Graphene-Based Sensing of Gas-Phase Explosives - ACS Applied

Subscriber access provided by WEBSTER UNIV

Article

Graphene-Based Sensing of Gas-Phase Explosives Jie Zhang, and Eric Fahrenthold ACS Appl. Nano Mater., Just Accepted Manuscript • DOI: 10.1021/acsanm.8b02330 • Publication Date (Web): 05 Feb 2019 Downloaded from http://pubs.acs.org on February 7, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Nano Materials

Graphene-Based Sensing of Gas-Phase Explosives Jie Zhang and Eric P. Fahrenthold∗ Department of Mechanical Engineering, University of Texas, Austin, TX 78712 * E-mail: [email protected] Phone: 512-471-3064. Fax: 512-471-8727 Abstract Graphene based sensors have shown excellent potential in the trace detection of specific gases which are hazardous to humans or environmentally toxic. Sensor designs incorporating pristine graphene, graphene oxide, and nano-patterned graphene have been the focus of much recent experimental and computational research. The application of graphene based sensors in the trace detection of explosives has seen relatively limited study, due in part to the difficulties of conducting experiments using the nitramine and aromatic explosives of central interest. Computational studies of explosive sensors are not subject to hazardous materials handling constraints, and may be used to complement experimental research on the development of low weight, low power, graphene based sensors. Ab initio models of five different graphene nanoribbon sensor configurations have been developed, and their chemiresistive response to three widely used explosives and four background gases has been investigated. The results indicate that the sensitivity and selectivity of nanoribbon devices exposed to mixtures of explosives and background gases will vary significantly with explosive type and sensor configuration. Keywords: graphene, nanoribbons, explosives, sensing, modeling

1

ACS Paragon Plus Environment

ACS Applied Nano Materials 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Introduction The development of light weight, low power sensors for the detection of gas phase explosives is a challenging task. Since many explosives have extremely low room temperature vapor pressures, 1 high sensitivity may be needed to enable trace detection. Since distinct explosive molecules may share similar chemical functional groups, highly selective sensors may be needed to identify particular explosives. Recent research has demonstrated the great potential of ‘solid state’ sensing devices 2,3 in some gas detection applications (e.g. CO and NO2), including high sensitivity, low cost, small size, and simplicity of operation. Despite this success, improvements in long-term stability, temperature sensitivity, and selectivity 4,5 are needed and motivate continued experimental and computational research in this field. 6 Previous research on explosives detection 7–10 has investigated a wide variety of sensing devices (solid state sensors, 11 nanomechanical sensors 12,13 ), sensing methods (mass spectrometry, 14,15 diffuse reflection spectroscopy 16 ), and sensing materials (fluorescence polymers, 17,18 carbon nanotubes, 19,20 graphene 21,22 ). Since much recent research has sought high mobility, low cost, and low power solutions to the trace detection problem, the development and application of explosive sensing materials or composites is a topic of considerable research interest. Since graphene and its derivatives offer the possibility to develop very high specific surface area sensors, they are promising candidates for miniature gas detectors, perhaps as components of wireless sensor networks. 23 Bendable gas sensors based on reduced graphene oxide have been successfully fabricated, 24 and chemically and structurally modified graphene has been shown to exhibit highly sensitive and selective performance in the experimental 22,25,26 detection of some toxic gases. Concentration detection limits as low as one ppb (for NO2) and one ppm (for NH3) have been demonstrated. 27 Graphene nanoribbons 28 (GNRs) and graphene films 29 have demonstrated trinitrotoluene (TNT) sensing in seawater, and graphene films used for chemiresistive sensing of NO2 have shown qualitatively similar conductance changes when exposed to the explosive simulant 2,4-dinitrotoluene (DNT). 21 2


Page 2 of 38

Page 3 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The present paper develops ab initio models of graphene based explosive sensors, in five distinct configurations, to evaluate the performance of graphene in chemiresistive detection of three widely used explosive molecules. The models quantify: (1) conductance changes in graphene platelets due to nitramine and aromatic explosives, (2) conductance changes in graphene platelets due to four different background gases, (3) adsorption energy for explosive and background gas molecules on graphene platelets, (4) charge transfer, changes in the density of states, and changes in the highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) distribution due to gas sensing, and (5) sensor chemiresistive performance in the presence of a combination of explosive molecules and background gases. The modeling results estimate the sensitivity of graphene based trace explosive detectors, as a function of both the target molecule and the sensor configuration, and evaluate the utility of commonly applied computational metrics (charge transfer, density of states, etc.) as indicators of chemiresistive sensor performance. The succeeding parts of this paper are organized as follows: (1) the computational models segment describes the numerical methods used, the modeled sensor configurations, the model boundary conditions, and the selected sensor performance evaluation criteria; (2) the sensor modeling results segment details the computed results for conductance and adsorption energy; (3) the discussion segment evaluates the use of additional computation metrics (charge transfer, density of states, etc.) to understand sensor chemiresistive performance; (4) the sensitivity model segment employs the ballistic conductance and energy adsorption calculations to estimate overall sensor performance, and (5) the conclusions section discusses the principal results of the study and suggests directions for future research.

Computational The macroscale schematic of Figure 1(a) depicts a sensor concept of the type which motivates the present study, with graphene GNRs serving as chemiresistive sensing elements. The

3



nanoscale graphic in Figure 1(b) depicts a computational model of the type used to estimate overall sensor performance. Conductance changes computed for individual target explosive and background gas molecules are used to estimate changes in sensor resistance, while energy adsorption values computed for individual target explosive and background gas molecules are used to estimate surface concentrations for adsorbed background gases, relative to the surface concentration of target explosive molecules. In all cases, the modeled sensor platelets were narrow zigzag GNRs. Figure S1 (Supporting Information) depicts the three modeled explosive molecules (TNT, RDX, HMX); the modeled background gases (N2, O2, CO2, H2O) are four principal components of air. Although the computational models employed here only approximate the complex physics of practical sensing devices, the fully ab initio modeling work presented here can complement experimental research by investigating sensor properties which are difficult to measure directly. Similar modeling work has proven to be of benefit in carbon based electrical conductor design. 30,31 All of the equilibrium, conductance, and post-processing (charge transfer, density of state, and HOMO-LUMO orbital) calculations were performed using the ab initio code suite SIESTA. 32 In the SIESTA suite, equilibrium modeling employs density functional theory (DFT) while the electrical transport properties are computed using a non-equilibrium Green’s function (NEGF) method, 33 incorporated in the module TranSIESTA. Initial equilibrium calculations considered both a local density approximation (LDA) and a generalized gradient approximation (GGA, specifically the Perdew-Burke-Ernzerhof 34 formulation) in the exchange-correlation functional. 35 Since the LDA calculations showed a better match (a 31 percent smaller RMS error) than the GGA calculations with experimental data 36–38 for bond lengths in the gas molecules (see Table S1, Supporting Information), the LDA formulation was used to determine the atomic positions for all of the models presented in this paper. All of the equilibrium models were relaxed until the maximum atomic force was less than 0.01 eV per Angstrom. In the conduction and energy adsorption calculations, which followed the equilibrium analyses, the aforementioned GGA exchange-correlation functional 4


