Single‐Atom Cu Stabilized on Ultrathin WO2.72 Nanowire for Highly Selective and Ultrasensitive ppb‐Level Toluene Detection

Abstract Various catalysts are developed to improve the performance of metal oxide semiconductor gas sensors, but achieving high selectivity and response intensity in chemiresistive gas sensors (CGSs) remains a significant challenge. In this study, an in situ‐annealing approach to synthesize Cu catalytic sites on ultrathin WO2.72 nanowires for detecting toluene at ultralow concentrations (R a/R g = 1.9 at 10 ppb) with high selectivity is developed. Experimental and molecular dynamic studies reveal that the Cu single atoms (SAs) act as active sites, promoting the oxidation of toluene and increasing the affinity of Cu single‐atom catalysts (SACs)‐containing sensing materials for toluene while weakening the association with carbon dioxide or water vapor. Density functional theory studies show that the selective binding of toluene to Cu SAs is due to the favorable binding sites provided by Cu SAs for toluene molecules over other gaseous species, which aids the adsorption of toluene on WO2.72 nanowires. This study demonstrates the successful atomic‐level interface regulation engineering of WO2.72 nanowire‐supported Cu SAs, providing a potential strategy for the development of highly active and durable CGSs.

XAFS spectra at the Cu K-edge were measured at 1W1B station in Beijing Synchrotron Ration Facility (BSRF). The Cu K-edge XAFS data were recorded in a fluorescence mode. Cu foil, CuO, and Cu2O were used as references. All spectra were collected at room temperature.
The acquired EXAFS data were processed according to the standard procedures using the ATHENA module implemented in the IFEFFIT software packages. The k 3 -weighted EXAFS spectra were obtained by subtracting the post-edge background from the overall absorption and then normalizing with respect to the edge-jump step. Subsequently, k 3 -weighted χ(k) data of Cu K-edge were Fourier transformed to real (R) space using a hanning windows (dk = 1.0 Å -1 ) to separate the EXAFS contributions from different coordination shells. To obtain the quantitative structural parameters around central atoms, least-squares curve parameter fitting was performed using the ARTEMIS module of IFEFFIT software packages.
The following EXAFS equation was used: where S 0 2 is the amplitude reduction factor, F j (k) is the effective curved-wave backscattering amplitude, N j is the number of neighbors in the j th atomic shell, R j is the distancebetween the X-ray absorbing central atom and the atoms in the j th atomic shell (backscatterer), λ is the mean free path in Å, ϕ j (k) is the phase shift (including the phase shift for each shell and the total central atom phase shift), σ j is the Debye-Waller parameter of the j th atomic shell (variation of distances around the average R j ). The functions F j (k), λ and ϕ j (k) were calculated with the ab initio code FEFF8.2.
Gas sensing performance testing.
The gas sensing properties were done in the laboratory. Figure S1 is a schematic diagram of the dynamic gas sensing performance testing system. The samples, prepared on alumina substrate (1 mm × 2 mm) with a printed gold electrode and a ruthenium oxide heater at the backside ( Figure S2), were put in a homemade chamber. All samples are drop-coated onto the alumina substrate with the gold electrode, until the gold electrode is completely covered by the sensing layer. The chamber is equipped with a gas inlet and an outlet, which are connected to a computer-controlled gas mixing system. The resistance of the sensor was continuously measured using a Keithley 2450 Source Meter with a computer interface. The sensor's operating temperature was determined by a voltage source, that is, the temperature of the heater can be altered by applying different voltages. The humidity was regulated by introducing dry synthetic air into a glass container filled with deionized water through bubbling. The relative humidity (RH) of the gas was regulated within a range of 0 to 70% by mixing the humid carrier gas with dry carrier gas at room temperature (25°C). For n-type semiconductors, the response was defined as Ra/Rg (Ra is the resistance of the sensor in air, and Rg is the resistance of the sensor in the target gas). The response and recovery time were defined as the time taken by the sensor to achieve 90% of the full magnitude change in resistance upon switching between target gas and air. [19] In situ infrared spectroscopy.

The experimental setup of Diffuse Reflaxions Infrared Fourier Transformations
Spectroscopy (DRIFTS) is combined with the dynamic gas sensing performance testing system to acquire the spectra continuously. The samples, prepared on alumina substrates, were put in a homemade chamber with a KBr window. The chamber was equipped with a gas inlet and an outlet connected to a computer-controlled gas mixing system. The DRIFTS spectra were acquired using a Vertex 80 v from Bruker with a nitrogen-cooled broad-band mercury cadmium telluride (MCT) detector with a spectral resolution of 4 cm -1 . The single-channel spectrum was continuously recorded every 10 min.

