Well-defined sub-nanometer graphene ribbons synthesized inside carbon nanotubes
Hans Kuzmany, Lei Shi, Miles Martinati, Sofie Cambré, Wim Wenseleers, Jenő Kürti, János Koltai, Gergő Kukucska, Kecheng Cao, Ute Kaiser, Takeshi Saito, Thomas Pichler
WWell-defined sub-nanometer graphene ribbons synthe-sized inside carbon nanotubes
Hans Kuzmany ∗ , Lei Shi , ∗ Sofie Cambr´e , Miles Martinati , Wim Wenseleers , Jen˝o K ¨urti ,J´anos Koltai , Gerg˝o Kukuczka , Kecheng Cao , Ute Kaiser , Takeshi Saito , & Thomas Pichler School of Materials Science and Engineering, State Key Laboratory of Optoelectronic Materialsand Technologies, Nanotechnology Research Center, Sun Yat-sen University, Guangzhou 510275,Guangdong, P. R. China Faculty of Physics, University of Vienna, 1090 Wien, Austria Experimental Condensed Matter Physics Laboratory, Physics Department, University of Antwerp,B-2610 Antwerp, Belgium Department of Biological Physics, ELTE E ¨otv ¨os Lor´and University, P´azm´any P´eter stny. 1/A,1117 Budapest, Hungary Central Facility for Electron Microscopy, Electron Microscopy Group of Materials Science, UlmUniversity, Ulm 89081, Germany Nanotube Research Centre, National Institute of Advanced Industrial Science and Technology(AIST), 305-8565 Tsukuba, Japan a r X i v : . [ c ond - m a t . m t r l - s c i ] A ug raphene nanoribbons with sub-nanometer widths are extremely interesting for nanoscaleelectronics and devices as they combine the unusual transport properties of graphene withthe opening of a band gap due to quantum confinement in the lateral dimension. Strongresearch efforts are presently paid to grow such nanoribbons. Here we show the synthesisof 6- and 7-armchair graphene nanoribbons, with widths of 0.61 and 0.74 nm, and excitonicgaps of 1.83 and 2.18 eV, by high-temperature vacuum annealing of ferrocene moleculesinside single-walled carbon nanotubes. The growth of the so-obtained graphene nanoribbonsis evidenced from atomic resolution electron microscopy, while their well-defined structureis identified by a combination of an extensive wavelength-dependent Raman scatteringcharacterization and quantum-chemical calculations. These findings enable a facile andscalable approach leading to the controlled growth and detailed analysis of well-defined sub-nanometer graphene nanoribbons.1 Keywords graphene nanoribbons, electronic structure, Raman scattering, resonance profiles, Albrecht theory,GW calculation 2 Introduction
Sub-nanometer graphene ribbons (graphene nanoribbons, GNRs) are promising structures forfuture electronic devices
1, 2 as they are considered of unifying the unique electronic propertiesof graphene with a reasonably sized gap in their electronic structure. The gap results fromquantum confinement in the lateral direction and consequently scales with the inverse width ofthe ribbons.
4, 5
The ribbons are strips of carbon atoms cut out from a graphene lattice. At presentmost common strips are of the armchair type, i.e., the edge of the ribbons consists of coaxialcarbon pairs oriented parallel to the direction of the ribbon axis (armchair graphene nanoribbons(AGNR)). They are characterized by the number n of such pairs across their width. The electronicstructures of AGNR can be classified into n= 3p, 3p+1 and 3p+2 species, where p is an integer. One of the most commonly investigated structures is the n=7 AGNR (7-AGNR)
6, 7 with bandgaps between 2.1 and 2.3 eV as reported from scanning tunneling spectroscopy (STS) and opticalstudies. More recently ribbons with more complex topologies were grown which have covedzigzag, chevron or chiral type structures.
7, 9, 10
The ribbons are usually grown on Au substratesfrom preselected and properly designed flat poly-aromatic hydrocarbon molecules (PAHs). In thiscase the PAHs are vacuum deposited on the substrate and subsequently transformed to polymericunits with nanoribbon structure. To take advantage of the ribbons grown in this way subsequenttransformation to insulating substrates is necessary. Raman scattering was used to evidence thatthe nanoribbons do not suffer in quality by this transformation.In general the carbons at the edge of the ribbons are saturated by hydrogen but by selecting3pecial PAHs, bandgap engineering , construction of ribbon heterojunctions
16, 17 and ribbonswith unusual electronic properties have been demonstrated where topological properties of theribbons under investigation play an important role. Most recently ribbons became relevant forapplications in photocatalytic hydrogen generation. Besides STS and electron microscopy, Raman scattering has repeatedly been used tocharacterize GNR. Several Raman active vibrational modes were identified to characterize theribbons. Such modes are among others the radial breathing like mode (RBLM), the CH in planebending mode (CH-ipb), the D line, and the graphene G line.
12, 13
The RBLM frequency scaleswith (the square root of) the inverse ribbon width,
20, 21 and is the pendant to the radial breathingmode of the carbon nanotubes.Filling and consecutive chemical reactions inside single-walled carbon nanotubes (SWC-NTs) is a promising technique to grow new nanoscale materials in general.
