The Ti_0.2V_1.8C MXene Ink-Prepared Chemiresistor: From Theory to Tests with Humidity versus VOCs

Abstract

The 2D structure of MXenes attracts wide research attention toward an application of these materials in gas sensors. These structures are extremely sensitive to minor variations in their composition, which are employed for tuning their functional properties. Here, we consider the partially substituted MXenes of the composition of Ti_xV_2-xC, where x = 0.2, via quantum chemical calculations, and test their chemiresistive characteristics as a receptor component of the planar-type sensor and on-chip multisensor array. We thoroughly discuss the synthesis process of Ti_0.2V_1.8AlC MAX-phase and the corresponding MXenes, to prepare functional inks and, furthermore, deposit the films by microextrusion printing over an array of planar multi-electrode structures at the surface of a pen-sized chip. The crystal structure of the obtained materials is evaluated via X-ray diffraction analysis. The developed chip has been exposed upon few gaseous analytes, of alcohol VOCs, NH₃, and H₂O, of a 500–16,000 ppm concentration, at room temperature to ensure that we could observe the positive chemiresistive effect matured from resistance enhancing, with up to 10% vs. water vapors. The calculations carried in the framework of the density-functional theory for V₂C, Ti₂C, and Ti_0.2V_1.8C crystals ensured that the variations in their electronic structure were almost consistent with the experiment fundings: the most prominent effect is observed in relation to the H₂O vapors. Therefore, these Ti_0.2V_1.8C structures could be considered for applying them in room temperature-operated hygrometers.

1. Introduction

Nowadays, metal oxides in the form of thick or thin films, which operate at elevated temperatures, ordinarily up to 400 °C, are widely used to design gas sensors of a chemiresistive type [1,2,3]. In order to reduce the working temperature and to advance the sensitivity, other materials are considered to meet the challenges of various current applications, including the Internet of Things [4]. Since the beginning of the XXI century, two-dimensional materials, primarily matured from a one atom-thick layer of carbon, called graphene [5], started to develop rapidly [1,6]. The lack of bulk in these structures allows for the electrical characteristics to entirely depend on gas adsorption and other environmental changes and, so, to develop sensors operating even at room temperature [7]. However, carbon structures have still much lower values of the chemiresistive response compared to oxide ones. Therefore, other two-dimensional materials are being intensively searched for an application in the sensors, like sulfides, selenides, nitrides, and carbides of transition metals [8,9,10,11,12,13,14]. The last two ones are called as MXenes [15]. The MXene structures are already shown to exhibit a suitable gas-sensing performance in forward to numerous gaseous analytes, including volatile organic compounds (VOCs) at room temperature [16,17,18].

The control of the humidity content in the atmosphere is essential for many processes and applications [19] that leads to intensive research of MXenes towards exposing the H₂O vapors [20,21]. In order to advance the performance of such sensors, the researchers attempt to employ modified MXenes primarily via designing heterostructures. For instance, He et al. demonstrated [22] that adding SnO₂ to Ti₃C₂T_x improves both the chemiresistive response of this heterostructure when compared to pure MXenes, and its stability. Alternatively, a transition metal substitution is another way to modify the chemiresistive characteristics of MXenes [23]. It was shown that the presence of vanadium in the MXenes structure is supposed to increase the sensor response to humid air [24,25,26], which was confirmed [24] by an example of the V₂CT_x MXenes-based sensor.

Still, the important aspect of the gas sensors is a selectivity to analytes. To date, all the available sensors of the chemiresistive type have a rather low selectivity. To advance this issue, sensors are assembled into sets or multisensor arrays whose vector signal depends on the analyte’s type in a higher degree following an adequate choice of sensor components [27,28]. For the ability to produce such arrays in a mass scale, they are frequently designed in an on-chip platform via forming multiple electrodes on the same substrate to be covered by a gas-sensitive layer (see, for instance, [29]). The array’s vector signals generated in the presence of analytes are processed via pattern recognition technologies to enable their identification, while the gas concentration could be extracted from a normal calibration of any of the sensor’s signal in the array [30].

