Single-step fabrication and work function engineering of Langmuir-Blodgett assembled few-layer graphene films with Li and Au salts

To implement large-area solution-processed graphene films in low-cost transparent conductor applications, it is necessary to have the control over the work function (WF) of the film. In this study we demonstrate a straightforward single-step chemical approach for modulating the work function of graphene films. In our approach, chemical doping of the film is introduced at the moment of its formation. The films are self-assembled from liquid-phase exfoliated few-layer graphene sheet dispersions by Langmuir-Blodgett technique at the water-air interfaces. To achieve a single-step chemical doping, metal standard solutions are introduced instead of water. Li standard solutions (LiCl, LiNO3, Li2CO3) were used as n-dopant, and gold standard solution, H(AuCl4), as p-dopant. Li based salts decrease the work function, while Au based salts increase the work function of the entire film. The maximal doping in both directions yields a significant range of around 0.7 eV for the work function modulation. In all cases when Li-based salts are introduced, electrical properties of the film deteriorate. Further, lithium nitrate (LiNO3) was selected as the best choice for n-type doping since it provides the largest work function modulation (by 400 meV), and the least influence on the electrical properties of the film.


Morphology of LPE GS films. Fabrication and doping of the GS films is schematically represented in
Morphology of LPE GS films is depicted in Fig. 1 consisting of both optical ( Fig. 1(b1-f1)) and Atomic Force Microscopy (AFM) topographic images ( Fig. 1(b2-f2)) for both undoped and metal doped LPE GS films. As can be seen from AFM images, the doping process does not change morphology of LPE films, except that the doped films contain more agglomerates (visible as bright particle-like domains). The following values for the surface roughness were obtained by AFM measurements averaged on ten 50 × 50 µm 2 areas: (a) 11.9 ± 1.5 nm for undoped LPE GS film, (b) 11.5 ± 3.5 nm for Li 2 CO 3 doped, (c) 13.3 ± 2 nm for H(AuCl 4 ) doped, (d) 13.7 ± 1.6 nm for LiCl doped, and (e) 13.8 ± 1.2 nm for LiNO 3 doped LPE GS films. Therefore, the surface roughness sligtly increases by around 2 nm after the doping, while for Li 2 CO 3 doped LPE GS film is practically the same as for the undoped film. Still, optical images recorded on larger scale depict formation of agglomerates in doped films which could degraded their optical (leading to an increased scattering and/or absorption of incoming lights on these clusters) and electrical properties (due to enhanced scattering of charge carriers).
The observed formation of the agglomerates is most likely not an inherent property of the particular metal-salt doping. Overcoming this would likely require further optimization of the LBA process. However, as a benchmark the LBA process in this study was optimized for an undoped film and was left unchanged for all of the metal-salt doped films. transmittance measurements. Using the different doping metal standard solution during LBA of graphene films was found to result in different transparency. In the UV region, the transmittance of graphene is dominated by an exciton-shifted van Hove peak in absorption 9,30 . Transmittance at 550 nm was 82% for undoped and 80%, 76%, 74%, 68% for H(AuCl 4 ), LiCl, LiNO 3 , Li 2 CO 3 doped LPE GS films, respectively (Fig. 2). It can be seen that transmittance decreases for doped LPE GS films. Metal salts decrease the transmittance of the graphene films regardless the type of the present metal (gold or lithium). The degree of the transmittance decrease was related to not only the metal cations but also the anions. Different lithium salts decrease transmittance in different amounts. Transmittance decrease of 14% was the highest for the LPE GS film doped with lithium carbonate (Li 2 CO 3 ). Similar results of the transmittance decrease for metal doped CVD graphene films were obtained in studies of Kwon et al. 22,23,25 . Transmittance decrease could be a consequence of the metal particles adsorption and agglomeration on doped films after the solvent evaporation process. Changes in the thickness of LPE GS films with doping could not be excluded because LBA process in this study was optimized for an undoped film and was left unchanged for all of the metal-salt doped films.
