Enhanced Ammonia Gas Adsorption through Site-Selective Fluorination of Graphene

Abstract

Graphene has been widely explored as an ideal platform for gas sensing owing to exceptional properties, such as its atom-thin two-dimensional conjugated structure and large specific surface area. Herein, we report that, by introducing covalent C-F bonds via site-selective ion-beam-induced fluorination, graphene sensing response to ammonia gas can be considerably improved due to the enhanced gas adsorption on the surface of fluorinated graphene. The response to the ammonia gas increased by a factor of eight together with the limit of detection approaching 65 ppb. The absorption kinetics between the ammonia gas and fluorinated graphene were analyzed by using the Langmuir isotherm model and the result shows that the enhanced sensitivity is mainly attributed to the strong binding energy of fluorinated graphene to ammonia gas molecules, which is consistent with previous theoretical predictions.

1. Introduction

Graphene has been extensively explored as an ideal material in solid state gas sensors due to its atomically thin two-dimensional structure, large specific area and superior electrical properties [1,2,3,4,5,6]. The past decade has witnessed the advancement of graphene sensors and a number of graphene-based gas sensors have been applied for sensing various gases, such as NO₂ [7,8], NO [9] and CO₂ [10] among others. However, graphene-based ammonia (NH₃) sensors [8,11,12] still attracted considerable interest as the sensor response is relatively much lower (typically less than 5%) when compared to other gas species sensed with pristine graphene [13]. Despite the fact that many methods, such as doping [14], defect insertion [15,16,17] and surface functionalization [18,19,20,21], have been applied, it is still a challenge to improve the sensitivity of graphene-based gas sensors to NH₃. Theoretical studies have predicted that, by covalently introducing fluorine onto graphene surfaces ( graphene fluorination), the sensitivity of graphene to NH₃ molecules can be remarkably increased due to the strong binding energy between NH₃ molecules and fluorine sites [22,23,24]. However, no experimental approach so far has been demonstrated that can verify the hypothesis that such a process can significantly change gas adsorption kinetics. Graphene fluorination can be achieved by a variety of methods, such as chemical exposure to a fluorine-contained substance [25,26,27], plasma treatment under fluorine-based atmosphere [28,29,30,31] and beam induced chemical reactions using XeF₂ as a precursor gas [18]. Among these, the ion-beam-induced functionalization technique displays significant advantages in localizing fluorination sites and controlling fluorine concentrations [18], and could thus provide ideal means for graphene fluorination. The comparison of key parameters for various graphene-based gas sensors for NH₃ detection is shown in

Table 1. Graphene-based gas sensors for NH₃ detection at room temperature.

Sensing Materials	Detection Limit	Response Time	Recovery Time	Ref.
CVD-Graphene	500 ppb	6 h	6 h	[8]
F-CVD-Graphene	1 ppm	20 min	20 min	[32]
F-CVD-Graphene	2 ppm	<100 s	<200 s	[22]
F-Graphene	<300 ppm	20 min	-	[33]
PPy/CVD-Graphene	0.1 ppb	10 s	-	[34]

.

In this work, the ion-beam-induced fluorination method is employed to achieve fluorinated graphene (FG), which is subsequently characterized by Raman spectroscopy, X-ray photoelectron spectroscopy (XPS) and scanning tunnelling microscopy (STM). Systematic NH₃ gas sensing experiments were performed, and our results found that the FG exhibits a high sensing response to NH₃ gas that is considerably enhanced, by a factor of eight, compared to pristine graphene under ambient pressure. According to the Langmuir isotherm model, the limit of detection (LOD) of the obtained FG samples has achieved 65 ppb, which belongs to the class of ultrasensitive gas sensors.

2. Materials and Methods

Commercially available chemical vapor deposited (CVD) graphene on a Cu foil (monolayer, Graphenea, San Sebastián, Spain) was used as the main material in this work. The graphene was transferred to Si substrate covered with 300 nm thick SiO₂ via a PMMA-assisted transfer process [35]. Graphene channels and electrical contacts were fabricated by employing electron beam lithography (NanoBeam nB5, NBL, London, United Kingdom) and low-power O₂ plasma etching (Vision 320 RIE, Advanced Vacuum, Bernin, France) followed by a metal evaporation/lift-off process with Ti (5 nm)/Au (50 nm). The obtained graphene device is shown in Figure 1a (inserted). Subsequently, the graphene is fluorinated by focused ion beam (FIB)-induced functionalization, as described in detail in our previous work [18]. The focused ion beam (1 pA) is scanned across the target surface with an ion dose of 10¹³ ions/cm² under a vacuum of 10⁻⁶ mbar. At the same time, the precursor gas XeF₂ streams out from the gas nozzle to functionalize the graphene. Under the ion beam, the XeF₂ is decomposed and F radicals react with the surrounding C atoms. Since this reaction is more efficient, the F atoms will predominantly bind to the defect sites in the graphene lattice that were caused by the ion bombardment, see Figure 1b. Raman spectroscopy is carried out on a Renishaw inVia spectrometer, with a laser wavelength of 532 nm, to investigate the structural evolution of graphene after ion beam fluorination.

