3. Results and Discussion
As the stability of the layer structure of MoO
3 is affected by the ion species and amount, 10 samples with different intercalated ions were prepared and tested during all of the experiments. The detailed information of the ten samples is shown in
Table 1 where 1 mmol, 2 mmol, and 5 mmol refer to the amount of the MCl (M: Li, Na, and K) in the precursor. As the experiment results in our previous study showed that the sensors exhibited worse performance when the amount of the precursor exceeded 5 mmol [
11], the precursor amount was controlled under 5 mmol in this work.
The sensors based on ion intercalated MoO
3 also exhibit good stability in their first test. The testing result is shown in
Figure 1b. It is found that the sensor exhibits an obvious and sharp response towards various hydrogen concentration and the detection limit is 400 ppm. Additionally, the sensor responds immediately as hydrogen concentration changes, the response time is usually 3 min, while the recovery time is a little longer. Moreover, the sensor shows good stability when it is exposed in a hydrogen concentration for 1 h, the fluctuation of the wavelength shifts during this period of time is within 2 pm. In addition, in both of the tests at 5000 ppm, the sensor has the same response, i.e., 139 ± 3 and 139 ± 1 pm, respectively.
Figure 2 shows the testing results of the sensors.
Figure 2a shows the sensor exhibited an obvious degradation phenomenon which is caused by the structural collapse of MoO
3 and agglomeration of Pt particles [
11]. Pt/MoO
3 is reduced by hydrogen and produced both water and heat. The wavelength shifts of Pt/MoO
3 decreased from 475 pm to 236 pm in the 50 tests. All the tests can be divided into four stages as three sharp declines occurred. It can be speculated that individual MoO
3 molecules will be degraded during each stage. Then, quantitative changes lead to qualitative change, whole MoO
3 layer degradation happens finally. For ion intercalated samples, the degradation phenomenon was obviously slowed down. In
Figure 2b, it can be found that the Li
+ intercalated samples still exhibit a degradation phenomenon but the rate of the degradation was significantly reduced. Li(1)-Pt/MoO
3 and Li(2)-Pt/MoO
3 exhibited almost the same response in the first 40 tests, a sharp decline in wavelength shift about 25 pm occurred in the second test for both two samples. In the last ten tests, the degradation rate of Li(1)-Pt/MoO
3 increased while it was reduced for Li(2)-Pt/MoO
3, which means that Li(2)-Pt/MoO
3 exhibited a better repeatability than Li(1)-Pt/MoO
3. In addition, Li(5)-Pt/MoO
3 exhibited the worst repeatability and the sensitivity was much lower than the other two samples. The testing results of Na
+ intercalated samples were shown in
Figure 2c. Similar to Li
+ intercalated samples, all of the three samples degraded during the tests. The degradation rate is much lower than that of Pt/MoO
3, which means Na
+ intercalation can improve the repeatability of the sensors. However, the optimal amount of NaCl added in a hydrothermal process lies in 1 mmol and the improvement of the repeatability reduced with the increasing amount of NaCl.
Figure 2d illustrates the cycling test results of K
+ intercalated MoO
3. Similar to Li
+ and Na
+ intercalated samples, K
+ intercalated samples exhibited better repeatability than Pt/MoO
3. K(1)-Pt/MoO
3 did not show a sharp drop in wavelength drift during the 50 tests, and its wavelength shift decreased 1–2 pm each time during the tests. During the 1–8 tests of K(2)-Pt/MoO
3, the wavelength shift decreased about 6–7 pm each time, the decreased rate slowed down to about 1–2 pm each time in the 9–45 tests. During the 46–50 tests, the wavelength shift decreased faster at a rate of about 3–5 pm each time. The wavelength shifts of K(5)-Pt/MoO
3 stepped down gradually during the 50 tests, the wavelength shift in each step slowly decreased, and the falling speed was 1–2 pm each time. K(5)-Pt/MoO
3 showed much lower sensitivity than K(1)-Pt/MoO
3 and K(2)-Pt/MoO
3. Combining the synthesis process, excess intercalated K
+ with a relatively large ionic radius may exfoliate the layer structure and result in lower sensitivity to hydrogen.
