Vanadium oxide-doped laser-induced graphene multi-parameter sensor to decouple soil nitrogen loss and temperature

Abstract

Monitoring nitrogen utilization efficiency and soil temperature in agricultural systems for timely intervention is essential to monitor crop health, promote sustainable and precision agriculture, and reduce environmental pollution. Therefore, it is of vital significance to develop a multi-parameter sensor for effectively and accurately decoupled detection of nitrogen loss and soil temperature, which is yet to be reported. Herein, this work presents a high-performance multi-parameter sensor based on vanadium oxide (${\text{VO}}_{\text{X}}$)-doped laser-induced graphene (LIG) foam to completely decouple nitrogen oxides (NO_X) and temperature. By exploiting the laser-assisted synthesis, the highly porous 3D ${\text{VO}}_{\text{X}}$-doped LIG foam composite is readily obtained by laser scribing vanadium sulfide (V₅S₈)-doped block copolymer and phenolic resin self-assembled films. Compared with the intrinsic LIG, the heterojunction formed at the ${\text{LIG/VO}}_{\text{X}}$ interface provides the sensor with a significantly enhanced response to NO_X, and an ultralow limit of detection (LOD) of 3 ppb (theoretical estimate of 451 ppt) at room temperature. The sensor also exhibits a wide detection range (from 3 ppb to 5 ppm), fast response/recovery (217/650 s to 1 ppm NO₂), good selectivity (ten-fold response to NO₂ over other interfering gases), and stability over 16 days. Meanwhile, the sensor can accurately detect temperature over a wide linear range of 10–110°C with a small detection limit of 0.2°C. The encapsulation of the sensor with a soft membrane further allows for temperature sensing without being affected by NO_X, presenting an effective strategy to decouple nitrogen loss and soil temperature for accurate soil environmental monitoring. The sensor without encapsulation but operated at elevated temperature removes the influences of ambient relative humidity and temperature variations for accurate NO_X measurements. The capability to simultaneously detect ultra-low NO_X concentrations and small temperature changes paves the way for the development of future multimodal electronic devices with decoupled sensing mechanisms for health monitoring and precision agriculture.

Graphical Abstract

ToC text:

The highly porous 3D vanadium oxide-doped laser-induced graphene foam composite is readily obtained by laser scribing vanadium sulfide-doped block copolymer and phenolic resin self-assembled films. The resulting foam composite provides a high-performance multi-parameter sensor to completely decouple nitrogen oxides and temperature, allowing for accurate detection of nitrogen loss and soil temperature for precision agriculture.

INTRODUCTION

One key issue in traditional agricultural practices is the enormous nitrogen fertilizers applied to soil for promoted food production¹, but resulting excessive nitrogen oxides (NO_X including NO and NO₂) emissions cause environmental pollution^2–4 (e.g., photochemical smog and acid rain³). Besides, soil temperature in the range from −10 to 50°C^{5, 6} also influences the physical, chemical, and microbiological processes in soil, which are key to the growth of plants and crops^{7, 8}. Continuous and real-time monitoring of soil conditions such as soil temperature and fertilizer emission will be essential to improve resource utilization, maximize agricultural yields, and minimize environmental hazards². Therefore, there is a demanding need to develop a multi-parameter sensor to monitor NO_X gas emission and soil temperature for efficient fertilization in smart or precision agriculture^{9, 10}. Despite their paramount importance in agriculture, gas-temperature sensors with decoupled sensing mechanisms especially those prepared by low-cost and scalable fabrication methods are rarely reported¹¹.

Sensors and sensing systems based on low-cost manufacturing approaches^{12, 13, 14} start to gain momentum in health monitoring and precision medicine due to their capability to collect big data from a large population and the potential to be integrated with artificial intelligence. Recent advances in gas sensors have explored various types of nanomaterials, including transition metal dichalcogenides (TMDs), metal-organic frameworks (MOFs), metal oxides, black phosphorus, as well as transition metal carbides and carbonitrides (MXene)^15–18. The synthesis strategies of these nanomaterials often involve solution phase reaction, hydrothermal reaction, chemical vapor deposition, and electrodeposition¹⁹. These nanomaterials prepared by relatively complex and costly methods also need to be integrated on high-resolution interdigitated electrodes that are fabricated with photolithography in chemiresistor gas sensors^20–22. Meanwhile, separately integrated heaters are commonly used to elevate the temperature for expedited gas adsorption/desorption toward real-time detection. As an alternative, the gas sensing platform based on 3D porous laser-induced graphene (LIG) by low-cost laser direct writing²³ can eliminate the need for separate heaters and high-resolution interdigitated electrodes due to large specific surface area and self-heating²⁴. The LIG has also been explored directly as a gas sensor and in many other sensors/devices^{23, 25–27}. However, it is challenging to control the porous structure of the LIG foam when using common carbon-containing materials during laser processes. Efforts to address this issue lead to the exploitation of self-assembled block copolymer during direct laser writing^{28, 29} in preparing for the LIG with hierarchically porous structures³⁰. Vanadium oxide (${\text{VO}}_{\text{X}}$) as a transition metal oxide exhibits excellent properties, including n-type conductivity good chemical and thermal stability, and excellent thermoelectric properties^{31, 32}. ${\text{VO}}_{\text{X}}$ and its composites also show excellent performance to physically adsorb and chemically interact with several gases, including nitrogen dioxide, ammonia, hydrogen, and methane, among others^33–36. However, the application of ${\text{VO}}_{\text{X}}$ and its composites is limited by complex synthesis methods, including spray pyrolysis³⁷, hydrothermal³⁸, sol-gel method³⁹, electrospinning⁴⁰, chemical vapor deposition⁴¹, and precipitation⁴². The recent report that directly synthesizes ${\text{VO}}_{\text{X}}$ with laser⁴³ provides inspiration and opportunities to directly laser write the ${\text{VO}}_{\text{X}}\text{/LIG}$ foam composites. Doping metal complexes in graphene is also found to improve the gas adsorption and detection sensitivity of resulting gas sensors^{44, 45}.

