*3.1.2 Hydrogen bonding*

*Solvents, Ionic Liquids and Solvent Effects*

region (at the left of −1.0 e/nm<sup>2</sup>

−1.0 and 1.0 e/nm2

1.0 e/nm2

*3.1.1 Interpretation of sigma profile graphs*

negative area and acts as an H-bond acceptor.

cation merely contributes very slightly in the process [57].

A sigma profile graph in COSMO-RS allows us to understand certain aspects of an IL-water system. The main information we can obtain from the graph is to learn about the hydrophobicity of IL and the tendency of IL to act as a hydrogen bond donor or hydrogen bond acceptor. According to Klamt [5, 58], the sigma profile graph can be divided into three regions. The first region is the hydrogen bond donor

), the second region is nonpolar region (between

), and the thirdly region is the acceptor region (at the right of

). By judging at which region the peak of an IL locates, the tendency of IL

to act as hydrogen bond donor or acceptor would be identified. Generally, a peak that locates at the right side of the sigma profile graph indicates the more electro-

Now, **Figure 4** shows the sigma profile graph of EMIM-Cl, BMIM-Br, and water molecules. Looking at the sigma profile of water molecules as shown in **Figure 4**, it is observed that water has two high peaks, one in the hydrogen bond donor region and another in the acceptor region [39]. This indicates that water has a high affinity toward both acceptor and donor. Furthermore, **Figure 4** shows the sigma profile of two ILs, which are EMIM-Cl and BMIM-Br. From the figure, it is observed that cations EMIM and BMIM both have their peak in the nonpolar region. However, water molecules which have peaks in the polar region tend to have higher affinity only with strong hydrogen bond donor or acceptor, but not cation that lays its peak in the nonpolar region [59]. As a result, cations do not interact much with water molecules. Meanwhile, anions that have their peaks in hydrogen bonding acceptor region are more attractive to water molecules. Hence, this inferred that anion is the main ion that interacts with water molecules to prevent hydrate formation, whereas

Moreover, we can see that EMIM, which has a shorter alkyl chain length, has its peak nearer to the polar region than BMIM. As consequences, EMIM is also more polarized and hydrophilic than BMIM, which is a desired characteristic of a good hydrate inhibitor. This also proves that a cation with shorter alkyl chain length is preferable during the tuning of IL inhibitor, as a shorter cation is less bulky and hence can more effectively interact with water molecules [49, 60]. For anion, Cl<sup>−</sup> proves itself to be a better H-bond acceptor as it has a peak at the right side of the graph, which is the indication of its further electronegative. This at the same time means that Cl<sup>−</sup> will be more effective in accepting H-bond from water molecules than Br<sup>−</sup>. Therefore, this makes Cl<sup>−</sup> more hydrophilic and serves as a better anion for hydrate inhibitor.

**150**

**Figure 4.**

*Sigma profile graph of EMIM-Cl and BMIM-Br.*

Although hydrogen bonding strength has been widely quoted to have a relationship with the effectiveness of IL as hydrate inhibitor [29, 30], so far, no work has been conducted to prove this relationship. In this work, validation is done and has successfully proven that a linear relationship exists between hydrogen bonding strength and the effectiveness of IL as hydrate. This linearity is validated through four different sets of data that comes from three papers [25, 30, 61]. All four sets of data show good linearity relationship, with the highest regression value as *R*<sup>2</sup> = 1 and the lowest as *R*<sup>2</sup> = 0.8926. As a result, this implies that the prediction of IL effectiveness could be made through the comparison of hydrogen bonding strength.

Besides proving this relationship, several interesting findings have also been observed throughout the process. Firstly, computation of COSMO-RS, in total, will calculate three kinds of energy value for an IL, namely, misfit energy (*E*MF), hydrogen bonding energy (*E*HB), and van der Waals energy (*E*vdW). The summation of these three energies leads to the value of total interaction energy (*E*int). Although hydrogen bonding strength is known to affect the effectiveness of IL, the significance of other energies could not be neglected yet. Hence, in **Figure 8**, all types of predicted energies including *E*MF, *E*vdW, *E*HB, and *E*INT are plotted against average depression temperature to determine if these energies could also affect the effectiveness of ILs as hydrate inhibitor.