Page 4 of 38

Page 5 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


was employed in order to avoid the overestimates of binding energy typically encountered with the LDA formulation. Non-local functionals, which include van der Waals interactions in computing the binding energies for adsorption processes, 39 are not considered here but are an appropriate focus for future work. Similarly, the modeling results presented here assume zero temperature; the application of finite temperature DFT methods 40 is of interest for future work. The basis set chosen for the calculations was double-zeta polarized (DZP), and was used for all atoms. The fineness of the real space mesh was determined by setting the energy cut-off value to 300 Ry (this determines the maximum quantum kinetic energy representable on the grid, using a plane wave interpolation, without aliasing). 32 A k-grid of 3 × 3 × 5 was employed to model the device (graphene platelet plus adsorbed gas) region and a k-grid of 3 × 3 × 20 was employed to model the electrodes. Graphene electrodes were modeled in all of the conductance calculations. The ballistic conductance (G) values quoted here are determined by the computed transmission at the Fermi energy T (Ef ), defined by: 41 T (Ef ) = Tr[t(Ef )† t(Ef )] =

2e2 G , G0 = G0 h

(1)

where t(E) is a matrix of transmission coefficients, G0 is the quantum conductance unit, e is the magnitude of the charge on an electron, and h is Planck’s constant. Note that the narrow pristine zigzag GNRs considered here (6-zGNR) exhibit a flat transmission band in the vicinity of (but not at) the Fermi energy (see Figure S2, Supporting Information), for a zero-bias spin-unpolarized calculation. The conductance peak of 3G0 at the Fermi level is explained as an edge effect, 42,43 and only smooth zigzag GNRs exhibit this property. 44 Since the peak in the transmission spectrum is narrow, some discussions of zigzag GNRs neglect its presence. In this paper, since all of the quoted conductance calculations are made at the Fermi energy, the value 3G0 is taken to be the Fermi energy ballistic conductance in discussing the chemiresistive performance of the various modeled systems.

5



Page 6 of 38

A second sensor property of central interest in this study is the energy associated with adsorption of the explosive or background gas molecule onto the GNR. These energies are used here to estimate the surface concentrations of background gases (relative to the surface concentrations of adsorbed explosive) present on a platelet sensor in thermal equilibrium. The adsorption energy is defined as

Eads = Egas+sen − Egas − Esen

(2)

where Egas+sen is the total equilibrium energy of the sensor platelet (including functional groups, if any) and the adsorbed gas molecule, while Egas and Esen are corresponding total energies for the isolated gas molecule and the sensor respectively. Note that all three energy values are computed for the same set of atomic coordinates.

Results and Discussion This section describes the results of the ab initio conductance and adsorption energy calculations, for five different sensor configurations: surface and edge sensing for pristine graphene platelets, surface sensing for C-OH and C-O-C functionalized platelets, and surface sensing for nanoholed platelets. Since pristine graphene tends to be stable and chemically inert, 45 the performance of the pristine edge, functionalized, and nanoholed configurations are of considerable interest for graphene based sensors. The platelet models are three unit cells in width, except that the ‘nanoholed’ sensors are six unit cells in width. Note that the reported successes of atomically precise bottom-up fabrication methods allow very thin GNRs, with widths less than ten nanometers, to be reliably produced. 46–48 The narrower the nanoribbon, the more edge length per unit mass, which may improve overall sensor performance. All of the modeled graphene platelets are edge terminated with single hydrogen atoms. Initial separation distances for the carbon atoms are 1.426Å, and the starting geometries for the background gas and explosives molecules are taken from published work. 36–38 The boundary 6


Page 7 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


conditions for the equilibrium calculations are periodic (in the conduction direction) and free (along the nanoribbon edges, by introducing a vacuum buffer space). Figure 2 shows the six modeled configurations; the atom count for the modeled systems is approximately 200. The computed conductance and adsorption energy results are provided in Table S2 (Supporting Information). Pristine GNR sensors. The first two modeled configurations considered surface and edge sensing with pristine graphene. These two sensing configurations are depicted in Figures 2(a) and 2(b). Figures 3 and 4 show the computed conductance changes and adsorption energies for all seven target species (three explosives and four background gases). Note that results were obtained for only one target molecule orientation. In the surface sensing configuration, all of the target molecules reduce the GNR conductance. The ballistic conductance changes due to RDX, HMX, N2, H2O, and CO2 adsorption are similar and on the order of ten percent. The TNT molecule produces a larger change in conductance than the nitramines, perhaps due to its aromatic hexagon structure and the associated π − π interaction with the graphene surface. 49 Overall, the O2 target produces by far the largest change in conductance. With regards to adsorption energy, values for the large explosives molecules are much greater than those for the background gases. In the edge sensing configuration, all of the target molecules again reduced the GNR conductance. As compared to the surface sensing configuration, all of the target molecules (except TNT and N2) show a larger change in conductance, consistent with published evaluations of GNR as most chemically active along its edges. 50,51 The exceptional response of TNT is likely the result of a reduced π − π interaction in the edge sensing geometry: the flat structure of the TNT molecules inhibits close interaction of all three nitrites with the GNR edge. By contrast, the deformation of the RDX molecule shown in Figure 2(b) facilitates the close interaction of all three target molecule nitrites with the graphene edge. Similar interaction physics is reported in Vovusha’s study 52 of binding energy and charge transfer for explosive molecules adsorbed on graphene nano-flakes. Although the change in conductance 7



for oxygen is larger than that for all of the other target molecules, note that its adsorption energy is much less than that for all of the explosive molecules. The adsorption energies for the explosives are also greater than those for the other background gas molecules. As compared to the explosive molecules, the background gas molecules are smaller and lack active functional groups. The differences in the adsorption energies for H2O and O2 are less in the edge sensing case than in the surface sensing case; since the water molecules are polarized, it appears that they are strongly attracted to the H-terminated edges. The conductance modeling results appear to be sign consistent with three published experimental studies, which measured an increase in resistance for graphene sensors exposed to N2, 53 CO2, 54 and H2O. 55 Overall, the response of the platelets to both explosive and background gases is qualitatively similar in the surface and edge sensing configurations. The results do however suggest that large area graphene is most suitable for TNT sensing, while nanoribbon based edge sensing appears to be a more suitable configuration for the detection of nitramine explosives. It should be noted that for nanoribbons which have been oxidized, the computed results suggest that sensor conductance may increase when explosive molecules approach an edge or surface sensing site and replace adsorbed oxygen molecules. Such a response would be consistent with published experimental research on carbon nanotube (CNT) 56 and graphene 57–59 based sensors. Experiments on the response of graphene films to DNT (a precursor to TNT) show a similar effect. Discussions of these experiments tend to focus on charge transfer between the sensing device and the target molecules, with the explosive molecules and O2 both characterized as electron acceptors. Functionalized GNR sensors. Decorating graphene with functional groups has been studied in both computational 60–62 and experimental 63 research, and is one design strategy aimed at influencing sensor performance. Graphene oxide, relatively easy to fabricate, is typically decorated with epoxy (C-O-C) and hydroxyl (C-OH) functional groups. Of these two, the C-O-C groups are more stable while the C-OH groups are more chemically active. 25,64 The densities of these functional sites can be modified through oxidation and thermal re8