In addition, theone-dimensional geometry of the CNTs is an excellent template for the controlled growth ofconventional or exotic low-dimensional compounds. . Here, we show that this in-tube synthesiscan be used to grow GNRs with well-defined sub-nanometer width. Growth of such narrow ribbonswith controlled width has previously been limited to the polymerization of specific PAHs on Ausubstrates. Although providing well-defined freestanding ribbons with relatively high yield, theflexibility of preparing GNRs with different widths and structures is limited by the availabilityof appropriate precursor molecules. When synthesizing GNRs inside SWCNTs, the width ofthe GNRs is determined by the diameter of the SWCNTs, hence could in the future be flexibly4uned by starting from CNT samples with different diameters (eventually even diameter/chirality-sorted SWCNTs). Moreover, the hybrid structures of encapsulated ribbons and a semiconductingSWCNT, both with a different band gap can be interesting also from a materials design perspective.The growth of GNRs inside SWCNTs has been demonstrated by first filling the SWC-NTs with flat precursor molecules, such as coronene
29, 30 or other PAHs , and subsequentlytransforming them at elevated temperature. Although these initial results were very promising,growth of high-quality specific types of GNRs with well-defined widths still remained a challenge.In particular the analysis of the electronic properties of the objects inside the tubes remaineddifficult due to the overlapping electronic transitions of the carbon nanotubes themselves. Toavoid this overlap, functionalization was necessary of the otherwise pristine tubes. As it will bedemonstrated below, Raman scattering, in particular in combination with evaluation of resonanceRaman excitation profiles, is an excellent tool to reveal the electronic structure of the objects insidethe tubes. This is due to the double selective nature of the resonance Raman scattering process.It is selective with respect to the geometrical structure of the objects by the vibrational mode butalso selective with respect to the electronic structure by the resonance profile. Small width of theRaman lines indicates well-defined geometrical structures and sharp excitation profiles provideevidence for uniform electronic configuration of the GNRs.In this work, we demonstrate the synthesis of two specific GNRs, i.e., 6- and 7-AGNRs,with well-defined geometrical structure, from the bulky molecule ferrocene (FeCp ). Evidencefor the growth of the AGNRs inside the tubes comes from aberration-corrected high-resolution5ransmission electron microscopy (AC-HRTEM) and from Raman scattering. We also showthat wavelength-dependent Raman spectroscopy gives direct access to the electronic and opticalproperties of the encapsulated objects by the evaluation of resonance Raman excitation profiles.This allows determining experimentally the excitonic gap and even more the electronic structurebeyond the gap. High level first principle calculations provide for the first-time relative intensitiesof the Raman lines of the GNRs by using the Placzek formulation with energy dependentpolarizabilities. The latter procedure is a fundamental progress in the evaluation of Ramanintensities.
The standard feature for the observation of encapsulated ribbons in TEM is an alternating patternof narrow and wide signals from the material inside.
28, 33
It originates from an electron beaminduced twisting of parts of the ribbons. Figure 1 depicts a collection of results from FeCp filled and subsequently transformed SWCNT species as acquired by the SALVE instrument withCs/Cc aberration corrector. It demonstrates the twisting of the graphene ribbons by the electronbeam (panel a-e) and provides images of the ribbons in atomic resolution with correspondingsimulations (panel f-j). The time-dependent AC-HRTEM images show clearly that the observedspecies are flat objects, that twist under the influence of the electron beam, and cannot be identifiedas inner nanotubes. The AC-HRTEM images (panels f-j) provide clear evidence of the atomicstructure of the ribbons, however, also demonstrate that the nanoribbons are very unstable under6he electron beam and defects along the length of the structure are quickly generated by the beam.Therefore, these AC-HRTEM images are not representative for the quality of the as-grown ribbonsand only serve to evidence that the structures are flat ribbons. Evidence for the well-definedelectronic structure of the as-grown ribbons comes from our detailed wavelength-dependent Ramanspectroscopic experiments (see below). More details on the AC-HRTEM are provided in theSupporting Information Section (a). 7igure 1: AC-HRTEM images and corresponding simulations for AGNR@SWCNTs. a-e, Timeseries of images showing a ribbon confined in a SWCNT twisting under 80 keV electron beam. Themodulation of the response with time is evidence for a flat ribbon. f,g, Two typical AC-HRTEMimages in atomic resolution showing a flat 7-AGNR dominated structure confined in a SWNT.Scale bar is 1 nm. The inserted models in red present the corresponding structure of the ribbon,showing also the instability of the ribbon under electron beam irradiation (i.e., f-g are the sameribbon but acquired after different exposure times). h, TEM image simulation for confirming thestructure of 7-AGNR@SWCNT in g. It depicts the tube (18,0), the ribbon, the combination of thetwo and the simulation. The most part of 7-AGNR and the wall of SWNT are AA stacked showingclear graphene structure for the former, while a small mismatch part as indicated by yellow arrowsresults in blurred contrast in the simulated TEM and in the raw TEM image in g. i,j, Intensityprofiles along the yellow line of the recorded pattern i and of the simulated pattern j. The round(outer) tubes exhibit a strong reduction of contrast at the edge whereas the contrast is weaker atthe edge of the flat ribbons. Indicated distances correlate with the diameter of a (18,0) tube andevidence a net ribbon width between 0.64 and 0.69 nm.8 aman Scattering Previously, we reported the observation of a set of Raman lines afterthermal conversion of FeCp filled SWCNT