It is worth mentioning that the surface-like 2D structure of MXenes allows one to perform the theoretical studying of the adsorption of the analytes from a gaseous phase. The major approach goes via applying quantum-chemical calculations in the framework of the density functional theory (DFT) method with a generalized gradient approximation (GGA) and Perdew–Berke–Ernzerhof (PBE) parameterization [31,32,33,34,35,36,37,38]. In the case of MXenes with a semiconductor conductivity type, researchers turn to employ Hubbard corrections [39,40,41] and hybrid functionals due to the presence of strongly correlated d-electrons in the transition metal elements [40,42,43]. This powerful method makes it possible to accurately interpret the experimental results and explain the mechanism of chemisorption in MXenes of a different chemical composition.

According to our survey of the current literature on the application of MXenes as the receptor components of chemiresistive gas sensors, the partially substituted V₂C-based MXenes have not previously been considered for gas sensors and multisensor arrays. Moreover, the quantum-chemical calculations to detail the adsorption features of various analytes over such a material have not been performed. Thus, the aim of this work was studying (i) the synthesis process of Ti_xV_2-xC MXene structures, where the degree of vanadium substitution with titan, x, is equal to 0.2, (ii) the DFT calculation of the adsorption of various VOCs and H₂O molecules over the Ti_0.2V_1.8C structure in comparison to Ti₂C and V₂C ones, (iii) the pilot measurements of the chemiresistive effect in Ti_0.2V_1.8C structures upon exposing to the VOCs and H₂O at room temperature, and (iv) applying a pattern recognition technique to the vector signal generated by an on-chip Ti_0.2V_1.8C-based sensor array for a selective discrimination of the test analytes.

2. Materials and Methods

The scheme of the technology route to synthesize the Ti_0.2V_1.8C MXene and to study its chemosensory properties is drawn in Figure 1. The stages are described in detail in the following subsections.

2.1. Materials

Graphite (99%, MPG-8, Technocarb, Chelyabinsk, Russia), titanium (99%, Lanhit, Moscow, Russia), vanadium (99%, Lanhit, Moscow, Russia) and aluminum powders (99%, Lanhit, Moscow, Russia), KBr (99.99%, Rushim, Moscow, Russia), hydrofluoric acid (50%, Honeywell International Inc., Charlotte, USA), hydrochloric acid (36%, Sigma Tek, Moscow, Russia), ethylene glycol (99%, TK Spektr-Chim, Moscow, Russia), and tetramethylammonium hydroxide N(CH₃)₄(OH) (water solution, Technic, Saint-Denis, France) of an analytical grade were used in this work without a further purification.

2.2. Synthesis of the Ti_0.2V_1.8AlC MAX-Phase

At the first stage, the Ti_0.2V_1.8AlC MAX-phase synthesis was carried out according to the earlier reported protocols [44,45]. The choice of an aluminum-containing MAX-phase, where the Al atoms are located between the (Ti,V)₂C layers, is based on their higher reactivity upon an interaction with HF for a subsequent MXene synthesis, when compared, for example, with Si-containing MAX-phases. For this purpose, the metallic powders of Ti, V, Al, and graphite, of a 99% purity, were mixed in the molar ratios of 0.2:1.8:1.2:0.8 following a number of preliminary experiments with the purpose of minimizing the forming of the by-products (Figure 1, pos.1). Then, KBr powder in the mass ratio of 1:1 was added to the obtained reaction mixture to be a subject of combined grinding in the planetary ball mill at ethanol under 100–500 rpm for 12 h to improve the homogenization of the material (Figure 1, pos.2). Next, the powder was dried under an advanced temperature of 70–80 °C to be formed into dense molds by pressing at 100 bar (Figure 1, pos.3). These samples were placed in ceramic crucibles covered with a protective layer of KBr powder and heat-treated in the muffle furnace at 1100 °C for 3 h (Figure 1, pos.4) following the inertial cooling down to room temperature (RT). After cooling, the samples were washed from the KBr residuals with a hot distilled H₂O and dried at 110 °C until the completion of the weight loss (Figure 1, pos. 5).

