Physical properties of reduced graphene oxide (rGO) strongly depend on the ratio of sp2 to sp3 hybridized carbon atoms, the presence of different functional groups, and the characteristics of the substrates. This research for the very first time illustrates successful wafer scale integration of 2D rGO with Cu/TiN/Si, employing pulsed laser deposition followed by laser annealing of carbon-doped copper layers using nanosecond excimer lasers. The XRD, SEM, and Raman spectroscopy measurements indicate the presence of large area rGO onto Si having Raman active vibrational modes: D, G, and 2D. A high resolution SEM depicts the morphology and formation of rGO from zone-refined carbon formed after nanosecond laser annealing. Temperature-dependent resistance data of rGO thin films follow the Efros-Shklovskii variable range hopping (VRH) model in the low-temperature region and Arrhenius conduction in the high-temperature regime. The photoluminescence spectra also reveal a less intense and broader blue fluorescence spectra, indicating the presence of miniature sized sp2 domains in the near vicinity of π* electronic states which favor the VRH transport phenomena. This wafer scale integration of rGO with Si employing a laser annealing technique will be useful for multifunctional integrated electronic devices and will open a new frontier for further extensive research in these functionalized 2D materials.
I. INTRODUCTION
Graphene has many desirable properties, including extremely high electron mobility (10 000–50 000 cm2 V−1 s−1), high Young's modulus (0.5 TPa), and excellent optical transparency (97.7%).1 These properties of graphene enable its use in FETs, memory devices, and energy storage applications. Graphene is a strong contender to ITO owing to its economic viability, high electrical conductivity, and optical transparency in the visible solar spectrum region. Commercial synthesis of graphene by CVD process requires a high thermal budget, and its transfer onto different substrates via chemical route (PMMA) or mechanical exfoliation (scotch tape method) leaves impurity residues and significant defects in the basal plane of graphene thereby thwarting its use in device manufacturing.
Transformation of graphene to graphene oxide (GO) and reduced graphene oxide (rGO) through chemical functionalization has generated significant research interests due to its potential in obtaining an optical band gap and subsequent tuning of its optoelectronic properties for device applications.1 GO, in general, has low conductivity and, therefore, has to be reduced to form more conductive rGO thin films. Additional chemical reagents and high temperature processing are required for the formation of rGO. There are several processes of manufacturing rGO, namely: chemical reduction by pre-mixed hydrazine during the spray coating of GO,2 thermal reduction at high temperatures of spin coated GO films,3 Meyer rod coating process and using premixed Palladium chloride as a reducing agent,4 meniscus dragging deposition followed by HI vapor reduction of a micro-liter scale solution, etc.5 The solution processing route for producing rGO flakes has a wide range of oxygen functionalities and high throughput manufacturing. Using this process both electrical and optical properties of rGO can be tuned by controlling the ratio of sp2 to sp3 hybridized carbon clusters.6 However, rGO films synthesized by CVD or solution based methods contain large-angle grain boundaries and lack wafer scale epitaxial integration, thereby limiting their usage in device fabrication. In addition, the reduction of graphene (to GO and rGO) creates disorder, and electron-localization-hopping phenomenon plays a significant role in determining its physical properties.7 A clear understanding of the electronic transport properties in rGO sheets is needed as recent studies report different conduction mechanisms such as Mott variable range hopping (VRH) and Efros-Shklovskii variable range hopping (ES-VRH).8
The change in physical properties of graphene based compounds with functionalization can be due to: (a) conversion of the hybridized state of carbon atoms, (b) creation of a barrier (scattering entity) at the functionalization site, (c) 2-D planar lattice distortion after functionalization or thermal treatment, and (d) introduction of sp2 clusters and defects thereby introducing different energy levels in the electronic band structure.9 Chemical, electronic, and structural properties of rGO can be modified by molecular interactions occurring due to: (1) covalent bonding, (2) π–π* energy levels interfacing, (3) physisorption, and (4) lattice substitution by different functional groups, namely, –OH, –O–, and –COOH.10 Covalent functionalization of sp2 hybridized carbon atoms in graphene converts it into a tetrahedral sp3 hybridized carbon, thereby losing the free π electrons associated with sp2 hybridized carbon atoms. Atomic substitution of carbon with other elements (such as N or B) helps to retain the sp2 character at the cost of disrupting the π-cloud continuum.11 Non-covalent functionalization of graphene does not distort the sp2 network. However, there is a change in the doping density, an increase in the density of electron–hole puddles, and scattering sites.11
Covalent bonding is the most studied form of graphene functionalization, and it causes significant changes in graphene's electrical and optical properties.7 Reducing GO produces rGO; however, rGO consists of unreduced and covalently bonded oxy groups.12 The optical band gap in GO and rGO develops owing to the loss of π electrons from the carbon atom. The functionalizing molecule introduces energy level in the form of edge states and functionalization states in the electronic band structure of covalently functionalized graphene (CFG), thereby rendering it to be either n-type or p-type semiconductor.11 In conclusion, a combination of charge carrier deficiency (due to removal of free π electrons), perturbation of the electron-potential continuum, and structural distortion in a 2D planar lattice causes a drastic reduction of the charge carrier mobility and polarity in graphene. As compared to graphene (room temperature mobility of 10 000–50 000 cm2 V−1 s−1 and intrinsic mobility limit 200 000 cm2 V−1 s−1), rGO exhibits a drastically reduced carrier mobility (0.05–200 cm2 V−1 s−1). Reduced graphene oxides have a few residual oxy groups and exhibit p-type behavior with a finite effective optical band gap of 0.2–2 eV depending upon the characteristics of these groups.7
The sp2 clusters in rGO exhibit Klein tunneling of electrons, thereby have high conductivity, and the sp3 regions exhibit insulating or semi-conducting properties. Electrical conduction between the sp2 clusters occurs via variable range hopping (VRH) mechanisms.7 There occurs a quantum confinement in the smaller sp2 regions thereby exhibiting semiconducting behavior. The dispersed functionalization on the basal plane of rGO leads to a reduction in the mean free path and Fermi velocity, and an increase in the elastic and inelastic scattering.13 The laser annealing technique is being widely employed to obtain exquisite carbon based structures.14,15 This paper addresses wafer-scale integration of rGO by using a Cu/TiN/Si platform and a laser annealing technique at room temperature in air at ambient pressure without any reducing agent. Specifically, the rGO films are created by laser annealing of carbon-doped Cu films which are grown epitaxially on TiN/Si(100) or Si(111) substrates. The laser melting of carbon-doped copper films zone refines carbon, where all the carbon segregates to the surface. Upon quenching, rGO films are formed on Cu/TiN/Si epitaxial heterostructures, which will open up new avenues for multifunctional solid state devices integrated on a computer chip.
