Chen, Z., Narita, A. & Müllen, K. Graphene nanoribbons: on-surface synthesis andintegration into electronic devices. Adv. Mater. 32, 2001893 (2020).4. Saraswat, V., Jacobberger, R. M. & Arnold, M. S. Material science challenges tographene nanoribbons. ACS Nano 15, 3674–3708 (2021).5. Wang, H. et al. Graphene nanoribbons for quantum electronics. Nat. Rev. Phys. 3,791–802 (2021).6. Pan, H. & Zhang, Y. W. Edge-dependent structural, electronic and magneticproperties of MoS2 nanoribbons. J. Mater. Chem. 22, 7280–7290 (2012).7. Slota, M. et al. Magnetic edge states and coherent manipulation of graphenenanoribbons. Nature 557, 691–695 (2018).8. Lin, J. et al. Enhanced electrocatalysis for hydrogen evolution reactions from WS2nanoribbons. Adv. Energy Mater. 4, 1301875 (2014).9. Karunadasa, H. I. et al. A molecular MoS2 edge site mimic for catalytic hydrogengeneration. Science 335, 698–702 (2012).10. Lihter, M. et al. Electrochemical functionalization of selectively addressed MoS2nanoribbons for sensor device fabrication. ACS Appl. Nano Mater. 4, 1076–1084(2021).11. Li, X. et al. Nickel particle enabled width-controlled growth of bilayer molybde-num disulfide nanoribbons. Sci. Adv. 7, eabk1892 (2021).12. Thomas, S. Gate-all-around transistors stack up. Nat. Electron. 3, 728–728(2020).13. Chen, M. L. et al. A FinFET with one atomic layer channel. Nat. Commun. 11, 1–7(2020).M.A. Aslam et al.7Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2022)    76 14. Xu, L. Can carbon nanotube transistors Be scaled down to the sub-5 nm gatelength? ACS Appl. Mater. Interfaces 13, 31957–31967 (2021).15. Jia, X., Campos-Delgado, J., Terrones, M., Meunier, V. & Dresselhaus, M. S. Gra-phene edges: a review of their fabrication and characterization. Nanoscale 2,86–95 (2011).16. Barone, V., Hod, O. & Scuseria, G. E. Electronic structure and stability of semi-conducting graphene nanoribbons. Nano Lett. 6, 2748–2754 (2006).17. Jolly, A., Miao, D., Daigle, M. & Morin, J. F. Emerging bottom-up strategies for thesynthesis of graphene nanoribbons and related structures. Angew. Chem. 132,4652–4661 (2020).18. Bennett, P. B. et al. Bottom-up graphene nanoribbon field-effect transistors. Appl.Phys. Lett. 103, 253114 (2013).19. Han, M. Y., Özyilmaz, B., Zhang, Y. & Kim, P. Energy band-gap engineering ofgraphene nanoribbons. Phy. Rev. Lett. 98, 206805 (2007).20. Chen, S. et al. Monolayer MoS2 nanoribbon transistors fabricated by scanningprobe lithography. Nano Lett. 19, 2092–2098 (2019).21. Tapasztó, L., Dobrik, G., Lambin, P. & Biro, L. P. Tailoring the atomic structure ofgraphene nanoribbons by scanning tunnelling microscope lithography. Nat.Nanotechnol. 3, 397–401 (2008).22. Moreno, C. et al. On-surface synthesis of superlattice arrays of ultra-long gra-phene nanoribbons. Chem. Commun. 54, 9402–9405 (2018).23. Moreno, C. et al. Bottom-up synthesis of multifunctional nanoporous graphene.Science 360, 199–203 (2018).24. Zhou, Y. & Loh, K. P. Making patterns on graphene. Adv. Mater. 22, 3615–3620(2010).25. Aljarb, A. Ledge-directed epitaxy of continuously self-aligned single-crystallinenanoribbons of transition metal dichalcogenides. Nat. Mater. 19, 1300–1206(2020).26. SLi, S. Vapour-liquid-solid growth of monolayer MoS2 nanoribbons. Nat. Mater.17, 535-542, (2018).27. Caridad, J. M. et al. A graphene-edge ferroelectric molecular switch. Nano Lett.18,4675–4683 (2018).28. Salazar, N., Rangarajan, S., Rodríguez-Fernández, J., Mavrikakis, M. & Lauritsen, J. V.Site-dependent reactivity of MoS2 nanoparticles in hydrodesulfurization of thio-phene. Nat. Commun. 11, 1–9 (2020).29. Wang, H. et al. Structural and electronic optimization of MoS2 edges for hydrogenevolution. J. Am. Chem. Soc. 141, 18578–18584 (2019).30. Koma, A. Van der Waals epitaxy for highly lattice-mismatched systems. J. Cryst.Growth 201, 236–241 (1999).31. Hlawacek, G., Khokhar, F. S., van Gastel, R., Poelsema, B. & Teichert, C. Smoothgrowth of organic semiconductor films on graphene for high-efficiency electro-nics. Nano Lett. 11, 333–337 (2011).32. Matković, A. et al. Epitaxy of highly ordered organic semiconductor crystallitenetworks supported by hexagonal boron nitride. Sci. Rep. 6, 1–10 (2016).33. Wang, J. Probing the crystallographic orientation of two-dimensional atomiccrystals with supramolecular self-assembly. Nat. Commun. 