2.3. Preparation of Ti_0.2V_1.8C MXene and Its Delamination

The MAX-phase powder (Figure 1, pos.6) was further used as a precursor to obtain the Ti_0.2V_1.8C MXene through a selective chemical Al etching. For this purpose, the Ti_0.2V_1.8AlC powder was introduced into a mixture of concentrated HF and HCl acids mixed in the volume ratio of 3:2, stirred at RT for 30 min, and then incubated at 40 °C for 90 h (Figure 1, pos.7). This resulted in a selective removal of the Al layers located between the M₂C, where M = Ti, V, the layers and the formation of the Ti_0.2V_1.8C MXene multilayer agglomerates, whose surface is functionalized mainly by F- and O- ions, and OH-groups (Figure 1, pos.7). After the chemical etching, the MXene-contained suspension was repeatedly washed from by-products with the distilled H₂O by a centrifugation at 3500 rpm until the hydrogen potential of the solution (pH) was equal to 5–6 at least.

The obtained multilayer agglomerates of Ti_0.2V_1.8C MXene (Figure 1, pos.8) were further delaminated into separate two-dimensional structures. The 12% aqueous solution of tetramethylammonium hydroxide, N(CH₃)₄(OH), in the ratio of 10 g/l was added to the corresponding powder, followed by ultrasonic treating for 20 min (Figure 1, pos.9). In order to isolate the single-layer nanosheets of Ti_0.2V_1.8C MXene and to eliminate the delamination agent, the resulting suspension was centrifuged at 3500 rpm for 60 min, the supernatant was removed, and the solid phase was washed twice with the distilled H₂O (Figure 1, pos.10).

2.4. Fabrication of Ti_0.2V_1.8C MXene-Based Multielectrode Chip

In order to employ the MXenes for the sensor, the functional inks were prepared as a suspension by mixing the synthesized MXene structures with ethylene glycol. For this purpose, the aqueous MXene suspension was centrifuged at 15,000 rpm for 1 h, the solid phase was washed with an absolute ethanol, and then ethylene glycol was added. The resulting suspension was dispersed in an ultrasonic bath for 15 min. The output inks were applied to the dielectric substrate made of oxidized Si via a microextrusion printing at 10 µm spot resolution (Figure 1, pos.11). For the printing, we employed a home-made three-coordinate positioning system equipped with the pneumatic dozer and the capillary dispenser in the form of a hollow needle with an inner diameter of 150 µm. The chip substrate has been prior equipped with multiple co-planar Pt strip electrodes of a 50 µm width with an inter-electrode distance of 50 µm, meander-shaped heaters, and thermoresistors [46], all at the frontal side. The latter two ones were not employed in the course of these studies because of the operation at RT. The substrate containing the MXene-based layer was then dried at 150 °C under a reduced pressure, below 15 kPa, to avoid a spontaneous oxidation of the material. The MXene layer confined between each couple of strip electrodes has been served as a sensor element of the chemiresistor type. All the sensor elements located over the electrode set have been considered as a multisensor array.

Next, the chip with the deposited layer of the Ti_0.2V_1.8C MXene structures was wired (Figure 1, pos.12) into the ceramic holder [47] containing electric tracks to connect the measuring electrodes with the output multi-pins of an interface socket (SMC 1.27 mm, Erni Electronics, Adelberg, Germany).

2.5. Quantum-Chemical Calculations

An experimental identification of the chemisorption processes in MXenes is a challenging task. However, the theoretical study using a density functional theory (DFT) could provide an in-depth understanding of the mechanism of interaction of MXenes with gaseous analytes [48,49,50,51]. In our work, the first principle study of the MXene structures was performed using the Siesta 4.1.5 DFT code [52,53,54,55,56] under Grimme corrections, DFT-D2, in order to consider the Van der Waals interaction. We have employed the split valence basis set, including the polarization functions, of double-zeta plus polarization, with a generalized gradient approximation (GGA) and Perdew–Berke–Ernzerhoff (PBE) parameterization, since these calculation approaches have been proved to be appropriate in terms of the accuracy of the calculation and the calculation time. The 5 $\times$ 5 $\times$ 1 Monkhorst–Pack grid [54] is applied for integration into an inverse space. To minimize the total energy, the geometry relaxation of the atomic coordinates and translation vectors of the MXene structures were approached via considering the Hellman–Feynman forces under Bullet-type’s corrections in the framework of the modified Broyden’s method [55]. When optimizing the MXenes cells, the force action between atoms was chosen to be 0.04 eV/Å, while the 50 Å vacuum layer over the MXene surface was taken into consideration. The geometry relaxation resulted in atomistic models with the lattice vectors equal to 15.38 Å for Ti₂C, 14.74 Å for V₂C, and 14.68 Å for Ti_0.2V_1.8C. The real-space grid cutoff was 350 Ry. A calculation of the chemiresistive response of the MXene structure has been carried out within the Landauer–Büttiker formalism using non-equilibrium Green–Keldysh functions [56] according to:

$$G=\frac{2e}{h}{{\displaystyle \int }}_{-\infty }^{\infty }T\left(E\right)F_T\left(E-E_F\right)dE$$

