Room‐Temperature Transport Properties of Graphene with Defects Derived from Oxo‐Graphene

Abstract In recent years, graphene oxide has been considered as a soluble precursor of graphene for electronic applications. However, the performance lags behind that of graphene due to lattice defects. Here, the relation between the density of defects in the range of 0.2 % and 1.5 % and the transport properties is quantitatively studied. Therefore, the related flakes of monolayers of graphene were prepared from oxo‐functionalized graphene (oxo‐G). The morphologic structure of oxo‐G was imaged by atomic force microscopy (AFM) and scanning tunneling microscopy (STM). Field‐effect mobility values were determined to range between 0.3 cm2 V−1 s−1 and 33.2 cm2 V−1 s−1, which were inversely proportional to the density of defects. These results provide the first quantitative description of the density of defects and transport properties, which plays an important role for potential applications.

graphene with various densities of defects (termeda sG D with density if defectsb etween 0.2 %a nd up to 1.5 %d etermined by the relation to I D /I G ratios, compare FigureS1, Supporting Information) obtainedb yw et-chemical preparation andr eduction with hydroiodic acid (HI). [20a] Surface morphologies of the samples were measured by atomicf orce microscopy (AFM) in tappingm ode. As shown in Figure 2a,t he average heighto f G 0% is determined to about 0.5 nm with lateral dimensions of 10-20 mm.C ontaminants are visible at the edge of the G 0% sheet, whichs tems from the used tape during the exfoliation and transfer processes. In contrast, the roughnesso fG D is with 1.0 nm twice as high as for G 0% due to bitopico xo-groups at the carbon basal plane and possible adsorbates (Figure 2b). The lateral size of the depicted G D flake was determined to be about 20 mm. To obtain information on the local atomic structure, scanning tunneling microscopy (STM) was used. Comparing the atomically-resolved honeycomb structure for defectfree highly ordered pyrolytic graphite (HOPG)i nF igure 2c,t he single-layer G D reveals distinguishable topographical features with the appearance of islands and rows at bright spots, as depicted in Figure 2d.T he amorphousn etworks arise from the presence of defects in the carbon lattices, such as residual oxo-groups, presumably at defect-sites, vacanciesa nd nmscale holes, as evidencedb efore. [10b, 17a, 22] The density of defects in single layers of graphene can be quantified by Ramans pectroscopy. Figure 3a showst he evolution of Ramans pectra obtained from single layers of graphene with various densities of defects. For the monolayer G 0% ,t here are two distinct peaks, the Gb and (at 1570 cm À1 ), associated with the in-plane stretching vibration of CÀCb onds, andt he 2D band (at 2670 cm À1 ), activated by ad ouble-resonant Ramans cattering. [23] The Ga nd 2D bands are sensitivet ot he structure of the carbon hexagonal lattice and, thus, they can   be used to characterize the quality of graphene-based materials. In addition, af aint Db and at around1 360 cm À1 is noticed, probably evolving from grain boundaries or other lattice imperfections. [24] The structural defects can be estimated by using the defect-activated Db and. Moreover, the intensity and shape of these peaks strongly depend on the nature of disorder.A st he amount of disorder in graphene increases, the Dband intensities enhance, whereas the 2D-bandi ntensities decrease.F urther,t he G-bands plits into two sub-bands, Gb and (1583 cm À1 )a nd D' band (1620cm À1 ). In addition, the broadening of all bands is observed with increasing density of defects. The full-width-at-half-maximum (FWHM, G)o ft he 2D peak (G 2D )i ncreases from approximately 24 cm À1 in the single-layer G 0% to about1 78 cm À1 in the single-layer G 1.5 % .I nT able 1, the details of Ramanp eak analyses are summarized for graphene samples with different densities of defects, namely G 0% ,G 0.2 % , G 0.4 % ,G 0.5 % ,G 0.9 % and G 1.5 % .
The intensity ratio of I D /I G is used for determining the density of defects in the G D samples. In the case of single-layer G 0% ,i t containsane xtremelyl ow density of defects, which belongs to the low-defect density regime according to the Raman model introduced by Lucchese, CanÅado and Ferrari et al.,w ith a ratio of I D /I G = 0.2 AE 0.07 corresponding to about2 4000 C atoms within the intact graphene lattice. According to Equation (1): in which the N C corresponds to intact carbon atoms and A cell = 0.246 2 sin (608) = 0.05239 nm 2 ,t he average distance between defects L D is about 25 nm. The relatedd efect density (n D )i s about 5.1 10 10 cm À2 ,u sing n D (cm À2 ) = 10 14 /(pL D 2 ). [19b] However, the investigated G D samples relate to the high-defect density regime. In this regime, the I D /I G ratio increases with an increase of L D ,b ased on the relation of I D /I G / L D / N C . [19b] Accordingly, the density of defects increases from 0.3 %t o1 .5 %, the corresponding I D /I G ratio decreasesf rom 4.5 to 1.0 and the L D values gradually decrease from 3.5 nm to 1.3nm, respectively.T he related n D increases from about 2.6 10 12 cm À2 to 1.9 10 13 cm À2 . As depicted in Figure3b, the evolution of qualities in the yielded graphene samples are illustrated by plottingt he I D /I G ratio versus G 2D .W ith increasing the density of defects, the G 2D values increase, which is consistent with our discussion above.
