Sub-6 nm Palladium Nanoparticles for Faster, More Sensitive H2

Jan 17, 2017 - Sub‑6 nm Palladium Nanoparticles for Faster, More Sensitive H2. Detection Using Carbon Nanotube Ropes. Xiaowei Li,. †. Mya Le Thai,...
0 downloads 0 Views 1MB Size
Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)

Article

Sub-6 nm Palladium Nanoparticles For Faster, More Sensitive H2 Detection Using Carbon Nanotube Ropes Xiaowei Li, Mya Le Thai, Rajen Kumar Dutta, Shaopeng Qiao, Girija Thesma Chandran, and Reginald M. Penner ACS Sens., Just Accepted Manuscript • DOI: 10.1021/acssensors.6b00808 • Publication Date (Web): 17 Jan 2017 Downloaded from http://pubs.acs.org on January 23, 2017

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 free 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 accessible to all readers and 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.

ACS Sensors 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 25

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 Sensors

Sub-6 nm Palladium Nanoparticles For Faster, More Sensitive H2 Detection Using Carbon Nanotube Ropes Xiaowei Li,† Mya Le Thai,‡ Rajen K. Dutta,¶ Shaopeng Qiao,¶ Girija T. Chandran,¶ and Reginald M. Penner∗,‡ †Department of Chemical Engineering and Materials Science, ‡Department of Chemistry, University of California, Irvine, California 92697, United States ¶Department of Physics, University of California, Irvine, California 92697, United States E-mail: [email protected]

Abstract Palladium (Pd) nanoparticle (NP)-decorated carbon nanotube (CNT) ropes (or CNT@PdNP) are used as the sensing element for hydrogen gas (H2 ) chemiresistors. In spite of the fact that Pd NPs have a mean diameter below 6 nm and are highly dispersed on the CNT surfaces, CNT@PdNP ropes produce a relative resistance change 20 - 30 times larger than is observed at single, pure Pd nanowires. Thus, CNT@PdNP rope sensors improve upon all H2 sensing metrics (speed, dynamic range, and limitof-detection), relative to single Pd nanowires which heretofore have defined the stateof-the-art in H2 sensing performance. Specifically, response and recovery times in air at [H2 ] ≈ 50 ppm are one sixth of those produced by single Pd nanowires with crosssectional dimensions of 40 × 100 nm Pd. The LODH 2 is 18 MΩ · cm), and air dried. In step 4, the sample was dipped in 0.8 M nitric acid for 5 – 10 minutes, in order to etch away exposed nickel and also to produce a horizontal undercut beneath the protective photoresist layer. The height of this undercut exactly equalled the thickness of the nickel evaporated in step 1. In step 5, the nickel edge within this horizontal trench was used as the working electrode for the dielectrophoretic deposition of SWNTs ropes. A 50 mL one-compartment two-electrode electrochemical cell was used for dielectrophoretic deposition of CNTs. The photolithographically patterned Ni electrode was immersed in aqueous solution of dispersed SWNTs while leaving the other edge of nickel out of the solution and connected to a sourcemeter (Keithley Instruments, model 2611A). The counter electrode was a pre-cleaned 1 cm2 platinum foil. CNTs were dielectrophoretically deposited from aqueous 10 mg/L SWCNTs and the 0.1 g/L sodium dodecylsulfate (SDS) by the application of a series of voltage pulses having an amplitude of 20V, a duration of 0.1 s, at a rate of one every two seconds. A total of 100 pulses were applied to generate the CNT ropes used in the devices described here. When the deposition was finished, the remaining photoresist layer was completely dissolved and rinsed off by acetone, and then nickel layer was totally etched away by 0.8 M nitric acid, leaving one single SWNTs bundle adhering strongly to the glass surface. These processes are also comprised in step 5. Subsequent steps shown in Figure 1 were associated with the preparation of electrical contacts, and the electrodeposition of Pd nanoparticles, as described below. In Figure 1 step 6 a layer of 2 nm Cr was thermally evaporated onto the as-made single SWNTs bundles, followed by the second thermal metal evaporation of 80 nm Au. Then, in step 7, a layer of S1808 photoresist was spin-coated and baked for as already described. After the photoresist layer was patterned, in steps 8 and 9 the uncovered part of Au/ Cr layers was etched away, leaving patterned metallic layers on both sides of a 50µm-long single

