3.1. Analysis of the Coating Process: Ultrasonic Parameters
The analysis of the coating performance is conducted by the interpretation of the ultrasonic parameters, the attenuation and the frequency shift. Both parameters are used as a practical approach to define whether the coated sensor will meet the requirements to work properly in the sensor system. The frequency shift, which is directly related to the mass deposited and the thickness of the deposited layer, provides information about the mass loaded upon the sensor surface. The attenuation provides information about the material distribution and homogeneity and about the viscoelastic properties of the coating material. Therefore, the interpretation of these integral parameters is very important for the analysis of SAW coating materials.
In the first part of this work [
44], based on the interpretation of the ultrasonic analysis and on the results of the CAT, some conclusions were drawn about the enhancements in the properties of the coating layers achieved by the PU polymer composites.
Scheme 1 represents a pictured visualization of the coating process to support the suppositions made based on the obtained ultrasonic results. In the same way,
Scheme 2 shows the situations before and after the CAT suggested to support the interpretation of the CAT effect over the sensing layers.
The results of the ultrasonic parameters obtained for the deposition of the PU polymer composites of each tested sensing polymer (PLMA, PIB, and PCTFE) before and after the application of the CAT are now presented and discussed.
3.1.1. PU-PLMA Composites: Analysis of the Ultrasonic Parameters
Figure 1 presents the frequency shift, before and after the CAT, obtained for the PU polymer composites with PLMA.
The results are strongly related to those obtained with the PU-PBMA composites [
44]. As observed with PBMA, after the CAT, the coating layer with the pristine PLMA was almost totally removed. Contrastingly, the presence of the PU in the composites promoted an increase in the mass deposited compared with the mass deposited with the pristine polymer, indicating that a cooperative effect occurs between PU and PLMA in the process of the composite deposition.
After the CAT, a substantial quantity of the mass deposited was removed from the surface of all the PU-PLMA composites, maintaining, however, a considerable fraction of the frequency shift originally obtained. In the pristine PU deposition, a small reduction in the frequency shift after the CAT was observed. All these results are in accordance with the previous results for the PU-PBMA composites [
44].
The quantitative relationship between the PLMA concentration in the coating formulations and the frequency shift obtained by the deposition of the coating materials (
Table 1) and after the CAT are shown in
Figure 2.
The curves of
Figure 2 reproduced the profiles obtained for the same relationship in the case of PU-PBMA composites [
44], which is a strong indication of the reproducibility and robustness of the method. The exact and reproducible quantitative relationship between the frequency shift due to the coating layer of the composites and the concentration of the PLMA in the coating solution, hereby indirectly represented by the volume of the sensing polymer solution added to the coating solution, and agreement with the results observed for the PU-PBMA composites suggest that the same mechanism is involved in the formation of PU polymer composites.
The results for the attenuation measured for the PU-PLMA coating materials (
Table 1) are presented in
Figure 3.
The results of the attenuation for the pristine PLMA and PU (
Figure 3) are in accordance with those observed for the frequency shift (
Figure 1) of both pristine polymers, which indicates almost complete removal of the PLMA coating, while the PU-coated sensor maintained its coating layer almost unchanged after the CAT. These observations, again, agree with those observed in the analysis of the PU-PBMA composites [
44].
For the PU-PLMA composites, the results of attenuation for the coating process in
Figure 3 show a regular increase in their values, which can be assigned to the successive increase in the mass deposited over the surface of the sensor elements (
Figure 1). The attenuation results after the CAT for the PU-PLMA composites, however, present a tendency toward the uniformization of their values. This observation suggests overall uniformity in terms of the coating layer structure as well as in terms of the material distribution over the sensor element surface.
Figure 4 presents the quantitative correlations of the attenuation results for the PU-PLMA composites, before and after the CAT.
A clear quantitative relationship between the attenuation and the PLMA concentration for the coating of the PU-PLMA composites is observed (
Figure 4). The results of the quantitative correlation between the attenuation and the concentration of the PLMA in the composites confirm the previous statements about the homogeneity in terms of the structure and material distribution, reflected by the regular increase in the attenuation before the CAT, which is correlated with the increase in the mass (frequency shift) of the coating material deposited (
Figure 1).