**Figure 5** demonstrates that for ILs with BMIM cation, it is evidently shown the anion contributes more to the total interaction energy than the cation. The reason behind this is virtually consistent; van der Waals energies are nearly constant for all of the tested ILs and have thus no effect on the temperature depression. The contribution of misfit energy, having only a regression value of 0.2247, is also negligible. This leaves the hydrogen bonding energy to be the only energy that plays an essential role in affecting the effectiveness of BMIM-ILs. Furthermore, the relationship between total interaction energy (*E*INT) and temperature depression is also not convincing. This graph hence supports the earlier statement that hydrogen

**Figure 5.** *Average temperature depression from Sabil et al. [25] work vs. types of predicted energy (binary components).*

**Figure 6.**

*Average temperature depression against predicted hydrogen bonding energy for both EMIM- and BMIM-based ILs (binary components).*

bonding strength between cation and anion is the most important type of energy that regulates IL interaction with water molecules [29, 30, 62]. The same pattern of relationship is then also observed in another two data sets from the work of Xiao et al. [30]. **Figure 6** now shows the relationship between average temperature depression of ILs and the predicted hydrogen bonding energy from COSMO-RS.

Clearly, the graph shows that the temperature depression value of IL-hydrate system is directly proportional to the hydrogen bonding energy (*E*HB) for both EMIM-based and BMIM-based ILs. The larger the absolute value of *E*HB, the higher the temperature depression of a hydrate system. For instance, for BMIM-based ILs in this graph, the rank of *E*HB from highest to lowest is as BMIM-Cl > BMIM-Br > B MIM-I > BMIM-BF4. The same ranking occurred to the average temperature depression as well, where BMIM-Cl has the highest temperature depression and BMIM-BF4 has the lowest depression. This ranking could be explained by the fact that among four anions, Cl<sup>−</sup> anion has the highest polarized charge and thus acts as the best hydrogen bond acceptor. BF4 <sup>−</sup> anion, on the other hand, has the lowest polarized charge after Br<sup>−</sup> and I<sup>−</sup> anion and thus shows the lowest hydrogen bond strength because it is the weakest hydrogen bond acceptor among all. This graph, however, also displays an interesting finding, which is the separation of EMIM- and BMIMbased ILs into two different data sets, instead of one. This step is necessary as the combination of all ILs into one data set may lower the linearity of relationship. This statement is supported by **Figure 10**, which shows a graph of average temperature depression against predicted hydrogen bonding energy.

**Figure 7** inferred that linear relationship only exists when ILs with the same cation are compared. An early deduction is that to ensure a linear relationship for a set of data, only one single ion, which is either cation or anion, can vary, while another one must be fixed. The relationship could not be applied to predict ILs with different cations and anions. This deduction is supported by **Figure 8**, which shows the regression value between average depression temperature and hydrogen bonding strength for a set of ILs with different cations but same Cl<sup>−</sup> anion.

With the regression value as high as 0.8976 from **Figure 8**, this supports our deduction earlier, where one ion must be fixed and another one could be varied to see the relationship. Furthermore, it is noticeable that when ILs with fixed anion but different cations are measured, the relationship between hydrogen bonding strength and average depression temperature is inversely proportional as before. The higher the absolute value of hydrogen bonding energy, the lower the average temperature depression. This could be explained by the sigma potential graph that has been

**153**

**Figure 8.**

**Figure 7.**

*different cations (binary components).*

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application…*

discussed earlier. Previously, it is explained that cations generally have their peaks located in the nonpolar region of sigma profile, which is from −1 to 1 e/nm2

*Average temperature depression against predicted hydrogen bonding energy for ILs with Cl- as anion but* 

*Average temperature depression against predicted hydrogen bonding energy for a single data set consisting of* 

On the other hand, water molecules show two high peaks, one at the region of hydrogen bond donor and another at hydrogen bond acceptor. As a result, water molecules tend to have higher affinity only with strong hydrogen bond donor or acceptor, but not cation that lays its peak in the nonpolar region. Hence, this inferred that anion is the main ion that interacts with water molecules to prevent hydrate formation, whereas cation merely contributes very slightly in the process [57]. Since cations have a low affinity with water molecules, this also indicates that most of the cations in water will continue to bond with anions. In that case, the excess hydrogen bonding energy provided by stronger cation (that has higher *E*HB) is unnecessary. This stronger hydrogen bonding energy will be used by cation to bond with anion, thus reduces the number of anions that are free to interact with water molecules. As a consequence, it will bring about an inverse effect on average temperature depression and reduce the effectiveness of ILs as a hydrate inhibitor. In short, linear relationship does exist between hydrogen bonding strength and the thermodynamic hydrate inhibition ability of an IL. For a set of ILs with fixed cation and

.