24, 34 but the origin of these lines remained unclear asthey did not fit to proper model calculations and high resolution TEM was not available. Here weidentified the origin of the Raman lines which turned out to be very sensitive to transformationtemperature. Figure 2a-d depicts this behavior in a plot where Raman intensities are characterizedby a color code as a function of Raman frequency and transformation temperature. Raman spectrawere normalized to the 2D band which is least influenced by the changes induced by the GNRencapsulation. Figure 2a,b depicts the responses measured for 568 nm excitation while Fig. 2c,dpresents those for 633 nm excitation. At low temperature, only the RBMs (around 200 cm − ) andthe G-line of the SWCNTs (around 1600 cm − ) are observed. When increasing the transformationtemperature beyond 500 ◦ C, depending on the excitation laser used two groups of Raman lines canbe observed. They exhibit a maximum Raman intensity between 600 and 700 ◦ C, indicated by thehorizontal white arrows in the figure. As shown further on, these two groups of lines correspondto the in-tube synthesis of AGNRs with different widths.The lower panels of Fig. 2 depict the Raman spectra explicitly. Panels e and f are spectrarecorded at 568 nm (transformation temperature 600 ◦ C) and g and h are spectra recorded at633 nm (transformation temperature 700 ◦ C), for two different Raman frequency regions. Whencomparing the Raman spectra of the ferrocene-filled and temperature-transformed samples (red)with the spectra of filled SWCNTs without transformation (black), the new Raman features appearvery well separated from the response of the SWCNTs in general. In the high-frequency region,the spectra of the SWCNTs can be subtracted straight forwardly, yielding the Raman response9igure 2: Raman scattering for temperature dependent transformation. a-d, Color-code maps forRaman line intensities observed after transformation of FeCp @SWCNTs as a function of Ramanfrequency and transformation temperature. The latter was increased in 50 ◦ C steps. Vertical whitearrows highlight the positions of the main new Raman lines. Horizontal white arrows are locatedat temperatures of maximum Raman response. Raman spectra in a and b and in c and d wereexcited with a yellow laser (568 nm) and red laser (633 nm), respectively. e-h, Raman spectra forFeCp @SWCNT tuned by transformation temperature to optimized response for the two groupsof lines. e, Spectra as observed for yellow laser excitation at 568 nm in the high frequency region,from bottom to top: tubes filled with FeCp (black) and subsequent transformation at 600 ◦ C(red), AGNR contribution by subtracting the Raman signals from the nanotubes (blue, dotted), andcalculation (green) for 7-AGNR. The Raman line marked by a down-arrow was also subtractedsince it originates form 6-AGNR as discussed below. f, Raman signals from low frequency region,from bottom to top: filled tubes, transformed at 600 ◦ C, and calculated for 7-AGNR. The arrowmarks the response from DWCNTs obtained during the transformation process. g,h, Similarspectra as in e,f but for transformation temperatures of 700 ◦ C, recorded with 633 nm laser, and ascalculated for 6-AGNR. 10f the newly synthesized objects (blue dotted spectra in panels e and g). For the low-frequencyregion, a subtraction is more difficult, since during the transformation also inner tubes are formedthereby changing the response of the SWCNTs as well (red arrows). Even though, due to theirhigher frequency, the lines from the RBLMs of the GNRs can be well separated from the responseof the CNTs. The figures demonstrate very narrow line widths of the order of 10 cm − for theresponse of the nanoribbons. This linewidth is equivalent to the Raman response of high-qualityribbons grown on Au substrates and subsequently transferred to semiconducting substrates.
2, 12, 35
Such narrow line widths (in combination with a clear resonance profile) can only be observedfor nanoribbons with a well-defined width along the entire length, as width variations would leadto inhomogeneous broadening of the lines or even disappearance of the RBLM mode. Figure 2also presents the calculated Raman spectra for 7-AGNR species (panels e and f, in green) and for6-AGNR (panels g and h, in green), which were obtained from first principle calculations usingthe dynamical Placzek formalism . More details about these calculations are in the Methodsection and in the Supporting Information Sections (b). They correspond very well to the observedexperimental lines, even in relative intensity. A blown up version of this comparison can befound in the Supporting Information Section (d). For the 7-AGNR, the high-frequency modesare accordingly identified as the CH-ipb mode (1258 cm − ), the D-line (1344 cm − ) and the Gline (1606 cm − ). The low frequency mode at 414 cm − (with small shoulder at 400 cm − ) can beidentified as the RBLM. Likewise, for the 6-AGNR the RBLM is located at 465 cm − , the CH-ipbmode at 1243 cm − , the D-line at 1358 cm − and the G line at 1595 cm − . Note that the latter isstrongly overlapping with the G line of the SWCNTs which prevents its direct determination from11he difference (blue) spectra. However, the frequency can be obtained by measuring the sum of theCH-ipb and the G line of the ribbons which is observed at 2839 cm − . This localizes the G line at1595 cm − .Table 1 presents the experimental Raman frequencies and line widths for both the 6-AGNRand 7-AGNR and compares it to our theoretical values as well as to values for the same modesmeasured on Au-substrates for 7-AGNR. For the 6-AGNR a comparison is made with the onlyreported data originating from the fusion of linear chains of poly-paraphyenlye at 800 K. Theagreement between calculated and experimentally determined Raman frequencies in the high-frequency region is better than 2% and also the agreement between calculated and experimentalrelative line intensities is excellent, except for the CH-ipb of the 6-AGNR which appears tooweak in the calculation. The table highlights the unusually narrow width of the Raman linesfrom the encapsulated species. In three very recent reports, Raman spectra for 7-AGNR grown onAu substrates and subsequently transferred to transparent substrates are shown to exhibit similarnarrow line widths as observed here.
2, 12, 35
Raman Excitation Profiles
Raman excitation profiles were measured for all Raman lines inthe low and in the high frequency region for the excitation wavelength range from 400 - 800 nmwith 5 nm steps. Figure 3a shows overall results in the form of a two-dimensional Raman map fora sample transformed at a temperature of 850 ◦ C. Raman map means Raman intensities are plottedon a color code versus Raman shift and excitation energy. The procedure to create the Raman mapfrom the experiments and to extract the excitation profiles for the various modes by wavelength-12 able 1: Raman lines of 7-AGNR and 6-AGNR; Frequencies ( ω ph ), intensities (weak(w), medium (m), strong (s), very strong (vs)), and linewidths W (FWHM, in parentheses)for the observed Raman lines (column Exp.) as compared to calculation (column Calc.)and to reference (column Ref.). For 7-AGNR and 6-AGNR the latter are from reference and reference , respectively. The column with the calculated frequencies depicts alsothe difference to the experiment in %. The last two columns depict the excited statefrequencies ( ω ∗ ph ) and the Huang-Rhys factor (HR) for the first excited state, both obtainedfrom a fit of the resonances to the Albrecht A-term. All frequencies are given in cm − androunded to integer values.Mode ω ph (W) ω ph ω ph (W) ω ph ∗ HRExp. Calc.;% Ref.