In this study, we report the one-step laser direct writing method to directly synthesize ${\text{VO}}_{\text{X}}$-doped 3D porous LIG foam nanocomposites by laser scribing the block copolymers (BCPs) doped with V₅S₈ precursor. Different from common carbon-containing materials (e.g., PI, PAI, PES, PPS) with only one monomer, the Pluronic F127-resols with a tunable mass ratio of Pluronic F127 copolymer to resols mixture in ethanol can leverage a bottom-up self-assembly process to control the resulting mesostructures and pore size distribution. The Pluronic F127-resols film can easily modulate the porous structure by the mass ratio of BCP to resins, whereas ${\text{VO}}_{\text{X}}$ particles can be anchored on the porous LIG without aggregation⁴⁶. The ${\text{VO}}_{\text{X}}\text{/LIG}\hspace{0.17em}\text{gas}$ sensor shows excellent selectivity for NO_X due to the lowest unoccupied molecular orbital (LUMO) of NO_X lower than that of other gases resulting in more electrons transferred from the ${\text{VO}}_{\text{X}}\text{/LIG}$⁴⁷. Meanwhile, the adsorption energy of ${\text{VO}}_{\text{X}}\text{/LIG}$ to NO₂ gas molecules is greater than that of other interfering gases. The heterojunction formed at the ${\text{VO}}_{\text{X}}\text{/LIG}$ interface significantly enhances the sensing performance of the multi-parameter sensor, allowing for the detection of the ultra-low concentration of 3 ppb NO₂ and a wide range of soil temperature with high sensitivity. The sensor encapsulated by a soft membrane can block the permeation of the gas molecules to respond only to temperature variations. The accurately measured temperature can, in turn, allow the dual-modal sensor to determine the NO_X emission from the soil. The proof-of-the-concept demonstration to decouple NO_X emission and temperature variations from the soil presented in this work can be leveraged to design and apply multimodal devices with decoupled sensing mechanisms for precision agriculture in all-weather conditions. Integrating the sensor with data processing and wireless transmission modules further results in a remote environmental monitoring system to wirelessly detect NO_X and temperatures for human health monitoring and precision agriculture.

RESULTS

Device structure and characterization

The ${\text{VO}}_{\text{X}}$-doped LIG sensor can be facilely obtained by laser scribing the carbon-containing material such as Pluronic F127-resols doped with a ${\text{VO}}_{\text{X}}$ precursor such as V₅S₈ (Fig. 1a, Fig. S1). In brief, spin coating the V₅S₈-doped Pluronic F127-resols solution on a Si substrate at different speeds forms a thin film with different thicknesses (Fig. S2, S3a). After drying the thermosetting phenolic resin, applying the CO₂ laser in a programmed pattern on the resulting thin film creates the ${\text{VO}}_{\text{X}}$-doped LIG sensor, which can be scalable for massive production (Fig. S3b). At a scanning rate of 12.7 μm/s, V₅S₈ in the thin film is instantaneously oxidized to${\text{VO}}_{\text{X}}$, providing uniform doping of ${\text{VO}}_{\text{X}}$ in the 3D LIG foam. The sensor capable of decoupling the NO_X and temperature (Fig. 1b) allows for NO_X and temperature monitoring in the soil (Fig. 1c).

Fig. 1. Schematic illustration showing.

a the fabrication, structure, b decoupling of temperature and gas, and c soil NO_X gas and temperature detection.