II. EXPERIMENTAL
TiN was epitaxially grown onto Si (100) and Si (111) using a KrF laser (pulse duration = 25 ns, wavelength = 248 nm, energy density = 3.0 J cm−2) using a domain matching epitaxy (DME) paradigm.17 Cu-2% (at) C was used as a target in the next step of pulsed laser deposition (PLD) growth of Cu-C/TiN/Si using a KrF laser without breaking the chamber vacuum. These films were irradiated in air with a KrF single laser pulse using varied laser energy density (energy density = 0.3–1.0 J cm−2). These films were characterized by XRD, Raman-photoluminescence (PL) spectroscopy, Time-of-Flight Secondary Ion Mass Spectrometer (TOF-SIMS), XPS, high resolution SEM, and temperature dependence resistivity measurements. Rigaku SmartLab X-ray diffractometer having CuKα X-ray source, interchangeable stages, and optical components was employed for collecting the XRD spectra. Bragg-Brentano reflection geometry was used during the data acquisition process. Alpha300 R—Superior Confocal Raman Imaging instrument was used for Raman spectroscopy measurements using a 532 nm laser excitation source. For better spectral line resolution, 600 mm gratings were employed during the data collection procedure. Crystalline Si (having a strong peak at 520.6 cm−1) was used to calibrate the instrument before and after the Raman spectrum acquisition process. High-resolution SEM measurements were carried out using the field emission gun technique in FEI Verios 460L SEM. A Time-of-Flight Secondary Ion Mass Spectrometer (TOF-SIMS) with a lateral resolution of <300 nm was used to detect the presence of O, OH, and C across the thickness of rGO thin films. XPS with X-ray energy 10–14 kV for Al/Mg and Al/Ag sources employing a superior analyzer (PHOIBOS 150) having <1 eV resolution were used for the collection of XPS data. Since the sample thickness was much less than the width and length, the van der Pauw method is used for temperature-dependent resistance measurements. To prevent heating and further alteration of the properties of rGO thin films, indium wire pressure contacts were used as the electrode material. A cryo-cooled vacuum chamber was used for four point temperature-dependent resistivity measurements. The LabVIEW programming was used as a user interface to communicate with a digital multimeter and a temperature controller.
III. RESULTS AND DISCUSSION
A. X-ray diffraction
Figure 1 shows an XRD pattern of the rGO thin films grown onto Cu/TiN/Si(100) by laser annealing with a laser energy density of 0.3 J cm−2 (sample A). The Cu films were doped with 2.0% (atomic fraction) C, whereby melting of Cu-C alloys leads to the formation of rGO by zone refining of C to the Cu surface. With the Cu-C alloy composites, the melting and zone refining of 22.44 nm thick layer will lead to the formation of monolayer of graphene. Thus, by increasing the depth of melting, we can tune the thickness of GO and rGO films. Laser annealing ambient determines the formation of graphene (vacuum), GO (oxygen, H2O), and rGO (vacuum). TiN with (a = 4.20 Å) has over 22% lattice misfit with Si (100), and Cu (a = 3.60 Å) has a lattice misfit of over 15% with TiN. This large lattice misfit can be handled by domain matching epitaxy, which addresses a new paradigm of epitaxy by matching of lattice planes across the misfit scale.16–18 The corresponding 2θ values for rGO samples are 11.8° (for sample A) and 12.2° (for sample B grown using a laser energy density of 0.6 J cm−2: not shown in Figure 1) which is in agreement with the previously reported values for rGO synthesized by chemical routes.19,20 Reports have shown that GO has the largest interlayer distance (5 Å–9 Å) due to the presence of intercalated water molecules and functional groups such as hydroxyl, epoxy, and carboxyl.21 The interlayer distances for rGO have been calculated by using the Bragg's relation22 which varies from 0.72 nm to 0.75 nm. The shift of the GO peak towards higher diffraction angle in the case of rGO is due to the vaporization of the water molecules present between the stable GO layers, which results in decrease of the interlayer distance.23 The crystallite size for the rGO films was calculated by using the Debye Scherer equation24
where represents the crystallite thickness, K (0.91) is a constant dependent on the crystallite shape, is the X-ray wavelength (1.5405 Å), is the corrected FWHM, and is the X-ray scattering angle. The corrected FWHM for rGO peaks have been calculated by utilizing a strain correction, i.e., subtracting the strain from the substrate used (in this case, Si).25 The strain correction is represented by the equation
where is the FWHM of the corresponding peak sample, and is the FWHM of the corresponding substrate peak. The change in the crystallite size explains about the growth of the rGO thin films. With the help of the Debye Scherer equation, Ju et al.26 calculated the number of layers present in GO using the following equation:
where is the interlayer distance of rGO thin films. The above calculation yielded a total number of rGO layers as 4. Since sample B is grown at a higher laser energy density of 0.6 J cm−2, the shift of the rGO peak in sample B to higher 2 angle signifies increased reduction of rGO with an increase in laser energy density.
XRD showing epitaxial integration of rGO (0002) onto Cu/TiN/Si(100) after laser annealing using a laser energy density of 0.3 J cm−2.
XRD showing epitaxial integration of rGO (0002) onto Cu/TiN/Si(100) after laser annealing using a laser energy density of 0.3 J cm−2.