8, 1–8 (2017).34. Vasić, B. et al. Molecules on rails friction anisotropy and preferential slidingdirections of organic nanocrystallites on two-dimensional materials. Nanoscale10, 18835–18845 (2018).35. Ha, T. J., Akinwande, D. & Dodabalapur, A. Hybrid graphene/organic semi-conductor field-effect transistors. Appl. Phys. Lett. 101, 033309 (2012).36. Wu, J. B. et al. Monolayer molybdenum disulfide nanoribbons with high opticalanisotropy. Adv. Opt. Mater. 4, 756–762 (2016).37. Ji, J. et al. Giant magneto-optical Raman effect in a layered transition metalcompound. PNAS 113, 2349–2353 (2016).38. Cong, C., Shang, J., Wang, Y. & Yu, T. Optical properties of 2D semiconductor WS2.Adv. Opt. Mater. 6, 1700767 (2018).39. Kwon, Y., Kim, K., Kim, W., Ryu, S. & Cheong, H. Variation of photoluminescencespectral line shape of monolayer WS2. Curr. Appl. Phys. 18, 941–945 (2012).40. Kratzer, M., Matkovic, A. & Teichert, C. Adsorption and epitaxial growth of smallorganic semiconductors on hexagonal boron nitride. J. Phys. D: Appl. Phys. 52,383001 (2019).41. Matković, A. et al. Light assisted charge propagation in networks of organicsemiconductor crystallites on hexagonal boron nitride. Adv. Funct. Mater. 29,1903816 (2019).42. Van Der Zande, A. M. et al. Grains and grain boundaries in highly crystallinemonolayer molybdenum disulphide. Nat. Mater. 12, 554–561 (2013).43. Zhu, D. et al. Capture the growth kinetics of CVD growth of two-dimensionalMoS2. npj 2D Mater. Appl. 1, 1–8 (2017).44. Zhou, W. et al. Intrinsic structural defects in monolayer molybdenum disulfide.Nano Lett. 13, 2615–2622 (2013).45. Li, F. et al. Strain improving the performance of a flexible monolayer MoS2photodetector. Adv. Electron. Mater. 12, 1900803 (2019).46. Arab, A. & Li, Q. Anisotropic thermoelectric behavior in armchair and zigzag mono-and fewlayer MoS2 in thermoelectric generator application. Sci. Rep. 5, 1–12 (2015).47. Liu, H., Gu, J. & Peide, D. Y. MoS2 nanoribbon transistors transition from depletionmode to enhancement mode by channel-width trimming. IEEE Electron DeviceLett. 33, 1273–1275 (2012).48. Lee, G. H. et al. Flexible and transparent MoS2 field-effect transistors on hex-agonal boron nitride-graphene heterostructures. ACS Nano 7, 7931–7936 (2013).49. Park, Y., Baac, H. W., Heo, J. & Yoo, G. Thermally activated trap charges responsiblefor hysteresis in multilayer MoS2 field-effect transistors. App. Phys. Lett. 108,083102 (2016).50. Kaushik, N. et al. Reversible hysteresis inversion in MoS2 field effect transistors.npj 2D Mater. Appl. 1, 1–9 (2017).51. Liao, A. D. et al. Thermally limited current carrying ability. Phys. Rev. Lett. 106,256801 (2011).52. Wang, H., Wu, Y., Cong, C., Shang, J. & Yu, T. Hysteresis of electronic transport ingraphene transistors. ACS Nano 4, 7221–7228 (2010).53. Singh, A. K. & Gupta, A. K. Reversible control of doping in graphene-on-SiO2 bycooling under gate-voltage. J. Appl. Phys. 122, 195305 (2017).54. Melios, C., Giusca, C. E., Panchal, V. & Kazakova, O. Water on graphene review ofrecent progress. 2D Mater. 5, 022001 (2018).55. Yu, Y. J. et al. Tuning the graphene work function by electric field effect. NanoLett. 9, 3430–3434 (2009).56. Matković, A. et al. Interfacial band engineering of MoS2/Gold interfaces usingpyrimidine-containing self-assembled monolayers toward contact-resistance-freebottom-contacts. Adv. Electron. Mater. 6, 2000110 (2020).57. Tanaka, N., Nishikiori, H., Kubota, S., Endo, M. & Fujii, T. Photochemical depositionof Ag nanoparticles on multiwalled carbon nanotubes. Carbon 47, 2752–2754(2009).58. Dong, B., Fang, Y., Xia, L., Xu, H. & Sun, M. Is 4-nitrobenzenethiol converted to p,p'-dimercaptoazobenzene or 4-aminothiophenol by surface photochemistryreaction? J. Raman Spectrosc. 