5. Synthesis of Graphene Design of Experiments

Transcription

1 CHAPTER 5 5. Synthesis of Graphene Design of Experiments 5.1 Introduction Large area graphene with good transparency and low sheet resistance is required for many applications including miniature transistors, piezo electric based transparent electrodes and solar cells, touch sensors and flat panel displays as a replacement for ITO indium tin oxide (ITO), [110, 111]. In general, graphene is synthesized by methods such as i) mechanical cleavage ii) chemical exfoliation iii) chemical vapour deposition, and iv) epitaxial growth on SiC layer. The mechanical cleavage and chemical exfoliation methods can produce only a small area of graphene (in the range of 5-10 um), but the yield is high compared to other two methods. In case of epitaxial growth on SiC layer, growth of Graphene layers is extremely hard to control and it is highly difficult to remove the developed graphene sheets. Compared to these methods, CVD synthesis of single-layer graphene with large-area on copper foils is widely preferred, using this method number of layers can be precisely controlled with better tunability of the synthesis parameters and it can be easily transferred to a suitable substrate[18, 11,11, 114]. At present, the CVD method is of great interest for achieving controlled and uniform growth of the graphene with single, two or a few layers on metal substrate by tuning the important parameters. In a CVD process, a number of factors such as the catalyst, gas flow rates with proper composition, chamber pressure, temperature, growth duration and cooling rate of the substrate will influence the morphology (shape), hybridization of carbon atoms and the number of graphene layers. A number of studies have attempted to study the effect of various factors on the graphene properties [115, 116, 117, ]. However, these studies varied one parameter at a time and did not vary multiple synthesis parameters simultaneously. Such experiments are generally unable to capture interactions between different factors. A design of experiments approach provides the ability to quantitatively relate the effect of each of the factors on the responses (graphene quality metrics and number of layers), including interaction effects. 7

2 Umair et al has investigated the effects of reaction conditions, growth time, annealing profile, and flow rate of various gases for improving quality of the graphene growth [117]. Bhaviripudi et al report that, the low pressure CVD favour the formation of large uniform graphene on copper foil with methane as carbon precursor and higher methane concentration with atmospheric pressure CVD (AP CVD) favour to formation multilayer graphene sheet. [116]. Recent reports on roll-to-roll synthesis of graphene over copper foil indicate that good quality graphene can be synthesized at atmospheric pressure. Atmospheric pressure CVD offers a cost advantage and is more suitable for industrial large scale production [110]. Copper catalyst is most suitable for obtaining uniform and high-quality single layered graphene compared to other transition metal catalysts because copper has lower carbon solubility [60]. There is a requirement for a systematic statistical study for tuning these parameters to obtain the best quality of graphene with controllable layers. Three important factors governing the quality of graphene in a CVD process are (i) temperature, (ii) nature of precursor and (iii) carbon precursor flow rate. Li et al and Gan et al indicate that the quality of graphene is highly dependent on the decomposition process of carbon precursor [94, 119]. Aromatic or non-aromatic hydrocarbons can be used as graphene precursors. The most commonly used carbon source is methane gas (non aromatic compound), it favours the formation of the methane radical at the high reaction temperatures that are normally used (~1000 C). CVD graphene growth at such high temperatures is undesirable for many practical reasons, such as difficulty in handling, storage and cost for maintaining purity. Aromatic carbon precursors are advantageous since they tend to adsorb better on the catalyst surface (less tendency to desorb) and more crucially, their dehydrogenation energies are lower than that of methane. This is because of the sp hybridisation and π π interaction in the aromatics which improves London dispersion force on metal substrates. Thus, using aromatic precursors, graphene growth becomes possible at much lower temperatures [10, 89]. Flow rate of carbon source is important for the synthesis of graphene sheet because the control of carbon radical concentration is determined through flow rate of carbon sources. Higher amounts of the carbon source gives more carbon radicals on metal surface and results in formation of graphite (increasing number of the graphene layers) and amorphous carbon. Amorphous carbon hinders the catalyst efficiency by blocking the catalyst surface and preventing catalysis. Higher hydrocarbon flow rate results in a shorter nucleation time with 7