B. Raman spectroscopy
Figure 2 shows the room-temperature unpolarized micro Raman spectra of rGO thin films formed after laser annealing of Cu-C/TiN/Si along with the as-deposited one. The inset figure shows fitted spectra of the 2D Raman peak. The Raman spectra of graphene based materials show almost all the characteristic features in the range between 1000 cm−1 and 3000 cm−1, and the corresponding Raman modes are known as D, G, and 2D vibrational bands.27 The A1g symmetry phonon near the K zone boundary involves the breathing modes of sp2 carbon atoms present in graphene and rGO, and it introduces D-peak around 1355 cm−1 which arises only due to the presence of defects and asymmetry in the lattice.13 The D peak position of the laser annealed samples is in accordance with the reported values of 1356 cm−1. The E2g phonon mode in the Brillouin zone center corresponding to the G peak is positioned in between 1500 cm−1 and 1630 cm−1. The origin of the G peak is due to the bond stretching of sp2 C atoms both for rings and chains.13 It has been observed that the G peak of rGO is shifted to lower frequencies to 1590 cm−1 due to the removal of functional groups by reduction.28 Stiffening of the G mode also indicates the increase in carrier density of either sign.29 The G peak position of the two laser irradiated samples (A and B) is in between 1590 cm−1 and 1592 cm−1, indicating the presence of sp2 C atoms in the rGO films. Minimal perturbations of the D and G peak positions in samples A and B indicate no stress difference between them though they are being formed at different laser energy densities. There is a considerable blue shift of ∼54 cm−1 in the G peak, when rGO is formed from the as-deposited sample after laser annealing. This Raman shift is related to the stress generated in the film during its formation. The Raman shift (Δω) is related to Δω (in cm−1) = 2.2 ± 0.10 cm−1 GPa−1 along the [111] direction, Δω (in cm−1) = 0.73 ± 0.20 cm−1 GPa−1 along the [100] direction, and Δω (in cm−1) = 3.2 ± 0.23 cm−1 GPa−1 for the hydrostatic component.30 The biaxial stress in thin films can be described as a combination of two-thirds hydrostatic and one-third uniaxial stress. The biaxial stress can be estimated using σ = 2μ (1 + ʋ)/(1 − ʋ)·Δα·ΔT, where μ is the shear modulus, ʋ is the Poisson's ratio, Δα is the change in thermal coefficient of expansion, and ΔT is the change in temperature. The magnitude Raman shift is related to the residual stress () formed in the film and is given by the equation
where is the shift in Raman wavenumber, is the Raman wavenumber corresponding to the reference state (in this case, the G vibrational mode in DLC), is the shear modulus, and is the Poisson's ratio of rGO. By considering elastic modulus as 207.6 GPa, as 0.197 the residual stress calculated from the Raman shift is 4.54 GPa in the rGO structure after it is being formed by laser annealing. The blue shift also indicates the residual stress to be of compressive nature.
Raman spectra of samples A and B along with DLC (before laser annealing). There is a clear indication of the formation of distinct Raman active vibrational modes D, G, and 2D after laser annealing process. The inset shows peak fitting of 2D Raman peak indicating the number of layers as 4.
Raman spectra of samples A and B along with DLC (before laser annealing). There is a clear indication of the formation of distinct Raman active vibrational modes D, G, and 2D after laser annealing process. The inset shows peak fitting of 2D Raman peak indicating the number of layers as 4.
The ratio between the peak intensities of D and G bands is an indicative of the degree of reduction.31 Ferrari and Robertson13 suggested that ID/IG increases with disorder according to the Tuinstra and Koenig (TK) relationship. However, further reduction reaction creates more disorder and distortion in the sp2 structure. As a result, ID decreases compared to IG and the TK relationship no longer holds true. The corresponding ID/IG ratio of samples A and B is 1.145 and 1.156, respectively. The higher ID/IG ratio signifies higher degree of reduction in sample B corresponding to higher angle x-ray diffraction in XRD of rGO. Sample A was prepared employing 0.3 J cm−2 and sample B using 0.6 J cm−2. Therefore, Raman spectroscopy and XRD show a direct correlation between reduction and laser energy density in rGO thin films without a considerable change in stresses generated (minimal shifts in the Raman peaks).
The second order overtone of the D peak reported around 2710 cm−1 is known as 2D vibrational mode. This peak is the characteristic of double resonance transitions resulting from the generation of two phonons with opposite momentum (+k and −k) to each other.32–34 The laser annealed A and B samples both show 2D peak at 2708 cm−1. Due to the absence of graphene like nature in as-deposited thin films, there is a complete absence of the 2D vibrational mode. For graphene, it has been reported that the 2D peak profile is sharp, while lower 2D peak intensity as compared to D and G peaks is an indication of disorder in the GO structure.35 The Gaussian peak fitting of 2D vibrational modes in the rGO films reveals two peaks at 2696 cm−1 and 2740 cm−1. This is consistent with the presence of 4 layers in rGO films.36 The laser annealed rGO samples show intense 2D peak, thereby indicating less disorder in its structure. A lower disorder in the basal plane of the rGO structure is due to the extremely fast processing time of nanosecond lasers, where atoms are quenched thereby getting less time to form disordered lattice structure at the expense of generating residual stresses in the films as seen by the Raman blue shift of the graphitic peak. The laser annealed rGO samples show FWHM value for 2D peak as 100 cm−1 and 92 cm−1 for samples A and B, respectively. Reduced values of FWHM for 2D peak in sample B are an indication of better conductivity as compared to sample A (which will be discussed in greater detail later). The intense and narrow 2D peak of laser annealed rGO samples indicate that the number of defects is less compared to the other rGO samples produced by thermal or solution based processes. The reducing trend of the FWHM of 2D peak from sample A to sample B suggests the quality of reduction, i.e., reduced structure with a few defects. The I2D/IG ratio can be used to predict the charge carrier mobility.37 This I2D/IG ratio for samples A and B is 0.85 and 0.92, respectively. An increased value of I2D/IG ratio in sample B is an indicative of the presence of free π electrons as charge carriers on its basal plane thereby reducing its resistivity.