6

ACS Paragon Plus Environment

Page 6 of 25

Page 7 of 25

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 Sensors

bundle SWNTs. Electrochemical Decoration of CNT ropes with Pd Nanoparticles. Pd nanoparticles were electrodeposited on 50 µm long sections of CNT ropes using a three-electrode electrochemical cell with saturated calomel reference electrode (SCE) and a platinum counter electrode. The aqueous plating solution contained 0.2 mM PdCl2 , 0.22 mM EDTA and 0.1 M KCl. Pd nanoparticles were electrodeposited by pulsing the applied potential as described in the text of this paper. For each pulse, the applied potential was -0.8 V vs. SCE, the pulse duration was 0.1 s. Between 50 and 400 deposition pulses were applied to achieve the range of QP d values explored in this paper. Integration of the net cathodic charge yielded QP d which included contributions from Pd deposition and H2 evolution. Scanning Electron Microscopy (SEM). Scanning electron micrographs were acquired using a FEI Magellan 400 XHR system. Energy dispersive spectroscopic (EDS) images were acquired by the same SEM system with EDS detector (Oxford Instruments, 80 mm2 , with Aztec software). Accelerating voltages of incident electron beams ranged from 1 kV to 5 kV, and probe currents ranged from 1.6 pA to 0.4 nA. All SEM specimens were mounted on stainless stubs and held by copper clips. Transmission Electron Microscopy (TEM). Transmission electron micrographs were acquired by using high resolution mode of a Phillips CM-20 system operating at an accelerating voltage of 200 kV. CNT ropes decorated with Pd NPs were held by 3 mm diameter a-carbon-coated copper TEM grids. Hydrogen Sensing. CNT@PdNP H2 sensors were mounted in a sealed flow cell (dead volume = 120 µL), equipped with two input ports – one for pre-mixed hydrogen/air the other for the air balance. The resistance of sensors was measured in situ as sensors were exposed to pulses of hydrogen gas at predetermined mixing ratios. Sensor resistance measurement was accomplished using a source-meter (Keithley Instruments, model 2611A) in concert with a multimeter (Keithley Instruments, model 2000). Flow controllers (MKS Inc., model 1479A) were used to control gas flow rates and to create pre-mixed hydrogen in air at

7

ACS Paragon Plus Environment

ACS Sensors

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

predetermined mixing ratios. A pair of switching valves (Parker Valve, switch cycle time = 25 ms) provided the means for switching between air balance and pre-mixed hydrogen/air pulses. These were controlled using a National Instruments interface (model BNC 2110) in conjunction with a computer. The gas composition, pulse parameters, and data acquisition were programmed and controlled using Labview. All hydrogen sensing experiments were carried out at ambient laboratory temperature (≈20 ◦ C) at a total gas flow rate of 1000 sccm.

Results and Discussion Preparation of Lithographically Patterned CNT Ropes. Hydrogen sensors were prepared using a variant of the lithographically patterned nanowire electrodeposition (LPNE) method (Figure 1). 27–29 LPNE was first used to prepare a single, linear CNT rope by pulsed, dc-dielectrophoresis (Figure 1, steps 1-5). This CNT rope was 100 µm in total length, and 1-2 µm in width. In prior work, patterned CNT ropes and rope arrays have been obtained

Figure 1: Process flow for the fabrication of a CNT@PdNP H2 sensor. Each process step is described in the Materials and Methods section.

by dielectrophoresis 30,31 using a variety of templates. 32–34 Alignment of individual CNTs with the axis of the rope is generally obtained because this direction coincides with the direction 8