The porous-structured LIG (Fig. 2a) doped with the laser-induced V₅S₈ and ${\text{VO}}_{\text{X}}$ particles (Fig. 2b, red box) can be observed in the scanning electron microscope (SEM) image. Several different peaks in the C 1s and V 2p of XPS spectra indicate the successful formation of graphene and ${\text{VO}}_{\text{X}}$ in the laser process (Fig. S4). The two V 2p peaks at the binding energy values of 517.60 eV and 525 eV in the spectrum correspond to V 2p_3/2 and V 2p_1/2, respectively (Fig. 2c). The observed binding energy values with a spin-orbit splitting of 7.4 eV are in good agreement with the V₅⁺ oxidation state^{37, 48, 49}. The spin-orbit splitting between V 2p_3/2 and V 2p_1/2 peak is 7.4 eV. The Raman spectra confirm the presence of few-layered graphene in ${\text{VO}}_{\text{X}}$-doped LIG (Fig. 2d) by exhibiting three prominent peaks: the D (∼1338 cm^–1), G (∼1524 cm^–1), and 2D (∼2673 cm^–1)⁴⁸. The enlarged Raman spectrum shows peaks at 142, 193, 281, 404, 695, and 991 cm^–1 to confirm the formation of ${\text{VO}}_{\text{X}}$^{50, 51}. The combination of LIG and ${\text{VO}}_{\text{X}}$ in ${\text{VO}}_{\text{X}}$-doped LIG abnormally shifts the peaks compared with the previously reported ${\text{VO}}_{\text{X}}$⁵¹. The XRD patterns of ${\text{VO}}_{\text{X}}$-doped LIG show peaks at 24.0°, 30.8°, 34.3°, 36.7°, 45.4°, 47.6°, 55.8°, and 58.2° (Fig. 2e), corresponding to (201), (011), (310), (002), (112), (212), (412), and (611) (JCPDS card No. 30–0286) to confirm the formation of ${\text{VO}}_{\text{X}}$^{52, 53}. The spectra confirm the presence of ${\text{VO}}_{\text{X}}$, although it is difficult to distinguish from V₅S₈ due to the overlap in the bands^{54, 55}. The distinct distributions of C, O, S, and V elements in the EDS spectrum in the sensing area demonstrate that V₅S₈ is partially transformed to ${\text{VO}}_{\text{X}}$ during laser processing (Fig. 2f).

Fig. 2. Characterizations of laser-heated resin structures.

Scanning electron microscope (SEM) images of a LIG and b ${\text{VO}}_{\text{X}}$-doped LIG. c The narrow scan X-ray photoelectron spectrum (XPS) of V 2p regions for ${\text{VO}}_{\text{X}}$-doped LIG. d Raman spectrum of ${\text{VO}}_{\text{X}}$-doped LIG with a zoom-in shown in the inset. e XRD patterns of ${\text{VO}}_{\text{X}}$-doped LIG. f Energy-dispersive spectroscopy (EDS) spectrum of ${\text{VO}}_{\text{X}}$-doped LIG for C, O, S, and V.

Thickness effect of F127-resols hybrid film on the gas sensing performance

The film thickness modulates the laser-induced gas sensing area and gas permeation through the thickness, affecting the sensor performance. The $\text{LIG}\hspace{0.17em}\text{gas}$ sensor with either a small or large thickness has a low response and signal-to-noise ratio (SNR) in the gas response curve to 1 ppm NO₂ at room temperature (Fig. 3a, Fig. S5, Fig. S6). On the other hand, with the increase of film thickness, the conductivity increases but gas permeability decreases⁵⁶ to give a reduced gas response. The $\text{LIG}\hspace{0.17em}\text{gas}$ sensor with a larger thickness can provide abundant gas adsorption sites and continuous gas diffusion pathways, facilitating gas analyte-induced charge carrier generation for enhanced gas-sensing properties⁵⁷. However, as the film thickness exceeds 35 μm, some structures in the resulting LIG become poor quality with uneven 3D microstructures, leading to unstable sensing performance⁵⁸. Therefore, the optimal film thickness of 35 μm is selected in the following studies.

Fig. 3. Gas sensing performance characterization of ${\mathrm{VO}}_{\mathrm{X}}$-doped LIG.

a The response curves of the gas sensor based on pure LIG with different film thicknesses to 1 ppm NO₂. b The comparison in the response between the gas sensors based on LIG and ${\text{VO}}_{\text{X}}$-doped LIG to 1 ppm NO₂. c Typical response curve of the gas sensor to determine the response/recovery time. d Dynamic response test in the presence of NO₂ from 1 to 5 ppm at room temperature. e Repeatability test to 0.5 ppm NO₂ for five consecutive cycles. f Dynamic response test to NO₂ from 300 to 700 ppb at room temperature and g its linear fit to the calibration curve (error bars from three samples). h Experimental demonstration of the ultralow limit of detection to 3 ppb NO₂ at room temperature. i Selectivity test to NO_X over a wide range of other gaseous molecules. (Note: results are obtained from the gas sensor based on ${\text{VO}}_{\text{X}}$-doped LIG unless specified otherwise).