C. Photoluminescence spectroscopy (PL)
Figure 3 shows the PL spectra of the rGO samples which were fitted with different peaks to extract the details. The PL spectrum shown is after background subtraction of the substrate under identical conditions. Reduced GO is a graphene sheet functionalized with oxygen groups on its basal plane and at the edges, thereby exhibiting interesting steady-state PL properties. The PL spectra show a broad PL region between 300 nm and 800 nm (1.550 eV and 4.135 eV). This is a characteristic PL for oxygen plasma-treated, mechanically exfoliated, single-layer graphene sheet.38 The prominent PL peaks seen in the case of sample A and sample B are at 470 nm (IP1) and 600 nm (IP2), respectively. The increased PL intensity in sample A is due to less reduction (decreased C/O ratio), as observed by XPS studies (as discussed later). The sharper peak profile for the 2D peak also influences the PL spectra. An increase in FWHM of the 2D peak also corresponds to an increase in the intensity of IP1 peak, thereby causing a blue fluorescence in the PL spectrum. This can be due to lower sharpness of 2D peak, which is an indicative of increased defect regions in the layered structure of rGO. Therefore, the radiative recombinations occur from defect states in the vicinity of π* electronic band. First-principle calculations suggest formation of strips of sp2 carbon atoms in rGO compounds by covalent bonding of a large fraction of C atoms in the hydroxyl and epoxy units.39 In rGO compounds, containing a mixture of sp2 and sp3 moieties, the optoelectronic properties are primarily determined by the π and π* electronic states of the sp2 sites, which lie within the σ–σ* gap.40 It is well known that π bonding is weaker and has lower formation energy. So, there are a large number of disorder-induced localized states in the 2D planar lattice of laser annealed rGO. In this structure, there are perturbed C atoms attached to functional groups containing oxygen. Projected dihedral angles dictate the interaction of π states. Therefore, the structural-disorder-induced localized states are located primarily in the band tails of π-π* energy level or lie deeper in the energy band gap.41 The broad absorption or emission bands are caused due to the optical transitions between these disorder-induced localized energy states. Reduction causes a decrease in the number of these disorder-induced electronic states by deoxygenation. Therefore, well defined sp2 clusters are formed thereby diminishing the PL intensity. The newly formed numerous sp2 clusters with reduction also broaden and reduce the intensity of the PL spectra as it is quite evident in the case of sample B which is more reduced than sample A. Interestingly, it is also observed from Gaussian peak fittings that increased reduction reduces the IP1/IP2 ratio. This is quite evident from the fact that the newly formed sp2 states after reduction are closer to the π electronic states, thereby causing a reduction in the effective optical band gap in rGO, compared to zero bandgap of graphene. The newly formed sp2 moieties in rGO provide percolation pathways between sp2 clusters already present. Thus, the reduction of GO to rGO results in the formation of zero band gap regions in the rGO layered structure with some of the functional groups remaining still unreduced even after the reduction process. This ensures that the ratio of the zero gap sp2 to sp3 clusters is comparatively high enough in the rGO sheets to result in quenching of photoluminescence intensity due to weak carrier confinement.40 The original structure of GO consists of numerous disorder induced defect states within the π–π* gap and exhibits a broad prominent PL spectrum centered at 500 nm–600 nm. After reduction/deoxygenation, the number of disorder-induced states within the π–π* gap decreases, and an increased number of cluster-like states in the form of small and isolated sp2 domains are formed. The electron–hole recombination among these sp2 cluster-like states exhibits weak fluorescence at higher wavelengths. The heterogeneous electronic structures of rGO with variable sp2 and sp3 hybridizations formed after laser annealing through reduction give rise to broader and interesting features in the PL spectra.
PL spectra of samples A and B. There is a clear indication of the presence of two PL peaks (IP1 and IP2) after deconvolution of the PL spectra.
PL spectra of samples A and B. There is a clear indication of the presence of two PL peaks (IP1 and IP2) after deconvolution of the PL spectra.
D. Time of flight-secondary ion mass spectroscopy (TOF-SIMS)
Figure 4 represents TOF-SIMS of the rGO thin film formed by laser annealing along with the as-deposited DLC. The concentration of C in the as deposited and laser annealed structure remains the same, whereas there is a considerable increase in the concentration of O followed by a slight increase in OH group in the laser annealed rGO thin film. Reduced graphene oxide consists of –OH and CO functional groups on its basal plane along with –COOH at its edges. As compared to DLC (as deposited) thin films, there is an overall increase in the concentration of O and OH in the laser annealed rGO thin films. So, TOF-SIMS study was quite useful to probe the formation of rGO structure after laser annealing by indicating an increase in OH and O concentration. It also helps to rule out any impurity phases formed after the laser annealing process.
TOF-SIMS of a rGO thin film formed after laser annealing along with the as deposited DLC showing an increase in O and OH concentration after the laser annealing technique.
TOF-SIMS of a rGO thin film formed after laser annealing along with the as deposited DLC showing an increase in O and OH concentration after the laser annealing technique.