Page 8 of 38

Page 9 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


duction 65,66 processes. Experiments on graphene oxide and reduced graphene oxide have been performed to study a variety of applications, including the separation of CO2 from CH4, 67 the sensing of NO2 and NH3 at room temperature, 22 humidity sensing, 68 and the design of wearable sensors. 24 This subsection describes modeling results for graphene platelets incorporating the functional groups C-O-C and C-OH. The functional groups are sited on the surface of the GNR, as indicated in Figures 2(c) and 2(d). Graphene oxide films are normally insulating, due to the presence of large numbers of sp3 -hybridized carbon atoms. 22 However the GNR models analyzed here incorporate only one oxygen group, hence they show essentially metallic properties in the transmission calculations which follow. The change in conductance and energy adsorption results for the two functionalized sensor configurations are plotted in Figures 5 and 6. With regards to the functionalization process, both the C-O-C and C-OH functionalized models (see Table S2, Supporting Information) show a reduction in conductance, as compared to pristine graphene, consistent with published ‘cleaning bake’ experiments. 63 With regards to sensing the background gases O2, H2O, and CO2, direct comparison of the conductance change results for the C-O-C and C-OH functionalized GNRs shows conductance changes which are similar in magnitude, but different in sign. It appears that the sign of the conductance change may depend on the relative strengths of the adsorbed molecule’s interactions with the graphene surface and with the functional group: 69 • In the C-O-C functionalized case, the O2 molecule appears to interact more strongly with the graphene surface, and the GNR conductance is reduced (as in the pristine sensing case). In the C-OH functionalized case, the O2 molecule appears to interact more strongly with the functional group, and the GNR conductance is increased. • In the C-OH functionalized case, the CO2 and H2O molecules appear to interact more strongly with the graphene surface, and the GNR conductance is reduced (as in the pristine sensing case). In the C-O-C functionalized case, the CO2 and H2O molecules appear to interact more strongly with the functional group, and the GNR conductance 9



is increased. These conductance modeling results appear to be sign consistent with three published experimental studies, which measured a decrease in resistance for graphene oxide sensors exposed to H2O, 70 an increase in resistance for reduced graphene oxide sensors exposed to H2O, 71 and an increase in resistance for reduced graphene oxide sensors exposed to DNT. 25 With regards to the explosive target molecules, the conductance changes in the functionalized configurations are on average less than those for the pristine graphene cases, while the adsorption energies are on average increased. The complex nature of the overall response of the functionalized sensors emphasizes the importance of additional modeling work, discussed in a later subsection, aimed at estimating the overall sensor performance in the presence of a gas mixture. Although some specialized configurations (such as TNT sensing in water 29 ) are of practical interest, the development of explosive sensors for operation in the presence of a target explosive molecule and multiple background gases is the application most commonly of interest. Nanohole patterned GNR sensors. Nanomeshed or nanoholed graphene has attracted considerable research attention, for a range of applications. The desired hole pattern may be obtained using lithography and ion etching, with the hole size and shape determined by template selection and etching time. 72–74 The introduction of nanoholes serves to increase the total edge length per unit mass in a graphene film, and is analogous to changing the surface area per unit mass in a porous sensor. Published experimental research on NO2 sensing has indicated that nanomeshed graphene offers improved sensing performance, 75 as compared to its graphene film counterpart, and that the performance of porous graphene oxide sensors improves with higher porosity. 76 In addition to gas sensing, nanoholed graphene has been employed in a filter configuration for DNA sequencing 77–79 and the selective identification of ions such as Na+ and K+. 80 In this application, when large molecules go through the nanomesh, the H-terminated hole edges play the role of a probe and can react to the chemical groups present in the target molecules. A holed structure may also facilitate target 10


Page 10 of 38

Page 11 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


gas adsorption and increase charge transfer between the target molecule and the detection device. This subsection describes modeling results for graphene platelets incorporating a 2 × 4 unit-cell nanohole, positioned in the nanoribbon center, as shown in Figures 2(e) and 2(f). Conductance changes and adsorption energies for this sensor configuration are plotted in Figure 7. As compared to the four sensor configurations perviously discussed, two results stand out: (a) the magnitudes of the conductance changes and adsorption energies for the full set of seven target gases vary over a smaller range, and (b) the signs of the conductance changes are positive for all of the explosives and negative for all of the backgound gases. These differences in sensor response may be due in part to molecule size, since the explosive molecules are similar in size to the graphene nanohole, while the background gas molecules are much smaller. The explosive molecules interact with numerous hole edge sites, increasing the adsorption energy, and appear to form a ‘bridge’ which improves electron transmission across the holed nanoribbon. By contrast, the background gas molecules interact with only a few edge sites, lowering the adsorption energy; their isolated interactions with the platelet reduce its conductance, as indicated for the pristine sensor configuration cases. Recognizing that nanohole size, nanohole shape, number of nanoholes per unit area, and other parameters may affect the conductance results presented in this subsection, it should be emphasized that the nanohole sensor modeling work included in this paper is limited in scope. A larger computational study varying a number of design parameters is needed to investigate fully the effects of nanohole patterning on graphene sensor performance. Conductance change mechanisms. An important objective of ab initio modeling of chemiresistive sensors is the development of an improved understanding of the nanoscale mechanisms responsible for changes in ballistic conductance. One example mechanism is charge transfer. The charge density difference plot of Figure 8(a) depicts charge transfer effects due to surface adsorption of an oxygen molecule on pristine graphene, indicating that O2 serves as an electron-acceptor during the adsorption process. Such ‘p-type’ doping is 11



commonly discussed in the experimental literature. Chen et al. 57 for example reported that UV light cleaning changes the reaction of graphene to NH3 adsorption, implying a p-doped configuration for graphene exposed to air. Similar results have been reported for CNT 56 and reduced graphene oxide 22 sensors. A second example mechanism is a change in the HOMOLUMO structure for the combined target molecule-sensor system. These two orbitals are associated with transmission at energies close to the Fermi energy, and are therefore considered to play important roles in determining the conductance of a chemiresistive sensing device. The edges of pristine zigzag graphene are commonly cited as important transmission paths identified by the HOMO-LUMO orbital distribution. 81 The continuity of these conduction pathways may be disrupted when a target molecule approaches a GNR device. A third example mechanism is changes in the density of states, 58,82–84 both the total density of states for the electron-dopant system and the partial density of states for the dopant. The discussion which follows considers all three of the aforementioned conductance change mechanisms, with reference to tables and plots provided in the Supporting Information. Note that the density of states plots include in each case results at all energies within 1.5 eV of the Fermi energy, as well as insets which focus on the results within 0.2 eV of the Fermi energy. This subsection discusses a set of charge transfer, density of states, and HOMO-LUMO structure calculations made for each of the five sensor configurations studied in this paper, to determine the correlation between changes in these electronic properties and changes in sensor conductance. Since these properties are generally considered to influence rather than dictate chemiresistive sensor conductance, their correlation with computed conductance involves a relative (not absolute) comparison. Hence the discussion which follows compares for each sensor configuration the changes in sensor conductance due to HMX and O2 adsorption with corresponding changes in charge transfer, density of states, and HOMO-LUMO structure due to those analytes. Consistency in the changes of these electronic property measures with changes in conductance is defined here as follows: (a) large charge transfers (estimated 12