(1)

where e is the charge of the electron, h is Planck’s constant, $E_F$ is the Fermi energy of the contacts, and T(E) is the probability of the electron transmittance through the channel to be defined as:

$$T\left(E\right)=Tr \left[{\Gamma }_L\left(E\right)G\left(E\right){\Gamma }_R\left(E\right)G^{†}\left(E\right)\right].$$

(2)

Here, $G^{†}\left(E\right),$ $G\left(E\right)$are the lead and lag Green’s functions which describe a contact with the electrodes, and Γ_L(E), Γ_D(E) are the level broadening matrices defined as:

$${\Gamma }_{L/R}\left(E\right)=i\left({\Sigma }_{L/R}\left(E\right)-{\Sigma }_{L/R}^{†}\left(E\right)\right)$$

(3)

where ${\Sigma }_{L/R}$ are the eigen energy matrices of the left and right electrodes. The Green’s matrices are calculated according to:

$$G\left(E\right)=\frac{1}{\left(E{S}_C-H_C-{{\displaystyle \sum }}_L\left(E\right)-{{\displaystyle \sum }}_R\left(E\right)\right)}$$

(4)

where $S_C$ is the matrix of the overlaps of the conducting channel atomic orbitals, E is the electron energy, and $H_C$ is the Hamiltonian of the conducting channel.

To calculate the chemiresistive response, the number of k-points, perpendicular to the current transfer, was chosen to be 100.

2.6. Instrumentation

X-ray diffraction (XRD) analysis of the materials under study, the MAX-phase powder, and the corresponding MXene film deposited on the glass substrate surface was performed using a D8-Advance diffractometer (Bruker, Billerica, MA, USA) equipped with a CuK_α source, 1.5418 Å, and Ni filter. The measurements were taken under 40 keV of energy at a current of 40 mA. The integration time was equal to 0.3 s/point with a step of 0.02°.

The microstructure and the chemical composition of the obtained Ti_0.2V_1.8C MXene was analyzed with the help of a scanning electron microscope NVision 40 three-beam workstation (Carl Zeiss, Inc., Oberkochen, Germany) combined with an energy-dispersive X-ray (EDX) spectrometer of INCA X-MAX 80 (Oxford Instruments, Abingone, UK). The scanning electron microscopy (SEM) images were acquired at an accelerating voltage U = 10 kV (secondary electron detector) and EDX spectrum at U = 20 kV.

The gas-sensing measurements were carried out in a home-made gas-mixing setup composing of a few bubblers containing the liquid analytes of an analytical grade, namely methanol, ethanol, isopropanol, ammonia solution, and distilled water (Figure 2). The bubblers were purged with dry air sourced from the corresponding generator (Peak Scientific, Glasgow, UK) with the constant flowrate of 400 sccm. The multi-line gas flows were managed via high-precision mass-flow controllers (Bronkhorst, Achterhoek, the Netherlands) to be mixed in the specified ratio. The gas mixtures were delivered to the chip under study installed into a sealed steel-stainless chamber equipped with entry and output tubes. The concentration of the analytes was calculated via [46] as:

$$C=\frac{F_{gas}{P}_{sat}}{F_{gas}{P}_{sat}+\left(P_{atm}-P_{sat}\right)F_{gas}+P_{atm}{F}_{air}}$$

(5)

where F_gas is the flow rate of gas through the bubbler (cm³/min), F_air is the flow rate of the background air (cm³/min), P_sat is the saturated vapor pressure of the analyte solution (mm Hg), and P_atm is the atmosphere pressure (mm Hg).