The precise quantitative correlation of the attenuation observed after the CAT (
Figure 4) confirms the tendency toward alike structures after immersion in the organic solvent. This can be inferred by the similarity of the attenuation values obtained for the PU-PLMA composites after the CAT (
Figure 4), strongly suggesting a high similarity of the properties of the remaining PU-PLMA composite coating layers. A possible interpretation for this observation is that after the removal of mass from the PU-PLMA coating layers by the CAT (
Figure 1), the remaining coating layers tend toward more similar structures, presenting, therefore, more similar attenuation values. This tendency can be seen in the profile of the attenuation after the CAT, presented in
Figure 4.
The same behavior for the attenuation before and after the CAT was observed for the PU-PBMA coating materials [
44]. Based on the present results and considering the PU as the one responsible for the adhesion, chemical resistance, and support of the remaining structure of the coating layer after the application of the CAT, it could be expected that the structures of the remaining coating layers become more alike after the CAT. After the removal of the excess of the nonbonded sensing polymer, the “CAT-resistant” PU-PLMA composite structure will remain as the coating layer, and its structure will be very similar.
The comparison of the results for the PU-PBMA and PU-PLMA composites as coating materials for the SAW sensor technology provided an interesting scenario because while the two sensing polymers are made of the same backbone, they do differ by their side radicals (butyl and lauryl, respectively). This difference in the structure of the sensing polymers did not present any distinction in the behavior of the ultrasonic parameters of the coating process of the SAW sensor elements. The results before and after the CAT presented the same behavior for both sensing polymers, including the obtention of very similar profiles for the quantitative relationship of the ultrasonic parameters in terms of the concentration of the sensing polymer in the coating solution. In future work, the absolute effect of the structure of the sensing polymers in the coating materials’ properties will be addressed.
3.1.2. PU-PIB Composites: Analysis of the Ultrasonic Parameters
PIB has a completely different chemical structure from the previously analyzed polyacrylate polymers. PIB has a fully aliphatic hydrocarbon structure and the absence of any organic functional group, heteroatoms, or aromatic radicals in its structure accord an apolar character to this polymer.
Figure 5 shows the results of the frequency shift for the PU-PIB coating materials (
Table 2), before and after the CAT.
The result for the pristine PU was the same as the one observed in the previous experiments where the polymer showed a deposition of approximately to 0.5 MHz of frequency shift and a little loss of this value after the CAT. For the pristine PIB, a considerable lowering of the frequency shift was observed compared with that observed for the PLMA (
Figure 1). This result may be explained by two reasons: first, the lower mass of the PIB in the coating solutions (
Table 2), and second, the lack of affinity of the polymer to the surface of the SAW sensor element since the lowering in the frequency shift was not proportional to the reduction in concentration compared with the previous polymer analyzed (
Figure 1).
The coating layer of the pristine PIB was almost completely washed out of the surface of the SAW sensor element after the CAT, while the results for the composites went in a completely different direction. For the coating of the PU-PIB composites, the frequency shift results presented similar values to those obtained for the PU-PLMA composites, even though the pristine polymers presented very distinct values of frequency shift (
Figure 1 and
Figure 5) and despite the fact that the PIB concentration in the composites was half that of PLMA (
Table 1 and
Table 2).
These results emphasize the role played by the PU in the formation of the composites. Even though PIB possesses a quite distinct chemical structure, and, therefore, different chemical affinities than the polymetacrylate polymers previously analyzed, its association with PU gives rise to composites that show the same properties as coating materials of the PU composites of the previous sensing polymers.
Figure 1 and
Figure 5 show that the behavior of the frequency shift in the PU composites with both polymers before the CAT is very similar. These results indicate that PU provides, at the same time, the formation of an aggregate with the sensing polymer, enhancing the mass of the sensing polymer available for the interaction with the analytes, and it enhances the adhesion of the composites to the surface even though the sensing polymer presents a lack of affinity and consequent poor adhesion to the surface of the SAW sensor element.
The results after the CAT show a similarity in the behavior of the PU-PLMA and PU-PIB composites (
Figure 1 and
Figure 5). Although the sensing polymers are chemically quite distinct, the similarity observed in their behavior suggests that the PU composites are probably formed by the same mechanism. In turn, this is an indication that the methodology for the formation of PU polymer composites is a general procedure for the sensitization of SAW sensors with polymeric coating layers, independently of the chemical nature of the sensing polymer used.