Page 12 of 38

Page 13 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


using the average of the Mulliken, Hirshfeld, and Voronoi measures) are expected to strongly affect conductance, (b) an increase in the density of states at the Fermi energy level is expected to improve conductance, and (c) disruption of the continuity of LUMO transmission pathways is expected to reduce conductance, and reductions in the overlap of the HOMO and LUMO orbitals are expected to reduce conductance. In the case of surface sensing by pristine graphene, Table S3 (Supporting Information) lists the change in conductance and the corresponding computed charge transfers due to O2 and HMX adsorption. A negative sign for the charge transfer indicates a loss of electrons by the sensor. Figure S3 (Supporting Information) shows the corresponding density of states plots associated with O2 and HMX adsorption. Each of the density of states plots shows: (a) the total density of states before adsorption (dotted line), (b) the total density of states after adsorption (solid line), and (c) the partial density of states for the analyte (shaded region) after adsorption. The discussion which follows focuses on changes in the density of states at the Fermi energy, since the conductance values reported in this paper are based on the computed transmission at the Fermi energy. Figure S3 (Supporting Information) also shows the corresponding HOMO-LUMO structure plots due to O2 and HMX adsorption (the direction of current flow is horizontal, parallel to the GNR axis). Comparing the various electronic properties metrics: • conduction is reduced for both O2 and HMX, with the change due to O2 greater by almost an order of magnitude, • the HMX n-dopes the graphene while O2 p-dopes the graphene (the later charge transfer magnitude is greater by a factor of approximately five), • the density of states plots show a substantial increase in the total density of states at the Fermi energy due to O2 adsorption, and a very small change due to HMX adsorption, and • the HOMO-LUMO structure plots show large changes in both the LUMO structure 13



and the HOMO-LUMO overlap due to O2 adsorption, and much more modest changes due to HMX adsorption. These results indicate that the charge transfer and HOMO-LUMO structure metrics are strongly correlated with the computed conductance changes for O2 and HMX surface adsorption on pristine graphene, but that the density of states metric is not. In the case of edge sensing by pristine graphene, Table S4 (Supporting Information) lists the change in conductance and the corresponding computed charge transfers due to O2 and HMX adsorption. Figure S4 (Supporting Information) shows the density of states plots associated with O2 and HMX adsorption, as well as the HOMO-LUMO structure plots corresponding to O2 and HMX adsorption (the direction of current flow is horizontal, parallel to the GNR axis). Comparing the various electronic properties metrics: • conduction is reduced for both O2 and HMX, with the change due to O2 greater by a factor of approximately six, • the HMX n-dopes the graphene while O2 p-dopes the graphene (the magnitude of the later charge transfer is less by a factor of approximately three), • the density of states plots show a small increase in the total density of states at the Fermi energy due to O2 adsorption, and small decrease due to HMX adsorption, and • the HOMO-LUMO structure plots show large changes in both the LUMO structure and the HOMO-LUMO overlap due to O2 adsorption, and modest changes due to HMX adsorption. These results indicate that for O2 and HMX edge adsorption on pristine graphene, the HOMO-LUMO distribution metric is correlated with the computed conductance changes, but that the charge transfer and density of state metrics are not. In the case of surface sensing by C-O-C functionalized graphene, Table S5 (Supporting Information) lists the changes in conductance and the corresponding computed charge 14


Page 14 of 38

Page 15 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


transfers due to O2 and HMX adsorption. Figure S5 (Supporting Information) shows the density of states plots associated with O2 and HMX adsorption, as well as the HOMO-LUMO structure plots corresponding to O2 and HMX adsorption (the direction of current flow is horizontal, parallel to the GNR axis). Comparing the various electronic properties metrics: • conduction is reduced for both O2 and HMX, with the change due to O2 greater by an approximate factor of nine, • the HMX n-dopes the graphene while O2 p-dopes the graphene (the later charge transfer magnitude is greater by an approximate factor of five), • the density of states plots show a reduction in the total density of states at the Fermi energy due to O2 adsorption, and a very small change due to HMX adsorption, and • the HOMO-LUMO structure plots show considerable changes in the LUMO distribution due to HMX adsorption, and a large change in the HOMO-LUMO overlap due to O2 adsorption. These results indicate that for O2 and HMX surface adsorption on C-O-C functionalized graphene, the charge transfer, density of states and HOMO distribution metrics are all correlated with the computed conductance changes. In the case of surface sensing by C-OH functionalized graphene, Table S6 (Supporting Information) lists the changes in conductance and the corresponding computed charge transfers due to O2 and HMX adsorption. Figure S6 (Supporting Information) shows the density of states plots associated with O2 and HMX adsorption, as well as the HOMO-LUMO structure plots corresponding to O2 and HMX adsorption (the direction of current flow is horizontal, parallel to the GNR axis). Comparing the various electronic properties metrics: • conduction is reduced by HMX and increased by O2, with the magnitude of change due to O2 greater by an approximate factor of eight,

15



• the HMX and the O2 both p-dope the graphene, with the charge transfers very similar in magnitude, • the density of states plots show a large increase in the total density of states at the Fermi energy due to O2 adsorption, and a small decrease due to HMX adsorption, and • the HOMO-LUMO structure plots show improved continuity in the LUMO distribution and improved HOMO-LUMO overlap due to both target molecules. These results indicate that for O2 and HMX surface adsorption on C-OH functionalized graphene, the density of states and charge transfer metrics are well correlated with the computed conductance changes, but that the HOMO-LUMO structure metric is not. In the case of sensing by nanoholed graphene, Table S7 (Supporting Information) lists the changes in conductance and the corresponding computed charge transfers due to O2 and HMX adsorption. Figure S7 (Supporting Information) shows the density of states plots associated with O2 and HMX adsorption, as well as the HOMO-LUMO structure plots corresponding to O2 and HMX adsorption (the direction of current flow is vertical, again parallel to the GNR axis). Comparing the various electronic properties metrics: • conduction is increased by HMX adsorption and decreased by O2 adsorption, with the magnitude of change due to O2 greater by an approximate factor of two, • the HMX n-dopes the graphene while the O2 p-dopes the graphene (the charge transfers differ in magnitude by an approximate factor of two), • the density of states plots show a small reduction in the total density of states at the Fermi energy due to O2 adsorption, and a significant increase due to HMX adsorption, and • the LUMO structure plots show large changes due to both HMX and O2 adsorption, and reduced HOMO-LUMO overlaps, for both analytes.