The resistance of the MXene layer located over the chip was measured via the corresponding multimeter (Keithley Instruments, Cleveland, OH, USA) assembled with National Instruments’ (Austin, TX, USA) multiple-entry NI-DAQ unit. All of the setup was driven by PC via Labview@ software (National Instruments, Austin, TX, USA).

3. Results and Discussion

3.1. Characterization of the MAX-Phase and MXene under Study

In order to study the MXene crystal’s structure, the functional ink was used to deposit the carbide film on the surface of a glass substrate. After drying the ink layer at 150 °C under a reduced pressure, the material has been found to have a crystalline structure, characteristic for MXene.

The XRD patterns characterizing the target Ti_0.2V_1.8AlC MAX-phase and the corresponding MXene of Ti_0.2V_1.8C are given in Figure 3. It could be seen that the obtained Ti_0.2V_1.8C MXene structure does not contain any of the crystalline impurities left after the initial components or byproducts.

The results of the EDX analysis showed the content of Ti and V atoms equal to 1.4 at.% and 13.2% at.%, correspondingly, which further confirms the given [Ti]/[V] ratio to be equal to approx. 0.11, and the appearance of MXene in the desired composition.

The microstructure of the Ti_0.2V_1.8C MXene layer placed over the surface of the glass substrate from the ink was studied by SEM (Figure 4). As can be seen from the electron microscopy images, the material appears as a homogeneous layer in a characteristic smooth morphology. Further, a detailed analysis of the carbide film’s structural features revealed that the material consists of an array made up of individual large-area thin nanosheets. Thus, these data indicate a high degree of delamination of the resulting Ti_0.2V_1.8C MXene at the stage of the preparation of the functional ink, as well as the retention of the two-dimensional structure of the corresponding disperse phase elements at the stage of the film’s formation on the substrate’s surface.

3.2. Quantum-Chemical Calculations

Three atomistic models of MXenes were considered for quantum-chemical calculations as Ti₂C, V₂C, and Ti_0.2V_1.8C MXene of the mixed composition at the 1T phase. Initially, we have employed the atomic configurations corresponding to the lowest possible supercells with lattice constants of 3.07 Å for Ti₂C and 2.95 Å for V₂C. Following the optimization of the atomic coordinates and lattice constants, the elementary structures were extended by translating the supercells. Thus, the atomistic models were obtained with lattice constants of 15.38 Å characterizing Ti₂C and 14.74 Å characterizing V₂C. By replacing the vanadium atoms in the V₂C structure with titanium ones, we have derived the atomistic model of Ti_0.2V_1.8C with lattice vectors equal to 14.68 Å. Such a rather large area of the supercells was required to consider a possibility for the analyte molecules to land over the surface of the MXenes. Originally, the analyte molecule was located above the surface of MXene at the distance of about 3.2 Å, where the total energy minimum of the MXene-analyte system was identified accounting for the geometry relaxation using Grimme dispersion corrections. The atomistic models under study are shown in Figure 5.

Further, in order to reveal the resistance of the supercell and its change due to the MXene interaction with analyte molecules, we have estimated the electron charge density distribution over the supercell atoms according to the Mulliken procedure [57]. The obtained values of the Mulliken charge are given in

Table 1. Mulliken charges, e, characterizing the interaction of analyte molecules with MXene surfaces.

MXene	Analyte
	Methanol	Ethanol	Isopropanol	Ammonia	Water
Ti0.2V1.8C	−0.01e	−0.11e	−0.006e	0.007e	0.003e
Ti2C	−0.028e	−0.061e	−0.016e	0.019e	0.012e
V2C	−0.006e	−0.009e	−0.011e	0.015e	0.004e

. Here, the negative value of the charge indicates a charge transfer from the MXene film to the analyte. As can be seen from the table, the transferred charge does not exceed the magnitude of 0.11e, which characterizes the interaction of Ti_0.2V_1.8C with the ethanol molecule.

The observed low values of the charge appear because all the analyte molecules are bound by Van der Waals forces to the MXene surface, as indicated by the binding energies yielded in

Table 2. The calculated binding energy, eV, characterizing the interaction of analyte molecules with MXene surfaces.