Field-effect transistors were fabricated by using the obtained monolayer graphene samples as conducting channels (Fig-ure 4a). The monolayer G 0% flakes were mechanically exfoliated from ab ulk graphite and transferred to ah eavily p-doped Si substrate with a3 00 nm thick SiO 2 layer (Si/SiO 2 ), [21] which acts as the reference sample. The G 2D value of 23.6 AE 2.6a nd I 2D /I G > 4w as determined by Raman spectroscopyt op rove the singlelayer nature of the produced flake ( Figure 3a). The monolayer G D flakes prepared by wet-chemistry were deposited on Si/SiO 2 substrates by using Langmuir-Blodgettt echnique and subsequent chemical reduction or thermal processing. [20a] An AFM image of G 0.5 % FET device is shown in Figure 4b and ah eight profileo fm onolayer G 0.5 % flake with about 1.2 nm is depicted in Figure 4c.The Si/SiO 2 substrate serves as ab ack-gate. The metal contacts Cr/Au (5 nm/70 nm), served as source and drain electrodes, were deposited ontos ingle-layer graphene channel materials by using e-beaml ithography ande vaporation pro- Table 1. Summary of the results of fitting Ramans pectra by Lorentz functions for the yielded monolayer graphene with defects densities of 0%,0 .2 %, 0.4 %, 0.5 %, 0.9 %a nd 1.5 %.  cesses. Avoiding any thermal decomposition of chemically-derived G D samples, no annealing process was performed for all prepared devicesa fter the lift-off process. Electrical transport measurementsw ere performed at ambient conditions in at wo-terminal configuration. The performance of transistors relies on the properties of channel materials,g ate dielectrics, electrodes and test conditions. Therefore, to reliably compare electrical performances for the obtained monolayer graphene samples, all transistors were prepared with parallel electrodes, the same manufacturing processesa nd test conditions.
The Figure 5p resentst he transfer characteristics of fabricated FET devices based on graphene samples with the defects from 0% to 1.5 %. Linearo utput relations (I DS -V DS )a re determined and visualized in the insets of Figure 5, indicating ohmic contacts betweent he graphene samples and the metal electrodes under ambient conditions. The G 0% device in Figure 5a shows V-shape transfer curves( I DS -V BG )w ith asymmetric Dirac voltage( corresponding to the minimum value of I DS )l ocated at + 20 V. The observed p-doping behavior was probably attributed to the heavily p-doped Si/SiO 2 substrate, impurity doping as ar esulto fe xfoliation and transfer processes or polar adsorbates like water or oxygen acting as charget raps between substrate and theg raphene surface. Furthermore, a hysteretic behavior between forward and reverses weeps are observed. For G D transistors (Figure5b-f), no Dirac point appears within the range of the scanned gatev oltages from À50 Vt o+ 50 V. All samples show unipolar p-type character. These defectiveG D samples are stronger p-doped than the G 0% sample due to the oxo-groups modification of the graphene networks. [25] The field-effect carrier mobilities weree xtracted by using Equation (2): [26] m ¼ L=W ðÞ Â 1=C ox V DS ðÞ Â dI DS =dV BG ðÞ ð 2Þ in which L and W are the channel length and width, respectively,a nd C ox = 1.15 10 À8 Fcm À2 is the capacitance per unit area of the gate dielectric material. Ther oom-temperature average mobility values of monolayers of G D are determined between 33.2 cm 2 V À1 s À1 and0 .3 cm 2 V À1 s À1 for densities of defectsb etween 0.2 %a nd 1.5 %. The mobility values are significantly lower than the value of 685 cm 2 V À1 s À1 obtained for our defect-free G 0% and not annealed reference sample. In addition, in Figure 6the field-effect mobilities are plotted as afunction of number of Ca tom (N C )a nd distance between defects (L D ), respectively.I tis found that the mobilities follow,w ithin Figure 5. Room-temperaturetransfer characteristics of graphene transistorswith densities of defects of 0%,0.2 %, 0.4 %, 0.5 %, 0.9 %a nd 1.5 %( a-f), respectively.The gate voltageiss wept continuouslyfrom À50 Vto5 0Vand back to À50 V. The inset shows the corresponding output curves. In summary,w eh ave studied the room-temperature electrical properties of single-layer graphene derived from oxo-G containing defect densities varying from 0.2 %to1.5 %. The defects give rise to ah eterogeneous topographical morphology of oxo-G. The isolated graphene domains (L D 3nm) in oxo-G were identified by Ramans pectroscopy.T he isolation of these domainsl imits the charge transport in reduced oxo-G. Therefore, the mobility values of chargec arriers of graphene with densities of defects between 0.2 %a nd 1.5 %, change by three orders of magnitude, from 0.3 cm 2 V À1 s À1 and 33.2 cm 2 V À1sÀ1 . More generally, the mobility of charge carriers varies by orders of magnitude,a lthoughi tl ooks like that the density of defects varies only al ittle. The fundamental findings reported here can explain the generally diverging results often reported for reduced graphene oxide used in applications.