6, 36 from fit from fit(cm − ) (cm − ) (cm − ) (cm − ) (cm − )7-AGNRRBLM 414m (7.5) 395m; 4.5 395m (22.1) 434 0.08CH-ipb 1258s (9.6) 1261s; 0.2 1263s (30) 1195 0.25D 1342vs (10.8) 1345vs; 0.2 1344vs (26) 1275 0.25G 1607s (13) 1580 s; 1.7 1607vs (31) no fit6-AGNRRBLM 466m (8) 457m; 1.7 not reportedCH-ipb 1243vs (9.6) 1236m; 0.6 1245s (100) 1281 0.171272m (10.6) 1272s; 0 1272 0.25D 1358m (11.4) 1352w; 0.4 1315m 1358 0.25G 1595 1568s; 1.7 1590vs no fit − comesmainly from a superposition of the the G line of the nanotubes and the AGNR. The circled signalsoriginate from double-walled carbon nanotubes grown during transformation. They correspondto the weak Raman lines in Fig. 2e,f assigned by the arrows. b, Experimental Raman excitationprofiles derived for the Raman peaks that are indicated by the dashed white lines in panel a arepresented as black squares together with the fitted Raman excitation profiles (in red) for the 6-AGNR (left) and 7-AGNR (right) (See also Supporting Information section (e)). The profilescorrespond to the CH-ipb and D-line, respectively.14ependent fitting of the experimental data, is described in detail in the Supporting InformationSection (e). The features between 1.7 and 2.1 eV excitation in the lower part of the map (R1 andR4) correspond to the 6-AGNR. The upper part has the resonances of the 7-AGNR (R2 and R5).R3 is the resonance of a weak and as yet unknown new line to be discussed below.For a selected number of modes such as the CH-ipb and the D line, experimental resonanceprofiles are plotted in Fig. 3b. The profiles for the 6-AGNR show only two vibronic peaks.In contrast, the resonances for 7-AGNR exhibit a highly structured profile which turned out tooriginate from vibronic sidebands as well as from higher exciton transitions.The red lines in Fig. 3b are fits from the Kramers-Heisenberg-Dirac (KHD) theory forresonance Raman scattering
37, 38 with wave functions in the adiabatic approximation. In this casethe resonance scattering intensity I s is given by I s = CI ω s (cid:88) ρσ | ( α ρσ ) fi | with (1) ( α ρσ ) fi = (cid:88) r (cid:104) f | µ ρ | r (cid:105)(cid:104) r | µ σ | i (cid:105) ω ri − ω L − i γ r where I , ( α ρσ ) fi , µ ρ,σ , ω ri , and γ r are the intensity of the incident light, the transition polarizabil-ity, the dipole moment, the transition energy from state i to state r, and an electronic dampingconstant, respectively. Within the adiabatic approximation for the wave functions and a transitiondipole moment independent of the vibrational normal coordinates (Albrecht A-term ) this resultseventually in the relation ( α ρσ ) fi = A = (cid:88) e,ν e µ e ρ µ e σ ω ri − ω L − i γ r (cid:104) ν f | ν e (cid:105)(cid:104) ν e | ν i (cid:105) . (2)15 e ρ,σ are the ρ , σ components of the pure electronic transition dipole moments which are assumedto be constant. r stands as abbreviation for the transitions to states e, ν e . The right part ofthe expression in Eq. 2 depicts the vibronic matrix elements given by the Frank-Condon (FC)integrals. The vibronic quantum numbers ν i were assumed zero except for i=0. This means allelectrons are in the vibronic ground state. ν f was assumed 1 throughout meaning that only onevibron processes were considered. The experimental results for the 7-AGNR required electronictransitions up to e = 2 , i.e., in solid state terminology E and E were needed for the fit and thusexperimentally determined. The most relevant parameters for the fit are the transition energies tothe first and to the second excited state, the Huang-Rhys factors, and the vibrational frequenciesin these states. Values for the transition energies are depicted in Tab. 2. Excited state frequenciesas they are listed in Tab. 1 were found to be very close to the ground state frequencies. For theresonance of the CH-ipb mode in the case of 7-AGNR the line width in the excited state for thetransition E was considerably smaller than the line width in the ground state. Therefore the peakfor the high energy resonance (outgoing resonance) is higher than the peak for the low energyresonance (incoming resonance). All parameters for the fitted resonances are summarized in theSupporting Information Section (f), Tab. S1. The observed transition energies were comparedto values calculated by solving the Bethe-Salpeter equation within the frame of a quasiparticleself-consistent GW calculation. These calculations go beyond the energy dependent Placzekapproximation as it was used to derive the Raman spectra described in Fig. 2. The calculationis described in detail in Methods and in Supporting Information Section (b). Table 2 lists theobserved transition energies as compared to our calculations and to reported values from optical16eflection measurements . The comparison depicted in the table further evidences the successfulgrowth of 7-AGNR and 6-AGNR inside SWCNT.The two peaks in the resonance for the 6-AGNR represent the transitions to the first and tothe second vibronic level in the excited state, or equivalently, the ingoing and outgoing resonance.Higher vibronic levels are neither observed for the 6-AGNR nor for the 7-AGNR.Calculated transition energies for 5-AGNR are included in the table since the values maybe relevant for the resonance of the Raman line observed at 533 cm − (R3 in Fig. 3) as discussedbelow. The low value of 0.84 eV calculated for the band gap is consistent with the 3p+2 family ofthese ribbons.For the analysis of AGNR@SWCNT the RBLMs are particularly important since theyexhibit a strong and characteristic response in a frequency region which is free from other Ramanlines. In the case of the 7-AGNR@SWCNT the response is unusual since it exhibits severalcomponents. Figure 4a shows a zoomed-in Raman map of region R2. Several peaks can beobserved in this region, which become also evident when plotting the Raman spectra for twodistinct laser excitations (Fig. 4b). The main peak (highest intensity) appears at a vibrationalfrequency of 414 cm − for 569.5 nm excitation with an exceptionally small line width of only7 cm − . In addition, a shoulder can be observed around 400 cm − , which is considerably broader(13 cm − ) and exhibits dispersion, i.e. line position shifts with changing excitation energy. Suchbehavior is well known for the D-line in CNTs but also for conjugated polymers like poly-acetylene. It is an indication for defective structures where vibrational frequencies and electronic17 able 2: Transition energies in eV as observed experimentally from resonance Ramananalysis, compared to calculated values (in parentheses) and to references for AGNR.Experimental values were rounded to two digits behind the decimal point. Line 1 and line3 were obtained for 7-AGNR from high frequency modes and from the RBLM, respectively.Column 5 and 6 are from references. The table also shows calculated transition energiesfor 5-AGNR, together with a value from a reference.Ribbon E E E E E Exp. Exp. Exp. Exp.