*DOI: http://dx.doi.org/10.5772/intechopen.86847*

*both EMIM- and BMIM-based ILs (binary components).*

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application… DOI: http://dx.doi.org/10.5772/intechopen.86847*

#### **Figure 7.**

*Solvents, Ionic Liquids and Solvent Effects*

hydrogen bond acceptor. BF4

**Figure 6.**

*ILs (binary components).*

bonding strength between cation and anion is the most important type of energy that regulates IL interaction with water molecules [29, 30, 62]. The same pattern of relationship is then also observed in another two data sets from the work of Xiao et al. [30]. **Figure 6** now shows the relationship between average temperature depression of ILs and the predicted hydrogen bonding energy from COSMO-RS. Clearly, the graph shows that the temperature depression value of IL-hydrate system is directly proportional to the hydrogen bonding energy (*E*HB) for both EMIM-based and BMIM-based ILs. The larger the absolute value of *E*HB, the higher the temperature depression of a hydrate system. For instance, for BMIM-based ILs in this graph, the rank of *E*HB from highest to lowest is as BMIM-Cl > BMIM-Br > B MIM-I > BMIM-BF4. The same ranking occurred to the average temperature depression as well, where BMIM-Cl has the highest temperature depression and BMIM-BF4 has the lowest depression. This ranking could be explained by the fact that among four anions, Cl<sup>−</sup> anion has the highest polarized charge and thus acts as the best

*Average temperature depression against predicted hydrogen bonding energy for both EMIM- and BMIM-based* 

charge after Br<sup>−</sup> and I<sup>−</sup> anion and thus shows the lowest hydrogen bond strength because it is the weakest hydrogen bond acceptor among all. This graph, however, also displays an interesting finding, which is the separation of EMIM- and BMIMbased ILs into two different data sets, instead of one. This step is necessary as the combination of all ILs into one data set may lower the linearity of relationship. This statement is supported by **Figure 10**, which shows a graph of average temperature

**Figure 7** inferred that linear relationship only exists when ILs with the same cation are compared. An early deduction is that to ensure a linear relationship for a set of data, only one single ion, which is either cation or anion, can vary, while another one must be fixed. The relationship could not be applied to predict ILs with different cations and anions. This deduction is supported by **Figure 8**, which shows the regression value between average depression temperature and hydrogen bond-

With the regression value as high as 0.8976 from **Figure 8**, this supports our deduction earlier, where one ion must be fixed and another one could be varied to see the relationship. Furthermore, it is noticeable that when ILs with fixed anion but different cations are measured, the relationship between hydrogen bonding strength and average depression temperature is inversely proportional as before. The higher the absolute value of hydrogen bonding energy, the lower the average temperature depression. This could be explained by the sigma potential graph that has been

ing strength for a set of ILs with different cations but same Cl<sup>−</sup> anion.

depression against predicted hydrogen bonding energy.

<sup>−</sup> anion, on the other hand, has the lowest polarized

**152**

*Average temperature depression against predicted hydrogen bonding energy for a single data set consisting of both EMIM- and BMIM-based ILs (binary components).*

#### **Figure 8.**

*Average temperature depression against predicted hydrogen bonding energy for ILs with Cl- as anion but different cations (binary components).*

discussed earlier. Previously, it is explained that cations generally have their peaks located in the nonpolar region of sigma profile, which is from −1 to 1 e/nm2 .

On the other hand, water molecules show two high peaks, one at the region of hydrogen bond donor and another at hydrogen bond acceptor. As a result, water molecules tend to have higher affinity only with strong hydrogen bond donor or acceptor, but not cation that lays its peak in the nonpolar region. Hence, this inferred that anion is the main ion that interacts with water molecules to prevent hydrate formation, whereas cation merely contributes very slightly in the process [57]. Since cations have a low affinity with water molecules, this also indicates that most of the cations in water will continue to bond with anions. In that case, the excess hydrogen bonding energy provided by stronger cation (that has higher *E*HB) is unnecessary. This stronger hydrogen bonding energy will be used by cation to bond with anion, thus reduces the number of anions that are free to interact with water molecules. As a consequence, it will bring about an inverse effect on average temperature depression and reduce the effectiveness of ILs as a hydrate inhibitor.