Experimental Section
Methods AFM characterization was performed by using aJ PK NanoWizard 4 Atomic Force Microscope in tapping mode at room temperature. Raman characterization was carried out with aH oriba Explorer spectrometer with a5 32 nm laser for excitation under air conditions. Scanning tunneling microscopy (STM) was conducted by using Omicron-STM1 microscope under ultra-high vacuum (< 10 À10 mbar). All transport measurements were performed under ambient conditions by at wo-probe station and two source-measurement units (Keithley 2450).

Preparation of defect-free G 0% flakes
The defect-free monolayer G 0% flakes were prepared by micromechanical exfoliation and then transferred on Si/SiO 2 substrates as reported methods. [21] Preparation of defective G D from oxo-G The defective G D flakes were prepared by low-temperature oxidation of graphite based by the before reported method of our group. [20a] Then, the oxo-G was dissolved in methanol/water 1:1 mixtures. The monolayer flakes of G D were deposited onto the Si/ SiO 2 substrate by using the Langmuir-Blodgett technique (LB, Kibron MicroTrough, 3mNm À1 with the surface tension of water as reference value of 72.8 mN m À1 ). Reduction was performed by vapor of hydriodic acid (HI) and trifluoroacetic acid (TFA) (1:1, v/v) at 80 8C( 10 min). Subsequently,t he surface of G D was cleaned with doubly distilled water (Carl Roth) to remove iodine species. The density of defects of individual flakes was determined by Raman spectroscopy (Horiba Explorer spectrometer with a5 32 nm laser for excitation under air conditions). Subsequently,f lakes with defined density of defects were selected for FET device fabrication.

Fabrication of FET devices
Standard electron beam lithography procedure (Raith PIONEER TWO) was used to define and expose the geometry of metal contacts. Subsequently,a5nm/70 nm Cr/Au layer was deposited with thermal evaporation (Kurt J. Lesker NANO 36) and lifted off in ace-tone to make electrode contact to monolayer G 0% and G D flakes, respectively.