The quantitative relationships of the frequency shift as a function of the PIB concentration in the coating solution, before and after the CAT, are presented in
Figure 6.
The profiles in
Figure 6 are similar to those obtained for the PU-PLMA composites (
Figure 2). Even the determination coefficients are in great agreement with both results, confirming the robustness of the coating process and the procedures of the CAT. The reproducibility of the profiles reinforces the argument that the formation of the PU polymer composites follows the same mechanism for both polymers, which is remarkable, considering the constitutional differences between PLMA and PIB.
The results after the CAT follow the same profile observed for the PU-PLMA composites (
Figure 6), which is also an indication that the formation of the composites follows the same mechanism, which, in turn, could lead to the formation of composites with similar structures but with quite different compositions, and, consequently, with differences in their properties like their viscoelastic behavior and their interactions with the analytes, for example.
The analysis of the attenuation provides more insights into the properties of the PU-PIB composites. The results of attenuation before and after the CAT are shown in
Figure 7.
The first important observation in
Figure 7 is the attenuation of the pristine PIB coating layer, which presents a similar attenuation to that observed by the pristine PLMA (
Figure 3) but shows a lower frequency shift than that observed with PLMA. In other words, it means that less mass of the PIB deposited leads to a comparable value of attenuation to that obtained with a higher mass of PLMA.
This observation should be interpreted as a difference in the intrinsic properties of the coating materials since a possible increase in the attenuation due to an irregular material distribution by the deposition has not been found so far. The reason is, therefore, in the differences in the constitution of the coating materials, which has a direct influence on the attenuation.
Again, for the pristine PU coating, the results reproduced the behavior observed in all the previous experiments. The results for the PU-PIB composites (
Figure 7) reproduced the general behavior observed with the other PU polymer composites analyzed when the increase in the concentration of the sensing polymer in the coating solution promoted an increase in the frequency shift, which, in turn, manifested itself as an increase in the attenuation values.
After the CAT, the results for the PU-PIB composites presented the same tendency toward uniformization in terms of the structure of their remaining coating layer, as previously observed for PLMA (
Figure 3). The reproduction of the relative behavior observed for the PU composites containing PLMA and PIB reinforces the argument that the same mechanism is involved in the formation of the composites and in its deposition on the surface of the SAW sensor elements.
Figure 8 shows the quantitative relationship between the attenuation and the PIB concentration of the coating solution.
The attenuation measured for the coating layers obtained with the PU-PIB composites also shows an exact quantitative relationship with the PIB concentration in the coating solution before the CAT; however, the profile of the correlation was different than that observed for PLMA (
Figure 4). The difference in the profiles arises from the differences in the constitution of the sensing polymers in the coating materials. As PIB has a quite different chemical constitution than that of the polyacrylate sensing polymers previously analyzed, the results of the attenuation reflect the differences in the properties between the sensing materials of each sensing polymer by producing different profiles for the attenuation for the respective PU composites as a function of the sensing polymer concentration.
The profiles obtained for the attenuation after the CAT for the PU-PIB composites are more alike to those observed for the PU-PLMA coating materials and both coating materials show the tendency toward uniformization of the structure of their remaining coating layers (
Figure 4).
3.1.3. PU-PCTFE Composites: Analysis of the Ultrasonic Parameters
PCTFE-Poly(chlorotrifluoroethylene-co-vinylidene fluoride)—is a highly halogenated polymer with the general formula [CF
2CF(Cl)]
x(CH
2CF
2)
y. Considering its constitution, it is important to evaluate the behavior of its combinations with PU as coating materials. The homogeneity of the structures of the resulting PU-PCTFE composites (
Table 3) and the uniformity of the material distribution by the coating process as well as the adhesion of the resulting coating layers to the highly polished quartz surface of the SAW piezoelectric element, with its interdigitated electrode structure, together with the sensor responses of the coating materials of the PU composites of this polymer, are now analyzed to confirm the general application of the methodology.
Figure 9 presents the frequency shifts achieved by the deposition of the pristine polymers (PCTFE and PU) and the PU-PCTFE composites (
Table 3), before and after the CAT.
The result of the coating with pristine PU can be taken as the control of the coating process methodology and gives the same results before and after the CAT (
Figure 9), as observed for all the other results previously analyzed, validating the coating process in the experiments with PCTFE.