Computational details.
In this work, we had employed the first-principles to perform all density functional theory (DFT) calculations within the generalized gradient approximation (GGA) using the Perdew-Burke-Ernzerhof (PBE) formulation. [1][2][3] We had chosen the projected augmented wave (PAW) potentials to describe the ionic cores and take valence electrons into account using a plane wave basis set with a kinetic energy cutoff of 400 eV. [4,5] Partial occupancies of the Kohn-Sham orbitals were allowed using the Gaussian smearing method and a width of 0.05 eV. The electronic energy was considered self-consistent when the energy change was smaller than 10 -5 eV. A geometry optimization was considered convergent when the energy change was smaller than 0.05 eV Å -1 . In our structural model, U-correction was applied to the Cu and W atoms.
The vacuum spacing in a direction perpendicular to the plane of the structure was 20 Å for the surfaces. The Brillouin zone integration was performed using 2×2×1 Monkhorst-Pack k-point sampling for a structure.
The adsorption energies (ΔE ads ) would be defined as follows: Where E gas is the energy of adsorbed gas molecules, E surface is the total energy of the material surface when the gas is not adsorbed, and E gas+surface is the total energy of the system after adsorbing the gas.
The free energy was calculated using the equation: where G, E ads , ZPE and TS are the free energy, adsorption energy, zero-point energy and entropic contributions, respectively.