(Calc.) (Calc.) (Calc.) (Calc.)(eV) (eV) (eV) (eV) (eV)7-AGNR 2.18 2.45 2.1 Ref. (2.37) (2.75) (3.03) (1.91) (2.3)7RBLM 2.18 2.436-AGNR 1.83 1.69 Ref. (1.82) (3.02) (3.24)5-AGNR (0.84) (2.19) (3.02) 0.1 Ref. Also, modified edges resulting in sample inhomogeneity can lead to slightly modifiedRBLM-modes and thus result in photo-selective resonance scattering. In addition to the shoulder avery weak Raman signal is observed at 375 cm − for 506.1 nm excitation and at shorter wavelengthof 460 nm a Raman line at 417 cm − . At even shorter wavelengths of 405 nm excitation a Ramanline is observed at 533 cm − . (The latter two are not shown in the figure.) The origin of these lineswill be discussed below and in some more detail in Supporting Informations Section (e).The strong line at 414 cm − is most appropriate to analyze resonance profiles. This profileis depicted in the Fig. 4c. It exhibits two strong resonances and a weak resonance at 2.7 eV, whichis not statistically significant and originates from a differentRaman frequency (see SupportingInformation (e)). Due to the low mode energy for the RBLM(51 meV) the two resonances donot explicitly show vibronic splitting. Fitting the strong resonances with the Albrecht A termyields the transition energies listed in Tab. 2. The resonances have very similar energies as theywere observed for the high frequency modes of the 7-AGNR and are therefore assigned to E andE . RBLMs for AGNR are known to depend on the width w of the ribbons. In several reportsthis dependence was observed to follow an / √ w behavior . A more recent work showed that theRBLM was found to scale with 1/w. Within the small frequency range one cannot discriminate19igure 4: Raman scattering for the RBLM of 7-AGNR. a, Blown up Raman map for the frequencyregion between 350 and 440 cm − ; b, Radial breathing like mode for 7-AGNR as excited with laserlines at 2,45 (blue) and 2,18 eV (red). Full drawn lines are fits from the experimental analysis. c,Resonance profile for the main peak of the RBLM at 414 cm − (dots, with errorbars) togetherwith the calculated values (red line); d, RBLMs frequencies versus ribbon width w . Blue starsare the experimental results from this work. Black squares are calculated values. The dashed linecorresponds to Vanderscuren et al. who found a relation of the form ω RBLM = a/ √ w − b withempirical parameters a and b for the dependence of the RBLM frequency on w , the ribbon widthwithout hydrogen. The dashed line in the figure was evaluated for a and b equal to 527.4 and 210.2,respectively. Red dots are for the additional Raman lines at 533 and 375 cm − if they are assignedto 5-AGNR and 8AGNR, respectively. The value for n = 9 is from a reference. − and 375 cm − fit also very well to the diagram if they are assigned to RBLMs for 5-AGNR and8-AGNR. Recently the RBLM for 5-AGNR was observed at 533 cm − .
12, 43
However, since bothribbons are narrow gap species of the type 3p+2, E resonances or even higher transitions must beinvolved for their observation. Interestingly for 8-AGNR very recently an effective gap of 2.3 eVwas reported from STS measurements. This is reasonably close to the 2.45 eV laser used to detectthe Raman line at 375 cm − . The Raman line at 533 cm − was observed in our case for excitationwith a considerable higher energy than the value calculated for the E transition. Shortening ofthe ribbons with respect to infinite long species could be the reason for the enhanced transitionenergies
41, 45 as they are observed here.
When using polyaromatic hydrocarbons as precursor molecules (such as coronene
29, 30 ), thegrowth of GNRs inside CNTs can be explained by an oligomerization reaction. In our case,however, and similarly as previously reported for tetrathiofulvalene molecules inside CNTs, which means 40 C/nm can be provided.A 7-AGNR needs 33.3 C/nm, hence providing sufficient carbon atoms to form the graphenenanoribbons and thus a high concentration of ribbons can be expected. We indeed observe veryhigh Raman intensities from the AGNRs compared to the Raman signatures of the surroundingCNTs, however, the Raman cross-section of encapsulated AGNRs with respect those of CNTs isnot yet known. Hence, one cannot use the Raman intensities to estimate the synthesis yield.The TEM analysis gives clear evidence for flat objects with ribbon structure inside the tubes.22he deviation from a perfect graphenic ribbon structure observed in the figure is a consequence ofthe electron beam irradiation rather than being intrinsic to the ribbon growth. The identificationof the ribbons comes from wavelength-dependent Raman scattering and comparison to high levelquantum-chemical calculations. The very narrow Raman lines as depicted in Tab. 1 and Fig. 2are evidence for clean and highly unperturbed ribbon material. For the 7-AGNR the experimentaland theoretical Raman spectra exhibit unprecedented agreement with respect to frequency of themodes and the relative Raman intensities.The very sharp and individual peaks of the resonance profiles are evidence for a well-definededge structure, which does not allow for defect induced modulation and consequently broadeningor splitting of the optical transitions. Conclusion
The results presented above demonstrate that graphene nanoribbons can be grown fromthermal decomposition of ferrocene encapsulated in carbon nanotubes. Aberration corrected highresolution transmission electron microscopy provides evidence for the ribbons inside the tubes.Raman scattering combined with first principle calculations and resonance excitation analysis isan excellent tool to identify the structure of GNR even if the ribbons are encapsulated in SWCNTs.Due to its nondestructive nature and frequency selective character the Raman method is superiorto any other optical method such as optical absorption or luminescence spectroscopy. In additionthe method allows determining the electronic structures of the ribbons, even beyond the HOMO-LUMO gap. The results provide a challenge that ribbons with larger or smaller width than those23eported here, can be grown in larger or smaller SWCNTs, respectively, and can be detected withRaman scattering. They thus open a new field in subnanometer graphene ribbon research.