Compared with pure $\text{LIG}\hspace{0.17em}\text{gas}$ sensors, the ${\text{VO}}_{\text{X}}$-doped $\text{LIG}\hspace{0.17em}\text{gas}$ sensors provide a higher response, e.g., over a two-fold increase from 1.2% to 2.8% when exposed to 1 ppm NO₂ (Fig. 3b). The response/recovery time of 217/650 s to 1 ppm NO₂ at room temperature is also relatively rapid (Fig. 3c). As the recovery time of 300 s is sufficient to capture the characteristics of the gas sensor, this value is selected in the following studies unless specified otherwise. As the concentration is gradually increased from 1 to 5 ppm, the continuous response curve to NO₂ shows an increase from 2.5% to 5.5% (Fig. 3d), indicating a good dynamic response/recovery at room temperature. The response and recovery time also increases with the increasing gas concentration from 1 to 5 ppm (Fig. S7). The incomplete recovery comes from the short time set for rapid testing. The cycling stability of the ${\text{VO}}_{\text{X}}$-doped LIG sensor is confirmed by exposure to 0.5 ppm NO₂ at room temperature (Fig. 3e). The consistent response and recovery of the sensor over five cycles indicate the good steady-state response of the sensor. To provide a more accurate estimation of the theoretical limit of detection (LOD), the gas sensor is exposed to NO₂ with a progressively increased concentration from 300 to 700 ppb with a step size of 100 ppb, which gives a continuous response curve from 0.9%, 1.2%, 1.5%, 1.8%, to 2.1% (Fig. 3f). The linear fit of the sensor response to the NO₂ gas concentration from 300 to 700 ppb yields a slope of 2.929 ppm⁻¹ with a correlation coefficient (R²) of 0.997 (Fig. 3g). The linear fit of the gas sensor response to the lower concentration (3 ppb, 5 ppb, 7 ppb, 9 ppb, to 11 ppb) gives a slope of 8.4×10⁻⁵/ppb, which further leads to the determination of the theoretical LOD, defined as $3\times RM{S}_{\text{noise}}\text{/slope}$⁵⁹, as 451 ppt (Fig. S8). As it is challenging to use the static gas test setup to measure the gas concentration below 1 ppb (additionally spiked gas in the ambient environment), testing of the sensor to 3 ppb NO₂ still shows a response of 0.3‰ with a signal-to-noise ratio (SNR) of 26.85 (Fig. 3h). The ${\text{VO}}_{\text{X}}$-doped $\text{LIG}\hspace{0.17em}\text{gas}$ sensor only gives a small response of 0.2%, 0.06%, 0.062%, 0.12%, 0.22%, and 0.09% to 1 ppm SO₂, 100 ppm CO₂, 10 ppm NH₃, 100 ppm acetone, 100 ppm methanol, and 100 ppm ethanol, respectively (Fig. 3i and Fig. S9). In comparison, the significantly higher response of the sensor to NO_X (e.g., 2.7%/1.1% to 1 ppm NO₂/NO) highlights the excellent selectivity. The exposure of the chemiresistive ${\text{VO}}_{\text{X}}\text{/LIG}\hspace{0.17em}\text{gas}$ sensor to an oxidizing NO_X gas results in the extraction of the electrons in the valence band of the LIG to the adsorbed NO_X and from ${\text{VO}}_{\text{X}}$ to LIG⁴⁷. The lowest unoccupied molecular orbital (LUMO) determines the number of transferred electrons. As the LUMO of NO_X gas molecules is lower than that of other gases⁶⁰, more electrons are transferred from the LIG to give a larger response and high selectivity to NO_X. As a result, the ${\text{VO}}_{\text{X}}$-doped LIG gas sensor with an ultralow LOD and a high selectivity outperforms the other NO_X gas sensors based on different nanomaterials (Tab. S1).

Effects of operating temperature from self-heating on the gas sensing performance

As the temperature is often used to modulate the gas sensing performance^61–63, self-heating of the LIG is explored to modulate the temperature by changing the applied voltage during the resistance measurements (Fig. 4a) with the infrared thermal images shown in Fig. S10. As the temperature is increased from 22°C (room temperature) to 50°C, the response/recovery time decreases from 217/650 s to 88/406 s, but the response is also reduced from 2.5% to 1.3% (to 1 ppm NO₂) (Fig. 4b, Fig. S11). The accelerated response/recovery at elevated temperature results from the promoted electron transfer (crossing the potential barrier), but the accelerated desorption rate of gas molecules also results in a smaller response⁶⁴. This temperature-dependent behavior is consistent with the results of room-temperature NO_X gas sensors in the previous literature reports^65–68.

Fig. 4. Temperature effect on the $\mathrm{gas}$ sensing performance.

a Real-time response curves and b response/recovery properties of the ${\text{VO}}_{\text{X}}$-doped $\text{LIG}\hspace{0.17em}\text{gas}$ sensor to 1 ppm NO₂ at different operating temperatures. c Response of the ${\text{VO}}_{\text{X}}$-doped $\text{LIG}\hspace{0.17em}\text{gas}$ sensor to 1 ppm NO₂ in different humidity levels at 22 and 50°C (error bars from three samples). d Long-term stability test of the gas sensor to 1 ppm NO₂ for 16 days (measurement of 2000 s in each day).