Raman measurements. Raman spectra for undoped and H(AuCl 4 ), LiCl, LiNO 3 , Li 2 CO 3 doped LPE GS films are given in Fig. 3(a). The four basic graphene/graphite peaks D (~1348 cm −1 ), G (~1579 cm −1 ), D ' (1614  D' peak can be used to get information on the nature of defects in graphene 32,33 . I(D)/I(D') was calculated, and the obtained results were presented in Fig. 3 Fig. 4. For the undoped LPE GS film FT-IR spectra is simple. It can be seen only a small peak assignable to C=C skeletal vibration [35][36][37] of the graphene basal planes at ~1560 cm −1 . This peak can also be seen in FT-IR spectra for   www.nature.com/scientificreports www.nature.com/scientificreports/ all investigated doped films at the same wavenumber indicating that graphene basal planes were not interrupt by doping. The strong peak at around ~3400 cm −1 and another, smaller one, near ~1630 cm −1 can be seen in all doped LPE GS films (Fig. (4a)) and corresponding metal standard solutions (Fig. (4b)). They are attributed to the water molecules and are assignable to the O-H stretching vibrations (~3400 cm −1 ) and H-O-H bending mode (~1630 cm −1 ) 38,39 . In the case of FT-IR spectra for LPE GS film doped with LiNO 3 the peak at ~1340 cm −1 and ~1390 cm −1 are assignable to the vibration mode of the NO 3 − ions and asymmetric stretch of O-NO 2 , respectively 38,40 . Similar vibration modes can be observed in the case of FT-IR spectra for LPE GS film doped with Li 2 CO 3 and can be assigned to the vibration mode of the CO 3 − ions (1340 cm −1 ) and asymmetric stretch of O-CO 2 (~1390 cm −1 ) 41 . The same vibrational modes could be seen for LiNO 3 and Li 2 CO 3 standard solutions ( Fig.  (4b)).
From the observed FT-IR results ( Fig. 4(a)) it is clear that additional peaks appear with LPE GS film chemical doping. These additional peaks match with vibrational modes of the anions in solution ( Fig. 4(b)). Considering that no new peaks are visible in the given spectra (which would indicate the formation of chemical bonds) the present peaks could be a consequence of the metal salts adsorption to the graphene lattice during the doping. In order to understand Li and Au doping mechanisms XPS measurements were performed and they are presented in separate section.

Work function modulation.
Results for the work function dependent on the different metal standard solution used in the LBA process are summarized in Fig. 5. The top row depicts an example with the topography ( Fig. 5(a)), corresponding contact potential difference (CPD) map measured by Kelvin probe force Microscopy-KPFM ( Fig. 5(b)), and the histogram of the CPD distribution measured on H(AuCl 4 ) doped graphene film (Fig. 5(c)). The histogram is characterized with a single peak, which is used for the averaging and calculation of the absolute value of work function. The same procedure was done for all considered films. More details about the measurements of CPD and WF calculations are given in Supplementary information in Supplementary Figs. S3-S5. As a result, the values of the absolute work function are presented in Fig. 5(d) for both, doped and undoped LPE GS films. As can be seen, n-doping of graphene films is achieved by Li-based salts, whereas Au-based salt leads to p-doping.