The results of the frequency shift due to the coating (corresponding to the results before the CAT) show consistently higher values either for the pristine PCTFE or for its PU composites, in comparison with all the other PU polymer composites previously analyzed (
Figure 1 and
Figure 5). Considering that the conditions of the coating procedure were the same for all the experiments, a possible explanation for increasing the mass observed by the coating with PCTFE and its corresponding PU composites could be a very favorable interaction between these materials and the surface of the sensor element, which would cause more mass to be retained on the surface after the spin coating process. This hypothesis will be discussed along with the further results.
Another reason could be due to a higher density of the PCTFE and, consequently, of its PU composites, leading to a higher mass deposited for the same volume of the coating layer. Indeed, the density of the PCTFE is approximately double that of the other polymers analyzed, which probably accounts for the significantly higher results of frequency shift observed for all the coating materials containing PCTFE from
Table 3.
The results for the frequency shift after the CAT for the PCTFE and its PU composites were also distinct compared with those observed in the previous experiments (
Figure 1 and
Figure 5). The pristine PCTFE preserved almost all its original value of frequency shift due to the coating, a quite distinct result in comparison with all the other polymers and PU composites, except for those of pristine PU. All the coatings with the PU-PCTFE composites also showed a significantly lower reduction in the frequency shift after the CAT than those observed in the PBMA, PLMA, and PIB experiments.
These observations can be visualized in
Figure 10 where the frequency shift values before and after the CAT are plotted as a function of the volume of the polymer solution in coating solutions.
The profiles in
Figure 10 are different than those obtained for PLMA and PIB (
Figure 2 and
Figure 6, respectively) and clearly show a different behavior of the PU-PCTFE composites before and after the CAT.
The results for the pristine PCTFE (
Figure 9) where the difference in the frequency shift before and after the CAT was much smaller than that observed for the other polymers (
Figure 1 and
Figure 5) could support the hypothesis of a stronger interaction of PCTFE with the SAW sensor element surface, as mentioned before, which could lead to less removal of mass from the coating layer by the CAT. The removal of PCTFE from its coating layer is much less than that observed for all the other polymers analyzed, except for PU.
In the same sense, the higher results of the frequency shift for the PU-PCTFE composites after the CAT could be explained by a favorable interaction between the PCTFE and the PU, which, in turn, increases the interaction with the surface of the SAW sensor element when compared with that observed for the pristine PCTFE. This leads the composites to be more stable with respect to the CAT, causing less removal of mass from the coating layers of the PU-PCTFE composite.
The results of the attenuation for the resulting coating layers with the PCTFE-based coating materials are shown in
Figure 11.
The results of the attenuation presented in
Figure 11 reveal a more significant effect of the CAT on the coating with pristine PCTFE, wherethe reduction in the attenuation after the CAT was more significant than that observed in the pristine PU. In general, the reduction in the attenuation suggests that after the CAT, the coating layer tends to show a more uniform structure, which, could be achieved by the removal of polymer units that were loose in the structure of the coating layer. These loose polymer units on the coating structure shall be responsible for a higher energy dispersion of the surface wave oscillation.
The attenuation values of pristine PU after the CAT reproduced the same results as those observed in the previous experiment with the other polymers (
Figure 3 and
Figure 7).
The attenuation results for the PU-PCTFE composites (
Figure 11) presented the same behavior as those for the other polymers, PLMA and PIB (
Figure 3 and
Figure 7) but showed much less influence of the CAT. The attenuation of the PU-PCTFE composites presented the same tendency toward the achievement of uniformity in the coating layers with the increment of the concentration of the PCTFE in the composites, as observed for all the PU polymer composites previously analyzed.
Figure 12 presents the quantitative relationship between the attenuation and the volume added of the polymeric solution to the coating solution for the PU-PCTFE composites, before and after the CAT.
The profiles in
Figure 12 are very exact and show a very clear quantitative correlation between the PCTFE concentration in the composite and the attenuation values. Much less influence of the CAT on the structure of the resulting PU-PCTFE composites can be clearly seen from the results of the attenuation, in this case presenting a very distinct behavior from those observed so far. It is also remarkable that the structure of the coating with pristine PU (corresponding to the zero volume in
Figure 12) is so closely correlated with those of the PU-PCTFE composites. This fact significatively suggests that the structure formed by the PU polymer composites is strongly correlated with that of the pristine PU, supporting the role proposed for PU in the mechanism of the formation of the PU polymer composites [
44].