ACS Paragon Plus Environment

Page 8 of 25

Page 9 of 25

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 Sensors

of the electric field gradient in the system, 32–34 but more disordered ropes have been obtained when the electric field gradient driving deposition is oriented orthogonal to the axis of the deposited rope. 31 This is the deposition geometry operating in our system for the deposition of CNT ropes using LPNE (Figure 2) and such ropes were used in this study. The LPNE patterning of dielectrophoretically deposited CNT ropes can be used to create arrays of linear ropes (Figure 2b), or ropes having more complex patterns (Figure 2c). The dielectrophoretic deposition of CNT ropes was accomplished from an aqueous solution containing 10 mg/L SWCNTs and the 0.1 g/L sodium dodecylsulfate (SDS) by the application of a series of voltage pulses having an amplitude of 20V, a duration of 0.1 s, at a rate of one pulse every two seconds. A total of 100 pulses were applied to generate the CNT ropes used in the devices described here, as shown in Figure 2b-e. A typical current-time trace acquired during CNT deposition (Figure 2a) shows an excess of anodic current consistent with the oxidation of water at +20V which occurred concurrently with the deposition of CNTs. The resulting CNT ropes consist of tangled assemblies of CNTs with a mean width of 1 - 2 µm (Figure 2d). The rope width dispersion of these CNT ropes is evident from these images. A 50 µm length was electrically isolated using evaporated gold contacts on a 2 nm thickness adhesive chromium interlayer (Figure 2e).

Palladium Nanoparticle Electrodeposition. Pd NPs were electrodeposited onto a 50 µm length section of CNT rope isolated between the lithographically patterned gold contacts (Figure 1, steps 10-13) using a multi-pulse electrodeposition scheme (Figure 3). Cyclic voltammograms for a single CNT rope in aqueous 0.2 mM PdCl2 , 0.22 mM EDTA, 0.1 M KCl show an onset for a irreversible reduction at 0.0V vs. SCE and an overpotential for palladium deposition, ηP d , on scan 1 of ≈ -300 mV. ηP d is the additional voltage, beyond the Nernst potential for Pd2+ reduction, that is required to form Pdo nuclei on CNT surfaces. Both Pd2+ reduction and hydrogen evolution can be expected to occur concurrently at this potential, and both are irreversible processes (Figure 3a). Pd nanoparticles were prepared

9

ACS Paragon Plus Environment

ACS Sensors

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 2: Patterning of CNT ropes prepared by LPNE-templated dielectrophoresis. a) Current versus time for the dielectrophoretic deposition of CNT ropes at an LPNE-patterned template induced by a train of +20 V × 0.10s pulses separated in time by 1.0 s. b) Scanning electron micrograph (SEM) of an array of CNT ropes on glass patterned at 20 µm pitch using LPNE. c) SEM image of loops of CNT rope patterned at 100 µm pitch using LPNE. d) Higher magnification SEM image of a section of CNT rope 1.0 - 1.5 µm in width. e) A Pd-CNT rope H2 sensor consisting of a single CNT rope decorated with Pd nanoparticles (not visible in this image) and evaporated gold electrical contacts.

10

ACS Paragon Plus Environment

Page 10 of 25

Page 11 of 25

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 Sensors

Figure 3: Electrodeposition of Pd NPs onto CNT ropes. a) Cyclic voltammograms at 20 mV/s in aqueous 0.2 mM PdCl2 , 0.22 mM EDTA, 0.1 M KCl showing reduction of Pd2+ and concurrent H2 evolution. b) Pd nanoparticles were electrodeposited onto a single CNT rope by applying a train of potential pulses of -0.80 V vs. SCE, using the pulse program shown here. c) Current (blue trace) and charge (green trace) versus time for two deposition pulses illustrating the build-up of cathodic charge corresponding to both Pd reduction (Pd2+ + 2e− → Pd0 ) and H3 O+ reduction (H3 O+ + 2e− → H2 + OH− ).

