Supporting Information for Graphene on Paper: a simple, low-cost chemical sensing platform Shishir Kumar*1, Swati Kaushik1, Rudra Pratap1, Srinivasan Raghavan1. 1 Centre for Nanoscience and Engineering, Indian Institute of Science, Bangalore, India, 560012. *shishirk@gmail.com. 1 Experimental Details Graphene growth: Graphene was grown on Cu foils by well established chemical vapor deposition (CVD) process.1 In brief, a Cu foil was placed in a quartz tube which was evacuated using a rotary pump. The tube was located inside a hot walled furnace, and was heated to 1000 under a flow of 50 sccm of H2. After waiting for 10 min at 1000 under the same H2 flow, CH4 was flown at 5 sccm and H2 flow was reduced to 2 sccm. This growth step was carried out for 2 min and then the furnace was cooled down under H 2 flow of 100 sccm. The foils were taken out of the tube at room temperature. Figure S1. Schematic views of (a) g-paper samples with ends coated with silver paste, (b) measurement setup for resistance of strip with applied strain and (c) the gas-sensing chamber. (d) shows a photograph of a sample for strain measurements. Transfer of graphene to paper: poly-(methyl methacrylate) (PMMA) was spin-coated on the copper foil on which graphene has been grown. The foil was etched in 0.25 M FeCl and the floating PMMA-graphene film was then transferred to deionised water. The process was repeated twice to ensure the removal of contaminants. The film was then transferred on paper and allowed to dry at room temperature for few hours. Normal A4 paper and glossy paper were used for transfers. Once S1
dried, the PMMA layer was dissolved by placing the sample in acetone for 5 min. For electrical measurements, strips were cut out from this g-paper, and conductive silver glue was deposited at the two ends of the strip (Fig. S1a). The root mean square roughness of normal and graphene coated paper was measured using Dimension ICON atomic force microscope in tapping mode over an area of 25 μm2. Raman point and area scans for the samples were recorded using a Horriba-Yvon Labram spectrometer, with 6 nm laser and 100X objective. The laser power was kept below 5 mw to avoid damage to samples. Gas sensing measurements: The measurements were carried out in a chamber whose inlet flow was controlled by MFCs (Fig. S1c).2 The measurements were made at room temperature and pressure. A g-paper strip was clamped on a glass slide which was placed inside the chamber. Two needle probes were then placed on the silver paste coated ends of the strip. The probes were connected to a Keithley 286 source-measurement unit, which was interfaced to a computer. Some samples were exposed to deep ultraviolet (DUV, ~254 nm) radiation using a clear fluorescent tube with flux of ~2.5 mw/cm2 at the position of samples. Immediately afterwards the samples were placed in the measurement chamber for measurements. During each measurement, a sample was exposed to synthetic air (a mixture of 80% N 2 and 20% O2) long enough for the response to stabilize. This interval was about ~10 min or lower. Subsequently, after stopping synthetic air flow, NO2 was allowed into the chamber, appropriately diluted with Ar. NO2 flow was stopped after desired time interval, and synthetic air flow was reestablished. During this period, the sample was at constant bias of 1 V and the current was continuously measured at 1 s intervals. Certified purity NO2 (.5% pure) was used in experiments. The starting concentration was.4 ppm, with Ar as carrier gas. Ultra high purity Ar (.8%) was used for dilution. The setup used two 5 lpm MFCs (MKS117A) for the two streams. Both MFCs have an accuracy of ±1% of full scale flow. The MFCs were controlled by analog voltage supply with 0.05% error of full scale range of 5 V. At 0. ppm (the minimum NO2 flow used in our experiments), the combined errors in S2
the two MFCs result in 26% error in determination of concentration of test gas. The errors were smaller at higher ppm settings. These errors can be used in determining slopes for Fig.a and Fig.b. The weighted least square curve line fitting used initial three data points in Fig.a and two data points in Fig.b. The resulting errors in value of slope of the lines were small compared to the error in flow rates. We have used the largest error in flow rates to estimate the deviation in LLD. Strain measurements: A 100 mm x 5 mm strip of paper whose middle 20 mm portion was covered with graphene was used for strain measurements (Fig. S1b and S1d). A digital voltmeter was connected to the silver paste at the end of paper strip to monitor the its resistance. The ends of paper strip were sitting on two glass slides which were moved towards each other to cause a bend in paper strip. The bent induced a strain