Although the results in
Figure 12 show an exact correlation between the attenuation and the concentration of the PCTFE, as was also observed in the previously analyzed PU polymer composites (
Figure 4 and
Figure 8), the profiles obtained for the composites of the three sensing polymers analyzed are quite distinct. This fact strongly suggests that even though the mechanism of formation seems to be the same and independent of the chemical nature of the sensing polymers used to form the composites with PU, the resulting structures of each distinct PU polymer combination will be quite different. The application of the methodology provided a reliable production of fully operational SAW sensors, and the analysis of the ultrasonic parameters before and after the CAT was able to characterize and differentiate each type of the PU polymer composites. The results demonstrate that the method is very robust and reproducible, minimizing the errors in the production of the sensors and therefore, reducing the significant cost of the loss of the expensive SAW piezoelectric quartz elements.
3.2. Analysis of the Relative Sensor Responses
The analysis of the relative sensor responses reveals the correlation between the nature of the composite coating materials of PU with varying quantities of the sensing polymer and their interaction with a given analyte from the gas phase. Once the results of the interaction with the analytes are directly correlated with the structure and the chemical composition of the coating layers obtained with the PU polymer coating materials, the relative sensor responses should be, therefore, correlated with the correspondent results of the ultrasonic parameters of the respective coating material. As the analyses of the sensing responses of the PU-polymer composites will be addressed in more detail in the sequence of this work, the analysis of the relative sensor responses will be presented for two analytes, chloroform and p-xylene, for the characterization of the PU polymer coating materials used in this work regarding their relative SAW sensor responses.
Figure 13 presents the results of the sensor responses to chloroform as the analyte for the PU-PLMA coating materials, before and after the CAT.
The results of the relative sensor responses to the chloroform as the analyte in
Figure 13 are in accordance with those of the frequency shift results for the PU-PLMA composites (
Figure 1), reproducing the same behavior before and after the CAT. The higher values of the relative sensor responses obtained before the CAT indicate an increasing capacity of sorption by the composite coating materials with the increment of the sensing polymer in the composite, in agreement with the increase in the frequency shift observed (
Figure 1). This fact confirms the role of the sensing polymer in the PU polymer composite coating materials as the one responsible for the responses of the SAW sensors and the sorption capacity of the analytes from the gas phase.
The relative sensor responses of the pristine PLMA as well as of the coating layers of the PU-PLMA composite coating materials originally deposited were affected by the CAT, showing the same behavior as that observed for the frequency shift (
Figure 1). The pattern showed by the relative sensor responses to chloroform after the CAT (
Figure 13) also suggest the same tendency toward the uniformity of the layer structure observed after the CAT with the PU-PLMA composites in the ultrasonic results (
Figure 1,
Figure 2,
Figure 3 and
Figure 4).
These statements can be visualized in the graphic of the relative sensor responses as a function of the PLMA concentration in the coating solution (
Figure 14), expressed by the volume of its polymer solution.
Very exact quantitative relationships regarding the content of the sensing polymer in the composites were also obtained for the relative sensor responses to chloroform, before and after the CAT (
Figure 14). These correlations are in perfect agreement with the quantitative profiles obtained for the frequency shift and the attenuation for the PU-PLMA composites (
Figure 2 and
Figure 4). These observations strongly suggest that the sensing mechanism is strongly related to the ultrasonic parameters. The same overall behavior of the ultrasonic parameters and their correlation to the relative sensor responses observed for the PU-PLMA coating materials were observed for the coating materials of PU-PBMA [
44], suggesting that the same mechanism for the formation of the PU-polymer composite coating layers is involved, resulting in coating layers with similar structures generating similar profiles for the interaction with the analytes.
The relative sensor responses for the PU-PIB composites before and after the CAT are shown in
Figure 15.
Here again, the same behavior as that for the PU-PLMA composites was observed. The relative sensor responses to chloroform as the analyte for the PU-PIB coating materials also agree with the behavior of its ultrasonic parameters (
Figure 5 and
Figure 7).
Figure 16 presents the quantitative relationship between the relative sensor responses to chloroform as the analyte as a function of the concentration of the sensing polymer in the PU-PIB composites.