3 more irregularity in graphene shapes. Controlling the hydrocarbon flow rate has been shown to control the number of layers. [59]. A statistical design of experiments [DOE] enables the determination on the influence of different synthesis parameters on the graphene properties and the mutual interactions between various parameters. The regression equation relates the effect of these parameters on graphene properties; it also enables optimization of process conditions for tuning certain properties. However, there are numerous factors that could potentially influence the end properties but practically there is a need to restrict the number of factors studied, to reduce the experiments needed to a reasonable number. Temperature, nature of the hydrocarbon sources and flow rate of carbon sources were varied for this DOE analysis. The rest of the factors were held constant throughout the study. Based on the literature and some preliminary results obtained from our experiments, we decided to maintain pressure, catalyst, reaction time and the catalyst preparation method constant in this DOE study. The discussion below outlines the rationale behind maintaining these factors as constant. The responses studied in this DOE were Raman spectroscopy features such as FWHM of D band and I G /I D of the samples obtained through micro Raman analysis. Raman spectroscopy is an important characterization tool for studying properties of graphitic materials such as graphene, carbon nanotube and other carbon nanomaterials [11]. It allows monitoring the quality of the graphene synthesized (extent of sp hybridization), the number of graphitic layers, etc. It also allows mapping of small areas on the sample surface to plot the distribution of the measured quantity over the selected sample surface. Coupled with Raman analysis, some morphology analysis was also conducted, which allows the estimation of the thickness and crystallinity, without indicating the nature of bonding the carbon atoms. 5. Experimental method 5..1 CVD Setup The experimental apparatus for the CVD synthesis of graphene consists of a 4 long and diameter quartz tube with an induction heating a coil wrapped around a 4 -length of the tube. 74

4 Rotameters were used to control the flow rate of the gases. For the liquid precursor (benzene), a syringe pump was attached to the quartz tube for controlling the benzene flow rate. For the solid precursors a custom arrangement was made and calibrated to yield desired flow rates of the precursors in the vapour phase. Excess quantity of the molten precursor (naphthalene or anthracene) was initially kept in a round bottom-flask (RB flask) which was immersed in a heated oil bath. An Ar-H gas mixture was bubbled through the molten carbon precursor and the vapor from the RB flask was introduced into the CVD quartz tube. The setup was calibrated to yield different precursor flow rates at different temperatures of the molten precursor. The feed-side of the quartz tube was preheated using tape heaters to prevent condensation of the carbon precursors on the quartz tube. 5.. Catalyst Preparation Commercially available Alfa-Aesar copper (Cu) foils of x cm with 5 µm thickness ( % pure) were used as the catalytic substrate for graphene growth. The Cu foils were cleaned with acetone followed by isopropanol and then heated in an acetic acid bath for 10mins at 60 o C (for the removal of the oxide layer from Cu). The foil samples were placed in the quartz tube and dried using argon at 400mL/min for 10 min. The temperature was ramped up to the growth temperature and hydrogen flow was started at 100mL/min for 0 min in order to further reduce the surface oxide. 5.. Graphene Synthesis Solid flow rate measurement Apparatus: Silicon Oil bath, Round bottom flask - 50 ml, Round bottom flask ml, Thermocouple, Condenser tube, Receiver tube, Argon Inlet tube, Ice pack, Dimmer stat- Nos, Tape Heater, Multimeter, Teflon tape. Solid flow rate measurement is performed using the principle of diffusivity. It is calculated as the amount of the precursor material diffusing in Argon and deposited during the cooling process.sufficient amount of the required precursor whose flow rate has to be measured is filled in a 50ml RB Flask. The Argon gas tube is inserted for purging through tubing and the tube inserted in a position such that the bottom of the tube will be below the level of the solid. The tubing is avoided by placing just 75