In summary, ion intercalated Pt/MoO3 shows better repeatability than pristine Pt/MoO3. The improvement in the sensor’s repeatability is affected by the amount of intercalated ions. The optimal amount of MCl lies in 2 mmol, 1 mmol, and 1 mmol for Li+, Na+, and K+, respectively. When the MCl amount exceeds the optimal value, the repeatability will decrease with the increase of the MCl amount.
Figure 3 shows the cycling test results of Li
+ intercalated Pt/MoO
3. It can be concluded that the growth rate of each fitting curve increases with the increase of hydrogen concentration, indicating that the sensitivity of the sensor becomes larger as the hydrogen concentration increases. According to the hydrogen sensing principle of Pt/MoO
3 [
8], when the hydrogen concentration is low, the hydrogen sensing reaction rate is controlled by the diffusion process of hydrogen. A lower hydrogen concentration results in a smaller hydrogen concentration gradient inside and outside of the sensitive material, which means a lower reaction rate [
11]. The heat exchange between the hydrogen sensitive material and the FBG is quickly balanced to stabilize the FBG’s temperature at a relatively low value. Therefore, at low concentrations, the sensitivity of the sensor is low; when the hydrogen concentration is high, the reaction is controlled by the catalytic rate. The higher hydrogen concentration gradient allows hydrogen to be quickly adsorbed onto the surface of the hydrogen sensitive material. The Mo
5+ generated by the hydrogen reduction process made MoO
3 have a higher affinity to hydrogen, which accelerated the adsorption of hydrogen by sensitive materials. In addition, the heat generated by the reaction increases the catalytic activity of the catalyst, these two factors result in high sensitivity in high hydrogen concentration. The dispersion degree of the 50 fitted lines reflects the repeatability of the sensors. The concentrated lines of Li(2)-Pt/MoO
3 indicate the best repeatability and Li(5)-Pt/MoO
3 exhibits the worst repeatability. The error analyses in
Figure 3d provide more accurate information about the sensor’s repeatability. It can be concluded that the relative errors of Li(1)-Pt/MoO
3, Li(2)-Pt/MoO
3, and Li(5)-Pt/MoO
3 are ±5.8%, ±3.0%, and ±10.5%, respectively. Additionally, the fitting curve of Li(5)-Pt/MoO
3 in
Figure 3d is always under those of Li(1)-Pt/MoO
3 and Li(2)-Pt/MoO
3, which means the sensitivity of Li(5)-Pt/MoO
3 is lower than Li(1)-Pt/MoO
3 and Li(2)-Pt/MoO
3.
Figure 4 illustrates the cycling test results of K
+ intercalated Pt/MoO
3, the relative errors of K(1)-Pt/MoO
3, K(2)-Pt/MoO
3, and K(5)-Pt/MoO
3 are ±5.5%, ±9.5%, and ±10.9%, respectively. With the increasing amount of KCl, the repeatability and sensitivity of the sensors decrease, which is similar to Na
+ intercalated samples. Based on the testing results and our previous study [
11], the error analyses results of Pt/MoO
3 and ion intercalated samples when the precursor amount is 1 mmol are shown in
Figure 5. It is found that the pristine Pt/MoO
3 exhibited the worst repeatability with a relative error of ±18%, the repeatability of the sensor can be significantly improved by ion intercalation. When the amount of MCl lies in 1 mmol, Na
+ intercalation improves the sensor’s repeatability the most, Li
+ ranges the second and K
+ has the lowest improvement in repeatability. This may be ascribed to the binding force between the layer structure.
Figure 5b illustrates the proposed mechanism of improving performance by ion intercalation [
14]. Three kinds of oxygen atoms on α-MoO
3 can be named as unshared oxygen, edge-sharing oxygen, and corner-sharing oxygen. The unshared oxygen atoms with non-bonding electrons form only one covalent bond, thus they will bond with the intercalated ions. The ions were intercalated into the double-layer structure of α-MoO
3 by a hydrothermal method, some of the intercalated ions will bond with the unshared oxygen atoms, transforming the van der Waals force into ionic bonds [
11], which is much stronger to maintain the layer structure during the reaction with hydrogen. However, the layer structure’s stability depends on the amount of M-O (M: Li, Na, and K) bond and the bonding force of the M-O bond. Ranked in order of the binding force, the Li-O bond is stronger than Na-O and K-O ranks at the bottom among them. So according to this assumption, the repeatability of the samples should be ranked as: Li(1)-Pt/MoO
3 > Na(1)-Pt/MoO
3 > K(1)-Pt/MoO
3 > Pt/MoO
3, which is inconsistent with the experiment. It should be mentioned that the amount of ionic bonds formed in the layer structure is not equal to the MCl amount in the precursor. The actual ion amount in the layer structure is dominated by the reaction conditions and the species of reactants. The element contents of the ion intercalated samples were analyzed by ICP and the results are shown in
Table 2.