The gas sensing measurements were carried out in a gas sensing probe station under ambient pressure, and the current-voltage (I-V) and current-time (I-t) characteristics of the gas sensing devices were recorded by using a Keithley 6430 sub-femtoamp source meter. Prior to the gas sensing experiment, the chamber was purged with dry N₂ gas for 1 h before introducing the target gas. The gas flow rate was kept to 150 mL/min by a mass flow controller. The sample was applied with a voltage of 50 mV and no gate bias. The gas sensing measurements were performed under an ambient condition with a temperature of 20 °C ± 1 °C and relative humidity 45% ± 3%.

3. Results

As shown in Figure 1c, the D band appears at ~1350 cm⁻¹, which indicates the disorder level [36]. The intensity of this band is significantly weak for pristine graphene, which refers to the good quality of the graphene sample with a level of defects. After graphene fluorination, the increase in the D band in the FG sample is clear to see. Moreover, the decrease in the double resonance 2D band and the broadening of the intrinsic G band is consistent with the previously reported doping effect [37], thus providing strong evidence of the presence of covalent C-F bonds and induced defects on the functionalized graphene. The presence of the covalent C-F bonds is further confirmed by the distinguishable F1s peak in XPS spectra shown in Figure 1c, which were taken with a Quantum 2000 Scanning ESCA system (Physical Electronics, Chanhassen, United States). The atomic fluorine concentration of the FG sample was calculated as 7.5 ± 0.6% after being corrected with the sensitivity factor and the transmission function of the spectrometer. The fact that C-F bonds are preferably formed in the vicinity of defects during the ion-beam-induced fluorination shown in our previous studies is further confirmed by the STM analysis under vacuum condition, as shown in Figure 1b. In the centre of the STM image, a structural defect is observed. Such defects are only present in high density in graphene layers exposed to the focused ion bombardment and are thus attributed to the damage caused by ion impact. The defect is surrounded by standing waves, which further emphasizes the presence of the defect missing atoms or a covalent bond with a molecule, which shows chemisorption of molecules [38]. The bright contrast in the immediate surroundings of the defect is attributed to the fluorine binding to the pristine graphene. Due to the loss of local conjugation (electron delocalization along C=C bonds), this part of graphene is rendered less conductive. The larger area scan in Figure S1 also shows that, providing that their density is large enough, dispersed defects/fluorinated areas can be “connected” by the standing waves in between, even if most of the graphene surface remains undamaged, which is consistent with previous reports [18]. The ion dose used to functionalize the graphene sample imaged via STM was chosen to be 10¹³ ions/cm². At this dose, most of the graphene area remains intact while defects are sufficient enough to be localized.

Figure 2a shows normalized conductance response of pristine graphene, defected graphene (DG) and fluorinated graphene (FG) as a function of time under the exposure of 10 ppm NH₃. Here, the sensitivity $\mathrm{S}$ of the gas sensor is defined as the normalized conductance response as follows:

$$\mathrm{S}=\frac{{\mathrm{G}}_{{\mathrm{N}}_2-}{\mathrm{G}}_{{\mathrm{NH}}_3}}{{\mathrm{G}}_{{\mathrm{N}}_2}}\times 100\%,$$

(1)

where ${\mathrm{G}}_{{\mathrm{N}}_2}$ is the initial conductance of the sensor and ${\mathrm{G}}_{{\mathrm{NH}}_3}$ is the final conductance of the sensor after gas exposure. From the comparison, it can be shown that FG exhibits much better response (16.2%) than pristine graphene, with only 2.1% towards NH₃ gas at a concentration of 10 ppm. It is worth mentioning that the gas reaction time in our work was found to be 380 s, much shorter compared to other graphene-based ammonia gas sensors [39,40,41], which implies a quite fast response time in real applications. It was also found in our work that the recovery time of the FG sample was 780 s with pure N₂ purging, which is slightly increased compared to 600 s for pristine graphene. To better understand the repeatability and stability of the gas sensing experiment, the NH₃ gas sensing was repeated five times in the FG sample, as shown in Figure 2b. It can be clearly observed that the fluorinated graphene has almost the same response, indicating excellent repeatability of performance for the sensor. To further investigate the sensitivity to other gases for the FG sensor, the FG was exposed to CO, which is an acceptor-like gas molecule comparable to NH₃ [7]. The experimental condition was kept unchanged and the concentration of CO was fixed to be 10 ppm. Figure 2d shows that pristine graphene and FG exhibit a nearly negligible response to CO at room temperature, indicating high selectivity to NH₃, which is an important example for environmental sensor applications.