5 above the level of the precursor because the boundary layer of the molten precursor will be impugned. The tube must be dipped inside precausor in order to facilitate the diffusion of the precursor into the argon gas. The tubing is free from any air leaks and seals the RB flask and finally it is sealed with Teflon tape. The condenser is connected to the Argon inlet tube, RB flask is setup and sealed. The condenser is tied with a tape heater to prevent the deposition of the precursor on the walls of the condenser and the heater is connected to a dimmerstat. The other end of the condenser is connected to the receiver tube (with an outlet for the argon gas) which is connected to another RB flask (100mL) for collecting the crystallized solids. The RB flask containing the precursor is placed in a silicon oil bath such that the oil level is higher than that of the level of the precursor. The RB flask is placed in the bath with the help of a stand and held firmly, this process maintained carefully for the reproducibility of the experiments to get accurate results. The receiving end of the setup has 100 ml RB flask and the receiver tube which is placed in an ice pack. The tape heater and the oil bath were connected to two different dimmerstats. A thermocouple is placed in the oil bath for monitoring the temperature levels. A voltmeter is placed in the dimmer stat and the potential is maintained to avoid the overheating of the oil bath. Extreme care has been taken to prevent the experimental setup even from mild movements. Hence after the calibration, the synthesis process of graphene is started immediately from a particular precursor to get the accurate results. Initially the mass of the empty receiver tube and RB flask setup is noted. It is made sure that the setup is sealed properly at the right places with Teflon tape. Teflon tape is because of its withstanding capacity at high temperatures. The heating process of the oil bath is started and the temperature is slowly increased. The temperature is increased till it reaches a few degrees below the boiling point of precursor and it is held constant by adjusting the temperature on the dimmer stat. The tape heater is set at a temperature above its boiling point such that the precursor doesn t condense along the walls of condenser tube. Once the precursor reaches molten state, the argon flow is started and is controlled by a rotameter and it is adjusted to a flow-rate of 400 ml/min. The argon gas is allowed to bubbled 76

6 for a period of 15 minutes. After 15 minutes, the flow is stopped and it is allowed to reach room temperature. Once the room temperature is attained, the receiving end of the condenser tube is removed carefully without moving the apparatus and the mass is measured. The difference in the weight gives the amount of precursor material which is reacted with the argon. While using naphthalene and anthracene as precursors, a tape heater was tied around the argon inlet tube to prevent the condensation of the evaporated precursor on the tube. This was especially important for anthracene because of high boiling point of anthracene. If the mixed vapors of argon-anthracene come in contact with the glass tube, which is at a much lower temperature than the boiling point of anthracene, anthracene can condense on the tube, leading to errors in the calculation of the flow rate of the precursor. Table 5-1 Experiment condition for synthesis of DOE Samples Precursor Temperature ( o C) Experiment Time (min) Wt. of beaker before expt (gm) Wt. of beaker after expt (gm) Wt. of precursor deposited (gm) Flow rate of precursor (gm/min) Benzene ml/min Benzene ml/min Benzene ml/min Benzene ml/min Naphthalene Naphthalene Naphthalene Naphthalene Naphthalene Naphthalene Naphthalene Anthracene Anthracene Anthracene Anthracene

7 The prepared catalyst was maintained at the synthesis temperature and the hydrogen flow rate reduced to 40mL/min. The carbon precursor was introduced (Table 5-1) into the quartz tube for 15 mins. After the synthesis of graphene, the quartz tube was allowed to cool to ambient temperature naturally under an Ar-flow of 100mL/min. 5. Characterization Figure 5-1 Synthesis of graphene from aromatic precursors using DOE Raman characterization was carried out on a Jobin Yvon Horiba LABRAM-HR 800 instrument with a laser source of 488nm (E=.5ev) and optical lens with optical magnification is 50x, a spot size of 0.59μm, a single monochromator and a Peltier cooled charge-coupled device. All the CVD graphene samples are etched from copper foil using ferric nitrate solution and then transfer to silicon wafer. Raman mapping was performed over 78