on CNT ropes in this solution by applying voltage pulses of -0.80 V (vs. SCE) × 100 ms separated by 2s wait times at 0.0 V (Figure 3b). This pulsed deposition scheme is intended to maximize the number density of Pd nanoparticles produced on the CNT rope because each pulse exceeds ηP d while the wait time between pulses allows Pd2+ ions to diffusively back-fill into the CNT rope after each pulse. Since Pd nanoparticle nucleation is first order in [Pd2+ ], refreshing the Pd2+ should enable the highest possible nucleation rate, resulting in size similar nanoparticles with a diameter that is correlated with the total coulombic loading. A plot of current and integrated charge versus time (Figure 3c) shows that cathodic currents exceed anodic currents, qualitatively as expected, and the build-up of the total excess cathodic charge, QP d , can be tracked as a function of time (Figure 3d). With successive pulses, the current amplitude also increases somewhat (Figure 3d), consistent with the elec-

11

ACS Paragon Plus Environment

ACS Sensors

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: TEM characterization of electrodeposited Pd nanoparticles. a) Low magnification TEM image of a CNT rope after the deposition of palladium nanoparticles (QP d = 42 µC). Palladium nanoparticles are distributed across this entire image, with some of the largest seen as dark spots in the lower right corner of this image. b) (Inset) Higher magnification view of the region shown in yellow in (a),. Blue circles highlight ≈40 Pd nanoparticles present in this region. A high resolution TEM image of one of these nanoparticles is shown in the main image. c) Histograms of Pd nanoparticle diameters obtained from TEM analysis of either 2 or 3 sensors prepared using the indicated QP d (Table 2).

12

ACS Paragon Plus Environment

Page 12 of 25

Page 13 of 25

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 Sensors

trocatalysis of both H2 evolution and Pd2+ reduction by the Pd NPs. This autocatalysis is also observed in the cyclic voltammograms shown in Figure 3a that show increasing peak currents on three successive voltammetric scans. Because the Pd2+ reduction current is augmented by H2 evolution, the measured QP d provides an upper limit on the actual Pd loading onto the CNT rope. QP d values of 15 µC to 102 µC were investigated here. Electrodeposited Pd nanoparticles distributed on the CNT sidewalls were characterized using transmission electron microscopy (TEM) (Figure 4a). Low magnification TEM images (Figure 4a) show tangles of CNTs within the rope. A few larger Pd NPs can be seen in this image on the lower right hand side, but most of the image appears to show none. However at higher magnification, it is apparent that very small, sub-6 nm diameter Pd NPs are distributed throughout the CNT rope. In the 120 nm × 120 nm region shown in yellow in Figure 4a, for example, ≈ 40 Pd NPs can be located and are highlighted with blue circles in the inset of Figure 4b. At still higher magnification (Figure 4b) an individual Pd NP is revealed at lattice resolution. Using TEM image data, histograms of Pd NPs were compiled as a function of the total deposition charge, QP d (Figure 4c). The mean diameters of Pd NPs ranged from 4.5 (± 1) nm for QP d = 15 µC, to 5.8 (± 3) nm for QP d = 102 µC. The relative standard deviation of the diameter, RSDdia , was in the range from 22% to 52% for these Pd NP distributions (Table 2). Compared with prior research in which Pd has been electrodeposited or vapor deposited onto CNT assemblies (Table 1), the pulsed plating method employed here yields smaller NPs that are also more narrowly distributed in diameter. We attribute the enhanced performance of the CNT@PdNP H2 sensors, as described below, to the small mean Pd NP diameter, the absence of larger Pd NPs in these distributions, and the cleanliness inherent to the electrodeposition process, which does not require the stabilization of Pd colloids by a ligand layer. 35

Detection of H2 in Air with CNT@PdNP H2 sensors. Four sets of CNT@PdNP H2 sensors, distinguished based upon the value of QP d , were tested. As already indicated,