in graphene sitting on paper. The strain ε, was calculated using the formula, ε=l 0 L / L0 =ΔL/ L= ( ( t+r ) θ rθ ) /rθ=t /r, where r is the radius of curvature of bent paper strip and θ is the angle of the arc so formed. The radius of curvature r can be calculated from the height h of the bend part of paper strip and the distance between the ends of the strip. The thickness of paper, t, was determined to be 100 μm. 2 Raman spectra for graphene transferred onto smooth paper Figure S2. Raman area map for single layer graphene transferred onto paper. (a) 2D peak to G peak ratio. (b) D peak to G peak ratio. The peak strength is the integral area around ±20 cm-1 around the peak. The center for integration for D, G and 2D peaks are 150 cm-1, 1584 cm-1 and 2665 cm-1 respectively. (c) shows several point spectra of graphene transferred onto SiO2 substrates, confirming monolayered graphene, whereas (d) shows SEM images of CVD graphene on the Cu foil on which it was grown. (scale bar is 2 μm). S
Raman spectroscopy of graphene transferred onto smooth paper confirms the presence of graphene (Fig.S2). The G (1584 cm-1) and 2D (2665 cm-1) peaks are characteristic of graphene, with their ratio determining the number of layer of graphene. 2D/G ratios larger than 2 are associated with monolayer graphene and lower values with multilayered graphene. Our Raman area maps have this ratio ~1.2 (Fig.S2a), which is due to interference from paper substrate. Similarly the D/G ratio which determines the presence of defects, is non-zero (Fig.S2b). Raman spectra obtained from CVD graphene transferred to SiO2 substrates (Fig.S2c) make these assertion clear, providing larger 2D/G ratios, sharp peaks and a very small D peak (possibly due to contaminants). The SEM image of graphene lying on Cu foil (Fig.S2d) indicates only very small density of bilayered islands (seen as faint draker areas). S4
Roughness comparison of normal and smooth pape Figure S. Optical profile of normal paper showing root mean square roughness of 7.7 μm S5
Figure S4. AFM area scan of smooth paper showing root mean square roughness of 55.1 nm. The optical and AFM profiles of normal paper and smooth papers are shown in Fig.S and Fig.S4, indicating the much smoother terrain on the latter. 4 Electrical characteristics of graphene transferred on paper Length (L) (mm) Width (W) (mm) Resistance (R) (kω) 12.552 4 5.744 8 4.86 8 4 7.771 6.421.5 10.28 7.512 8 5.58 4 6.22 7.48 Table T1. Two point resistance (R) of various strips of g-paper as a function of their lengths (L) and widths (W). Sheet resistivity is then given by WR/L. S6
Figure S5. Due to low resistance of the g-paper strip, an LED requiring ~10 ma current can be sustained with g-paper acting as a resistive element in circuit (also shown in inset). A measure of quality of graphene is its sheet resistance. High quality CVD grown graphene transferred on to SiO2 substrates show sheet resistance as low as hundreds of Ω/. In order to determine the sheet resistance of graphene on paper, we measured resistance R of 10 g-paper strips with length and widths similar to the samples used below (Tab.T1). The sheet resistance was then WR/L, where W and L are width and length of the samples respectively. The mean sheet resistance came out to be 2.8 ± 0.8 kω/, which indicates that the quality of graphene transferred on paper is good. For comparison, graphene transferred on SiO 2 substrates had sheet resistance of 1.2 kω/. A g-paper sample with width and length of few mm, could also sustain a light emitting diode which needs currents in the range of ma (Fig.S5). 5 Behavior of response parameters and time-constants across different samples Figure S6. (a) The estimates of parameters S 1, S 2, τ1 and τ2 for various samples on exposure to 2.5 ppm of NO2. Errorbars are also shown for all the estimates. (b) S and τ values for samples shown in (a). S7
Figure S7. (a) The estimates of parameters S 1, S 2, τ1 and τ2 for various samples after 10 mins of DUV exposure and flow of 0.5 ppm of NO2. (b) S and τ values for samples shown in (a). Figure S6 and S7 show the various fitting parameters for response-time curves across several samples. 6 Raman characterisation of samples exposed to DUV Figure S8. Raman characteristics of graphene transferred on to SiO 2 substrate before and after DUV exposure. Each point denote datum from one point spectrum. (a) Position of G peak, (b) FWHM of G peak, (c) position of 2D peak and (d) ratio of 2D to G peaks. Graphene was transferred onto SiO 2 substrate and Raman spectra were taken on several points all over the sample before and after 10 min of DUV exposure. Two samples were analyzed. The position of G peak (~1585 cm-1), its full width at half the maximum (FWHM), the position of 2D peak (~2680 cm-1) and the ratio of 2D to G peaks can be used to determine the doping level of graphene.4 Before the exposure, the samples were p-doped as shown by G and 2D peaks shifted to wave numbers higher than 1584 cm-1 and 2685 cm-1(si, Fig.S8a). On exposure to DUV radiation S8