As the relative humidity (RH) level can vary in a large range in greenhouse and soil environments, it is essential to analyze its influence on the gas sensing performance^{69, 70}. Because of the adsorption competition between NO₂ and water molecules on the sensor surface in the high RH range⁷¹, the response of the sensor to 1 ppm NO₂ decreases from 1.92% to 0.54% as the RH level is increased from 50% to 80% at room temperature (Fig. S12). However, the influence of RH on the NO₂ response can be drastically reduced when the sensor is operated at elevated temperatures⁶⁸ (e.g., a response of 1.43/1.41/1.4/1.31% in the RH of 50/60/70/80% at 50°C) (Fig. 4c, Fig. S13). The elevated temperature in the sensing area creates thermal radiation to drive the water molecules away from the region⁷². Additional strategies such as superhydrophobic coating can be further explored to minimize the effects of humidity^73–75. The sensor is also highly stable over time, as evidenced by the almost unchanged response to 1 ppm NO₂ for 16 days (Fig. 4d), demonstrating a high potential for practical applications.

Gas-sensing mechanisms and theoretical calculations

The gas sensing mechanism of the chemoresistive ${\text{VO}}_{\text{X}}$-doped LIG relies on the direct charge transfer between the absorbed O₂ and NO_X gas molecules and the sensing material. The large density of oxygen vacancy defects and dangling bonds in both LIG and ${\text{VO}}_{\text{X}}$ allow easy adsorption of oxygen molecules onto the composite structure at room temperature. The NO_X adsorbed on the ${\text{VO}}_{\text{X}}\text{/LIG}$ surface continuously extracts electrons and extends the hole (main carriers) accumulation zone on the ${\text{VO}}_{\text{X}}\text{/LIG}$ surface to lower the resistance, resulting in negative values in the relative changes as in the previous literature reports⁷⁶. In the ${\text{VO}}_{\text{X}}$-doped LIG sensor, the work functions of LIG and ${\text{VO}}_{\text{X}}$ are ca. 4.7 (W_G) and 6.8 eV (W_M), respectively. Due to the difference in work function, the majority charge carriers (holes and electrons) in the p-type LIG^{77, 78} and n-type ${\text{VO}}_{\text{X}}$⁷⁹ migrate across the heterojunction established at the interface forming a depletion layer by energy band bending (Fig. 5a). Heterojunction systems have been recognized to explain the enhanced gas sensing characteristics of ZnO/rGO⁸⁰, ZnO/NiO⁸¹, CuO/ZnO⁸², CoO/SnO₂⁸³, PdO-ZnO⁸⁴, ZnFe₂O₄-ZnO⁸⁵ in previous papers. The exposure of ${\text{VO}}_{\text{X}}$-doped LIG to NO₂ traps electrons from the conduction band and further bends the energy band (Fig. 5b) to decrease the resistance. Compared with pure LIG, the ${\text{VO}}_{\text{X}}$-doped LIG shows enhanced carrier transfer, leading to higher conductivity⁸⁶ as shown in the linear I-V curves (Fig. S14). Small changes in charge carrier concentration can also lead to great enhancement in the sensor response⁷⁸. The decoration of ${\text{VO}}_{\text{X}}$ also decreases the initial concentration of electrons in LIG, providing a larger change in NO₂ gas adsorption and response per electron⁸⁷.

Fig. 5.

Schematic showing band diagrams of the ${\text{VO}}_{\text{X}}$-doped LIG a before and b after contact. c The sensing mechanism with corresponding d band structures and DOS of ${\text{VO}}_{\text{X}}$-doped LIG. e Deformation charge density of NO₂ adsorbed on ${\text{VO}}_{\text{X}}$-doped LIG. f The adsorption energy of different gas molecules adsorbed on ${\text{VO}}_{\text{X}}$-doped LIG.

The difference in the structure and electronic properties between pristine LIG and ${\text{VO}}_{\text{X}}$-doped LIG can be revealed by the density functional theory (DFT) calculations^{88, 89}. Compared to pristine LIG, the ${\text{VO}}_{\text{X}}$-doped LIG with heterostructures (Fig. 5c) exhibits enhanced electronic levels near the Fermi level to elevate electron transfer (Fig. 5d) due to the ${\text{VO}}_{\text{X}}$ decoration^90–92. Deformation charge density elucidates large charge transfer between ${\text{VO}}_{\text{X}}$ and LIG and the adsorption of NO₂ on the sensor surface (Fig. 5e). The adsorption energy $E_{ad}$ of gas molecules on the sensor surface can be calculated as⁹³:

$$E_{\mathrm{ad}}=E_{{\mathrm{VO}}_X\mathrm{/LIG+gas}}-E_{{\mathrm{VO}}_X\mathrm{/LIG}}-E_{gas},$$