E. X-ray photoelectron spectroscopy
Figure 5 depicts C1s XPS plots of rGO samples A and B synthesized using 0.3 J cm−2 and 0.6 J cm−2 laser energy density. The O1s spectra are also shown in Figure 5. The curve fitting of the C1s spectra was performed using a Pseudo-Voigt peak fitting function after performing a Shirley background correction in Origin Pro 8.5.1 data plotting and acquisition software. The binding energy of the C–C and C–H bonds are assigned between 284.5 and 285 eV and chemical shifts to 288.5 eV and 286.7 eV are typically assigned to the C=O and –O-CH3 functional groups, respectively.42–45 According to the literature, graphite oxide has epoxide groups (C-O-C), which have similar C1s binding energy to that of C-OH.43 But, there can be large chemical shifts than expected which move the C-O-C XPS peak to the C=O peak range thereby causing a complete disappearance of the epoxide peak. The C1s spectra in Figures 5(a) and 5(b) predominantly have peaks of C-C, C-H, and C=O. Larger FWHM and broader tail of C1s spectra signify contributions from a variety of different carbon bonding configurations. The similarity of the C1s peak profiles and positions in samples A and B suggests insignificant change in the surface as well as in the depth of the sample with increasing laser energy density. The photoelectron kinetic energies of C1s are larger than O1s and so the sampling depth (for C1s) is larger. The rGO samples show a marked reduction in the intensity of C=O, indicating that the delocalized π conjugation was restored in the samples. The C/O ratio was calculated to be 17 and 25 for samples A and B, respectively. This ratio is an indicative of the reduction in the graphene oxide structure, and it establishes that sample B is more reduced than sample A. However, it should be pointed out that during thermal reduction, an extreme high pressure is created in the stacked GO layers, thereby causing disruption of the graphitic network by loss of carbon. Calculations using state equations show that a pressure of 40 MPa and 130 MPa are generated at 300 °C and 1000 °C reaction temperatures, respectively.41 A pressure of only 2.5 MPa, as predicted by the evaluation of the Hamaker constant, is sufficient to delaminate two stacked GO platelets.41 During the process of thermal exfoliation, considerable structural damage is caused by the release of carbon dioxide.46 Recent studies predict a 30% loss of the GO mass during exfoliation, thereby rendering the rGO structure with vacancies and topological defects throughout the 2D planar lattice.30,41 These defects adversely affect the electronic properties of rGO by reducing the ballistic transport path length of electrons and introducing scattering sites (sp3 moieties). Carbon-oxygen signatures are detected in the XPS spectrum of the samples, but these are far diminished as reported for thermal or solution based reduction processes and are also dwarfed by the C–C and C-H signals.
XPS of (a) sample A (laser annealed at 0.3 J cm−2) and (b) sample B (laser annealed at 0.6 J cm−2). The C/O ratio increases in sample B, indicating increased reduction with an increase in laser energy density. O1s XPS spectra are also shown in the inset.
XPS of (a) sample A (laser annealed at 0.3 J cm−2) and (b) sample B (laser annealed at 0.6 J cm−2). The C/O ratio increases in sample B, indicating increased reduction with an increase in laser energy density. O1s XPS spectra are also shown in the inset.
F. High resolution scanning electron microscopy
Figures 6 and 7 depict the high resolution SEM images of rGO thin films and flakes formed on Cu/TiN/Si(100) after laser annealing using different laser energy densities. The nanosecond pulsed laser heating leads to formation of super undercooled carbon at the film-substrate interface. Depending on the thermal conductivity of the substrate, nanosecond laser annealing can form different forms of carbon, for example, rGO (in this case) or nano, micro, and large area diamond.52 Large area rGO thin films are formed from filamentary structures consisting of nano globular carbon as shown in Figure 6(d). The filaments of nano globular carbon formed after laser heating process act as nucleating seeds for rGO thin films, as shown in Figures 6(b) and 6(c). The formation of the cellular structure is a result of interfacial instability at the solid-liquid interface, driven either by solute segregation or strain.47 Figure 7(a) is a rGO thin film at higher magnification, showing grains and grain boundaries with grain size measuring ∼200 nm. Increasing the laser energy density causes enough interfacial instability, thereby causing the delamination to form rGO flakes as shown in Figure 7(b) (using 0.7 J cm−2), Fig. 7(c) (using 0.9 J cm−2), and Fig. 7(d) (using 1.0 J cm−2). The SEM of rGO reveals nano globular carbon which acts as seed for the formation of rGO thin films. Figure 8 depicts high resolution SEM images of the rGO thin film formed onto Cu/TiN/Si(111). It is well known that there occurs a large-area growth of graphene related compounds on Cu(111) owing to symmetry compatibility. Different laser energy densities were used for synthesizing rGO thin films. Reduced GO thin films formed on Cu/TiN/Si(111) using a laser energy density of (a) 0.3 J cm−2, (b) 0.4 J cm−2, (c) 0.5 J cm−2, (d) 0.6 J cm−2, and (e) 1.0 J cm−2 are shown in Figure 8. Figure 8(f) shows the formation of large area rGO thin films using 0.6 J cm−2. The formation of uniform large area rGO thin films depends on laser parameters and the substrate used. As seen from the SEM images, formation of non-uniform (serrated/nanoglobular/nanoflakes) C nanostructures is observed when we used a laser energy density of 0.3 J cm−2, 0.4 J cm−2, and 0.5 J cm−2. At 0.6 J cm−2, we obtain more uniform large area rGO when Si(111) substrates were used. With the increase of laser energy density to 1.0 J cm−2, the uniform rGO thin film breaks up to form flakes and carbon nanotubes (CNT)-like structures.
(a) rGO thin film formed by using a laser energy density of 0.3 J cm−2. Figure (b) shows the interface between filamentary “stream-like” structure and the thin film, (c) shows a higher magnification of the interface revealing the formation of larger sized rGO grains, and (d) showing nano globular carbon in the filaments formed by laser annealing process which seeds the formation of rGO thin film.
(a) rGO thin film formed by using a laser energy density of 0.3 J cm−2. Figure (b) shows the interface between filamentary “stream-like” structure and the thin film, (c) shows a higher magnification of the interface revealing the formation of larger sized rGO grains, and (d) showing nano globular carbon in the filaments formed by laser annealing process which seeds the formation of rGO thin film.
(a) rGO thin film formed by using a laser energy density of 0.3 J cm−2. Increasing the laser energy density causes interfacial instability thereby causing flaking off as shown in (b) using 0.7 J cm−2, (c) 0.9 J cm−2, and (d) 1.0 J cm−2 laser energy densities.
(a) rGO thin film formed by using a laser energy density of 0.3 J cm−2. Increasing the laser energy density causes interfacial instability thereby causing flaking off as shown in (b) using 0.7 J cm−2, (c) 0.9 J cm−2, and (d) 1.0 J cm−2 laser energy densities.
rGO thin films formed on Cu/TiN/Si(111) using a laser energy density of (a) 0.3 J cm−2, (b) 0.4 J cm−2, (c) 0.5 J cm−2, (d) 0.6 J cm−2, and (e) 1.0 J cm−2. Figure (f) shows the formation of large area rGO thin films using 0.6 J cm−2. The formation of uniform large area rGO thin films depends on laser parameters and the substrate used.
rGO thin films formed on Cu/TiN/Si(111) using a laser energy density of (a) 0.3 J cm−2, (b) 0.4 J cm−2, (c) 0.5 J cm−2, (d) 0.6 J cm−2, and (e) 1.0 J cm−2. Figure (f) shows the formation of large area rGO thin films using 0.6 J cm−2. The formation of uniform large area rGO thin films depends on laser parameters and the substrate used.