8 an area of 0 x 16 μm on each sample, with each map consisting of 0 points of measurement. 5.4 Design of Experiments In this study, three-factors, three-level experimental Box-Behnken design is used in order to investigate the effect of synthesis parameters on the responses for the graphene quality and quantity. The factors varied in the study were temperature of deposition ( 1 ), carbon flow rate ( ), and the number of benzene rings in the carbon precursor ( ). In this design, the model relating to the response of Y with the input parameters 1,, were derived as: Y a 0 a1 1 a a a 4 1 a5 a6 1 a7 1 a8 a9, Where a 1, a, a are the linear coefficients; a 4, a 5, a 6 are the interaction coefficients; a 7, a 8, a 9 quadratic coefficients and ε is the error term. The factor levels were coded as 1 (for high), 0 (middle) and -1 (low). The regression analysis, analysis of variance (ANOVA), response surface, contour plots and optimizations plots were generated using the statistical software MINITAB 15. The input parameters were selected and their ranges are shown in Table 5-. In an atmospheric CVD set up operating isothermally, the deposition and growth of graphene are governed by the relative rates of gas-phase decomposition of the carbon precursor, the nature of the decomposed product, the deposition of the decomposed product on the copper surface, surface diffusion of the product, dehydrogenation/rearrangement of the product to form sp -hybridized structures, and cyclization to form graphene domains [10]. These processes are governed by the temperature of operation ( 1 ). Further, the amount of carbon in the feed ( ) determines the amount of decomposed product, which further determines the structure and quality of the graphene film produced. Considering the amount of carbon fed to the reactor rather than the flow rate of the precursor helps in comparison across different precursors. Finally, the structural differences in the carbon precursors chosen are captured by the third parameter for study, namely the number of benzene rings in the precursor ( ). It is known that the presence of benzene rings in the carbon precursor, at least at lower temperatures, results in little ring opening but formation of radicals that form better-quality graphene [94]. Thus, these three parameters studied are expected to reasonably capture not only the relative rates of the sequential/parallel processes involved in graphene synthesis but also the different possible mechanisms involved in graphene synthesis. In such complex 79

9 processes involving the interplay of different processes that could lead to entirely different mechanisms for the same graphene formation, it is to be anticipated that only preliminary indications may be obtained on how to design graphene synthesis for desired structure and quality. To that end, this work throws light on the nature of more detailed studies that need to be conducted to better determine graphene deposition/growth mechanisms under different conditions as well as control relevant process parameters for desired structure and quality. Table 5-: Levels of parameters chosen in the design of experiments Factors Different levels in Box-Behnken design Coded levels Abbreviation Low (-1) Middle (0) High (1) Temperature C Carbon flow rate (mg/ min) No. of the benzene rings in the carbon precursor. 1 Table 5-: Responses with R and adjusted R were indicated S.NO Response Surface Regression Abbreviation R % Adj- R % 1 Average FWHM of D peak Y Ratio of the intensities of the G and D Y peaks in the Raman spectrum I G /I D No. of points in the sample with Single layer graphene (SLG). 4 No. of points in the sample with Bi layer graphene (BLG). 5 No. of points in the sample with Few layer graphene (FLG). Y Y Y Benzene,. Naphthalane,. Anthracene 80

10 5.5 Results and Discussion Table 5-4: Raman features of the all the DOE samples Factors Responses Run Order Temperature ( C) Carbon flow rate (mg/min) No. of benzene rings I D /I G ratio D (FWHM cm -1 ) I G /I D ratio G (FWHM cm - 1 ) SLG BLG FLG

11 5.5.1 Response surface regression analysis The synthesis conditions (factors) and the corresponding responses for each of the experiments in the Box-Behnken design are shown in Table 5-. Regression equations that relate the factors with each of the responses were generated. Only the significant terms (p<0.5) were retained in the model equations. The fitness of each model was analyzed by the value of coefficient of determination (R ). High R values indicate that the model is a good predictor of the response. The adjusted coefficient of determination (adjusted R ) is generally lower than R and a small difference between the two indicates that the model does not contain unnecessary terms, which tend to spuriously inflate R. Table 5- reports the R and adjusted R values for the model equations for each of the responses Regression analysis and interaction plot of FWHM of D band Y , Equation 5-1 Y 1 = Response for D_FWHM 1 = Reaction Temperature ( C) = Flow rate As per Eqn (5 1) the large negative coefficients of and 1 indicate that low temperatures are highly favorable for the formation of single layer graphene (low Y 1 values). Similarly, the negative coefficient of 1 also indicates that low carbon flow rates are favorable for the formation of single layer graphene (low Y 1 values). The interaction plots between 1 and (Figure 5- top-left) for the Y 1 response show that the FWHM of the D band value decreased, which indicates the formation of the single or bilayer graphene at the low (800 C) and high (1000 C) temperatures with medium carbon flow rate, whereas the FWHM of D bands increased at low and high levels of carbon flow rate with the same temperatures, indicating the formation of few-layer graphene. The number of layers monotonically decreased at the medium temperature value (900 C), with carbon flow rate. 8