both these peaks become sharper (Fig.S8b), whereas the position of 2D peak is shifted to lower wave numbers by ~ cm-1 (Fig.S8c). The ratio of 2D to G peak also improves from 1.4 to 1.5 (Fig. S8d). 7 Calculations for determining the noise level in response-time curves The estimation of LLD values involves signal to noise ratio in the response of the sensors. For Fig., the background noise was determined from the noise in the two point conductance obtained just before the test gas was flown on the samples. The conductance was first converted to the response S ( t ) =ΔG ( t ) /G0 where G0 is the mean conductance for a given time interval and ΔG(t) is change in conductance from this mean a time t in the interval. To calculate the noise during the resting interval, S(t) values for mins taken at intervals of 1 s were used. In first method, a polynomial of 2nd order was fit to this data, i.e. the coefficient of following polynomial were determined by least square fitting: P(xi) = p0 + p1xi + p2xi2 for xi S(i), i {1, 2,,, 180} The fitting procedure results in P(xi) = 2. 10-4 1.1 10-5 xi + 6.6 10-8xi2, the root mean square (r.m.s.) noise was then calculated using the formula: σ = ((S(i) P(xi))2/180) for I {1, 2,,, 180} This yields σ = 0.042%. Ammu et al.5 used response values picked up at 10 s intervals over 140 s for a sample size of 14. They also used a fifth order polynomial for fitting this data. Following their algorithm we have used: P(xi) = p0 + p1xi + p2xi2 + pxi + p4xi4 + p5xi5 for xi S(i), i {15, 0, 45,, 180} The fit procedure yields P(xi) = 1.5 10-4 + 8.8 10-4xi 5.7 10-4xi2 + 1.4 10-4xi 1.4 10-5xi4 + 5.2 10-7xi5, resulting in σammu = ((S(i) P(xi))2/12) = 0.028%. The LLD was calculated comparing the slope of response vs concentration plots shown in Fig.. To achieve a minimum SNR for detection (=), the concentration needed is σ/slope. In Fig.a, the slope is 167% ppm-1, which gives a concentration of 0.000754 ppm (= (0.042)/167 ppm), or 754 ppt. For Fig.b, with slopes of 2% ppm-1 and 275% ppm-1, we have LLDs of 87 ppt and 458 ppt. Using σammu these values come down to 258 ppt and 05 ppt. We note that the noise calculated above are higher than the noise reported Ammu et al.5 and by Chen et al.6 This is due small temporal drift in conductance values measured in our setup. For S
example, using smaller interval of 1 min for calculations cuts noise to half the values displayed above. A better design of chamber and instrumentation can improve the LLD by an order of magnitude. 8 Correlations of response-time fit parameters to the induced strain Figure S. The change in conductance of a sample as strain is applied on it under 2.5 ppm flow of NO2. The parameters S, S 1 and S 2 and the resistance of the sample R, change with strain, as shown in the inset. As noted in main text, the baseline resistance of the sample increases with strain, and with it the response improve as depicted in the inset in Fig.S. The inset shows that the increase in response is almost entirely caused by increase in the response associated with long time constant (S 2), there is almost no change in S 1. References (1). Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.; Ruoff, R. S. Large-Area Synthesis of High-Quality and Uniform Graphene Films on Copper Foils. Science 200, 24, 112 114. (2). Rudraswamy, S. B.; Basu, P. K.; Bhat, N. Sensitivity Characteristics of Ag Doped BaTiO-CuO Mixed Oxide as Carbon-Dioxide Sensor. In 2014 IEEE International Conference on Electronics, S10
Computing and Communication Technologies (IEEE CONECCT); 2014; pp. 1 4. (). Bae, S.; Kim, H.; Lee, Y.; Xu, X.; Park, J.-S.; Zheng, Y.; Balakrishnan, J.; Lei, T.; Ri Kim, H.; Song, Y. I.; Kim Y.-J.; Özyilmaz, B;, Ahn, J.-H.; Hong, B.-H.; Iijima, S. Roll-to-Roll Production of 0-Inch Graphene Films for Transparent Electrodes. Nat. Nanotechnol. 2010, 5, 574 578. (4). Das, A.; Pisana, S.; Chakraborty, B.; Piscanec, S.; Saha, S. K.; Waghmare, U. V.; Novoselov, K. S.; Krishnamurthy, H. R.; Geim, A. K.; Ferrari, A. C.; et al. Monitoring Dopants by Raman Scattering in an Electrochemically Top-Gated Graphene Transistor. Nat. Nanotechnol. 2008,, 210 215. (5). Ammu, S.; Dua, V.; Agnihotra, S. R.; Surwade, S. P.; Phulgirkar, A.; Patel, S.; Manohar, S. K. Flexible, All-Organic Chemiresistor for Detecting Chemically Aggressive Vapors. J. Am. Chem. Soc. 2012, 14, 455 4556. (6). Chen, G.; Paronyan, T. M.; Harutyunyan, A. R. Sub-Ppt Gas Detection with Pristine Graphene. Appl. Phys. Lett. 2012, 101, 0511. S11