MXene	Analyte
	Methanol	Ethanol	Isopropanol	Ammonia	Water
Ti0.2V1.8C	−0.051	−0.044	−0.041	−0.058	−0.062
Ti2C	−0.101	−0.128	−0.112	−0.118	−0.082
V2C	−0.084	−0.098	−0.073	−0.106	−0.063

. As one can see, the binding energies do not exceed −0.128 eV, characteristic of an ethanol interaction with Ti₂C. In the case of Ti_0.2V_1.8C MXene, the lowest values of the binding energy are observed in range of 0.044–0.062 eV depending on the analyte. It indicates the easier adsorption/desorption of molecules over this surface when compared to Ti₂C and V₂C where the energies are higher. It is worth noting that these energies match well the conditions of an RT operation of the Ti_0.2V_1.8C MXene-based sensors.

In order to evaluate the contribution of the analytes to the change in the electronic structure of MXene, we have estimated the electronic density of the states (DOS) accounting for setting the smearing parameter equal to zero. In this case, even a minor DOS change due to some charge overflow is clearly noticeable than one taking the conventional DOS pattern with the smearing of the states. Figure 6 yields the DOS patterns calculated for Ti_0.2V_1.8C MXene without smearing at a pristine state and following the adsorption of the analytes under interest (alcohols, ammonia, and water).

It is well known that the resistance of a material depends on the number of electronic states near the Fermi energy. The contribution of electronic states into the conductivity decreases with a displacement along the energy axis to be relative to the Fermi energy, E-E_f, with the major one to lie in the approximate energy band of E-E_f = [−0.5 eV, +0.5 eV].

It can be seen from Figure 6 that the analyte adsorption results in a shift of the DOS peak positions, and the modulation of their magnitude near the Fermi energy, especially in the conduction band. In case of pristine Ti_0.2V_1.8C MXene, the three DOS peaks are concentrated in the range of ~50 meV off the Fermi energy, and the amplitude of one of the maxima is equal to 1780 (eV)⁻¹. The nearest peak is located in the conduction band, as a positive region of the energy axis, with the energy of ~25 meV and an amplitude of 1146 (eV)⁻¹. In this regard, we observe the lowest resistance value of 4.118 kΩ. When the H₂O molecule approaches the MXene surface, the peak at the energy of ~25 meV disappears and the amplitudes of the peaks closest to the Fermi energy are reduced down to values of about 750 (eV)⁻¹. A similar picture is also observed for an ethanol interaction with the MXene. The nearest peak arises at 51 meV in energy with an amplitude of higher than 500 (eV)⁻¹. An interesting picture is observed with the isopropanol molecule, whose interaction yields the nearest DOS to be at 17 meV with an amplitude of 1489 (eV)⁻¹. This process should lead to a significant reduction in the electrical resistance of the crystal under study, but this peak is the only one in the energy range up to 0.24 eV which levels out its contribution to the electrical conductivity. All these features in the DOS’ behavior lie behind the change in the MXene resistance upon the analyte adsorption.

We have estimated the resistance of the Ti_0.2V_1.8C MXene in a pristine state and following the interaction with the analytes via the method of nonequilibrium Green–Keldysh functions. Primarily, the introduction of Ti atoms into the V₂C crystal structure enhanced approx. two-fold the resistivity of the Ti_0.2V_1.8C structure. Figure 6b, the bottom plot, summarizes the resistance values derived for the Ti_0.2V_1.8C supercell in a pristine state and upon an analyte adsorption. Obviously, we cannot compare the exact values of the supercell resistance with the experimental data which relate to much larger structures. However, we could estimate the change in the MXene resistance upon the analyte’s appearance according to the definition of the conventional chemiresistive response as:

$$S=\left(\frac{R\left(analyte\right)}{R\left(pristine\right)}-1\right)·100\%$$

(6)

where R(analyte) and R(pristine) are the resistance of the MXene supercell with and without analytes on the surface. Figure 6b, the upper plot, shows the calculated S values which well correlate with the observed shifts in the DOS. For the pristine layer, the peak of the DOS shifts from the Fermi energy by approx. 25 meV, while the adsorption of the analytes leads to its larger shifting by approx. 50 meV. Such a shift yields an increase in the supercell’s resistance according to the above discussion regarding the regularities of the location of the DOS’ peaks. A similar pattern is observed for the chemiresistive response values in dependence on the test analytes.