16


Page 16 of 38

Page 17 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


These results indicate that for O2 and HMX adsorption on nanoholed graphene, the density of states changes are correlated with the computed conductance changes, but that the charge transfer and HOMO-LUMO structure metrics are not. Overall the results on correlation of the three electronic property metrics with conduction changes were mixed. Each metric was found to be well correlated with conductance changes in three of the five modeled configurations. Only in the C-O-C functionalized configuration were all three metrics indicators of the computed conductance changes. Note that Figures S3 through S7 (Supporting Information) also list the HOMO-LUMO energy gaps for all of the modeled sensor configurations. The computed energy gaps are for the combined sensortarget molecule system (they are not comparisons of HOMO and LUMO orbitals for distinct adsorbates and hosts 84 ), hence the energy gaps are very small and reflect the substantial conductivity of the combined systems modeled in this paper. Additional insight is provided by the charge density difference plots provided in Figures 8 and 9. Figure 8 compares charge density difference distributions for the interaction of pristine graphene with the four different analytes (HMX, RDX, TNT, and O2), while Figure 9 compares charge density difference distributions for the interaction of TNT with four different sensor types (C-O-C functionalized, C-OH functionalized, edge sensing, and nanohole patterned). The adsorption energy and charge density difference results suggest that aromatic explosives are more strongly bound to pristine graphene sensors (Figure 8(d)) than to the C-O-C and C-OH functionalized sensors (Figures 9(a) and 9(b)) due to the flat structure of the TNT molecules. Sensitivity model. The platelet modeling results presented in the preceding subsections emphasize the complexity of the graphene sensor design problem. This subsection develops a sensitivity model for graphene devices in the five configurations previously analyzed, one which estimates sensor performance under simultaneous exposure to a single explosive molecule type and all four of the background gases previously discussed. The tabulated ballistic conductance changes quantify each sensor’s chemiresistive response to each 17



Page 18 of 38

target molecule, while the tabulated adsorption energies are used to estimate adsorbed surface concentrations for each background gas, relative to the surface concentration of adsorbed explosive molecules. Although the sensitivity model developed here is certainly approximate, it is considered to be an essential first step in relating the ab initio modeling results to the performance of practical devices, and may assist experimental research in analyzing the performance of alternative sensor configurations. The sensor’s total resistance R is defined as the sum of a reference (vacuum) resistance Rref and a resistance change ∆R due to the adsorption of gas molecules (3)

R = Rref + ∆R

The change in resistance is taken to be sum of resistance changes due to Ns adsorbed species,

∆R =

NS X

∆Ri =

NS X

α i ci

(4)

i=1

i=1

where αi is a species sensitivity for the ‘ith’ species and ci is a surface concentration (in mass per unit area) for the ‘ith’ species. The species sensitivities are estimated using the computed ballistic conduction results for each sensor configuration,

αi =

∆Ri A mi

(5)

where the A is the sensor surface area in the ab inito model and mi is the mass of a target molecule of the ‘ith’ species. Writing ci as the product of moles per unit area fi and a molar mass Mi , and assuming that the adsorbed mole ratios at equilibrium are determined by the adsorption energies per unit mass fi ei = , fj ej

ei = Ei /mi

18


(6)

Page 19 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the total change in the sensor resistance may be expressed as

∆R = c1

NS X i=1

αi

ci = S c1 , c1

S=

NS X

αi

i=1

ei Mi e1 M1

(7)

where species ‘1’ is taken to be the explosive and S is the overall sensitivity of the device. The computed sensitivities describe the chemiresistive response of the five modeled sensors to changes in the mass per unit area of adsorbed explosives, in the presence of all four modeled background gases. Since the relative surface concentrations of the background gas molecules are assumed to be determined by their mass specific adsorption energies, the computed sensitivities estimate the performance of sensors in thermal equilibrium. Figure 10(a) compares the computed sensitivities as a function of sensor type and explosive type, while Figure 10(b) compares the relative sensivities Sr , defined by ∆R = S r c1 , Rref

Sr =

S Rref

(8)

These plots suggest four principal conclusions: • The change in resistance varies linearly with the change in the explosive’s surface concentration; this result is qualitatively consistent with experimentally measured sensor performance in studies of NO2 85 and CO2 86 detection. • The relative sensitivities are largest in magnitude for the edge sensing configuration, perhaps not surprising since a distinguishing property of GNRs is their edge reactivity. • The lowest sensitivities are shown by the pristine and C-O-C functionalized sensors, perhaps not surprising since they are the most chemically stable of the modeled sensor configurations. • The sensitivities for the C-OH functionalized devices are unique in that they differ in sign from those of all four other sensor configurations. 19



Although the analysis results and discussion presented in this section offer basic insights, previous work has demonstrated that the design of effective explosive sensors will require additional analytical 10 and experimental 7 work, to address issues such as response and recovery times, 9 doping effects, 8 and the application of multiple sensors in discriminating arrays, which might employ graphene-based sensors in combination with other sensor types.

Conclusions Graphene based devices have garnered extensive research attention as gas sensors, and have shown considerable promise in experimental studies of some trace detection applications. They have also shown promise for use in the trace detection of explosives, although published experimental research on this application has been much more limited. The reasons include low vapor pressures for the explosives of primary interest and the hazards and constraints associated with explosives handling, even for very small sample sizes. Computational research on the development of graphene based explosive sensors avoids hazardous materials handling problems, and can complement experimental research by investigating nanoscale properties and mechanisms which are difficult to measure directly. The five sensor configurations analyzed and evaluated in this paper show significant differences in sensitivity, selectivity, and conductance change mechanisms. The computational results presented here suggest that the use of multiple sensor configurations, in combination, may offer opportunities to develop improved light weight, low power graphene based sensors for both nitramine and aromatic explosives.

Acknowledgements This work was supported by the Office of Naval Research (Grant Number N00014-16-2357). Computer time support was provided by the Texas Advanced Computing Center at the University of Texas at Austin (project group number G-815029) and the Department of 20


Page 20 of 38

Page 21 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Defense High Performance Computing Modernization Program (project ONRDC40983493).

References (1) Senesac, L.; Thundat, T. G. Nanosensors for Trace Explosive Detection. Mater. Today 2008, 11, 28–36. (2) Goswami, K.; Saini, D. P.; Klainer, S. M.; Ejiofor, C. H. Solid State Sensor for Carbon Monoxide. 1995; US Patent 5,405,583. (3) Zampolli, S.; Elmi, I.; Ahmed, F.; Passini, M.; Cardinali, G.; Nicoletti, S.; Dori, L. An Electronic Nose Based on Solid State Sensor Arrays for Low-Cost Indoor Air Quality Monitoring Applications. Sens. Actuators, B 2004, 101, 39–46. (4) Azad, A.; Akbar, S.; Mhaisalkar, S.; Birkefeld, L.; Goto, K. Solid-State Gas Sensors: A Review. J. Electrochem. Soc. 1992, 139, 3690–3704. (5) Varghese, S. S.; Lonkar, S.; Singh, K.; Swaminathan, S.; Abdala, A. Recent Advances in Graphene Based Gas Sensors. Sens. Actuators, B 2015, 218, 160–183. (6) Singh, E.; Meyyappan, M.; Nalwa, H. S. Flexible Graphene-Based Wearable Gas and Chemical Sensors. ACS Appl. Mater. Interfaces 2017, 9, 34544–34586. (7) Guo, L.; Yang, Z.; Dou, X. Artificial Olfactory System for Trace Identification of Explosive Vapors Realized by Optoelectronic Schottky Sensing. Adv. Mater. 2017, 29, 1604528. (8) Wu, Z.; Zhou, C.; Zu, B.; Li, Y.; Dou, X. Contactless and Rapid Discrimination of Improvised Explosives Realized by Mn2+ Doping Tailored ZnS Nanocrystals. Adv. Funct. Mater. 2016, 26, 4578–4586.