The change of the WF due to the doping can be explained according to the schematic presentation in Fig. 5(e), illustrating that Li (Au) as a lower (higher) work function material compared to GS films. Therefore, presence of Li-based salts into the graphene film results in a reduction of the work function of the entire film. This behavior can be interpreted as an increase in the Fermi level of GSs -compared to the value for the undoped filmsindicating predominantly a charge transfer from Li-based salts to graphene (n-doping), as expected when considering that Li has lower WF than graphene (graphite). In contrast to Li-based salts, the Au-based salt shows an opposite trend for the relative change of the work function. This indicates charge transfer from graphene to www.nature.com/scientificreports www.nature.com/scientificreports/ Au-based salt and a relative reduction of the Fermi level in GSs (p-doping). It is also worth mentioning that poly-crystalline nature of LPE based GS films, large amount of sheet edges and presence of the residual solvent (NMP) results in p-doped films 9 , as was also observed in the electrical measurements presented in the following subsection. Therefore, WF values are lower for the LPE-based films by at least 200 meV, than for the pristine exfoliated single-crystals 42 . p-type doping is also reflected on the WF of the reference samples (undoped LPE GS), and therefore on the whole accessible range for the WF modulation by this method. This was also highlighted in Sheet resistance. The schematic cross-section of the devices used for the electrical characterization is shown in Fig. 6(a), also indicating electrical connections. An optical microscopy image for one of the devices without PDMS encapsulation (for clarity) is shown in Fig. 6(b) illustrating source (S) and drain (D) contact geometries. One characteristic set of transport and output curves for H(AuCl 4 ) and LiNO 3 doped film is presented in Fig. 6(c-f). Here linear fits were used to extract sheet resistances and apparent linear hole mobilities. Transfer curves for all four salt-treatments and for the reference LPE GS film are presented in the Supplementary information ( Supplementary Fig. S1).
In the cases of a reference (undoped) and H(AuCl) 4 doped LPE GS samples, output curves barely deviate from a perfect linear behavior in a rather large bias range, indicating that the contact resistance is negligible in comparison to the channel. This is in contrast to all samples doped with Li-based salts, where a significant deviation from the linear output curves were observed at higher bias, indicating non-negligible contact resistance. This can be attributed to large WF differences with Au bottom contacts in the case of Li-based salt doping of the films. www.nature.com/scientificreports www.nature.com/scientificreports/ apparent hole mobility of the devices. While the type of majority carriers was not affected by the doping, a significant (over one order of magnitude) suppression of the field-effect was observed for Li salt dopings of the films. Figure 7 summarizes electrical properties obtained for all of the measured devices as a function of the different metal based doping.
The results indicate that anions also play a significant role. In the case of Li-based salts, a large variation of the electrical properties was obtained by the different choice of the anion species. Nonetheless, the experiments point out that metal cations dictate the direction of the WF shift (see Fig. 5), as is apparent in the case of H(AuCl 4 ) and LiCl where only cation species is varied. Our results of metal based doping of LPE graphene films demonstrate a tradeoff between enhancement of the electrical performance and modulation of the WF. Similar results were obtained for CVD doping with Li and Au salts 23,25 . Of a particular technological relevance is large reduction of the WF of graphene. While many methods for chemical modulation of graphene result in p-type doping [43][44][45][46] , stable and simple n-type doping is much harder to achieve [47][48][49] . For an efficient electron injection, a significant reduction of graphene's WF is required. As pointed out by WF measurements and electrical characterization, LiNO 3 is the best choice from the tested Li-based salts with respect to both the largest WF reduction (by 400 meV) and least deterioration of the electrical properties of the films with ~2-3 times increase in sheet resistance compared to the reference (undoped LPE GS).
In contrast, doping of LPE GS films by HNO 3 vapor results in an increase of the apparent mobility 9 . However, using a LiNO 3 solution reduces the mobility by one order of magnitude. Therefore, Li + cations -and not anionsare likely responsible for the deterioration of the electrical properties upon n-doping. An increase of sheet resistance was observed in doping of CVD graphene with alkali metal carbonates and chlorides 23,25 . There, a significant increase in the sheet resistance was related to the combination of carbon atoms and dopant metals because electron donation occurred 23,25 . Also, Chen et al. observed that the mobility of the charge carriers decreases with the increase of the potassium doping concentration which they attributed to additional scattering caused by ionized potassium atoms 49,50 . It is most likely that Li + cations are acting as scattering centers for the carriers, or provide traps at the boundaries between neighbouring GSs and effectively increase contact resistance between the overlapping GSs.