3.3. Chemiresistive Characterization

Prior to measuring the gas response of the Ti_0.2V_1.8C MXene-based sensor elements, we have tested the I-V characteristics of the local layer confined between two measuring Pt electrodes at the chip. Figure 7a shows the curves recorded for five sensor elements under RT conditions. As one can see, the I(V) follows a linear function under various slopes depending on the MXene-layer resistance. The observed differences belong to a spatial variation between the elements, primarily in a volume of ink-deposited material as a prerequisite for the differentiation of the sensor elements for combining a sensor array, as discussed later. The resistance magnitudes lie in a sub-MOhm range which is easy to record and process with conventional read-out electronics.

The Ti_0.2V_1.8C MXene-based sensors were exposed to five volatile analytes of three alcohols (methanol, ethanol, and isopropanol), ammonia, and water vapors at a concentration, C, range of 500–16,000 ppm. The exposures to analytes at certain concentrations were repeated twice to ensure the reproducibility of the records. The typical resistance transients recorded for the exemplary sensor element operated under RT are drawn in Figure 7c regarding all five of the analytes. When the vapors appear, the resistance goes up with a magnitude defining by the analyte’s concentration; the higher the concentration, the larger the resistance gain. This behavior is quite characteristic to match the known one in MXenes chemiresistors [19,21] and fully match the quantum-chemical calculations given in Section 3.2. Still, we may note that the sensor response to methanol present at the concentrations below 2000 ppm is rather negligible, while other analytes deliver a clear chemiresistive response to all the test analytes in the whole concentration range under study. This response is reproducible and reversible. It is worth noting that we observe some residual steady enhancing of the resistance when the concentrations of the analytes exceed 1000 ppm, which indicates that not all the adsorbed species are fast desorbed from the surface under RT conditions. Normally, to accelerate these processes, some activation via a slight heating or ultra-violet irradiation is required [58].

We have estimated the recorded chemiresistive response, S, according to Formula (6) and plotted it versus an analyte concentration, as a major characteristic of the gas sensor, in Figure 7b. The curves go linear in a double log scale according to the $S~C^n$ function, which is consistent with the Freundlich isotherm. The slope of the curves depends on the power index, n, which is distinctive for various analytes. We have collected the observed values of n in

Table 3. The characteristics of Ti_0.2V_1.8C MXene-based sensor versus volatile test analytes: power index, n, characterizing S~Cⁿ fitting of experimentally observed chemiresistive response and calculated limit of detection (LOD).

Analyte	Methanol	Ethanol	Isopropanol	Ammonia	Water
n	0.38 ± 0.1	0.55 ± 0.01	0.49 ± 0.02	0.51 ± 0.02	0.61 ± 0.03
LOD, ppm	181	24	39	27	14

. As one can see, these values lie in the range of 0.38–0.61 which are similar to the ones observed in previous experiments with MXene structures [21].

Further, we have taken the magnitude of the electrical noise of the sensor recorded upon background air exposures and estimated the signal-to-noise ratio (SNR) for finding a limit of detection (LOD). Thus, the LOD value is determined as the lowest analyte concentration which yields a $\times$3 magnitude of SNR [59]. The calculated LOD values for various analytes are collected in Table 3. It is worth noting that the weakest chemiresistive effect is observed in case of methanol vapors with an LOD value equal to ca. 181 ppm, while the water vapors influence the MXene resistance in most degrees, yielding an LOD of more than an order lower, to be about 14 ppm. The humidity exhibits also the highest value of n equal to ca. 0.61 and the higher magnitude of the response. Therefore, we have compared the observed response of Ti_0.2V_1.8C MXene-based sensors to H₂O vapors with the ones previously published in the literature on titanium carbide MXenes as sensors (Figure 8a).