13

ACS Paragon Plus Environment

ACS Sensors

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 5: CNT@PdNP H2 sensor responses, for devices prepared using four QP d values, as indicated. H2 densing data for three [H2 ] concentration ranges, low, medium, and high, are shown: Left, 10 ppm < [H2 ] < 100 ppm; Middle, 100 ppm < [H2 ] < 1000 ppm, Right, 0.2% < [H2 ] < 4%. R0 values, in the 80-90 kΩ range, are found in Table 2.

these four sensors had QP d ranging from 15 µC to 102 µC, corresponding to mean Pd NP diameters of 4.5 nm to 5.8 nm (Figure 3c). The resistance of a CNT@PdNP sensor was measured by applying a constant applied bias of Vapp = 0.50 - 1.0 V and measuring the current. The response of each sensor was evaluated across a H2 concentration of 10 ppm to 40,000 ppm (4%) at room temperature in air (Figure 5). CNT ropes that were not decorated with Pd NPs showed no detectable response to H2 at any concentration in this range (c.f. middle panel, black trace). All Pd CNT@PdNP H2 sensors produced a prompt, reversible increase in resistance, ∆R, upon exposure to H2 . We normalize ∆R by the initial resistance, R0 , to obtain a dimensionless signal, ∆R/R0 . Before we undertake comparisons between these four sensors, the amplitude of ∆R/R0 is striking because it is significantly larger than what is seen for single Pd nanowire sensors, which constitute some of the fastest and most sensitive H2 sensors so far reported. 8–11,13 A comparison with a Pd nanowire with lateral dimensions of 40 nm × 100 nm shown in Figure 6a reveals for example that ∆R/R0 is a factor of 30 higher for the least sensitive CNT@PdNP (QP d = 15 µC) at 1000 ppm. 8 If the Pd nanowire is heated, to accelerate response and recovery to H2 exposure, then the disparity

14

ACS Paragon Plus Environment

Page 14 of 25

Page 15 of 25

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 Sensors

is even larger because the signal amplitude is decreased (Figure 6a). All four CNT@PdNP H2 sensors are able to detect H2 at 10 ppm with a signal-to-noise of at least 10 (Figure 5, left panel) while a Pd nanowire produces a LODH 2 of 400 ppm with a signal-to-noise ratio of 2. 8 We conclude that CNT@PdNP H2 sensors are much more sensitive than Pd nanowires, producing significantly lower LODH 2 and expanded dynamic range for the detection of [H2 ]. Sensing metrics and Pd NP statistics are both summarized in Table 2. For CNT@PdNP H2 sensors, QP d influences sensor performance in two ways: First, the amplitude of the relative resistance change, ∆R/R0 , increases with QP d . QP d = 102 µC sensors produced ∆R/R0 values that are 6-8 times as high as those seen for QP d = 15 µC sensors across this entire [H2 ] range. This is readily apparent in the raw sensing data of Figure 5. Second, response and recovery times improve (decrease) with decreasing QP d across the concentration range we explored (Figure 6b,c). As shown in Figure 7, the effect of Pd NP diameter is strongest on recovery. Again, CNT@PdNP H2 sensors are significantly faster than Pd nanowires operating at room temperature. The influence of QP d on sensor speed is most evident in the lowest H2 concentrations (Figure 5, left). For QP d of 23 µC, 42 µC, and 102 µC, ∆R/R0 does not reach a steady-state, time-invariant value on the time scale of the 700 s exposures. But a time-invariant response is seen for QP d = 15 µC sensors because of the more rapid response and recovery of ∆R/R0 . Since even QP d = 15 µC CNT@PdNP sensors have an abundance of signal-to-noise and can readily detect 10 ppm, these sensors out-perform the three with larger QP d for the detection of H2 . In view of the fact that the Pd NP diameter is modestly changed from 4.5 nm to 5.8 nm across the range of QP d explored here, the strong and reproducible influence of QP d on response and recovery rate, as well as on ∆R/R0 , is somewhat surprising. The Pd NP diameter distributions overlap one another (Figure 4c), except at the large end of the distribution. For QP d = 102 µC sample, for example, there are nanoparticles larger than 8 nm in diameter present within the distribution where as these are not present in TEM images of QP d = 42 µC, 23 µC, or 15 µC sensors. Thus, it is possible that even though