In short, linear relationship does exist between hydrogen bonding strength and the thermodynamic hydrate inhibition ability of an IL. For a set of ILs with fixed cation and

**Figure 9.**

*Average temperature depression from Sabil et al. [25] work vs. types of predicted energy (quaternary components).*

different anions, stronger hydrogen bonding between ILs lead to higher average depression temperature of an IL-hydrate system. Vice versa, for a set of ILs with fixed anion but different cations, stronger *E*HB produces lower depression temperature. A lower depression of temperature subsequently signifies that the IL is less capable of shifting the equilibrium curve and is thus a weaker THI inhibitor. Predicted hydrogen bonding energy computed by COSMO-RS through a binary system consisting of only cation and anion has thus proven to be useful in predicting the effectiveness of ILs as inhibitors.

However, the above method of computation in COSMO-RS involves only the interaction between cation and anion, and it does not represent the hydrate system fully. Thus, the second computation of the quaternary system containing cation, anion, water, and methane gas has been conducted. Similar graphs have been plotted to find out how consistent *E*HB is in predicting the effectiveness of IL. **Figure 9** shows the graph of average depression temperature plotted against a different type of predicted energies.

As observed from **Figure 12**, when quaternary components are involved, which include cations, anions, water, and methane, it is still obvious that hydrogen bonding energy (*E*HB) is the main energy that influences the hydrate inhibition effect. Meanwhile, misfit energy and van der Waals energy have only a low regression value that is below 0.10. However, it is noticed that total interaction energy (*E*INT) provides a slightly higher regression value than EHB which is 0.6848 than 0.6671, which does not occur in binary component simulation. This could be because while involving more components such as methane and water, the van der Waals energy and misfit energy between different components are now more significant and influential. As compared to binary component regression value, the highest regression value that is obtained here is only 0.6848, which is extracted from the EINT. Nevertheless, this low regression value could be improved to 0.8276 by removing the outlier which is BMIM-HSO4 (1-butyl-3-methylimidazolium hydrogen sulfate) as shown in **Figure 10**. This is because of the nature of HSO4 − anion, which has an extra hydrogen bonding functional group, OH<sup>−</sup> (hydroxide), and thus resulting in stronger inhibition effect [62].

**Figure 11** then shows the regression value of two more data sets from the work of Xiao et al. [30]. For both sets of data, total interaction energy (*E*INT) gives the highest regression value too.

Similarly, from **Figure 12**, when the anions are fixed and cations are varied to study, the temperature depression value also decreases as the hydrogen bonding energy becomes more negative (stronger) which leads the increase in total

**155**

**Figure 10.**

**Figure 11.**

*components).*

*components, without BMIM-HSO4).*

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application…*

interaction energy. Therefore, generally, COSMO-RS simulation of binary components and quaternary components both work well as a quick prediction for the effectiveness of IL as a thermodynamic hydrate inhibitor. **Table 4** shows the regres-

*Average temperature depression from Xiao et al. [30] work vs. predicted total interaction energy (quaternary* 

*Average temperature depression from Sabil et al. [25] work vs. predicted total interaction energy (quaternary* 

Here, it is shown that the simulation of hydrogen bonding energy (*E*HB) of binary components simulation provides a more consistent regression value. On the other hand, total interaction energy (*E*INT) of quaternary components simulation more accurately reflects out the hydrated state which involves not only the IL itself but also water molecules and methane gas. To determine whether binary or quaternary components simulation is more effective in predicting ILs effectiveness, more sets of experimental data should be validated using the above approach. However, experimental work that tested ILs set with fixed anion or cation is very limited. Therefore, it is hard to conclude here whether binary or quaternary components simulation is more superior. Nevertheless, since real hydrate system consists of the interaction between water, methane, and IL, quaternary components simulation

sion value of both binary and quaternary components simulation.

will be further studied, and correlation will be developed in this work.