It can be concluded that the actual ion amount in the intercalated samples varies from ion species, the molar ratio of M/Mo of Li(1)-Pt/MoO3, Na(1)-Pt/MoO3, and K(1)-Pt/MoO3. The amount of Li+ intercalated in the sample is much lower than Na+. Although the Li–O bond is stronger than the Na-O bond, the lower concentration results in fewer ionic bonds. Combining these two factors, the binding force between the layer structure of Li(1)-Pt/MoO3 is lower than that of Na(1)-Pt/MoO3. The relative low stability of the Li(1)-Pt/MoO3’s layer structure is easier to collapse during the reaction with hydrogen and results in lower repeatability, which has been confirmed by the experiments. As for the K+ intercalated samples, the relatively large ionic radius and high concentration may expand the layer distance and results in lower stability of the layer structure compared to the other samples.
To investigate the structure change during the ion intercalated process, XRD patterns of the samples were collected and the results are shown in
Figure 6.
Figure 6a exhibits the XRD patterns of Pt/MoO
3 after cycling, Pt/MoO
3, Li(1)-Pt/MoO
3, Na(1)-Pt/MoO
3, and K(1)-Pt/MoO
3. The diffraction peaks of the XRD pattern for all of the samples can be indexed to be orthorhombic with lattice constants of a=3.962 Å, b=13.85 Å, c=3.697 Å (JCPDSNo.05-0508) and the cubic phase of Pt (JCPDS No.00-004-0802), which means the intercalation process did not change the major phase of MoO
3. However, the layer structure diffraction peaks corresponding to the (020) plane varies from each sample. The (020) diffraction peak of Pt/MoO
3, Li(1)-Pt/MoO
3, Na(1)-Pt/MoO
3, and K(1)-Pt/MoO
3 are 12.94°, 12.84°, 12.69°, and 12.39°, respectively. For the sample Pt/MoO
3, diffraction peak at 12.94° disappeared after the cycling test, which corresponds to the (020) plane. It indicates that the layer structure collapses after the cycling test. The layer structure diffraction peaks show a small shift towards low degree with the increasing of ion radius, which means the distance between the interlayer was expanded after ion intercalation. Combining with the ICP results, high concentration of intercalated K
+ did not result in a stronger structure stability than the Li
+ and Na
+ intercalated samples. In contrast, the higher concentration of K
+ with a large radius expanded the interlayer distance. Additionally, the relatively weak K–O bond didn’t provide enough binding force with the interlayer compared to the other samples, resulting in lower repeatability of the sensor.
In addition, it is found in the experiments that when the amount of MCl exceeds 5 mmol, the average wavelength shifts decreased sharply whatever ion species [
11]. The average wavelength shifts of Li(5)-Pt/MoO
3, Na(5)-Pt/MoO
3, and K(5)-Pt/MoO
3 at a hydrogen concentration of 15,000 ppm are 325, 334 and 219 pm. K(5)-Pt/MoO
3 exhibited a much lower sensitivity than Li(5)-Pt/MoO
3 and Na(5)-Pt/MoO
3. Combining with the analyses above, K
+ with a large radius will expand the layer structure and may change the crystalline structure of MoO
3. XRD analyses were conducted to investigate the potential structure change of MoO
3 and the results are shown in
Figure 6b. XRD patterns of Li(5)-Pt/MoO
3 and Na(5)-Pt/MoO
3 are almost the same and they show typical diffraction peaks of α-MoO
3, which means the ion intercalation of Li
+ and Na
+ did not change the major phase of MoO
3. However, K(5)-Pt/MoO
3 exhibited different diffraction peaks compared to the others, it can be concluded that the sample mainly consists of a mixture of MoO
2 and MoO
3, which means some of MoO
3 in the sample was reduced to MoO
2, which is similar to the previous work [
14]. As the affinity to hydrogen of MoO
2 is much lower than MoO
3, the sensor exhibited extremely lower sensitivity.