21



(9) Qu, J.; Ge, Y.; Zu, B.; Li, Y.; Dou, X. Transition-Metal-Doped p-Type ZnO Nanoparticle-Based Sensory Array for Instant Discrimination of Explosive Vapors. Small 2016, 12, 1369–1377. (10) Zhou, C.; Wu, Z.; Guo, Y.; Li, Y.; Cao, H.; Zheng, X.; Dou, X. Ultrasensitive, Real-time and Discriminative Detection of Improvised Explosives by Chemiresistive Thin-film Sensory Array of Mn 2+ Tailored Hierarchical ZnS. Sci. Rep. 2016, 6, 25588. (11) Gole, B.; Shanmugaraju, S.; Bar, A. K.; Mukherjee, P. S. Supramolecular Polymer for Explosives Sensing: Role of H-Bonding in Enhancement of Sensitivity in the Solid State. Chem. Commun. 2011, 47, 10046–10048. (12) Thundat, T.; Chen, G.; Warmack, R.; Allison, D.; Wachter, E. Vapor Detection Using Resonating Microcantilevers. Anal. Chem. 1995, 67, 519–521. (13) Thundat, T. G. Microcantilever Detector for Explosives. 1999; US Patent 5,918,263. (14) Yinon, J. Mass Spectrometry of Explosives: Nitro Compounds, Nitrate Esters, and Nitramines. Mass Spectrom. Rev. 1982, 1, 257–307. (15) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Mass Spectrometry Sampling under Ambient Conditions with Desorption Electrospray Ionization. Science 2004, 306, 471–473. (16) Liu, H.-B.; Chen, Y.; Bastiaans, G. J.; Zhang, X.-C. Detection and Identification of Explosive RDX by Thz Diffuse Reflection Spectroscopy. Opt. Express 2006, 14, 415– 423. (17) Zhou, Q.; Swager, T. M. Fluorescent Chemosensors Based on Energy Migration in Conjugated Polymers: The Molecular Wire Approach to Increased Sensitivity. J. Am. Chem. Soc. 1995, 117, 12593–12602.

22


Page 22 of 38

Page 23 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(18) Gao, D.; Wang, Z.; Liu, B.; Ni, L.; Wu, M.; Zhang, Z. Resonance Energy TransferAmplifying Fluorescence Quenching at the Surface of Silica Nanoparticles Toward Ultrasensitive Detection of TNT. Anal. Chem. 2008, 80, 8545–8553. (19) Masoumi, S.; Hajghassem, H.; Erfanian, A.; Molaei Rad, A. Design and Manufacture of TNT Explosives Detector Sensors Based on CNTFET. Sens. Rev. 2016, 36, 414–420. (20) Zhang, H.-X.; Hu, J.-S.; Yan, C.-J.; Jiang, L.; Wan, L.-J. Functionalized Carbon Nanotubes as Sensitive Materials for Electrochemical Detection of Ultra-Trace 2,4,6Trinitrotoluene. Phys. Chem. Chem. Phys. 2006, 8, 3567–3572. (21) Fowler, J. D.; Allen, M. J.; Tung, V. C.; Yang, Y.; Kaner, R. B.; Weiller, B. H. Practical Chemical Sensors from Chemically Derived Graphene. ACS Nano 2009, 3, 301–306. (22) Lu, G.; Ocola, L. E.; Chen, J. Reduced Graphene Oxide for Room-Temperature Gas Sensors. Nanotechnology 2009, 20, 445502. (23) https://www.techbriefs.com/component/content/article/1198-ntb/news/ news/26011?eid=333222515&bid=1607931. (24) Yun, Y. J.; Hong, W. G.; Choi, N.-J.; Kim, B. H.; Jun, Y.; Lee, H.-K. Ultrasensitive and Highly Selective Graphene-Based Single Yarn for Use in Wearable Gas Sensor. Sci. Rep. 2015, 5, 10904. (25) Robinson, J. T.; Perkins, F. K.; Snow, E. S.; Wei, Z.; Sheehan, P. E. Reduced Graphene Oxide Molecular Sensors. Nano Lett. 2008, 8, 3137–3140. (26) Chung, M. G.; Kim, D. H.; Lee, H. M.; Kim, T.; Choi, J. H.; kyun Seo, D.; Yoo, J.-B.; Hong, S.-H.; Kang, T. J.; Kim, Y. H. Highly Sensitive NO2 Gas Sensor Based on Ozone Treated Graphene. Sens. Actuators, B 2012, 166, 172–176. (27) Lv, R.; Chen, G.; Li, Q.; McCreary, A.; .etc, Ultrasensitive Gas Detection of Large-Area Boron Doped Graphene. PNAS 2015, 112, 14527–14532. 23



(28) Goh, M. S.; Pumera, M. Graphene-Based Electrochemical Sensor for Detection of 2,4,6Trinitrotoluene (TNT) in Seawater: the Comparison of Single-, Few-, and Multilayer Graphene Nanoribbons and Graphite Microparticles. Anal. Bioanal. Chem. 2011, 399, 127–131. (29) Tang, L.; Feng, H.; Cheng, J.; Li, J. Uniform and Rich-Wrinkled Electrophoretic Deposited Graphene Film: A Robust Electrochemical Platform for TNT Sensing. Chem. Commun. 2010, 46, 5882–5884. (30) Li, Y.; Fahrenthold, E. Ab Initio Study of Iodine-Doped Carbon Nanotube Conductors. J. Eng. Mater. Technol. 2018, 140, 021008. (31) Li, Y.; Fahrenthold, E. Quantum Conductance of Copper-Carbon Nanotube Composites. J. Eng. Mater. Technol. 2018, 140, 031007. (32) Soler, J. M.; Artacho, E.; Gale, J. D.; García, A.; Junquera, J.; Ordejón, P.; SánchezPortal, D. The SIESTA Method for Ab Initio Order-N Materials Simulation. J. Phys.: Condens. Matter 2002, 14, 2745. (33) Brandbyge, M.; Mozos, J.-L.; Ordejón, P.; Taylor, J.; Stokbro, K. Density-Functional Method for Nonequilibrium Electron Transport. Phys. Rev. B 2002, 65, 165401. (34) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865. (35) Perdew, J. P.; Zunger, A. Self-Interaction Correction to Density-Functional Approximations for Many-Electron Systems. Phys. Rev. B 1981, 23, 5048. (36) Vrcelj, R. M.; Sherwood, J. N.; Kennedy, A. R.; Gallagher, H. G.; Gelbrich, T. Polymorphism in 2-4-6 Trinitrotoluene. Cryst. Growth Des. 2003, 3, 1027–1032. (37) Choi, C. S.; Boutin, H. P. A Study of the Crystal Structure of Beta-Cyclotetramethylene Tetranitramine by Neutron Diffraction. Acta Crystallogr., Sect. B 1970, 26, 1235–1240. 24