*DOI: http://dx.doi.org/10.5772/intechopen.86847*

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application… DOI: http://dx.doi.org/10.5772/intechopen.86847*

#### **Figure 10.**

*Solvents, Ionic Liquids and Solvent Effects*

*Average temperature depression from Sabil et al. [25] work vs. types of predicted energy (quaternary* 

different anions, stronger hydrogen bonding between ILs lead to higher average depression temperature of an IL-hydrate system. Vice versa, for a set of ILs with fixed anion but different cations, stronger *E*HB produces lower depression temperature. A lower depression of temperature subsequently signifies that the IL is less capable of shifting the equilibrium curve and is thus a weaker THI inhibitor. Predicted hydrogen bonding energy computed by COSMO-RS through a binary system consisting of only cation and anion has thus proven to be useful in predicting the effectiveness of ILs as inhibitors. However, the above method of computation in COSMO-RS involves only the interaction between cation and anion, and it does not represent the hydrate system fully. Thus, the second computation of the quaternary system containing cation, anion, water, and methane gas has been conducted. Similar graphs have been plotted to find out how consistent *E*HB is in predicting the effectiveness of IL. **Figure 9** shows the graph of average depression temperature plotted against a different type

As observed from **Figure 12**, when quaternary components are involved, which include cations, anions, water, and methane, it is still obvious that hydrogen bonding energy (*E*HB) is the main energy that influences the hydrate inhibition effect. Meanwhile, misfit energy and van der Waals energy have only a low regression value that is below 0.10. However, it is noticed that total interaction energy (*E*INT) provides a slightly higher regression value than EHB which is 0.6848 than 0.6671, which does not occur in binary component simulation. This could be because while involving more components such as methane and water, the van der Waals energy and misfit energy between different components are now more significant and influential. As compared to binary component regression value, the highest regression value that is obtained here is only 0.6848, which is extracted from the EINT. Nevertheless, this low regression value could be improved to 0.8276 by removing the outlier which is BMIM-HSO4 (1-butyl-3-methylimidazolium hydrogen sulfate) as shown in **Figure 10**. This is

anion, which has an extra hydrogen bonding func-

(hydroxide), and thus resulting in stronger inhibition effect [62].

**Figure 11** then shows the regression value of two more data sets from the work of Xiao et al. [30]. For both sets of data, total interaction energy (*E*INT) gives the

Similarly, from **Figure 12**, when the anions are fixed and cations are varied to study, the temperature depression value also decreases as the hydrogen bonding energy becomes more negative (stronger) which leads the increase in total

**Figure 9.**

*components).*

of predicted energies.

because of the nature of HSO4

highest regression value too.

tional group, OH<sup>−</sup>

−

**154**

*Average temperature depression from Sabil et al. [25] work vs. predicted total interaction energy (quaternary components, without BMIM-HSO4).*

#### **Figure 11.**

*Average temperature depression from Xiao et al. [30] work vs. predicted total interaction energy (quaternary components).*

interaction energy. Therefore, generally, COSMO-RS simulation of binary components and quaternary components both work well as a quick prediction for the effectiveness of IL as a thermodynamic hydrate inhibitor. **Table 4** shows the regression value of both binary and quaternary components simulation.

Here, it is shown that the simulation of hydrogen bonding energy (*E*HB) of binary components simulation provides a more consistent regression value. On the other hand, total interaction energy (*E*INT) of quaternary components simulation more accurately reflects out the hydrated state which involves not only the IL itself but also water molecules and methane gas. To determine whether binary or quaternary components simulation is more effective in predicting ILs effectiveness, more sets of experimental data should be validated using the above approach. However, experimental work that tested ILs set with fixed anion or cation is very limited. Therefore, it is hard to conclude here whether binary or quaternary components simulation is more superior. Nevertheless, since real hydrate system consists of the interaction between water, methane, and IL, quaternary components simulation will be further studied, and correlation will be developed in this work.

#### **Figure 12.**

*Average temperature depression against predicted total interaction energy for ILs with Cl- as anion but different cations (quaternary components).*


#### **Table 4.**

*Comparison of regression values produced by binary and quaternary components simulation.*

From the previous analysis for quaternary simulation, it is observed that the total interaction energy of anion and cation has a different effect on average temperature depression. Anion with higher interaction energy shows a higher average temperature depression, while the stronger interaction energy of cation reduces the average temperature depression. Due to the opposite effect of these two types of interaction energies (cation and anion), it is thus a must to consider them separately during the development of correlation. This results in the splitting of total interaction energy (*E*INT) into two variables, which are EINT contributed by anion (*E*INT, A) and *E*INT contributed by cation (*E*INT, C). Both of them are available and obtainable from COSMO-RS simulation. **Table 5** shows an example of *E*INT, C, *E*INT, A, and *E*INT calculated by COSMO-RS for the ILs from the work of Sabil et al. [25].