G. Temperature dependent resistance measurement
Figure 9 represents temperature dependent resistance characteristics of A and B rGO thin films. Preliminary electrical measurements indicate improved electrical conductivity of the laser synthesized rGO thin films as compared to GO. Within the voltage range of −100 mV–100 mV, the I–V curves are Ohmic, which allows us to measure the resistance of the samples at different temperatures. Four-probe setup also helps to avoid the Schottky barrier effects during the electrical measurements. The resistance data points are collected and analyzed in the temperature range of 20 K–350 K. In both rGO samples, the value of resistance increases nonlinearly with a decrease in temperature which implies semiconducting nature of the thin films. The Hall mobility and resistivity of the rGO thin films prepared by our method vary between 3.5 and 8.5 cm2/V s and 7.5 × 10−3 and 1 × 10−2 Ω cm at 300 K, respectively. The resistivity is much lower in this case as the carbon gets zone refined (in the liquid state) by the laser to form rGO, thereby reducing the structural defects which impede the flow of electrons. Previous studies report resistivity of rGO as 1 × 10−2–1 Ω cm,48,49 and Hall mobilities as 2–200 cm2/V s depending on the chemical reduction technique.50,51 With decreasing temperature from 350 K to 20 K, we have noticed two completely different types of rate of change of resistance, i.e., at a high temperature region, the resistance increases slowly and at low temperature region it increases rapidly. So the transport mechanism follows two different kinds of model. Muchharla et al.52 experimentally found that the electrical transport mechanism is dominated by the band-gap scattering at high-temperature and variable-range hopping at low-temperature. Resistance vs temperature data of all the samples in the temperature range 20 K–190 K elucidate that the transport mechanism is dominated by Efros-Shklovskii variable range hopping (ES-VRH) and from range 190 K to 350 K the plot fits best with the Arrhenius equation. Owing to the fact that at sufficiently low temperatures, activated type of conduction is not possible, therefore, the variable range hopping mechanism plays a significant role in the conduction process in rGO thin films. The sp3 hybridization of C atoms and discontinuous sp2 clusters in the sp3 matrix play a major role in the reduced electrical conductivity in GO. The conductivity is improved after reduction of GO to rGO due to the creation of new sp2 clusters at the expense of defects in the form of vacancies and restructuring of atoms (Stone Wales defect). Eda et al.40 suggested that due to the reduction of GO, new domains of sp2 clusters are formed by the removal of oxygen groups. This provides percolation paths between sp2 domains already present. Initial sp2 clusters of GO do not take part in the conduction due to its sparse distribution in the sp3 (insulating) matrix. The new sp2 clusters, which are formed by the laser irradiation technique, help in electron transport. Newly formed sp2 clusters can be considered electronically as isolated states that aid to the hopping of the free electrons. As the number of sp2 clusters is fairly significant so the conductivity in the as grown rGO film is measurable.
(a) Resistance vs temperature characteristics, (b) the ES-VRH model fitting in a low temperature range, and (c) the Arrhenius conduction model fitting in a high temperature range of rGO thin films. The red arrow in figure (a) marks the crossover point from the ES-VRH model to the Arrhenius model.
(a) Resistance vs temperature characteristics, (b) the ES-VRH model fitting in a low temperature range, and (c) the Arrhenius conduction model fitting in a high temperature range of rGO thin films. The red arrow in figure (a) marks the crossover point from the ES-VRH model to the Arrhenius model.
To study the conduction mechanism in the rGO samples, different conductivity mechanism models are used to fit the experimental data. At a low-temperature regime, we found that our data fit best to a straight line with the ES-VRH model, shown in Figure 5(b) {ln(R) vs T−0.5} as compared to Mott-VRH or 3D models.53 The ES-VRH model considers specific form of single particle density of states in Coulomb gap to include the effects of Coulomb interactions and a modified Mott's argument of variable range hopping.54 The ES-VRH relationship can be expressed in terms of resistance, R as a function of temperature, T by55
where T0 is the characteristic temperature and the exponent ½ is independent of dimensionality of the sample. This characteristics temperature can be used to find out the localization length, ξ using the following equation:55
where = 2.8 is a constant, kB is the Boltzmann constant, e is the charge of an electron, is the vacuum permittivity, and is the effective dielectric constant of rGO thin films. From different studies, the value of is found to be 3.5 for rGO samples.56,57 From the fitted data demonstrated in Figure 5(b), we obtain characteristic temperatures which allow us to calculate the localization lengths for both rGO samples. From the values of localization length reported in Table I, we can infer that the longest localization length has been found in the sample with larger conductivity, i.e., sample B. Reduced localization length is found in less conducting samples.
Electrical measurement data of rGO thin films formed by laser annealing.
Sample name . | Localization length (nm) . | Activation energy (meV) . | Hopping energy (meV) . |
---|---|---|---|
Sample A | 916.21 | 4.67 | 4.21 |
Sample B | 4227 | 2.31 | 3.65 |
Sample name . | Localization length (nm) . | Activation energy (meV) . | Hopping energy (meV) . |
---|---|---|---|
Sample A | 916.21 | 4.67 | 4.21 |
Sample B | 4227 | 2.31 | 3.65 |
Hopping energy (Eh) is an important parameter calculated by the ES-VRH model in the case of 2D materials. The hopping energies for the samples are calculated by using the following formula and also tabulated in Table I:
where A is a constant, and is the conductivity. As lower hopping energy ease the transport of electrons in a sample, we found the maximum conductivity in sample B where minimum hopping energy is calculated. On the other hand, as sample A demonstrates increased hopping energy, more resistivity is observed in this sample. The hopping energy is a parameter which is generally used to determine the electron conduction by variable range hopping phenomena in between sp2 clusters in a matrix of sp3.