12 Figure 5-: Interaction plot for FWHM of D band The interaction plot between 1 and for Y 1 (Figure 5- top-right) shows that the low and the high values of (benzene and anthracene) and the high and low temperatures favor the formation of graphene with low number of layers. For the medium value of (naphthalene), for the same temperatures, a higher number of layers of graphene are formed. For the medium value of temperature, however, the number of layers monotonically increases with. The interaction plot between (carbon flow rate) and (no. of benzene rings) (Figure 5- bottom-right) shows that benzene and anthracene at low and high carbon flow rates favor an increase in the number of layers, while with naphthalene, at the same low and high carbon flow rates, a low number of layers are obtained. For the medium value of carbon flow rate (7.5mg/min), the numbers of layers are highest at the medium value of (naphthalene), while they are low for benzene and the lowest for anthracene. Though the above observations may be made from the interaction plots, the best-fit model is unable to capture any effect of the number of benzene rings (either as a single variable or as an interaction effect) on the number of layers of graphene deposited. It is probably because of this that the R and the adjusted R values for the regression model are not high. A more elaborate design of experiments (ex. more levels) may be needed to capture the effect of the number of benzene rings in the carbon source on the number of layers of graphene formed. 8

13 5.5. Quality of the graphene measured by Intensity ratio of I G /I D Y Equation Y = I G /I D ratio 1 = Reaction Temperature ( C) = Flow rate = No. of benzene rings in the carbon source - The model predicts that the intensity ratio of I G /I D decreases with increasing temperature, flow rate of carbon precursor, and the number of benzene rings in the carbon precursor. While it predicts a positive influence of the combination of flow rate and number of benzene rings, this causes a qualitative change only for anthracene at low flow rates. Figure 5-: Interaction Plot for Intensity ratio of G and D bands (I G /I D ) Interaction plot between 1 (temperature) and (carbon flow rate) for Y (Figure 5- topleft ) shows that the low (14 mg/min) and high (57mg/min) values of at low (800 C) and medium (900 C) temperatures favor the formation of graphene from naphthalene with high crystallinity (I G / I D ratio > ), while medium (5.5mg /min) carbon flow rate leads to the formation of graphene with relatively lower crystallinity (I G /I D ratio 1) at all three synthesis temperatures (800, 900, 1000 C). The interaction plot between 1 (Temperature) and (No. of the benzene rings) for Y Figure 5- top-right) indicates that the medium no. of the benzene rings (naphthalene) shows that the I G /I D ratio is 1 at all the three reaction temperatures ( C) and the I G / I D ratio is >4 was observed from anthracene at 900 C. 84

14 We observed more defective graphene samples at medium temperature with low (1) and high () number of benzene rings. Interaction plot between and For Y (Figure 5- top bottom )indicates that the low carbon flow rate (14mg /min) with lower no of the benzene ring (benzene )and higher carbon flow rate (57 mg /min) with higher no of the benzene ring (anthracene) have I G / I D ratio is > 4. The same I G / I D ratio value observed at naphthalene with all three flow rate (14 57 mg/min). Naphthalene at 14 mg /min at 800 C shows better quality compared to other samples with naphthalene precursor. This indicates that lower temperature with lower flow rate favour for graphitization was very good. Because of low flow rate of the naphthalene at 800 C decomposition and formation of naphthalene radical, and uniform diffusion on the surface of the copper foil is likely to occur, and this leads to improved graphitization [1, 1]. At higher (1000 ) temperature and higher flow rate (57mg/min) the carbon free radical concentration and growth rate increase. After graphene growth, extra carbon radicals remaining will form amorphous carbon on the copper foil and this reduces the G/D ratio values. Benzene at 900 C with flow rate of 14 mg /min gives better result of G/D ratio compared to naphthalene. Anthracene at 900 C with higher flow rate 57mg /min show best result of G/D ratio (8.16) value compared to others sample synthesis from benzene and anthracene carbon precursor. At higher temperature, saturated active carbon related radicals have more energy, which is easily overcome the nucleation barrier energy and then for bi few layer graphene layer. DOE equation indicate that low temperature with higher flow rate of anthracene have good quality of graphene ring. According to literature survey increasing benzene ring, will improve the London force between precursor and copper foil [10] No. of points in the sample with SLG Y 4 Equation 5- Y = No. of points in the sample with SLG 1 =Reaction Temperature ( C) = flow rate = No. of the benzene rings 85