From **Table 5**, it is clear that the summation of *E*INT, A and *E*INT, C would result in the value of *E*INT. In comparison, it is also evidently shown that anion contributes more to the total interaction energy than the cation. Now after obtaining the two variables, Minitab is used to assist in developing a suitable correlation for the prediction of average temperature reduced by each IL. Generally, the model could be described as.

$$Y = \beta\_1 + \beta\_2 X\_2 + \beta\_3 X\_3 + \dots \tag{2}$$

$$
\Delta T = \beta\_1 + \beta\_2 E\_{\text{INT},\text{C}} + \beta\_3 E\_{\text{INT},\text{A}} \tag{3}
$$

**157**

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application…*

*Type of interaction energies predicted by COSMO-RS for the work of Sabil et al. [25].*

Δ*T* = 1.758 + 0.0643*E*INT,C − 0.00559*E*INT, <sup>A</sup> (4)

**Table 6** then shows the experimental value obtained from the literature review, as well as the predicted temperature using the above equation. It listed 25 ILs with their experimental average temperature depression value from 4 literature review [25, 30, 49, 54]. Using the values of ions' interaction energy (*E*INT,A and *E*INT,C) obtained from COSMO-RS, average temperature depression has been predicted for each IL. Absolute error between the experimental and predicted value is then calculated and shown at the last row of the table, without considering the three

**Table 6** shows several interesting findings and limitations of the model. First, regarding the three extreme outliers, all three of them are substituted cations that have a hydroxyl (OH<sup>−</sup>) group. This type of substituted cation, as calculated by COSMO-RS, has an overly high *E*INT, C (42.17 kJ/mol for [OH-C2MIM]-Cl as compared to 20.60 kJ/mol for EMIM-Cl), which is supposed to reduce their inhibition ability. But, in truth, hydroxyl group-substituted cation has constantly performed better than common cation because the OH<sup>−</sup> serves as a strong hydrogen bond donor that will react with water [61, 63]. The increased interaction with water molecules will thus improve the average temperature depression [64]. Due to this reason, a large discrepancy is observed between experimental and predicted temperature depression for OH<sup>−</sup>-substituted cation-based ILs. This also signifies that the model developed earlier does not apply to hydroxyl group-substituted cations or

Next, a pure error which is caused by inconsistency between experiments has also limited the accuracy of this model. **Table 7** shows the simplified list of ILs which have different experimental average temperature depression value obtained

As observed from **Table 8**, the experimental value obtained from literature review does not agree with each other. They are inconsistent, and this has thus hindered the development of a fully accurate model that could predict the inhibition ability of ILs as THI inhibitors. For instance, the inhibition ability of BMIM-BF4 was reported in three different papers, and the difference of experimental value from each paper is fairly large, ranging from 0.270 to 0.858°C. Nevertheless, Zare et al. [54] reported an experimental value of 0.460°C, which only presents a 2.92%

Next, looking at BMIM-HSO4 and EMIM-HSO4, it is experimentally proven that EMIM-HSO4, which has a smaller alkyl chain length for cation, would serve as a better inhibitor [49, 60, 62]. However, because experimental values are

obtained from two different papers [25, 54], BMIM-HSO4 recorded a higher average

*DOI: http://dx.doi.org/10.5772/intechopen.86847*

extreme outliers that are highlighted in red.

possibly any other substituted cations ILs.

error when compared to the predicted value.

from the literature review.

Model:

**Table 5.**

Among the many equations that have been tested, the best equation is listed below. It involves both *E*INT, A and *E*INT, C as independent variables.