Finally, considering that the main potential application of these LPE GS films lies in transparent electrodes, direct current conductivity to optical conductivity ratio (σ DC /σ OP ) is presented in Fig. 7(c) for all metal standard solution doping cases and for the reference (undoped). σ DC /σ OP is a parameter frequently reported in order to characterize the relative performance of the films in terms of transparency and sheet resistance 11,33,51 . The higher the ratio the better the quality of transparent electrodes 33 . Compared to the changes in the electrical properties ( Fig. 7(a)) the changes in the optical properties (Fig. 2) are minor. Therefore, the dependence of the σ DC /σ OP on the type of the metal-ion doping clearly follows the trend set by 1/R □ .

X-ray photoemission Spectroscopy (XpS) measurements. In order to understand Au and Li ion
doping mechanisms XPS measurements were performed. C 1 s, Au 4 f and Li 1 s core-level XPS spectra are shown in Fig. 8. N 1 s, Cl 2p and O 1 s spectra are presented as Supplementary Fig. S2. The C 1 s peak of undoped and LiCl, LiNO 3 , Li 2 CO 3 , H(AuCl 4 ) doped LPE GS films is shown on Fig. 8(a). The C 1 s peak is deconvoluted using Gaussian profile into 4 components for undoped and doped films: C=C/C-C in aromatic rings (284.5 eV); C-C sp 3  www.nature.com/scientificreports www.nature.com/scientificreports/ dopant in graphene 21 . Thus, different shifts of C=C/C-C peak for different metal-salt doping materials could be also a consequence of anions influence on graphene films. Figure 8(c) show the Li 1 s core-level XPS spectra. Literature values for Li 1 s core-level for different Li compounds are: LiCl (56.2 eV), Li 2 CO 3 (55.5 eV) and LiNO 3 (55.8 eV) 56 and they correspond well to the values obtained in this work. Li 1 s peak at 55.0 eV is assigned to Li-O bond 57 . Vijayakumar and Jianzhi have shown that lithium ion tends to bind with the oxygen rather than the carbon on graphene surface, and interacts by forming Li-O ionic bond 58 . Also Kwon et al. have proposed that C-O-X complexes can be formed during doping treatment and can act as an additional dipole to further reduce the value of WF [23][24][25]59 . The intensity ratio between sum of the intensities of C=C/C-C and C-C peaks, and the intensity of C-O (I (C=C/C-C+C-C) /I (C-O) ) is shown in Fig. 8(b). Also, the ratio of Li 1 s intensity from Li salts to Li-O intensity (I Li /I Li-O ) can be seen in Fig. 8(b). In both cases, intensity ratios decrease for Li 2 CO 3 , LiNO 3 , LiCl, respectively and this implies increased formation of C-O and Li-O bonds. Increased number of Li-O bonds follow the increasing trend of C-O bonds, which is in correlation with the WF change (Fig. (5d)). The above mentioned results strongly suggest that the mechanism of n-type doped LPE GS films with lithium-salts could be explained with formation of Li complexes (C-O-Li). Figure 8(d) show the Au 4 f peak of gold-chloride doped LPE GS film. The peak is composed of metal (Au 0 ) and metal ion (Au 3+ ). The peaks at 84.2 eV and 87.9 eV are assigned to neutral Au (Au 0 4f 7/2 and Au 0 4f 5/2 , respectively), and the peaks at 86.5 eV and 90.2 eV are assigned to Au ion (Au 3+ 4f 7/2 and Au 3+ 4f 5/2 , respectively). Au ions (Au 3+ ) have positive reduction potential and have tendency to spontaneously accept charges from other materials (graphene) and reduce to Au 0 21,22,25,60 . Therefore, the mechanism of p-doped LPE GS film can be explained as spontaneous electron transfer from graphene film to Au 3+ , resulting in depletion of electrons in the graphene networks, thus increasing the WF of doped graphene.