15 Figure 5-4: Interaction plot for No. of points have single layer graphene in the sample As per Eqn (5-) it is observed that while and have a positive influence on Y, 1 has an inverse relation with Y. In other words, increasing the carbon flow rate and the number of benzene rings in the carbon source favors the formation of single-layer graphene. On the other hand, decreasing the reaction temperature favors single-layer graphene. In addition, the model shows a strong inverse dependence on the combination of the number of benzene rings and the reaction temperature. Therefore, one could infer that the number of points with single layer graphene is expected to be more when either benzene is used at high temperature or anthracene is used at low temperature. This is indeed observed in the interaction plots as well No. of points in the sample with BLG a 86

16 b Figure 5-5: (a) Interaction plot for presence of No. of the bi layer graphene (a) and few layer graphene (b) in the samples Y Equation 5-4 Y 4 = No. of points in the sample with BLG 1 = Reaction Temperature ( C) = flow rate = No. of the benzene ring The model, given by Equation (5-4), predicts that the formation of bilayer graphene is favoured by lower reaction temperatures, lower carbon flow rates, and the choice of benzene over naphthalene or anthracene as the carbon source. On the other hand, the interaction plots, shown in Figure 5-5 (a), along with Table 5-4 indicate that, overall, the experimental conditions mostly favour the formation of bilayer graphene. More specifically, the interaction plots for bilayer graphene seem to indicate no strong dependence on temperature, but a preference for low carbon flow rates and anthracene. The model does not clearly bring out the dependence of the three process variables chosen on the outcome, viz., formation of bilayer graphene. 87

17 5.5.6 No. of points in the sample with FLG Y Equation 5-5 Y 5 = No. of points in the sample with FLG 1 = Reaction Temperature ( C) = flow rate = No. of the benzene ring The model predicts that higher temperatures, high carbon flow rates, and with more number of benzene rings in the carbon source, few-layer graphene should be formed. It further indicates an inverse effect by the combination of (carbon flow rate) and (number of benzene rings) but not strong enough to counteract the direct relationship of the individual variables on the amount of few-layer graphene formed. That few-layer graphene would form when more carbon is fed to the system is intuitive. However, the interaction plots in Figure 5-5 (b) present a more complex picture. First of all, few-layer graphene is formed mostly at 900 C rather than 800 or 1000 C. Further, while napthalene or anthracene as carbon source seems preferable for few-layer graphene, the high carbon flow rate (57 mg/min) is clearly not preferred Mechanism of Graphene Growth Graphene growth from aromatic precursors (benzene, naphthalene, and anthracene) in an atmospheric pressure RF-CVD reactor involves the following steps: Flow of the carbon source/precursors in the boundary layer above the graphite susceptor Temperature-dependent gas-phase decomposition of carbon precursors into active radical species Deposition of the active radicals onto the surface of the copper catalyst Surface diffusion of the radicals on copper Dehydrogenation of the radicals Cyclization of the radicals into 'benzene' rings Stabilization of the radicals/rings on copper surface through secondary bonding interactions 88

18 Covalent bonding of the radicals/rings with a growing graphene sheet. The rates of each process are dependent on a number of factors including temperature, residence time of the carbon precursors/radicals in the gas phase, the nature of the carbon precursor (ex. number of rings in the precursor), and the surface structure of copper catalyst. Therefore, the resultant growth of graphene (ex. number of layers of graphene and the extent of defects) is dependent on the complex interplay of the rates of various processes. Depending on which step(s) is slowest and therefore rate limiting, the graphene quality is determined. Possibly the best way to grow large-area, uniform films of single-layer graphene is to design the process such that the deposition of active radicals on the copper surface is the ratelimiting step, even while ensuring that stabilization of the deposited radicals/rings is not very strong so that there is sufficient time for them to find the growing graphene sheet. This is, however, a major challenge, as the preliminary experiments bear out. It can be concluded from these studies that the carbon flow rates have to be lower than the lowest flow rate (14 mg/min) used in this study to grow large-area, single-layer graphene films. In other words, with benzene, naphthalene or anthracene as carbon precursor, at temperatures ranging from 800 to 1000 C, it is essential to have lower pressures (possibly through vacuum) than those maintained in the experiments. In order to draw accurate prescriptive conclusions about how the relative rates of the various processes affect the quality of the obtained graphene, the individual steps and their rates and mechanisms have to be examined in detail. This is essentially a simulation problem that has to be fed with results from designed experiments. Given these limitations, an attempt is made below to interpret the results obtained. Temperature and velocity profile of the gases above the precursor In this problem, at the flow rates and temperatures studied, radiation is the most significant form of heat transfer. Boltzmann Number (Bo = C P R/ T ) for benzene at 107 K in the RF-CVD reactor is of the order of 10 7, which signifies its dominance over convection. Furthermore, Prandtl Number is also less than one. This implies that conduction effects are stronger than convection, though of similar order of magnitude. Therefore, it is safe to 89