where $E_{{\mathrm{VO}}_X\mathrm{/LIG}}$ and $E_{{\mathrm{VO}}_X\mathrm{/LIG+gas}}$ are the total energy of the system before and after the adsorption of gas molecules and $E_{\mathrm{gas}}$ is the energy of the isolated gas molecule. The larger negative values of $E_{\mathrm{ad}}$ correspond to the stronger interaction between the sensor surface and the gas molecule. The adsorption energy of −1.434 eV of NO₂ on ${\text{VO}}_{\text{X}}$-doped LIG is almost 7 times of −0.219 eV on pristine LIG, indicating improved interaction between NO₂ and ${\text{VO}}_{\text{X}}$-doped LIG⁸⁹. Meanwhile, the adsorption energies of other gas molecules (e.g., NH₃, SO₂, CO₂, and acetone) on ${\text{VO}}_{\text{X}}$-doped LIG are lower in magnitude than that of NO₂ (Fig. 5f), which supports the highly selective detection of NO₂ over interfering gas molecules.

The temperature sensing performance of the multi-parameter sensor

Due to the high electron mobility, superior thermal conductivity, and structural stability at high temperatures of LIG^{68, 94}, the multi-parameter sensor shows a sensitive response with high repeatability (Fig. 6a) over a wide temperature range from 30 to 110°C (Fig. 6b). The fitting of the linear calibration curve gives a negative temperature coefficient and a sensitivity of 4.52×10⁻⁴°C^–1 (R² = 0.97) (Fig. 6c). Further increased linearity (R² = 0.99) is observed in the temperature range from 30 to 50°C (relevant for soil, human body, and infant formula temperatures) (Fig. S15). The sensor also exhibits a low limit of detection of 0.2°C (Fig. 6d) and good repeatability over heating-cooling cycles (Fig. 6e, f) to monitor both subtle and large temperature changes in practical applications. The proof-of-the-concept demonstrations include the detection of formula milk temperature in the bottle for the infant (Fig. 6g) or simulated fever (Fig. 6h), with a response time of 60 s that is much shorter than that (>6 min)⁹⁵ of most commercial mercurial thermometers. The above results confirm that the ${\text{VO}}_{\text{X}}$-doped LIG temperature sensor exhibits high sensitivity, wide detection range, fast response, and good reliability, which is suitable for dynamic temperature detection with favorable performance over the previously reported literature studies (Tab. S2).

Fig. 6. The temperature sensing performance of the multi-parameter sensor.

a The normalized relative resistance change ($\text{Δ}{\text{R/R}}_0$, %) of the sensor at different temperatures over multiple cycles. b The calibration curve of the sensor to temperature from 30 to 110°C and c its linear fit to determine the sensitivity (error bars from three samples) with the inset to show the resistance change during continuous heating. d The demonstration of the sensor to detect a small differential temperature change ΔT of 0.2°C. Repeatability test of the sensor e over five heating-cooling cycles between 40 and 70°C and f in various heating-cooling cycles. Dynamic response of the multi-parameter sensor to different g formula bottle and h skin temperatures (with the response curve shown in the inset).

Application of the ${\mathrm{VO}}_{\mathrm{X}}$-doped LIG multi-parameter sensor for soil monitoring

The ${\text{VO}}_{\text{X}}$-doped LIG sensor shows excellent sensing performance to both NO_X gas and temperature, so it is imperative for the sensor to decouple gas and temperature when the two stimuli are simultaneously present in large-scale soil monitoring (Fig. 7). Therefore, the PDMS membrane that is impermeable to gas is introduced as an encapsulation layer to break the symmetry in gas and temperature response (Fig. 7a). The ${\text{VO}}_{\text{X}}$-doped LIG sensor encapsulated with a 10 μm-thick PDMS membrane exhibits a significantly diminished response to NO₂ at room temperature (RH of 80%) (Fig. 7b, Fig. S16). Meanwhile, the rapid heat transport in the 10 μm thick PDMS membrane has minimal effect on the temperature sensing, resulting in a negligible difference with the unencapsulated sensor (Fig. 7c, Fig. S17). The encapsulated sensor can also be combined with the unencapsulated sensor operated at elevated temperature from self-heating to completely decouple the temperature and NO_X gas (Fig. 7d, Fig. S18). The temperature can be accurately captured by the encapsulated sensor, whereas the NO_X gas can be determined by the self-heated sensor as it removes the influence from the temperature changes, with no interference between the two input signals. In the proof-of-the-concept demonstration, the encapsulated sensor does not show any response to the NO₂ gas from 1 to 3 ppm (at room temperature), which is captured by the self-heated sensor operated at 50°C (blue shaded region in Fig. 7d). Similarly, the progressively increased temperature from 22 to 50°C and then decreased back to 22°C (to 3 ppm NO₂) does not cause any response in the self-heated sensor. Meanwhile, the temperature change is accurately detected by the encapsulated sensor (yellow shaded region in Fig. 7d).