SWCNTs as grown by the eDIPS technique with mean diameteraround 1.3 nm were used as a starting material. All tubes were purified by first etching in air andsubsequent treating with HCl as described previously to remove amorphous carbon and catalyticparticles from tube growth. SWCNT bucky paper was obtained by washing and filtering the tubeswith distilled water and ethanol.For the filling process the tubes were first opened by etching in air at 420 ◦ C and then exposedto ferrocene at 400 ◦ C for two days in previously evacuated quartz tubes. The transformation ofthe FeCp molecules to the AGNRs inside the tubes was performed by vacuum annealing forseveral days and at various temperatures as described in the main text. This process resultedin the appearance of new lines in the Raman spectra from some so far unknown objects.Control experiments were performed without opening the tubes but otherwise treating the materialidentically. In this case almost no new Raman lines were observed as depicted explicitly in theSupporting Information Section (c). This can be considered as evidence for the growth of theobjects inside the tubes. Analysis by Transmission Electron Microscopy
To obtain information on the grownobjects TEM investigations were performed with a Thermo Fisher (formerly FEI) TITAN G2 80-2400 (Fig. 1a-e) at 80 kV. Figure 1f, g and i are acquired by the specific Thermo Fisher (formerlyFEI) SALVE (subangstrom low voltage electron microscopy) transmission electron microscopefitted with a CEOS CETCOR spherical aberration corrector (axial 5th order, off-axial 3rd order), achromatic aberration corrector, and a Thermo Fisher Ceta 4K CMOS camera. The exposure timeof the images is 1.0 s and the dose rate is 6.85*10 electrons s − nm − . Theoretical evaluation vibrational frequencies and mode-specific Raman intensities
Vibrational frequencies of the ribbons were calculated at the Γ point using the Vienna Ab initioSimulation Package (VASP). Raman intensities I s were evaluated using the frequency dependentPlaczek approximation. In this case the intensity of the Raman lines is proportional to the square ofthe derivative of the frequency dependent polarizability with respect to the phonon normal mode: I s ( ω s , ω L ) = ω s ω L (cid:88) ρ,σ (cid:12)(cid:12)(cid:12)(cid:12) ∂α ρ,σ ( ω L ) ∂Q ph (cid:12)(cid:12)(cid:12)(cid:12) Γ( ω − ω ph )( n ( ω ph ) + 1) (3)where α ρ,σ ( ω L ) is the dynamic (frequency dependent) polarizability tensor evaluated from firstprinciple calculations at the laser energy ω L . Calculation of the dynamic polarizability was donewithin the linear response theory with wave functions and structural parameters used during thefrequency calculation and geometrical optimization. Q ph and ω ph are the phonon normal modesand frequencies, respectively, and ω s is the frequency of the scattered light. ω is the differencebetween ω L and ω s . n ( ω ph ) is the Bose-Einstein distribution at room temperature and Γ( x ) is anormalized Lorentzian function with full width at half maximum of 10 cm − . Numeric derivativesof the polarizability tensor were calculated using symmetric derivatives by manually shifting theatoms according to normal modes in both positive and negative directions. From this the following25xpression can be obtained for the derivative of the polarizability: ∂α ρ,σ ( ω L ) ∂Q ph = (cid:88) r,s (cid:104) f | H e − p | s (cid:105)(cid:104) s | H e − ph | r (cid:105)(cid:104) r | H e − p | i (cid:105) ( ω L − ω r − i γ e − p )( ω r − ω s − i γ e − ph ) (4)where r,s are intermediate virtual electronic states with energy ω r,s obtained from first principlecalculations, H e − ph (derivative of the electron-ion potential with respect to the normal modes) and H e − p are the Hamiltonians for the electron-phonon and electron-photon coupling, respectively and γ e − ph and γ e − p are the corresponding damping constants (life times). Details of the derivation canbe found in ref. . The calculation was successfully used previously for the analysis of dopingand strain induced changes of the Raman spectra and gap structure of MoS and silicene. Optical excitation energies were calculated by solving the Bethe-Salpeter equation withinthe frame of a quasiparticle self-consistent GW calculation as built in the QUESTAAL code with ladder diagram corrections. Excitation energies were determined by taking the energyvalues at the maximum of the peaks present in the imaginary part of the macroscopic dielectricfunction. Explicitly, calculations were performed for the low gap 3p+2 ribbon 5-AGNR, for themoderate gap 3p ribbon 6-AGNR, and for the large gap 3p+1 ribbon 7-AGNR. More details on thecalculation are in supporting Information Section (b).
Raman scattering
Raman spectra were excited at room temperature and at ambientconditions with various lasers in the visible spectral range. Spectra were recorded with a Dilorxy800 and with a Labram HR800 microscope. In Figure 2 spectra with a spectral resolution of lessthan 2 wavenumbers. Thus the observed linewidths are intrinsic to the ribbon material. Ramanintensities were normalized to the 2D line of the carbon nanotubes, since the 2D-line is known to26e highly independent from defects. In the case of resonance analysis normalization was performedto the fundamental Raman line of Si.
Evaluation of Excitation Profile
Resonance Raman spectra were evaluated for wavelengthdependent excitation in the range between 400 and 800 nm with 5 nm step-size. To cover therequested energy range for excitation the following three laser systems were used. 400-526 nm:A frequency-doubled Ti:Sa laser (Msquared SolsTis external cavity frequency doubled ECD-Xmodule) which was pumped by an 18 W Sprout-G diode pumped solid state laser (532 nm). 534-605nm and 610-690 nm: A dye laser (spectra Physics model 375) pumped by an Ar+ ion laser(Spectra physics 2020) equipped with either Rhodamine 110 or DCM laser dyes. 690-800 nm:A tunable Ti:Sa laser (Spectra Physics 3900S). The Raman spectra were recorded with a highresolution triple grating Dilor XY800 spectrometer and a liquid nitrogen cooled CCD detector.More details about the recording and evaluation of the Raman spectra can be obtained fromSupporting Information Section (e).