At a high temperature regime (190 K–350 K), the data are analyzed based on the Arrhenius temperature-dependent resistance equation
where Eg is the activation energy. The plotted data ln(R) versus inverse of temperature (T−1) fits best to a straight line in Figure 9(c). This implies that in a higher temperature regime, band-gap dominated Arrhenius-like temperature dependence plays the major role in electronic transport mechanism. Using the slope of the fitted lines, the calculated activation energy is found to be in between 2.35 meV and 37.54 meV, which is comparable with the previously reported values.58,59
The activation energy is less in case of sample B. So, at room temperature, it is easier for the electrons in the donor atoms to reach the conduction band. Completely opposite phenomena occur in the case of sample A, which has higher activation energy and, therefore, increased value of resistance at higher temperatures. Previous research reports indicate that at lower temperatures, the electrical conduction in rGO is by VRH phenomena and at higher temperatures by the Arrhenius model.58,60 This suggests that at higher temperatures, thermally excited carriers begin to dominate the electrical conduction. The temperature of this crossover is also indicated in Figure 9(a) by a red arrow. It has also been shown in earlier reports that increasing the extent of reduction in rGO decreases the crossover temperature due to the restoration of percolating sp2 network, thereby facilitating band-like transport.58,61 Thus, the activation energy calculated from the Arrhenius model corresponds to band transport. The work of Pearson and Bardeen62 also shows a decrease in activation energy that results an increase in carrier concentration. For each kind of defect/impurity, there should be a specific characteristic ionization energy also termed as activation energy. This energy is fixed under normal operational conditions.63 So the value of the activation energy is dependent only on the nature of the intrinsic and extrinsic defects/impurities in the samples. The variance in the activation energy of laser annealed rGO samples implies that the nature and the level of the defects and impurities are dependent on the growth parameters, namely, the laser energy density, wavelength, and pulse duration.
H. Laser-solid interaction
Laser annealing is accomplished by illuminating the thin film with laser of appropriate wavelengths, energy density, and pulse durations. Most of the laser light is absorbed within a thin surface layer (inverse of absorption coefficient) in a few hundreds of angstroms deep into the solid. This produces high temperature and melting in some cases, which helps to anneal out damage and zone refines the substrate. Laser annealing is carried out in air as the surface layer cools so rapidly (few hundreds of nano seconds) that the introduction of impurities from atmosphere is minimized. The highly localized annealing process can help formation of patterned features onto the substrate and will have profound implications on device processing.
The annealing process involves melting of the crystal in the near surface region, dopant diffusion in the molten state, and subsequently, liquid phase epitaxial growth from the underlying substrate. The laser annealing indirectly helps the dopants to zone refine to the surface for certain impurities with distribution coefficients much less than unity. The mechanism of annealing differs on the type of laser used. The diffusion coefficients in liquid counterparts are much higher than its solid phase and thereby give a reasonable understanding of the dopant profile after laser annealing.64,65 It has been seen that there are significant modifications taking place in the implanted region of the Si lattice as a result of laser annealing. There occurs a substitution of the dopant atoms in Si lattice sites. Ion back scattering and ion channeling techniques help to determine this lattice substitution.64 The lattice contraction (after substitution of lower atomic radii elements) and expansion can be measured by X-ray diffraction and ion channeling techniques. The lattice always expands or contracts 1D in out of plane conditions. High dopant concentrations can also cause a measurable strain in the substrate lattice. The time associated with laser annealing is too small (less than a micro-second) to cause formation of misfit dislocations which tend to relieve the strain and destroys the one dimensional change of a lattice parameter. The ease of transportation of dopants to surface after laser annealing is dependent on the dopant segregation coefficient in the film. During the process of solidification, the concentration of dopants in the liquid melt builds up until it exceeds that in solid. Since the surface is the last region to be solidified, the dopant segregates and stays there.64
Calculation of threshold energy is crucial for this study, as it will determine the minimum laser energy needed for melting the surface (Cu) thereby causing zone refining of solute (C in this case). The threshold energy () required for melting few layers of Cu-2%C thin film can be calculated using the equation66
where is the thermal conductivity of Cu-2%C alloy (=0.33171 W cm−1 K−1 at 1085 °C), is the difference between the substrate temperature (TiN) and melting temperature of the Cu-2%C alloy (=1065 K), is the pulsed laser width (=25 ns and 50 ns), is the reflectivity of the liquid Cu-2%C alloy at 248 nm (=0.4733), and is the diffusivity of Cu-2%C at room temperature (=0.342 cm2 s−1).
The maximum molten depth of the material as a function of laser energy density (J/cm2) is shown in Figure 10(a). Two different pulse widths corresponding to 25 ns and 50 ns are plotted for a comparative study. The thickness of the melted regions varies linearly with the pulse energy density. The slopes of the curves are related to the sensitivity of the melt depth to the pulse energy density. The x intercept measures the minimum energy required by the melt to propagate into the substrate, commonly referred to as Eth. As quite evident from the plot that threshold energy increases with the increase in pulse width. This can be rationalized as less heat is utilized in melting (more heat is being conducted away) as pulse duration increases. Eth for 25 ns and 50 ns is measured to be 0.167 J/cm2 and 0.247 J/cm2, respectively.
(a) Maximum melt depth vs pulse energy density (for two different pulse widths, i.e., 25 nm and 50 ns) as a result of laser irradiation using 248 nm a KrF laser and (b) melt depth vs time for different pulse duration for a fixed laser energy density of 0.3 J cm−2.
(a) Maximum melt depth vs pulse energy density (for two different pulse widths, i.e., 25 nm and 50 ns) as a result of laser irradiation using 248 nm a KrF laser and (b) melt depth vs time for different pulse duration for a fixed laser energy density of 0.3 J cm−2.