Fig. 7. Demonstration of the ${\mathrm{VO}}_{\mathrm{X}}$-doped LIG sensor to decouple $\mathrm{gas}$ and temperature.

a Schematic of the encapsulated sensor to block the permeation of the gas molecules. b Response of the ${\text{VO}}_{\text{X}}$-doped LIG sensor with and without the encapsulation membrane to 1 ppm NO₂. c Response of ${\text{VO}}_{\text{X}}$-doped LIG sensors with varying encapsulation thicknesses to the temperature of 30, 40, and 50°C. d Application of the encapsulated ${\text{VO}}_{\text{X}}$-doped LIG sensor and the unencapsulated one operated at 50°C from self-heating to completely decouple NO₂ gas and temperature. The top illustrations show the changes in gas concentration and temperature.

The ${\text{VO}}_{\text{X}}$-doped LIG sensor capable of decoupling NO_X and temperature can be applied to accurately monitor the soil temperature and NO_X emission for future precision agriculture. The encapsulated sensor on the soil surface can accurately monitor the soil temperature without being affected by the NO_X gas emission (Fig. 8a). By heating the soil sample in the oven to simulate overheating, the soil temperature cycled from 35°C (suitable for crop growth) to 40°C (unfavorable to crop growth) and then back to 35°C and room temperature is accurately captured by our sensor (Fig. 8b). Meanwhile, the sensor without encapsulation but operated at an elevated temperature such as 50°C from self-heating can be used to detect NO_X volatilized from the soil at the specified temperature after applying urea (Fig. 8c). In the representative demonstration, the NO_X gas emission from the soil sample (23 cm × 19 cm × 4 cm) fertilized with 5 g urea can be detected at a much larger response than the un-fertilized control sample after three days (Fig. 8d), which is consistent with previous literature reports^{7, 96}. Although the sensor operated at room temperature is largely influenced by the RH (50–60% in the soil environment) to give a positive response (Fig. S19), the sensor operated at 50°C from self-heating largely eliminates the humidity effect to allow for accurate gas detection. As a result, the positive response is reduced to only 0.2‰ in unfertilized soils (RH of 50–60%) to demonstrate a very weak influence of humidity. The decoupled measurements of NO_X and temperature from the multi-parameter sensor can provide accurate, real-time monitoring of the crop growth environment for smart or precision agriculture.

Fig. 8. Monitoring of soil NO_X and temperature.

a Schematic diagram of excessive soil temperature and b the sensor response to suitable and over-heated soil. c Schematic diagram of urea application and d the difference in the sensor response to different fertilizer levels (operated at 50°C from self-heating).

Remote environmental monitoring system

The ${\text{VO}}_{\text{X}}\text{/LIG}$ sensor can be integrated with data processing and wireless transmission modules to yield a remote environmental monitoring system for human health monitoring and precision agriculture (Fig. 9). The signal measured by the ${\text{VO}}_{\text{X}}\text{/LIG}$ sensor and processed by a low pass filter to remove the noise is first digitized using a 16-bit delta-sigma modulator and then wirelessly transmitted to the smartphone via a Bluetooth module (Fig. 9a). When the real-time monitored NO₂ in the local environment around the human subject exceeds the safety threshold, the system automatically triggers on a red light and send alert to the smartphone for timely protection (Fig. 9b and Video S1). Additionally, the integrated wireless monitoring system can also be used to wirelessly monitor NO_X concentration after fertilization and soil temperature in the greenhouse for promoting plant growth in smart agriculture (Fig. 9c and Video S2, 3).

Fig. 9. The design and demonstration of the integrated remote environment monitoring system based on the ${\mathrm{VO}}_X/\text{LIG}$ sensor.

a The circuit design of the remote monitoring system for b real-time monitoring of NO₂ in the local environment of the human subject and c soil gas and temperature detection for smart agriculture.

DISCUSSION

In summary, this work reports the design, fabrication, and application of ${\text{VO}}_{\text{X}}$-doped LIG nanocomposites to decouple NO_X and temperature for soil environment monitoring. Created from a single-step laser scribing of V₅S₈-doped block copolymer and phenolic resin self-assembled films, the ${\text{VO}}_{\text{X}}$-doped LIG exhibits an ultra-low detection limit to NO_X and high sensitivity/precision over a wide temperature range. Introducing a soft membrane as an encapsulation layer on the sensor blocks the permeation of the gas molecules to provide accurate temperature measurements, which further helps decouple the NO_X gas from temperature. Additionally, the influence of the RH in the ambient environment can be effectively removed by operating the sensor at elevated temperatures from self-heating. The unencapsulated sensor operated at elevated temperatures also allows accurate measurements of the NO_X gas without being affected by environmental temperature variations. Therefore, the combination of the encapsulated ${\text{VO}}_{\text{X}}$-doped LIG sensor and the unencapsulated sensor operated at elevated temperatures can completely decouple temperature and NO_X without interference. The proof-of-the-concept demonstration of the multi-parameter decoupled sensors is showcased to detect nitrogen loss and soil temperature for smart agriculture. The design strategies and demonstrations from this work can also be leveraged to help create the next-generation multi-parameter stretchable sensors with decoupling sensing mechanisms.