Page 24 of 38

Page 25 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(38) Choi, C. S.; Prince, E. The Crystal Structure of Cyclotrimethylenetrinitramine. Acta Crystallogr., Sect. B 1972, 28, 2857–2862. (39) Carter, D. J.; Rohl, a. L. Noncovalent Interactions in SIESTA Using the VDW-DF Functional: S22 Benchmark and Macrocyclic Structures. J. Chem. Theory Comput. 2011, 8, 281–289. (40) Ruiz-Serrano, Á.; Skylaris, C.-K. A Variational Method for Density Functional Theory Calculations on Metallic Systems with Thousands of Atoms. J. Chem. Phys. 2013, 139, 054107. (41) Imry, Y.; Landauer, R. Conductance Viewed as Transmission. Rev. Mod. Phys. 1999, 71, S306. (42) Nakada, K.; Fujita, M.; Dresselhaus, G.; Dresselhaus, M. S. Edge State in Graphene Ribbons: Nanometer Size Effect and Edge Shape Dependence. Phys. Rev. B 1996, 54, 17954. (43) Han, M. Y.; Özyilmaz, B.; Zhang, Y.; Kim, P. Energy Band-Gap Engineering of Graphene Nanoribbons. Phys. Rev. Lett. 2007, 98, 206805. (44) Li, X.-F.; Wang, L.-L.; Chen, K.-Q.; Luo, Y. Tuning the Electronic Transport Properties of Zigzag Graphene Nanoribbons via Hydrogenation Separators. J. Phys. Chem. C 2011, 115, 24366–24372. (45) Mowbray, D.; Morgan, C.; Thygesen, K. S. Influence of O2 and N2 on the Conductivity of Carbon Nanotube Networks. Phys. Rev. B 2009, 79, 195431. (46) Kimouche, A.; Ervasti, M. M.; Drost, R.; Halonen, S.; Harju, A.; Joensuu, P. M.; Sainio, J.; Liljeroth, P. Ultra-Narrow Metallic Armchair Graphene Nanoribbons. Nat. Commun. 2015, 6 .

25



(47) Cai, J.; Ruffieux, P.; Jaafar, R.; Bieri, M.; Braun, T.; Blankenburg, S.; Muoth, M.; Seitsonen, A. P.; Saleh, M.; Feng, X.; Mullen, K.; Fasel, R. Atomically Precise BottomUp Fabrication of Graphene Nanoribbons. Nat. Lett. 2010, 466, 470–473. (48) Ruffieux, P.; Cai, J.; Plumb, N. C.; Patthey, L.; Prezzi, D.; andrea Ferretti,; Molinari, E.; Feng, X.; Mulllen, K.; Pignedoli, C. A.; Fasel, R. Electronic Structure of Atomically Precise Graphene Nanoribbons. ACS Nano 2012, 6, 6930–6935. (49) Ershova, O. V.; Lillestolen, T. C.; Bichoutskaia, E. Study of Polycyclic Aromatic Hydrocarbons Adsorbed on Graphene Using Density Functional Theory with Empirical Dispersion Correction. Phys. Chem. Chem. Phys. 2010, 12, 6483–6491. (50) Fujii, S.; Ziatdinov, M.; Ohtsuka, M.; Kusakabe, K.; Kiguchi, M.; Enoki, T. Role of Edge Geometry and Chemistry in the Electronic Properties of Graphene Nanostructures. Faraday Discuss. 2014, 173, 173–199. (51) Jia, X.; Campos-Delgado, J.; Terrones, M.; Meunier, V.; Dresselhaus, M. S. Graphene Edges: A Review of Their Fabrication and Characterization. Nanoscale 2011, 3, 86–95. (52) Vovusha, H.; Sanyal, B. DFT and TD-DFT Studies on the Electronic and Optical Properties of Explosive Molecules Adsorbed on Boron Nitride and Graphene Nano Flakes. RSC Adv. 2015, 5, 4599–4608. (53) Panchal, V.; Giusca, C. E.; Lartsev, A.; Martin, N. A.; Cassidy, N.; Myers-Ward, R. L.; Gaskill, D. K.; Kazakova, O. Atmospheric Doping Effects in Epitaxial Graphene: Correlation of Local and Global Electrical Studies. 2D Mater. 2016, 3, 015006. (54) Sun, J.; Muruganathan, M.; Mizuta, H. Room Temperature Detection of Individual Molecular Physisorption Using Suspended Bilayer Graphene. Sci. Adv. 2016, 2, e1501518.

26


Page 26 of 38

Page 27 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(55) Rumyantsev, S.; Liu, G.; Potyrailo, R. A.; Balandin, A. A.; Shur, M. S. Selective Sensing of Individual Gases Using Graphene Devices. IEEE Sens. J. 2013, 13, 2818–2822. (56) Zhang, Y.; Xu, M.; Bunes, B. R.; Wu, N.; Gross, D. E.; Moore, J. S.; Zang, L. OligomerCoated Carbon Nanotube Chemiresistive Sensors for Selective Detection of Nitroaromatic Explosives. ACS Appl. Mater. Interfaces 2015, 7, 7471–7475. (57) Chen, G.; Paronyan, T. M.; Harutyunyan, A. R. Sub-ppt Gas Detection with Pristine Graphene. Appl. Phys. Lett. 2012, 101, 053119. (58) Shao, L.; Chen, G.; Ye, H.; Niu, H.; Wu, Y.; Zhu, Y.; Ding, B. Sulfur Dioxide Molecule Sensors Based on Zigzag Graphene Nanoribbons with and without Cr Dopant. Phys. Lett. A 2014, 378, 667–671. (59) Schedin, F.; Geim, A.; Morozov, S.; Hill, E.; Blake, P.; Katsnelson, M.; Novoselov, K. Detection of Individual Gas Molecules Adsorbed on Graphene. Nat. Mater. 2007, 6, 652. (60) Ghaderi, N.; Peressi, M. First-Principle Study of Hydroxyl Functional Groups on Pristine, Defected Graphene, and Graphene Epoxide. J. Phys. Chem. C 2010, 114, 21625– 21630. (61) Milowska, K.; Birowska, M.; Majewski, J. Structural and Electronic Properties of Functionalized Graphene. Acta Phys. Pol., A 2011, 120, 842–844. (62) Dobrota, A. S.; Pašti, I. A.; Mentus, S. V.; Skorodumova, N. V. A General View on the Reactivity of the Oxygen-Functionalized Graphene Basal Plane. Phys. Chem. Chem. Phys. 2016, 18, 6580–6586. (63) Dan, Y.; Lu, Y.; Kybert, N. J.; Luo, Z.; Johnson, A. C. Intrinsic response of graphene vapor sensors. Nano Lett. 2009, 9, 1472–1475.

27



(64) Bagri, A.; Mattevi, C.; Acik, M.; Chabal, Y. J.; Chhowalla, M.; Shenoy, V. B. Structural Evolution During the Reduction of Chemically Derived Graphene Oxide. Nat. Chem. 2010, 2, 581. (65) Basu, S.; Bhattacharyya, P. Recent Developments on Graphene and Graphene Oxide Based Solid State Gas Sensors. Sens. Actuators, B 2012, 173, 1–21. (66) Zhu, Y.; Murali, S.; Cai, W.; Li, X.; Suk, J. W.; Potts, J. R.; Ruoff, R. S. Graphene and Graphene Oxide: Synthesis, Properties, and Applications. Adv. Mater. 2010, 22, 3906–3924. (67) Lin, L.-C.; Grossman, J. C. Atomistic Understandings of Reduced Graphene Oxide as an Ultrathin-Film Nanoporous Membrane for Separations. Nat. Commun. 2015, 6, 8335. (68) Borini, S.; White, R.; Wei, D.; Astley, M.; Haque, S.; Spigone, E.; Harris, N.; Kivioja, J.; Ryhanen, T. Ultrafast Graphene Oxide Humidity Sensors. ACS Nano 2013, 7, 11166– 11173. (69) Guo, L.; Hao, Y.-W.; Li, P.-L.; Song, J.-F.; Yang, R.-Z.; Fu, X.-Y.; Xie, S.-Y.; Zhao, J.; Zhang, Y.-L. Improved NO2 Gas Sensing Properties of Graphene Oxide Reduced by Two-beam-laser Interference. Sci. Rep. 2018, 8, 4918. (70) Yun, S.-W.; Gong, M.-S. Preparation of Flexible Resistive Micro-Humidity Sensors Using Quaternary Ammonium Salt-Modified Graphene Oxide and Their HumiditySensing Properties. Macromol. Res. 2014, 22, 1043–1049. (71) Jung, I.; Dikin, D.; Park, S.; Cai, W.; Mielke, S. L.; Ruoff, R. S. Effect of Water Vapor on Electrical Properties of Individual Reduced Graphene Oxide Sheets. J. Phys. Chem. C 2008, 112, 20264–20268.