4. Discussion

To better understand the gas physisorption kinetics, as well as the limitations regarding detection, the NH₃ sensing experiment of the FG sample with various gas concentrations was performed as shown in Figure 2d. The Langmuir isotherm model was applied to investigate the superior NH₃ gas sensing performance of the fluorinated graphene [42]. The Langmuir isotherm model can be applied for the FG gas sensing mechanism and is summarized by the following formula [43,44,45]:

$$\frac{P}{S}=\frac{P}{S_0}+\frac{1}{K{S}_0},$$

(2)

where $P$ is the partial pressure of NH₃, $S$ is the response, $K$ is the equilibrium constant and $S_0$ is the saturated response. Figure 3a shows the relationship between $P$ and sensitivity. The values of $S_0$ and $K$ can be derived, from the best linear fit, as 0.19 and 0.40 Pa⁻¹, respectively, as shown in Figure 3a. The sticking probability of NH₃ on FG can be estimated by using the following relation between molecule coverage $\theta$ and time $t$:

$$\theta \left(t\right)=\theta (\infty )-\theta (\infty )exp\left(\frac{-K_{a}Pt}{\theta (\infty )}\right),$$

(3)

where $\theta (\infty )$ is molecule coverage at equilibrium under pressure $P$. $K_a$ is the adsorption constant defined as $K_a=r_a{P}^{-1}{\left(1-\theta \right)}^{-1}$ where $r_a$is the adsorption rate. Since, in the Langmuir isotherm model, sensitivity $S\left(t\right)$ is proportional to coverage as $\theta \left(t\right)=S\left(t\right)/S_0$, then Equation (3) can be written as:

$$S\left(t\right)=S(\infty )-S(\infty )\exp \left[\frac{-S_0{K}_{a}Pt}{S(\infty )}\right],$$

(4)

where $S(\infty )$ is the sensor response at the state of equilibrium. The equilibrium constant $K_a$ can be further written as the function of sticking probability:

$$K_a=\frac{s\sigma }{\sqrt{2\pi k_{B}mT}},$$

(5)

where $\sigma$ is the molecular cross-section, $m$ is the mass of NH₃ molecule, $k_B$ is the Boltzmann constant and $T$ is the temperature (300 K). Here, we take $\sigma$ and $m$ as 10⁻¹⁹ m² and 7.64 × 10⁻²⁶ kg, respectively. Figure 2b shows the FG sensor’s response to 10 ppm NH₃ under ambient pressure, and the solid line and dashed line represent the experimental and best-fitted value according to Equation (4), respectively. $K_a$ is found to be 7.37 × 10⁻³ s⁻¹ Pa⁻¹ according to the fitted value. By substituting the value of $K_a$ into Equation (5), the sticking probability is calculated to be 1.98 × 10⁻⁶. The adsorption energy $E_a$ can be expressed as:

$$E_a=-k_{B}TIn\left(\frac{K_a}{\upsilon K}\right),$$

(6)

where υ (10¹² s⁻¹) is attempt frequency. From the equation above, it can be obtained that the adsorption energy between NH₃ molecules and the fluorinated graphene gas sensor is −0.79 eV (−0. 51 eV for pristine graphene). Remarkably, the calculated value is in line with the theoretically predicted values due to the presence of strong hydrogen bonding between the C-F bond and N-H bond [46], leading to a strong sensitivity enhancement of graphene to ammonia gas. Moreover, we also calculate the limit of detection of the FG sensor to NH₃ by $LOD=3S_{rms}/l$, where $S_{rms}$ is the root mean square of base noise and $l$ is the slope of response as a function of concentration [47,48]. By substituting $S_{rms}$, which is calculated to be 6.65×10⁻⁴, from the base signal of FG and $l$ (3.2% ppm⁻¹) by fitting the linear part of response change as a function of partial pressure (as shown in Figures S2 and S3) into the previous formula, the LOD is calculated to be 64 ppb, which is in the class of ultrasensitive ammonia gas sensors.

The enhancement of ammonia gas sensing properties of graphene can be attributed to the improved bonding energy between ammonia gas molecules and fluorination sites [23]. The ammonia molecule as the acceptor would attract electrons from graphene, which results in the current decrease. It is also worth mentioning that defects, which are generated in the fluorination process, would also contribute to the gas sensing properties of graphene [17].

5. Conclusions

In this work, we have investigated the effect of fluorination on the NH₃ molecule adsorption properties of graphene through a site-selective ion-beam-induced functionalization technique. Our results show that, compared to pristine graphene, fluorinated graphene exhibits an eight-fold increase in sensing response to NH₃ with LOD down to 64 ppb, which belongs to the class of ultrasensitive ammonia gas sensors. These improved gas sensing properties are mainly attributed to the enhanced adsorption energy between NH₃ molecules and C-F sites of graphene after fluorination, which is revealed from sensing kinetics by the Langmuir isotherm model. Therefore, our work not only enables an ammonia gas sensor with superior sensitivity, but also contributes to studies involving interactions between gas species and graphene functional groups. The reported approach also has the potential to be employed in a variety of gas detection applications.