conclusion
We demonstrate a straightforward single-step method for forming and doping of LPE GS films by metal standard solutions through charge transfer processes. Chemical doping of graphene allows to modulate its WF in a very large range, and therefore potentially enables to use the same electrode material for both, the injection and for the extraction of the electrons. n-doping of graphene films is achieved by Li-based salts, whereas Au-based salt leads  www.nature.com/scientificreports www.nature.com/scientificreports/ to p-doping. Furthermore, solution-processed graphene films are in particular suited for the chemical modulations, since a large number of the sheet edges opens up many adsorption sites and enhances the doping effects when compared to many other types of graphene.
The morphology of the LPE GS films does not change with the doping process, except that doped films contain agglomerates. FT-IR measurements point out that graphene basal planes stay chemically unchanged with metal doping and the charge transfer process is enabled with adsorption of the metal salts. Li-based salts decrease the WF, while Au-based salts increase the WF of the entire film. The maximal doping in both directions gives a significant range of around 0.7 eV for the work function modulation. Changing the dopant (Au or Li based salts) significantly affects the electrical properties of the films. In the case of the Li-based salts doping of the film, a significant suppression of the field-effect mobility and the increase of the sheet resistance was observed. This indicates that adsorbed Li-anions act as scattering centers for the charges. XPS data indicated that different mechanisms exist in the case of Au and Li doping. For Au ions spontaneous charge transfer occurred from graphene, thus increasing WF. In the case of Li doping, potential adsorption sites are a large number of the sheet edges where C-O bonds are preferential sites for lithium ions and for forming of C-O-Li complexes. In all cases graphene films are p-type, which is in accordance with KPFM measurements. Also, tradeoff between Li complex which reduce the value of WF and anion which increase the value of WF could be a reason of such a doping.
Metal salts charge transfer doping -which happens with this single-step method -provides a facile and effective method to tune the WF of LPE graphene therefore extending the potential use of these materials in low-cost transparent electrode applications.

Methods preparation of GS dispersion and doping solutions.
A dispersion of GS in N-methyl-2-pyrrolidone (NMP, Sigma Aldrich, product no. 328634) has been used. GS dispersion was prepared from graphite powder (Sigma Aldrich, product no. 332461) of initial concentration 18 mg/mL. The solution was sonicated in a lowpower ultrasonic bath for 14 h. The resulting dispersion was centrifuged for 60 min at 3000 rpm immediately after the sonication.
Stock standard solutions used in our work for n-doping are 1 mg/mL LiCl, LiNO 3 and Li 2 CO 3 and for p-doping is 1 mg/mL gold standard solution (Merck, H(AuCl 4 ), product no. 170216). Lithium standard solutions were prepared from originated Li salts (LiCl, LiNO 3 and Li 2 CO 3 , Merck, product no. 105679, 105653 and 105680, respectively). By appropriate dilution of the stock solution with deionized water we obtained 0.1 mg/mL metal water solution which is then used in doping process.
Deposition on a substrate and doping of LPE GS films. GS dispersion in NMP was used to fabricate transparent and conductive films by LBA technique at a water-air interface, like in our previous work 9, 29,61 . A small amount of GS dispersion was added to the water-air interface and after the film was formed it was slowly scooped onto the target substrate. Applying the same process of fabricating the GS films and using the appropriate metal standard solution instead of water, chemical doping was achieved. As substrates SiO 2 /Si wafer were used for electrical and WF measurements, while quartz and CaF 2 substrates were chosen for optical and FT-IR spectroscopy, respectively.
Characterization of undoped and doped LPE GS films. The Morphology of LPE GS films was studied by optical and atomic force microscopy (AFM). Topographic AFM measurements were done by NTEGRA Prima AFM system and NSG01 probes with a typical tip radius of around 10 nm. The surface roughness of LPE GS films was calculated as a root-mean square of the height distribution and averaged on ten 50 × 50 μm 2 areas.