28


Page 28 of 38

Page 29 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(72) Bai, J.; Zhong, X.; Jiang, S.; Huang, Y.; Duan, X. Graphene Nanomesh. Nat. Nanotechnol. 2010, 5, 190–194. (73) Liang, X.; Jung, Y.-S.; Wu, S.; Ismach, A.; Olynick, D. L.; Cabrini, S.; Bokor, J. Formation of Bandgap and Subbands in Graphene Nanomeshes with Sub-10 nm Ribbon Width Fabricated via Nanoimprint Lithography. Nano Lett. 2010, 10, 2454–2460. (74) Wang, M.; Fu, L.; Gan, L.; Zhang, C.; Rümmeli, M.; Bachmatiuk, A.; Huang, K.; Fang, Y.; Liu, Z. CVD Growth of Large Area Smooth Edged Graphene Nanomesh by Nanosphere Lithography. Sci. Rep. 2013, 3 . (75) Paul, R. K.; Badhulika, S.; Saucedo, N. M.; Mulchandani, A. Graphene Nanomesh as Highly Sensitive Chemiresistor Gas Sensor. Anal. Chem. 2012, 84, 8171–8178. (76) Han, T. H.; Huang, Y.-K.; Tan, A. T.; Dravid, V. P.; Huang, J. Steam Etched Porous Graphene Oxide Network for Chemical Sensing. J. Am. Chem. Soc. 2011, 133, 15264– 15267. (77) Schneider, G. F.; Kowalczyk, S. W.; Calado, V. E.; Pandraud, G.; Zandbergen, H. W.; Vandersypen, L. M.; Dekker, C. DNA Translocation Through Graphene Nanopores. Nano Lett. 2010, 10, 3163–3167. (78) Garaj, S.; Hubbard, W.; Reina, A.; Kong, J.; Branton, D.; Golovchenko, J. Graphene as a Subnanometre Trans-Electrode Membrane. Nature 2010, 467, 190–193. (79) Sathe, C.; Zou, X.; Leburton, J. P.; Schulten, K. Computational Investigation of DNA Detection Using Graphene Nanopores. ACS Nano 2011, 5, 8842–8851. (80) He, Z.; Zhou, J.; Lu, X.; Corry, B. Bioinspired Graphene Nanopores with VoltageTunable Ion Selectivity for Na+ and K+. ACS Nano 2013, 7, 10148–10157. (81) An, Y.; Ji, W.; Yang, Z. Z-Like Conducting Pathways in Zigzag Graphene Nanoribbons with Edge Protrusions. J. Phys. Chem. C 2012, 116, 5915–5919. 29



(82) Mowbray, D. J.; Garca-Lastra, J.; Thygesen, K. S.; Rubio, A.; Jacobsen, K. W. Designing Multifunctional Chemical Sensors Using Ni and Cu Doped Carbon Nanotubes. Phys. Status Solidi B 2010, 247, 2678–2682. (83) Huang, B.; Li, Z.; Liu, Z.; Zhou, G.; Hao, S.; Wu, J.; Gu, B.-L.; Duan, W. Adsorption of Gas Molecules on Graphene Nanoribbons and its Implication for Nanoscale Molecule Sensor. J. Phys. Chem. C 2008, 112, 13442–13446. (84) Wehling, T.; Novoselov, K.; Morozov, S.; Vdovin, E.; Katsnelson, M.; Geim, A.; Lichtenstein, A. Molecular Doping of Graphene. Nano Lett. 2008, 8, 173–177. (85) Ricciardella, F.; Massera, E.; Polichetti, T.; Miglietta, M.; Di Francia, G. A Calibrated Graphene-Based Chemi-Sensor for Sub Parts-Per-Million NO2 Detection Operating at Room Temperature. Appl. Phys. Lett. 2014, 104, 183502. (86) Yoon, H. J.; Yang, J. H.; Zhou, Z.; Yang, S. S.; Cheng, M. M.-C. Carbon Dioxide Gas Sensor Using a Graphene Sheet. Sens. Actuators, B 2011, 157, 310–313.

30


Page 30 of 38

Page 31 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figures

Figure 1: Schematic descriptions of: (a) a graphene nanoribbon-based sensing device, and (b) the graphene nanoribbon-based sensing process, with a functionalized nanoribbon exposed to both explosive and background gas molecules.

31



Figure 2: Schematic descriptions of five sensor types: (a) surface sensing of RDX by a pristine GNR, (b) edge sensing of RDX by a pristine GNR, (c) surface sensing of TNT by a C-O-C functionalized GNR, (d) surface sensing of TNT by a C-OH functionalized GNR, (e) sensing of HMX by a nanohole patterned GNR, oblique view, and (f) sensing of HMX by a nanohole patterned GNR, top view.

Figure 3: Conductance change at the Fermi energy and adsorption energy for three different explosive molecules and four different background gases, for pristine GNR surface sensing.

32


Page 32 of 38

Page 33 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4: Conductance change at the Fermi energy and adsorption energy for three different explosive molecules and four different background gases, for pristine GNR edge sensing.

Figure 5: Conductance change at the Fermi energy and adsorption energy for three different explosive molecules and four different background gases, for GNR sensing with a C-O-C functional group.

Figure 6: Conductance change at the Fermi energy and adsorption energy for three different explosive molecules and four different background gases, for GNR sensing with a C-OH functional group. 33



Figure 7: Conductance change at the Fermi energy and adsorption energy for three different explosive molecules and four different background gases, for nanoholed GNR sensing.

34


Page 34 of 38

Page 35 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 8: Charge density difference plots for four different gas molecules, sensed by a pristine GNR: (a) O2, (b) HMX, (c) RDX, and (d) TNT.

35



Figure 9: Charge density difference plots for TNT sensing by four different sensor types: (a) C-O-C functionalized GNR, (b) C-OH functionalized GNR, (c) pristine GNR (edge sensing), and (d) nanoholed GNR.

36


Page 36 of 38

Page 37 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 10: Computed sensitivity as a function of (a) sensor type and (b) explosive type, for GNR sensors adsorbing a mix of one explosive and four background gases (mole ratios determined by the absorption energies per unit mass for each species). Computed relative sensitivity as a function of (c) sensor type and (d) explosive type, for GNR sensors adsorbing a mix of one explosive and four background gases (mole ratios determined by the absorption energies per unit mass for each species).

37



Table of Contents graphic:

38


Page 38 of 38

Graphene-Based Sensing of Gas-Phase Explosives - ACS Applied

Recommend Documents