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application… DOI: http://dx.doi.org/10.5772/intechopen.86847*


**Table 5.**

*Solvents, Ionic Liquids and Solvent Effects*

*different cations (quaternary components).*

From the previous analysis for quaternary simulation, it is observed that the total interaction energy of anion and cation has a different effect on average temperature depression. Anion with higher interaction energy shows a higher average temperature depression, while the stronger interaction energy of cation reduces the average temperature depression. Due to the opposite effect of these two types of interaction energies (cation and anion), it is thus a must to consider them separately during the development of correlation. This results in the splitting of total interaction energy (*E*INT) into two variables, which are EINT contributed by anion (*E*INT, A) and *E*INT contributed by cation (*E*INT, C). Both of them are available and obtainable from COSMO-RS simulation. **Table 5** shows an example of *E*INT, C, *E*INT, A, and *E*INT

From **Table 5**, it is clear that the summation of *E*INT, A and *E*INT, C would result in the value of *E*INT. In comparison, it is also evidently shown that anion contributes more to the total interaction energy than the cation. Now after obtaining the two variables, Minitab is used to assist in developing a suitable correlation for the prediction of average temperature reduced by each IL. Generally, the model could

*Y* = β<sup>1</sup> + β2*X*<sup>2</sup> + β3*X*<sup>3</sup> + … (2)

*T* = β<sup>1</sup> + β2*E*INT,C + β3*E*INT,A (3)

Among the many equations that have been tested, the best equation is listed

below. It involves both *E*INT, A and *E*INT, C as independent variables.

calculated by COSMO-RS for the ILs from the work of Sabil et al. [25].

*Comparison of regression values produced by binary and quaternary components simulation.*

*Average temperature depression against predicted total interaction energy for ILs with Cl- as anion but* 

**156**

be described as.

**Table 4.**

**Figure 12.**

*Type of interaction energies predicted by COSMO-RS for the work of Sabil et al. [25].*

Model:

$$
\Delta T = \text{1.758 + 0.0643 E}\_{\text{INT,C}} - \text{0.00559 } E\_{\text{INT,A}} \tag{4}
$$

**Table 6** then shows the experimental value obtained from the literature review, as well as the predicted temperature using the above equation. It listed 25 ILs with their experimental average temperature depression value from 4 literature review [25, 30, 49, 54]. Using the values of ions' interaction energy (*E*INT,A and *E*INT,C) obtained from COSMO-RS, average temperature depression has been predicted for each IL. Absolute error between the experimental and predicted value is then calculated and shown at the last row of the table, without considering the three extreme outliers that are highlighted in red.

**Table 6** shows several interesting findings and limitations of the model. First, regarding the three extreme outliers, all three of them are substituted cations that have a hydroxyl (OH<sup>−</sup>) group. This type of substituted cation, as calculated by COSMO-RS, has an overly high *E*INT, C (42.17 kJ/mol for [OH-C2MIM]-Cl as compared to 20.60 kJ/mol for EMIM-Cl), which is supposed to reduce their inhibition ability. But, in truth, hydroxyl group-substituted cation has constantly performed better than common cation because the OH<sup>−</sup> serves as a strong hydrogen bond donor that will react with water [61, 63]. The increased interaction with water molecules will thus improve the average temperature depression [64]. Due to this reason, a large discrepancy is observed between experimental and predicted temperature depression for OH<sup>−</sup>-substituted cation-based ILs. This also signifies that the model developed earlier does not apply to hydroxyl group-substituted cations or possibly any other substituted cations ILs.

Next, a pure error which is caused by inconsistency between experiments has also limited the accuracy of this model. **Table 7** shows the simplified list of ILs which have different experimental average temperature depression value obtained from the literature review.

As observed from **Table 8**, the experimental value obtained from literature review does not agree with each other. They are inconsistent, and this has thus hindered the development of a fully accurate model that could predict the inhibition ability of ILs as THI inhibitors. For instance, the inhibition ability of BMIM-BF4 was reported in three different papers, and the difference of experimental value from each paper is fairly large, ranging from 0.270 to 0.858°C. Nevertheless, Zare et al. [54] reported an experimental value of 0.460°C, which only presents a 2.92% error when compared to the predicted value.

Next, looking at BMIM-HSO4 and EMIM-HSO4, it is experimentally proven that EMIM-HSO4, which has a smaller alkyl chain length for cation, would serve as a better inhibitor [49, 60, 62]. However, because experimental values are obtained from two different papers [25, 54], BMIM-HSO4 recorded a higher average


#### **Table 6.**

*Experimental and predicted average temperature depression of ILs for selected literature review.*

temperature depression. This contradiction due to inconsistency again hardened the process of model development. Two factors could probably explain this inconsistency between experimental values: (i) purity of ILs being used in an experiment and (ii) experimental procedure and atmospheric condition.