Here, we employ the S/C value because of the variations in the presentation of the data in the literature. This parameter is close to a sensitivity one, which is normally defined as ΔS/ΔC. It might be seen from Figure 8a that the Ti_0.2V_1.8C MXene slightly outperforms the literature data on the chemiresistive sensors based on titanium carbide MXenes in the range of water concentrations below 15,000 ppm. At a higher H₂O, there are few reports on the capacitive-type and quartz microbalance sensors whose response is still higher but was, however, recorded in the narrower range of the concentrations. The finding on so high a response of the Ti_0.2V_1.8C MXene-based sensor to the water vapors fully corresponds to our DFT calculations. In order to match the experimental data with the DFT ones, we have plotted the observed chemiresistive response to the test analytes at the highest concentration of 16,000 ppm in Figure 8b. While comparing this plot to the DFT data present in Figure 6b, we should note a large alignment between the experimental and theoretical estimations of the response. Still, the largest difference is the response to methanol vapors which has to deliver the effect comparable to the ethanol ones according to the DFT, while the experiment shows its effect as the lowest one. These differences should be further clarified.

Furthermore, the same (positive) sign of the chemiresistive response of Ti_0.2V_1.8C MXene-based sensors to various analytes does not allow us to selectively distinguish the test analytes without a prior knowledge on the probe content similar to other researches [67]. Indeed, if we take only the response magnitude as a measure, it does not provide the selectivity as discussed in numerous works [30]. Say, the response equal to ca. 1% might be induced (Figure 7b) by water vapors at 317 ppm or ethanol at 1219 ppm, or ammonia at 2119 ppm, or isopropanol at 3793 ppm. Therefore, to expand the sensor’s performance, we have collected the resistances of the five sensor elements in the Ti_0.2V_1.8C MXene-based on-chip array recorded in the stationary conditions upon exposing to various analytes. These multisensory data were processed with a linear discriminant analysis (LDA) protocol to build an artificial space with a dimensionality equal to N-1, where N is the number of classes to be recognized [68]; in our case N = 5. For framing the classes, we have employed a normal distribution of the vector data within each class with a 0.9 confidence level. The obtained LDA plot is shown in Figure 8c in a 2D cross-section of the first two LDA components. While the physical meaning of the LDA components is unknown, they reflect the specific features of the classes under interest, in our case the analytes, in relation to the vector input composing of the resistances of Ti_0.2V_1.8C MXene-based sensors. In Figure 8c, the gravity centers of all the classes are well distanced from the center of the coordinate system, which means that the analytes yield some specific output which links to the variations in the charge exchange in the adsorbate/adsorbent system in the framework of the DFT consideration discussed in Section 3.2. Moreover, these differences are significant enough to separate the analyte-related classes in the LDA space; the average Mahalonobis distance here is 73.6 un. It shows the ability to selectively distinguish the analytes according to the known E-nose approach [69].

4. Conclusions

In this study, quantum-chemical calculations and experimental chemosensory measurements were carried out to evaluate the possibility of employing partially substituted Ti_xV_2-xC (where x = 0.2) MXenes as a receptor component of the gas sensor and gas-analytical multisensor array. For this purpose, the synthesis process of the precursor, the Ti_0.2V_1.8AlC MAX phase, was investigated, and the corresponding MXene (Ti_0.2V_1.8C) was obtained following a selective etching of aluminum. The multilayer MXene agglomerates were delaminated into individual two-dimensional structures using tetramethylammonium hydroxide, and the resulting stable disperse system was utilized as the functional ink for the microextrusion printing of Ti_0.2V_1.8C layers on the surface of the multielectrode chip. The formation of the target MAX-phase and MXene was confirmed by X-ray diffraction analysis, and the results of the energy dispersive X-ray microanalysis indicated the given ratio of the metals. The performed quantum-chemical calculations allowed us to estimate the contribution of various analytes adsorption on the surface of MXenes with a V₂C, Ti₂C, and Ti_0.2V_1.8C structure into changing their electronic structure and, accordingly, the chemiresistive response. It was demonstrated that the greatest change in the electrical resistance of the Ti_0.2V_1.8C MXene is observed upon the adsorption of H₂O molecules when compared to other test analytes (methanol, ethanol, isopropanol, and ammonia). The theoretical and experimental results are in a good agreement, except for the methanol effect which needs a further clarification. To further improve the gas sensitivity, these structures could be combined into heterojunctions with oxides [70], including a partial oxidation of the pristine material [71].

The selectivity to the analytes has been approached via the multisensory processing of the array signal recorded from Ti_0.2V_1.8C MXene-based sensors. These findings show that designing complex MXene structures with a substitution of a major matrix with foreign atoms, here the V₂C with some Ti ones, might be an effective strategy to optimize the functional properties of this layered material.