19 assume that heat is transferred to the carbon precursor and inert gases through radiation. Assuming the graphite susceptor to be a perfect black body, the rate of radiation of heat is so fast that the gas phase in the boundary layer reaches the susceptor temperature within few millimeters from the inlet. Therefore, a uniform temperature profile in the gas phase boundary layer may safely be assumed. Gas-phase decomposition of carbon precursor It is known that the temperature in the gas phase determines the radicals that are formed. At lower temperatures, ring opening is unlikely and more phenyl-type radicals are formed. On the other hand, at 1000 C and above, it is likely that ring opening becomes dominant and smaller alkenes and their radicals are the common species. As yet, there are few well-defined studies of the radical species formed and their life times during the gas-phase decomposition of benzene, naphthalene, and anthracene as a function of temperature and residence time [14-17]. In the high temperatures such as those established in the gas phase, a dynamic equilibrium exists for the gas-phase decomposition reaction of carbon source into active radical species. The equilibrium conversion of the carbon source into active radical species is a function of temperature, flow rate and the number of rings in the carbon precursor. It is expected that gas-phase decomposition of the carbon precursors is endothermic. Thus with higher temperatures the rate of decomposition should increase. However, since the heats of the decomposition reactions are unknown, it is unclear as to how these reactions affect the temperature profile of the gas phase. With increase in the number of benzene rings in the carbon source, the rate of decomposition should decrease. Deposition of active radicals on the substrate A concentration gradient develops within the boundary layer due to the deposition of the radical species on the catalyst surface and the consequent diffusion of these radical species radially. An estimate of the Schmidt Number (< 1) reveals that diffusion is the dominant mechanism in the boundary layer compared to convection. Rate of deposition of the active radical species onto the surface ought to be proportional to the temperature and the size of the 90

20 radical molecule. Thus, one ought to observe faster rates of deposition (and consequently larger number of layers, with more defects) at high temperatures and for carbon sources with higher molecular weight. Surface diffusion, stabilization, dehydrogenation, cyclization, and bonding of active radicals on copper The deposited active radicals have to then diffuse along the surface to the growing edge of graphene; resistance is encountered by the stabilization of the aromatic rings by way of London dispersion forces. As the number of rings in the radical increases, its diffusion is slower and it is more likely to be stabilized by the dispersive interactions between the radicals and copper. On the other hand, when the rate of deposition of the active radicals on the surface is high (for example, due to higher carbon flow rate), the radicals are also less likely to diffuse long distances on the surface because they are more likely to encounter another radical or a growing graphene sheet to bond to. Once the surface is reasonably saturated with monolayer graphene, the rates of surface diffusion change significantly. The radicals have much stronger π-π interactions with the underlying graphene layer. Therefore they are likely to diffuse shorter distances, form bonds sooner, and increase the defects further. Higher temperatures weaken the dispersive interactions of the radicals with copper and favour surface diffusion. However, they also hasten the rates of dehydrogenation, cyclization, and attachment to a growing graphene sheet. Therefore higher temperatures can favour growth of defective graphene sheets as well as multiple layers of graphene. Studies on the energetics (adsorptive energies) and rates of surface diffusion, dehydrogenation, and cyclization of the various radicals possibly arising from the gas-phase decomposition of carbon precursors will help interpret the experimental results more conclusively. However, it may be concluded that among the three carbon precursors studied, benzene (as the smaller precursor) is probably more favourable for surface diffusion, dehydrogenation and cyclization (which require bending of the aromatic ring away from the planar surface to achieve reaction [94, 10, 119]. Therefore, one would expect benzene to form better crystalline single- or bilayer graphene at high temperatures. Defects in graphene sheets The following are the types of defects that could arise in a growing graphene sheet: 91