MATERIALS AND METHODS

Preparation of V₅S₈-doped Pluronic F127-resols hybrid film.

The Pluronic F127-resols solution was obtained by mixing 4 g resols and 6 g Pluronic F127 copolymer in ethanol in a water bath (40°C, 4 h) (Fig. S1). Adding V₅S₈ to the above solution followed by stirring prepared the V₅S₈-doped Pluronic F127-resols solution. V₅S₈ were purchased from Nanjing MKNANO Tech. Co., Ltd. (www.mukenano.com). After vacuum treatment for 20 min, spin coating the solution on a silicon (Si) substrate at varying speeds (i.e., 200, 500, 850, 1000, 1500, 2000, 2500, 3000, 3500, and 4000 rpm), followed by drying in a vacuum oven at 150°C for 48 h, formed a uniform thin film (Fig. S3a).

Characterization.

Field-emission scanning electron microscopy (FESEM) (JSM 7100F, JEOL) was used to characterize the structure and morphology. X-ray photoelectron spectroscopy (XPS) was applied by ESCALAB 250 photoelectron spectrometer (Thermo Fisher Scientific, USA). Raman scattering was performed on a laser micro Raman spectrometer (Renishaw, in Via Reflex). X-ray diffraction (XRD) was obtained by a D8 Discover X-ray diffractometer.

Sensor fabrication and measurement.

The sensing region (width of 150 μm and length of 4.5 mm) with two square electrodes was directly created by scribing the V₅S₈-doped Pluronic F127-resols hybrid film with a CO₂ laser (Universal Laser, 10.6 μm, spot size of 127 μm, and power of 9 mW) (Fig. S3b). The laser processing parameters were fixed (power of 3.0%, speed of 1%, and PPI of 500), unless specified otherwise. Copper tapes with silver paste on the square electrodes connected the sensor to the data acquisition system. The PDMS solution was prepared at room temperature with stirring using 2 g of prepolymer (a) and 0.1 g of crosslinking agent (b) in a mass ratio of 20:1. Then, the PDMS solution was spin-coated onto the sensing area of ${\text{VO}}_{\text{X}}\text{-LIG}$ sensor at varying speeds (i.e., 6000, 7000, 8000, and 9000 rpm) and drying in a vacuum oven at 85°C for 1 h. Target gas are first collected into an aluminum foil gas collecting bag and then injected into a closed chamber for static detection. Different concentrations of NO_X were prepared by diluting and fully mixing the commercial calibration gas of 50 ppm NO_X with air in the chamber (volume of 5 L). The different relative humidity values in the chamber were prepared by the saturated salt solution method⁹⁷. The concentration of the $\text{VOC}$ was obtained by injecting the needed quantity of anhydrous liquid analytes into a sealed glass container using a microliter syringe. The concentration ($\text{C}$ in ppm) of the $\text{VOC}$ in the chamber was calculated using the following equation:

$$\text{C = }\frac{22.4\rho {\text{TV}}_{\text{S}}}{237\text{MV}}\text{,}$$

where $\rho$, ${\text{V}}_{\text{S}}$, and $\text{M}$ are the density (g ml⁻¹), volume $\left(\text{μL}\right)$, and molecular weight (g mol⁻¹) of the anhydrous liquid $\text{VOC}$, $\text{T}$ is the testing temperature $\left(\text{K}\right)$, and $\text{V}$ is the volume of the glass container $\left(\text{L}\right)$ filled with the $\text{VOC}$. Demonstration of the sensor performance is provided in Video S4.

The real-time resistance was recorded by a SourceMeter (Keithley 2400) at a constant voltage of 0.05 V. The sensor response is defined as $S=\text{Δ}R/R_0$ with $\text{Δ}R=R-R_0$, where $R_0$ is the initial resistance in air and $R$ is the resistance in the target gas. The response (or recovery) time is the time taken for the sensor response to reach 90% of the response (or recovery) at saturation in the target gas (or air).

Computational Methods.

We have employed the Vienna Ab initio Simulation Package (VASP) to perform all the spin polarized density functional theory (DFT) calculations within the generalized gradient approximation (GGA) using the Perdew-Burke-Ernzerhof (PBE) formulation. The projected augmented wave (PAW) potential was selected to describe the ionic cores and take valence electrons into account using a plane wave basis set with a kinetic energy cut-off of 520 eV. The density of k-meshs grids for Brillouin zone samping was set as $0.04\times 2\text{π/Å}$. 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 10⁻⁶ eV. The geometry optimization was considered convergent when the force on each atom was smaller than 0.02 eV/Å.