MD simulations.
In this work, two confined cases (pristine WO2.72 and Cu SA/WO2.72) were built for molecular dynamic (MD) simulations. Each confined case was constructed by using solid substrates and filled with 100 toluene. The substrates were set as rigid to ensure that the atoms of the substrates were fixed during the simulation. The temperature of each confined case was maintained at 300 K by using a Nosé-Hoover thermostat with a damping factor of 100 fs. The systems were equilibrated for 10 ns to obtain a stable system. [6] After that, we added 20 molecules of carbon dioxide and 20 molecules of water to each system. And another 10 ns of simulation was performed for whole-case relaxation. The tip3p model was used for water. [7] The opls-aa force field was applied for toluene and carbon dioxide. [8] The Interface Force Field was used for solid substrates. The Lorentz-Berthelot mixing rule was adopted for the van der Waals interactions of different kinds of atoms. All the cases were placed in periodic orthogonal boxes.
And all the MD simulations were performed by using LAMMPS software package. [9] Figure S1. Schematic diagram of the dynamic gas sensing performance testing system. (The mass flow controller is denoted by MFC) Figure S2. Structure diagram of the sensor plane electrode: (a) gold electrode and (b) ruthenium oxide heater at the backside.                                                            All as-synthesized samples display type IV isotherms with H3 hysteresis loops at relatively high pressure, indicating the presence of mesoporous in the nanowires. [10] As is evident from Pore diameter (nm) WO       For XRD patterns, the appearance of WO2.72 is supported by the visible diffraction peaks from (010) and (020) planes. Meanwhile, the XRD patterns agree with monoclinic WO2.72 phase (space group P2m, a = 18.318, b = 3.784, and c = 14.028 Å). [11] With the increasing Cu SA loadings, no apparent peaks of Cu NPs or other phases could be detected. However, for Cu NPs/WO2.72, the typical diffraction peaks of Cu NPs, corresponding to the (111) and (200) planes, are observed. [12] In addition, XRD analysis results reveal that introduction of Cu SAs does not alter the crystal structure of pristine WO2.72. Notably, the existence of Cu SAs has a certain effect on the crystallinity.   [13,14] We could see that, with the increasing Cu SA loadings, the peak intensity of the as-synthesized samples gradually decreases, which is deterioration of the crystallinity. It is worth noting that, compared with pristine WO2.72, the positions of the peaks around 256, 325 and 703 cm -1 shift to a high wavenumber after being loaded with Cu SAs, representing a blue-shift. [15] This phenomenon indicates that the introduction of Cu SAs increases the number of oxygen vacancies, i.e. the local atomic structure disorders/defects. [16][17][18]  samples.
An apparent EPR signal at a g factor of 2.0046 is observed, demonstrating the unpaired electrons in the oxygen vacancy sites. [19] Compared with pristine WO2.72, the EPR signals are stronger after Cu SAs introduction, implying more oxygen vacancies. The more oxygen vacancies could promote toluene gas adsorption, which is conducive to the enhancement of gas sensing performance. .72 previously reported. [20] In addition, the bands (1630 and 3410 cm -1 ) correspond to the bending vibration of W-OH groups from adsorbed water molecules and -OH groups in absorbed water molecules, respectively. [21] Compared with pristine WO2.72, the peak shapes of Cu SA/WO2.72-4%, Cu SA/WO2.72-6% and Cu SA/WO2.72-8% samples at about 823 cm -1 become wider, which may be due to the decrease in crystallinity. [22] 600 1200 1800 2400 3000 3600   shows the easy aggregation of Cu NPs reduces active sites exposed for toluene gas sensing reaction. However, with the increase of Cu SA loadings, the response values first increase and then decrease, which could be attributed to that excessive Cu SAs reduce the surface area of sensitive materials overall, therefore reduction in the reaction site results in lessened resistance change. Note that the Cu SA/WO2.72-4% sensor has the largest response value. Next, the gas sensing performance evaluation of the sensors in the manuscript is performed at an optimum operating temperature of 160°C.   According to the International Union of Pure and Applied Chemistry (IUPAC), the theoretical limit of detection (LOD) of the sensor is defined as: [23,24] LOD(ppb) = 3 RMS noise Slope where RMSnoise is the noise of the sensor in the baseline phase, Slope is the slope of the linear region of the sensor response-gas concentration function curve.
We replot 500 data points at the baseline before the toluene exposure and calculate the RMSnoise The standard deviation formula is as follows: [25,26]  The attenuation ratio (α) of the response signal is defined as: [27,28] α = R initial −R final R initial where R initial and R final are the initial and final response values of the sensor, respectively.
Langmuir isothermal model: [29] µ = R i R max = K L C 1 + K L C where R i is sensor response to toluene with a certain concentration of C (ppb), K L and R max are equilibrium constant and maximum theoretical response of the sensor, respectively.
Freundlich isothermal model: [30] R i = K F C 1 n ⁄ lnR i = lnK F + 1 n lnC where K F is equilibrium constant, the larger the K F value, the better the sensor's performance,    Thermodynamic equation of response: [29,30] The Gibbs free energy (ΔG 0 ) of the sensor is calculated as follows: Where R i is sensor response, C is the gas concentration (C = 2.5 ppm), ΔH 0 is enthalpy change, ΔS 0 is entropy change, ΔG 0 is gibbs free energy change, T(K) is kelvin temperature, and R α is the Avogadro constant, 8.314 J mol -1 K -1 .
The amount of the surface-adsorbed gas is directly proportional to the sensor response. Here, the sensor response values are fitted by thermodynamic equation. The values of ΔH 0 and ΔS 0 are calculated from the slope and intercept, respectively. As shown in Table S8, ΔH 0 >0 and ΔG 0 <0 (433 K) indicate that the gas sensing process is endothermic and spontaneous.   As we all know, the number of carriers in semiconductor increases exponentially with temperature. The higher the temperature, the higher the number of carriers, the higher the conductivity. In addition, the basic characteristics of semiconductor gas sensor resistance are temperature-dependent, which is closely related to the conduction mechanism of semiconductor materials. Thus, the sensor resistance decreases exponentially with temperature, satisfying the following the Arrhenius relation: [31,32] R air = R 0 exp where K B is the Boltzmann constant, T is absolute temperature in degrees Kelvin, E A is charge transport activation energy, R air is resistance value of sensor and C 0 is a constant related to the      is the debye-Waller factor to account for both thermal and structural disorders; d ΔE0 is the inner potential correction; R is the factor indicates the goodness of the fit; S0 2 was fixed to 0.91 according to the experimental EXAFS fit of Cu foil by fixing CN as the known crystallographic value. Fitting range: 3.0 ≤ k(/Å) ≤ 10 and 1.5 ≤ R(Å) ≤ 3.2 (Cu foil); 3.0 ≤ k(/Å) ≤ 11.0 and 1.0 ≤ R(Å) ≤ ~2.6 (Cu SA/WO2.72-4%). A reasonable range of EXAFS fitting parameters: 0.700 < Ѕ0 2 < 1.000; CN > 0; σ 2 > 0 Å 2 ; ΔE0 < 10 eV; R factor < 0.02.