The exact quantitative relationship obtained for the relative sensor responses of the PU-PIB coating materials (
Figure 16) reproduces almost exactly the profile obtained for the frequency shift, and all the observations made for the PU-PLMA coating materials can be applied to the case of the PU-PIB coating materials. The results for both polymer composites agree with the ultrasonic parameters, and in both cases, the behavior was very alike, strongly suggesting that the formation of the PU-PIB, PU-PLMA and PU-PBMA composites follow the same mechanism as that postulated for the formation of the PU-PBMA composites [
44].
Figure 17 shows the relative sensor responses to chloroform as the analyte for the PU-PCTFE composites, before and after the CAT.
The relative sensor responses of the PU-PCTFE composites to chloroform follow the same pattern as the results for the corresponding frequency shifts measured for these coating materials (
Figure 9), reproducing once more the same behavior observed for the previous two PU polymer composites. The results of the relative sensor responses before and after the CAT agree with the frequency shift results, confirming the argument about the resulting composite structures and the respective resulting effect of the CAT on the original deposited sensing layers for this type of composites.
Additionally, the analysis of the quantitative relationship observed between the relative sensor responses to chloroform and the concentration of the sensing polymer in the composite presented in
Figure 18 show very exact correlations before and after the CAT.
The exactness of the correlations and their agreement with the frequency shift patterns indicate that the formation of the sensing layers of the composites follows the same mechanism, and their structures are correlated with the concentration of the sensing polymer in the composites. The same was observed for all the composites analyzed so far, for all the families of composites, even though they are made of chemically distinct sensing polymers in combination with PU.
The relative sensor responses to chloroform as the analyte were very high for all sensing layers of the three PU polymer composites analyzed, which denotes a very favorable and intense interaction.
The next analysis of relative sensor responses of the coating materials uses p-xylene as the analyte.
Figure 19 presents the relative sensor response patterns for the PU-PLMA coating materials to p-xylene as the analyte, before and after the CAT.
Despite the fact that the relative sensor responses were lower for p-xylene than those obtained with chloroform (
Figure 13), the same patterns for the relative sensor responses were observed for p-xylene as those obtained for the frequency shift of the PU-PLMA composites (
Figure 1) and for the relative sensor response to the chloroform as the analyte (
Figure 13), once again indicating the correspondence between the structures of the composites and the frequency shift parameter and their correlation with the relative sensor responses.
Consistently, the same profiles for the quantitative relationship between the relative sensor responses and the sensing polymer concentration in the composite were observed (
Figure 20). These profiles were in perfect agreement with the quantitative profile obtained for the relationship between the frequency shift and the concentration of PLMA in the composites with PU (
Figure 2).
Figure 21 shows the results for the relative sensor responses of the PU-PIB composites as the coating materials.
The relative sensor responses for the p-xylene as the analyte were much lower than those measured for chloroform with the coating of the PU-PIB composites (
Figure 15). The patterns of the relative sensor responses for the p-xylene for the PU-PIB composites agree with the patterns obtained for the frequency shift results for the PU-PIB composites (
Figure 5).
Exact quantitative profiles for the relative sensor responses as a function of the PIB concentration in the composites were consistently obtained for the results before and after the CAT (
Figure 22).
The results in
Figure 22 reproduce the same profiles obtained for the quantitative relationship of the frequency shift with the PIB concentration in the PU-PIB composites (
Figure 6), as observed for all the PU polymer composites investigated, strongly indicating that the process of formation of the PU-PIB composites follows the same mechanism observed for all the PU polymer composites previously analyzed.
Figure 23 presents the relative sensor responses to p-xylene obtained with the PU-PCTFE composites as coating materials.
For p-xylene, the relative responses of the PU-PCTFE composite sensors (
Figure 23) reproduce the same behavior observed for the results for the frequency shift for the PU-PCTFE composites (
Figure 9), following the same behavior observed for the previous PU polymer composites.
The quantitative profiles, before and after the CAT, for the relative sensor responses as a function of the PCTFE concentration on the composites are presented in
Figure 24.
In the same way, the quantitative profile correlating the relative sensor responses and the concentration of PCTFE in the composites (
Figure 24) resembles the quantitative profiles obtained for the frequency shift for the PU-PCTFE composite coating before and after the CAT (
Figure 10).