Kelvin probe force microscopy (KPFM) -established almost three decades ago 62 and in the meantime frequently applied to graphene 42,63-65 -was employed in order to characterize changes in the electrical surface potential and corresponding Fermi level shifts due to doping. For this purpose, we measured the contact potential difference (CPD) between AFM tip and the sample surface 66 by using Pt covered NSG01/Pt probes with a typical tip curvature radius of 35 nm. In the first pass of KPFM, the sample topography was measured in tapping AFM mode. In the second pass, the probe was lifted by 20 nm, and moved along the trajectory measured in the first pass. Simultaneously, the sum of AC and DC voltage was applied between the sample and the probe. The AC voltage excites AFM probe oscillations during its movement, while the CPD between AFM tip and the sample surface in every point is then equal to the value of variable DC voltage which cancels the AFM probe oscillations. For all samples, the CPD was measured on five 5 × 5 μm 2 areas, and then averaged. In order to obtain the absolute value of the work function, the following procedure was applied 42 . The CPD is equal to the work function difference between AFM tip (WF t ) and sample (WF s ), CPD = WF t -WF s . The calibration of the WF t was done by a standard procedure consisting of KPFM measurements on a freshly cleaved HOPG with a well known work function of 4.6 eV 42 . Finally, the sample work function was calculated as WF s = WF t -CPD, where CPD is measured by KPFM for all, undoped and doped LPE GS films.
The effect of chemical doping on optical properties of LBA GS films was investigated with measurements of optical transmittance, using UV-VIS spectrophotometer (Beckman Coulter DU 720 UV-VIS Spectrophotometer).
Electrical measurements were performed under ambient conditions in a standard field-effect device configuration with Si substrate acting as a back gate electrode, using Keithley 2636 A SYSTEM SourceMeter. Devices were based on bottom-contact gold pads defined by a shadow mask with L/W = 30 µm/1000 µm, and SiO 2 as a gate dielectric with thickness of 285 nm. Graphene films were deposited using the same LBA method as described above. The top surface of the devices was encapsulated by polydimethylsiloxane (PDMS) films (GelPak X4) to ensure stable performance and minimize any adsorption/desorption during electrical measurements that could occur from the surroundings (e.g. water vapor). Electrical characterization was performed on several devices of each doping with metal standard solution, and for undoped films as a reference. For each device ten subsequent forward and (2020) 10:8476 | https://doi.org/10.1038/s41598-020-65379-1 www.nature.com/scientificreports www.nature.com/scientificreports/ backward transfer and output curves were measured, using low sweeping rate (~0.005-1 Hz per point in a voltage sweep) to minimize parasitic capacitance. Sheet resistance and apparent linear field-effect mobility were extracted using fits to output and transfer curves, respectively. For the output measurements source-drain bias was varied in a range between −10 V and +10 V, with the gate electrode grounded. For transfer measurements, the gate voltage was varied between 0 V and 50 V, with source-drain bias at 1 V in all cases except for Li 2 CO 3 where due to a very weak field-effect (very low mobility) 10 V bias was used.
The room-temperature micro-Raman spectra of undoped and metal salt doped LPE GS films were collected using Tri Vista 557 triple spectrometer coupled to the liquid nitrogen-cooled CCD detector. Nd:YAG laser line of 532 nm was used for the excitation and 50 magnification objective was used for focusing the beam onto the sample. Low laser power (less than 1 mW) was applied to prevent the thermal degradation of the sample. Each LPE GS film sample was measured at eight different positions.
Fourier transform infrared absorbance spectra (FT-IR spectra) of undoped and metal salt doped LPE GS films were measured over a range of 400-4000 cm −1 with Nicolet Nexus 470 FT-IR spectrometer. Standard solutions which were used for the preparation of doped films were measured too and they were prepared by drop casting method on the CaF 2 substrate. XPS spectra were recorded using a Thermo Scientific instrument (K-Alpha spectrometer, Thermo Fisher Scientific, Waltham, USA) equipped with a monochromatic Al Kα X-ray source (1486.6 eV). High-resolution scans were performed with a pass energy of 50 eV and a step size of 0.1 eV. All analyses were performed at room temperature.

Data availability
The datasets obtained and analysed during the current study that are not included in this article are available from the corresponding authors on reasonable request.