42, 1205–1206. (2011).59. Marion, I. D. et al. Atomic-scale defects and electronic properties of a transferredsynthesized MoS2 monolayer. Nanotechnology 29, 305703 (2018).60. Liang, L. & Meunier, V. First-principles Raman spectra of MoS2, WS2 and theirheterostructures. Nanoscale 6, 5394–5401 (2014).ACKNOWLEDGEMENTSThe authors would like to thank Prof. Roman Gorbachev from the University ofManchester and Prof. Jose0 Manuel Caridad from University of Salamanca for theiruseful input in improving the manuscript. This work is supported by the AustrianScience Fund (FWF) under grants no. I4323-N36 and Y1298-N, and by the RussianFoundation for Basic Research under the project no. 19-52-14006. K.W. and T.T.acknowledge support from the JSPS KAKENHI (Grant Numbers 19H05790, 20H00354and 21H05233). A.S. and M.K. acknowledge support from the European RegionalDevelopment Fund for the “Center of Excellence for Advanced Materials and SensingDevices” (No. KK.01.1.1.01.0001). Also, the bilateral Croatian-Austrian project fundedby Croatian Ministry of Science and Education and the Centre for InternationalCooperation and Mobility (ICM) of the Austrian Agency for International Cooperationin Education and Research (OeAD-GmbH) under project HR 02/2020 is acknowl-edged. Further, the bilateral French-Austrian project funded by the Ministère de laRecherche et des Nouvelles Technologies (Amadeus PHC under project no. 42333PL,France) and the Centre for International Cooperation and Mobility (ICM) of theAustrian Agency for International Cooperation in Education and Research (OeAD-GmbH) under project FR 12/2019 is acknowledged.AUTHOR CONTRIBUTIONSM.A.A., supervised by A.M., prepared the samples, carried out experiments and dataanalysis, with exception of NP decoration and related experiments which werecarried out by T.H.T. under supervision of E.S. M.A.A. and A.M. prepared the figures.M.A.A, A.M., and R.D.R. wrote the manuscript. T.H.T., E.S., and R.D.R. interpreted theresults related to NP decoration. V.M. and E.S. provided interpretations and supportfor the Raman spectroscopy experiments. K.W. and T.T. provided hBN crystals. O.S.provided DHTA7 molecular source. A.S. and M.A.A. prepared CVD MoS2 samplesunder supervision of M.K. C.T., with M.A.A. and A.M. interpreted AFM related data. C.T.assisted in the final MS preparation. A.M. proposed the concept of NRN fabricationand with R.D.R. acquired the main source of funding for the study. All the authorsdiscussed the results and reviewed the manuscript.COMPETING INTERESTSThe authors declare no competing interests.M.A. Aslam et al.8npj 2D Materials and Applications (2022)    76 Published in partnership with FCT NOVA with the support of E-MRSADDITIONAL INFORMATIONSupplementary information The online version contains supplementary materialavailable at https://doi.org/10.1038/s41699-022-00356-y.Correspondence and requests for materials should be addressed to MuhammadAwais Aslam or Aleksandar Matković.Reprints and permission information is available at http://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard to jurisdictional claimsin published maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in anymedium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the CreativeCommons license, and indicate if changes were made. The images or other third partymaterial in this article are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is not included in thearticle’s Creative Commons license and your intended use is not permitted by statutoryregulation or exceeds the permitted use, you will need to obtain permission directlyfrom the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2022M.A. Aslam et al.9Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2022)    76 https://doi.org/10.1038/s41699-022-00356-yhttp://www.nature.com/reprintshttp://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/ Single-crystalline nanoribbon network field effect transistors from arbitrary two-dimensional materials Introduction Results Fabrication of NRNs Predominant crystallographic orientations of nanoribbons TMDC-NRN field effect transistors Graphene NRN-FETs and edge-induced ferroelectric effect Decoration of nanoribbons with plasmonic particles Discussion Methods 2D materials, organic masks, and device fabrication Reactive ion etching Electrical characterization AFM and FM-KPFM Measurements Micro-PL Measurements NP edge-decoration and photocatalysis experiments DATA AVAILABILITY References Acknowledgements Author contributions Competing interests ADDITIONAL INFORMATION