15

ACS Paragon Plus Environment

ACS Sensors

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 larger particles are in the minority, the slow response of these NPs are responsible for retarding the overall response and recovery of ∆R/R0 of the entire Pd-decorated CNT rope, functioning much like H2 capacitors. The hyper-sensitivity of response and recovery speed to the Pd NP diameter - seen in the data of Figure 6b,c - readily explains why the still larger Pd deposits investigated in earlier papers produced dramatically slower sensor performance in many cases (Table 1). In the response data of Figure 6b, in particular, the influence of the α-to-β phase transition for PdHx is readily observed as a “glitch”-like decrease in the response time at 1% H2 , which is seen for all Pd NP diameters and for the Pd nanowire. A less obvious discontinuity in the response data (Figure 6c) is also seen at this concentration for the same reason. While many H2 sensors based upon ensembles of CNTs and ensembles of Pd particles have already been described (Table 1), no consensus has emerged on the mechanism responsible for the resistance increases induced by H2 exposure. The most definitive work, involving the single carbon nanotube experiments of Collins and coworkers, 18 showed that giant, 1000-fold resistance increases were observed for Pd NP-decorated SWCNTs were Pd nanoparticles were nucleated at sidewall defects. It is reasonable to assume that in our experiments, Pd NPs nucleate both at pristine CNTs and at sidewall defects. It is not surprising, then, that the maximum ∆R values we measure, up to 3R0 at high H2 concentrations (Figure 5, right), are larger than those measured by Collins et al. 18 for pristine SWCNTs decorated with Pd NPs, but smaller than those workers reported a single Pd particle nucleated at a single sidewall defect. A detailed description of the sensing mechanism awaits future work involving welldefined, single SWCNT chemFETs.

Conclusions The CNT@PdNP H2 sensor architecture approaches the amazing H2 sensing performance promised by the single SWCNT experiments conducted in 2010. 18 Specifically, this H2 sensor operates across a broad dynamic range of 3.5 orders of magnitude, has a limit-of-detection of

16

ACS Paragon Plus Environment

Page 16 of 25

Page 17 of 25

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 Sensors

Figure 6: Hydrogen sensing metrics for CNT@PdNP H2 sensors, and a comparison to a single Pd nanowire sensor (40 nm × 100 nm × 50 µm) operated at three temperatures. 8 a) ∆R/R0 , the relative resistance change, versus [H2 ] in air for five sensors, as indicated. Parameters for the Pd-CNT rope H2 sensors as summarized in Table 2. ∆R/R0 increases monotonically with Pd loading, and the mean Pd particle diameter. b) Response time versus [H2 ] in air, and, c) Recovery time versus [H2 ] in air for the same five sensors described in (a). Data for the Pd nanowire are adapted with permission from: Xiaowei Li, Yu Liu, John C. Hemminger, and Reginald M. Penner*, Catalytically Activated Palladium@Platinum Nanowires for Accelerated Hydrogen Gas Detection, ACS Nano 9 (2015) 3215, DOI: 10.1021/acsnano.5b00302 by the American Chemical Society.

17

ACS Paragon Plus Environment

ACS Sensors

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 7: Comparison of response (a) and recovery (b) for a single Pd nanowire H2 sensor (40 nm (h) × 100 (w) × 50 µm) and CNT@PdNP H2 sensors with QP d values of 15 µC, 23 µC, 42 µC, 102 µC, as indicated. [H2 ] = 1000 ppm. The horizontal dashed line indicates ∆R/Rmax = 0.90.