21 i) Temperature-dependent point defects These point defects are independent of the flow rate of the carbon precursor and its structure. Their probability is merely a function of temperature. ii) Ring closure point defects These arise from faster cyclization than surface diffusion/alignment. They are dependent on temperature, surface coverage (which in turn is dependent on carbon precursor flow rates, among other variables), and the nature of the radicals. Particularly near the edge of growing graphene, the radicals are constrained more severely for diffusion and alignment. Due to this loss of entropy as well as stronger energetic interactions with existing graphene sheet and/or other radicals, the force required to cause next directed diffusion for alignment becomes higher, and the rate of alignment becomes smaller, facilitating cyclization that creates/retains defects. iii) Line defects These arise from the origin of a new layer of graphene. Each existing ring-closure or temperature-dependent point defect site can act as a new nucleation site for the start of a new layer. A newly formed point defect may either immediately initiate a new layer of graphene or over a period of time. iv) Sheet boundary defect Where a growing graphene sheet terminates, sheet boundary defects arise akin to grain boundaries in bulk crystal. They could arise from the underlying copper surface's defects/surface profile; it has been demonstrated that a graphene sheet can grow across copper grain boundaries. However, how much of change in orientation of atoms/planes in adjoining copper grains can be tolerated by a growing graphene sheet is yet to be understood. 9

22 5.6 Conclusions Depending on the relative rates of these different steps, the nature of the obtained graphene film characterized, for instance, by D Raman mapping, would be substantially different. It is seen from our experiments that, at atmospheric pressure, the obtained graphene is predominantly bilayer rather than single layer. The loading of carbon source is, in our opinion, the primary reason for the obtained result. It is likely that lower pressures would yield single-layer graphene samples. The amount of single-layer graphene is highest for benzene at 1000 C and anthracene at 800 C. At 1000 C, phenyl radicals depositing on a surface are likely to move easily, and undergo rapid dehydrogenation and cyclization, due to weaker London dispersion forces. On the other hand, anthracene at lower temperatures is also a likely candidate to move less, because of steric hindrance. In both these cases, it is seen that the control of the three important processes in graphene deposition, namely, surface diffusion, dehydrogenation, and cyclization, is not fine enough at the conditions were predominantly single-layer graphene is formed. Therefore, at these conditions, the obtained single layer has more defects. On the other hand, the best observed crystallinity is for a temperature of 900 C, where bi- or fewlayers of graphene are favoured. Further, dehydrogenation of naphthalene and anthracene are endothermic reactions. These will slowdown the decomposition rate, deposition, and dehydrogenation processes, thereby allowing a period for graphene to eliminate defects. There are two cases where maximum regions of single-layer graphene are obtained one at high temperature (1000 C) with medium flow rate (5.5 mg/min), benzene precursor, and one at low temperature (800 C) with same flow rate, anthracene precursor. These trends suggest that there are two possible different mechanisms involved in the formation of singlelayer graphene, depending on the number of benzene rings in the aromatic carbon source. This is because benzene is known to decompose at temperature (above 700 C), and this allows the ring opening process to take place in the bulk gas phase. The resultant, however, seems to be higher molecular weight, polyaromatics such as diphenyl and other compounds. This reaction is reversible, and exhibits linear increase of reaction rate with initial concentration of benzene, and negative order with respect to hydrogen concentration [18]. Therefore, benzene decomposition must be a series of self-limiting reversible reactions, 9

23 which must reach equilibrium depending on temperature, initial benzene concentration, and hydrogen concentration. Since the deposited compounds are likely polyphenyls, they form good surface coverage and good crystallinity with rapid surface diffusion and desorption and weak vdw interactions. The objective of using DOE is to get a clear understanding of chemical vapour deposition of graphene from aromatic carbon precursors of a homologous series. The experiments give a preliminary indication of how the quality of graphene synthesized through CVD could be controlled. These results throw light on further experiments, simulations, and analysis needed to precisely determine how to control graphene synthesis. 94