Variation of pulse duration affects the maximum melt depths (increasing the pulse duration decreases melt depth). So, the solidification velocities change considerably as we vary pulse duration. Figure 10(b) illustrates the position of melt front at different laser pulse durations for a fixed laser energy density (E = 0.3 J/cm2). The curves indicate that the surface starts to melt slowly due to the sudden change in reflectivity during the phase transition. As the laser pulse terminates, the melt front recede back to the surface. The initial stages of solidification see a low velocity but quickly reach the maximum when the steady state condition is achieved. With the increase in pulse duration, the onset of melting decreases with a decrease in solidification velocity. The maximum depth of melt also decreases with the increase in pulse duration. The time for peak melt depth increases with an increase in the laser duration.
The maximum melt depth () is calculated by
where
where is volume heat capacity of copper at its melting point (=3.7651 J cm−3 K−1), and L is latent heat of fusion of copper (=1828.6 J cm−3).
The melt in velocity () is expressed by
where is the temperature difference between maximum temperature and melting point of Cu.
Again is calculated by the relation
It was also calculated that the melt depth increases with the increase in substrate temperature. In all the above mentioned calculations, the substrate temperature was considered as 293 K. The depth of layer to be melted in Cu-2%(at)C thin film to form one layer, two layer, and three layer graphene is calculated to be 22.44 nm, 44.88 nm, and 67.32 nm, respectively.
IV. CONCLUSIONS
Reduction of GO into rGO reduces the resistance in thin films due to the formation of sp2 clusters which act as percolation pathways for electrons in a sp3 matrix. The laser energy density plays a crucial role in determining the reduction of rGO thin films as a decrease in resistance by 50% can be achieved with increasing laser energy density from 0.3 J cm−2 to 0.6 J cm−2. The decrease in resistance with increasing sp2 fraction demonstrates that the restoration of π-π* bonding improves charge percolation pathways in the rGO basal plane. Figure 11 is a schematic representation of the rGO structure showing sp2 clusters in the sp3 matrix along with the defect states formed between π-π* electronic bands. Erickson et al.66 and Gómez-Navarro et al.67 observed that such improvement occurs at the expense of increasing topological defects in the samples. The salient features of improved conductivity can be accounted for by (i) less hopping energy, (ii) longer localization length, (iii) minimal defect density, and (iv) smaller activation energy. Better conductivity of laser annealed samples suggests the formation of numerous sp2 clusters by the removal of oxygen groups in the sp3 matrix. The presence of a large number of small sp2 clusters is also consistent with reduced PL emission and an increase in the IP2/IP1 ratio. These studies provide a better understanding of the factors (covalent bonding, adsorption, π-π interactions, and lattice incorporation) influencing physical properties of functionalized graphene. These factors control the electronic properties of rGO synthesized by the novel nanosecond laser annealing technique. The ratio of sp2 to sp3 hybridized carbon atoms in functionalized graphene plays a critical role in controlling the electronic properties. At lower temperatures, the charge carriers are having limited thermal energy to move in the plane, which can be drastically reduced with the increase in defect density with reduction. Increased sp2 fraction helps the ES-VRH mechanism thereby introducing enhanced electrical mobility. This variable range hopping conduction mechanism can lead to exciting electrical properties at low temperatures. As compared to other 2D materials, synthesis of rGO is scalable and, therefore, holds tremendous importance in electronics industry. Recently, rGO flakes are synthesized utilizing less toxic and non-explosive reducing agents to alleviate environmental issues.68–70 Reduced graphene oxide having a C/O ratio as high as 7.15 is successfully prepared using Caffeic acid as the reducing reagent.68 It has also been shown that rGO can be used as gas sensors and supercapacitors.68 The principle of a gas sensor is based on efficient transfer of charge/electron between the adsorbed gas and the sheets of rGO. Fast responses (changes in resistance) are observed in rGO based gas sensor with the introduction of 100 ppm NO2 and 1% NH3 in diluted air.68 This is due to the fact that NO2 and NH3 act as electron acceptor and donor thereby decreasing and increasing the resistance, respectively, in rGO based gas sensors.68 The huge specific area and high electrical conductivity favor the use of rGO sheets as supercapacitors. Research study shows that the cyclic voltammetry (CV) curves of a rGO based working electrode using KCl and tetra-ethylammonium tetrafluoroborate in acetonitrile solvent (TEABF4/AN) have more enclosed area and better quasirectangular shape as compared to GO based devices.68 These indicate faster charging and discharging response in rGO based supercapacitor devices. Hybrid mixtures of rGO and MoS2 are used as microwave absorbing electronic materials owing to its effective microwave absorption bandwidth of 5.72 GHz for a thickness less than 2 mm.71 Composites of MoS2 nanoflower and rGO are prepared by hydrothermal and ultrasonic exfoliation techniques and are also used as anodes in sodium-ion batteries.72 The composite shows an extremely high specific heat capacity of 575 mAhg−1 at 100 mA g−1 in the range of 0.01 V–2.6 V and 218 mAh g−1 at 50 mA g−1 when discharged in the range of 0.4 V–2.6 V.72 Owing to its high electrical conductivity, high surface area, and porous structure, rGO thin films are also used as an electrode material in enzymatic biofuel cells.73 This interesting correlation with the structural and electrical properties in rGO thin films and epitaxial integration of rGO with Si will open up new avenues for better design and function of graphene oxide based electronic devices. Using this technique, we are able to create epitaxial graphene, graphene oxide, and reduced graphene oxide integrated with silicon at ambient temperatures and pressures for next-generation multifunctional solid-state devices.
Schematic diagram depicting the structural, optical, and electrical properties of the rGO thin film formed by laser annealing. The dark gray balls in the structural drawing represents C, the yellow ones H, and the red ones O.
Schematic diagram depicting the structural, optical, and electrical properties of the rGO thin film formed by laser annealing. The dark gray balls in the structural drawing represents C, the yellow ones H, and the red ones O.
ACKNOWLEDGMENTS
We are grateful to Fan Family Foundation Distinguished Chair Endowment for J. Narayan. We are also very pleased to acknowledge the support of National Science Foundation (ECCS-1306400) and technical help and useful discussions with John Prater, Roger Narayan, Fred Stevie, Elaine Zhou, Sudhakar Nori, and Sandhyarani Punugupati.