Work supported by the NSFC (51902353), the FWF project P21333-N20, and the NKFIH, GrantNo. K-115608. L.S. acknowledges the financial support from the Natural Science Foundationof Guangdong Province (Grant No. 2019A1515011227) and the Sun Yat-Sen University (GrantNo. 29000-18841218, 29000-31610028) J.K., J.K. and G.K. further acknowledge the [NIIF] forawarding access to resource based in Hungary at Debrecen, they further acknowledge support27y the National Research Development and Innovation Office of Hungary within the QuantumTechnology National Excellence Program (Project No. 2017-1.2.1-NKP-2017-00001), and theELTE Excellence Program (1783-3/2018/FEKUTSTRAT) supported by the Hungarian Ministryof Human Capacities. K.C. acknowledges the China Scholarship Council (CSC) for financialsupport. U.K. acknowledges the support of the Graphene Flagship and DFG SPP Graphene as wellas the DFG and the Ministry of Science, Research and the Arts (MWK) of Baden-Wuerttembergwithin the frame of the SALVE project. M.M., S.C., and W.W. acknowledge funding from theFund for Scientific Research Flanders (FWO projects No. G040011N, G02112N, G035918N,G036618N and the EOS-charming project G0F6218N [EOS-ID 30467715] ). M.M. acknowledgesfunding of a DOCPRO4 PhD scholarship from the UAntwerp research fund (BOF) and S.C. alsoacknowledges funding from the European Research Council Starting Grant No. 679841.
H.K. and L.S. contributed equally to this work. H.K. and L.S. designed and supervised theexperiments. L.S. prepared the samples and did the characterization with spontaneous Ramanscattering in Vienna with lasers at wavelength of 633 and 568 nm. S.C., M.M., and W.W.performed the wavelength-dependent resonance Raman scattering experiments and their analysis.H.K. analyzed the resonance profiles. K.C. and U.K. performed HRTEM characterization andsimulations. J.K., J.K., and G.K. performed the first principles DFT calculations. T.S. providedthe SWCNTs. T.P. provided the laboratory facilities and the Raman setup in Vienna. All authorsdiscussed the results and commented on the manuscript at all stages.28
Competing Financial Interests
The authors declare that they have no competing financial interests associated to the publication ofthis manuscript.
Corresponding Author [email protected], [email protected]
References
1. Bennett, P. B. et al. Bottom-up graphene nanoribbon field-effect transistors. Appl. Phys. Lett. , 253114 (2013).2. Martini, L. Graphene nanoribbon devices. Carbon , 36–42 (2019).3. Novoselov, K. S. et al. Electric field effect in atomically thin carbon films. Science ,666–669 (2004).4. Son, Y. W., Cohen, M. L. & Louie, S. G. Energy gaps in graphene nanoribbons.Phys. Rev. Lett. , 216803 (2006).5. Barone, V., Hod, O. & Scuseria, G. E. Electronic structure and stability of semiconductinggraphene nanoribbons. Nano Lett. , 2748–2754 (2006).6. Cai, J. et al. Atomically precise bottom-up fabrication of graphene nanoribbons. Nature ,470–473 (2010). 29. Talirz, L., Ruffieux, P. & Fasel, R. On-surface synthesis of atomically precise graphenenanoribbons. Adv. Mater. , 6222–6231 (2016).8. Denk, R. et al. Exciton-dominated optical response of ultra-narrow graphene nanoribbons.Nat. Commun. , 4253 (2014).9. Durr, R. A. et al. Orbitally matched edge-doping in graphene nanoribbons. J. Am. Chem. Soc. , 807–813 (2018).10. Sun, Q. et al. Massive dirac fermion behavior in a low bandgapgraphene nanoribbon near atopological phase boundary. Adv. Mater. , 1906054 (2020).11. Gigli, L. et al. Detachment dynamics of graphene nanoribbons on gold. ACS Nano , 689–697 (2019).12. Overbeck, J. et al. Optimized substrates and measurement approaches for Raman spectroscopyof graphene nanoribbons. Phys. Stat. Sol. B 1900343 (2019).13. Verbitskiy, I. A. et al. Raman fingerprints of atomically precise graphene nanoribbons.Nano Lett. , 3442–3447 (2016).14. Pedramrazi, Z. et al. Concentration dependence of dopant electronic structure in bottom-upgraphene nanoribbons. Nano Lett. , 3550–3556 (2018).15. Narita, A., Chen, Z. P., Chen, Q. & Mullen, K. Solution and on-surface synthesisof structurally defined graphene nanoribbons as a new family of semiconductors.Chemical Science , 964–975 (2019). 306. Chen, Y. C. et al. Molecular bandgap engineering of bottom-up synthesized graphenenanoribbon heterojunctions. Nat. Nanotechnol. , 156–160 (2015).17. Cao, T., Zhao, F. & Louie, S. G. Topological phases in graphene nanoribbons: junction states,spin centers, and quantum spin chains. Phys. Rev. Lett. , 076401 (2017).18. Groning, O. et al. Engineering of robust topological quantum phases in graphene nanoribbons.Nature , 209–213 (2018).19. Akilimali, R. et al. Graphene nanoribbon-tio2-quantum dots hybrid photoanode to boost theperformance of photoelectrochemical for hydrogen generation. Catal. Today , 161–169(2020).20. Vandescuren, M., Hermet, P., Meunier, V., Henrard, L. & Lambin, P. Theoretical study of thevibrational edge modes in graphene nanoribbons. Phys. Rev. B , 195401 (2008).21. Liu, D., Daniels, C., Meunier, V., Every, A. G. & Tom´anek, D. In-plane breathing and shearmodes in low-dimensional nanostructures. Carbon , 364–370 (2020).22. Lim, H. E. et al. Growth of carbon nanotubes via twisted graphene nanoribbons.Nature Comm. , 2548 (2013).23. Khlobystov, A. N. Carbon nanotubes: From nano test tube to nano-reactor. ACS Nano ,9306–9312 (2011).24. Kuzmany, H. et al. The growth of new extended carbon nanophases from ferrocene insidesingle-walled carbon nanotubes. Phys. Stat. Sol. RRL , 1700158 (2017).315. Kitaura, R. et al. High-yield synthesis of ultrathin metal nanowires in carbon nanotubes.Angew. Chem. Int. Ed. , 8298–8302 (2009).26. Shi, L. et al. Confined linear carbon chains as a route to bulk carbyne. Nature Mater. ,634–639 (2016).27. Pham, T. et al. Torsional instability in the single-chain limit of a transition metaltrichalcogenide. Science , 263–266 (2018).28. Chamberlain, T. W. et al. Size, structure, and helical twist of graphene nanoribbons controlledby confinement in carbon nanotubes. ACS Nano , 3943–3953 (2012).29. Chernov, A. I. et al. Optical properties of graphene nanoribbons encapsulated in single-walledcarbon nanotubes. ACS Nano , 6346–6353 (2013).30. Anoshkin, I. V. et al. Coronene encapsulation in single-walled carbon nanotubes: Stackedcolumns, peapods, and nanoribbons. Chemphyschem , 1660–1665 (2014).31. Lim, H. E. et al. Fabrication and optical probing of highly extended, ultrathin graphenenanoribbons in carbon nanotubes. ACS Nano , 5034–5040 (2015).32. Kukucska, G., Z ´olyomi, V. & Koltai, J. Frequency-dependent ab initio resonance ramanspectroscopy. arXiv preprint arXiv:1903.00253 (2019).33. Chuvilin, A. et al. Self-assembly of a sulphur-terminated graphene nanoribbon within a single-walled carbon nanotube. Nat. Mater. , 687–692 (2011).324. Kuzmany, H., Shi, L. & Pichler, T. The origin of nondispersive raman lines in the d-bandregion for ferrocene@hipco swcnts transformed at high temperatures. Phys. Stat. Sol. B ,2530 – 2535 (2015).35. Borin Barin, G. et al. Substrate transfer and ex situ characterization of on-surface synthesizedgraphene nanoribbons. ACS Appl. Nano Mater. , 2184–2192 (2019).36. Basagni, A. et al. Molecules-oligomers-nanowires-graphene nanoribbons: A bottom-upstepwise on-surface covalent synthesis preserving long-range order. J. Am. Chem. Soc. ,1802–1808 (2015).37. Kiefer, W. Nonlinear raman spectroscopy infrared and raman spectroscopy ed b schrader(1995).38. Haroz, E. H. et al. Resonant Raman spectroscopy of armchair carbon nanotubes: Absence ofbroad G- feature. Phys. Rev. B , 121403 (2011).39. Tang, J. & Albrecht, A. Developments in the theories of vibrational raman intensities. InRaman Spectroscopy, 33–68 (Springer, 1970).40. Merino-Diez, N. et al. Width-dependent band gap in armchair graphene nanoribbons revealsfermi level pinning on Au(111). ACS Nano , 11661–11668 (2017).41. Kimouche, A. et al. Ultra-narrow metallic armchair graphene nanoribbons. Nat. Commun. ,10177 (2015). 332. Kuzmany, H., Imhoff, E. A., Fitchen, D. B. & Sarhangi, A. Frank-condon approach for optical-absorption and resonance Raman-scattering in trans-polyacetylene. Phys. Rev. B , 7109–7112 (1982).43. Chen, Z. P. et al. Lateral fusion of chemical vapor deposited N=5 armchair graphenenanoribbons. J. Am. Chem. Soc. , 9483–9486 (2017).44. Sun, K. & Ji, P. On-surface synthesis of 8- and 10-armchair graphene nanoribbons. Small ,1804526 (2019).45. Talirz, L., Sde, H. & Kawai, S. Band gap of atomically precise graphene nanoribbons as afunction of ribbon length and termination. Chem. Phys. Chem. , 2348–2353 (2019).46. Plank, W. et al. Electronic structure of carbon nanotubes with ultrahigh curvature. ACS Nano , 4515–4522 (2010).47. Saito, T. et al. Selective diameter control of single-walled carbon nanotubes in the gas-phasesynthesis. J. Nanosci. Nanotechn. , 6153–6157 (2008).48. Kuzmany, H. et al. Nondispersive raman lines in the d-band region for ferrocene functionalizedcarbon nanotubes. Phys. Stat. Sol. B , 2457 – 2460 (2014).49. Kresse, G. & Furthmuller, J. Efficient iterative schemes for ab-initio total-energy calculationsusing a plane-wave basis set. Phys. Rev. B , 11169–11186 (1996).50. Gajdos, M., Hummer, K., Kresse, G., Furthmuller, J. & Bechstedt, F. Linear optical propertiesin the projector-augmented wave methodology. Phys. Rev. B , 045112 (2006).341. Pet˝o, J. et al. Moderate strain induced indirect bandgap and conduction electrons in mos 2single layers. npj 2D Mater. Appl. , 1–6 (2019).52. Kukucska, G., Zolyomi, V. & Koltai, J. Characterization of epitaxial silicene with Ramanspectroscopy. Phys. Rev. B , 075437 (2018).53. Kotani, T. & van Schilfgaarde, M. Quasiparticle self-consistent gw method: A basis for theindependent-particle approximation. Phys. Rev. B , 165106 (2007).54. Cunningham, B., Gruning, M., Azarhoosh, P., Pashov, D. & van Schilfgaarde, M. Effect ofladder diagrams on optical absorption spectra in a quasiparticle self-consistent gw framework.Phys. Rev. Materials2