In short, hydrogen bonding energy is the main type of energy that affects the interaction of ions with water and subsequently the inhibition ability of ILs. For a quaternary component simulation, however, total interaction energy shows a better linear relationship with average temperature depression. The model developed which considers cation interaction energy and anion interaction energy sufficiently predicts average temperature depression with an average error of 20.49%. It is to be noted that to a certain degree, the inconsistency between experimental values also contributed to the average error. **Table 9** shows the regression statistics and *P*-value from the ANOVA test for the equation developed. The confidence level for the model is set at 95%, and thus a *P*-value of 0.000 (<0.05) signifies a reliable model.

**159**

**Table 7.**

**Table 8.**

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application…*

*3.1.3 Effect of temperature on predicted inhibition ability*

*ILs with inconsistent experimental average temperature depression.*

sion against simulation temperature.

*Model summary and ANOVA for the model developed.*

temperature is shown at the table.

From the earlier section, it is mentioned that the simulation work in this study is fixed at a temperature of 10°C, which is a common temperature where hydrates start to form. In this section, the effect of temperature is further examined to investigate if the predicted inhibition ability of ILs changes dramatically with temperature. **Figure 13** shows the graph of predicted average temperature depres-

Nevertheless, if the percentage of difference is calculated out, it will be noticed that the effect of temperature is very insignificant. For example, for EMIM-Cl, using simulation of 10°C as reference state, the percentage difference for each

As shown in the table, the range of predicted average temperature depression is between 1.079 and 1.151°C, where the difference is really small. Furthermore, it is found out that most experimental studies involve only temperature range of −3.15

*DOI: http://dx.doi.org/10.5772/intechopen.86847*


*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application… DOI: http://dx.doi.org/10.5772/intechopen.86847*

#### **Table 7.**

*Solvents, Ionic Liquids and Solvent Effects*

temperature depression. This contradiction due to inconsistency again hardened the process of model development. Two factors could probably explain this inconsistency between experimental values: (i) purity of ILs being used in an experiment

*Experimental and predicted average temperature depression of ILs for selected literature review.*

In short, hydrogen bonding energy is the main type of energy that affects the interaction of ions with water and subsequently the inhibition ability of ILs. For a quaternary component simulation, however, total interaction energy shows a better linear relationship with average temperature depression. The model developed which considers cation interaction energy and anion interaction energy sufficiently predicts average temperature depression with an average error of 20.49%. It is to be noted that to a certain degree, the inconsistency between experimental values also contributed to the average error. **Table 9** shows the regression statistics and *P*-value from the ANOVA test for the equation developed. The confidence level for the model is set at 95%, and thus a *P*-value of 0.000 (<0.05) signifies a

and (ii) experimental procedure and atmospheric condition.

**158**

**Table 6.**

reliable model.

*ILs with inconsistent experimental average temperature depression.*


#### **Table 8.**

*Model summary and ANOVA for the model developed.*

#### *3.1.3 Effect of temperature on predicted inhibition ability*

From the earlier section, it is mentioned that the simulation work in this study is fixed at a temperature of 10°C, which is a common temperature where hydrates start to form. In this section, the effect of temperature is further examined to investigate if the predicted inhibition ability of ILs changes dramatically with temperature. **Figure 13** shows the graph of predicted average temperature depression against simulation temperature.

Nevertheless, if the percentage of difference is calculated out, it will be noticed that the effect of temperature is very insignificant. For example, for EMIM-Cl, using simulation of 10°C as reference state, the percentage difference for each temperature is shown at the table.

As shown in the table, the range of predicted average temperature depression is between 1.079 and 1.151°C, where the difference is really small. Furthermore, it is found out that most experimental studies involve only temperature range of −3.15


### **Table 9.**

*Percentage difference of predicted* ∆T *for EMIM-Cl due to temperature difference.*

**Figure 13.** *Graph of predicted average temperature depression against simulation temperature.*

to 16.85°C (270–290 K) [3, 29, 30, 49, 63]. This means that the highest percentage difference is just around 3.12% (for −5°C). Hence, it can be concluded that the effect of temperature is insignificant and would not affect the screening process of ILs using the correlation.

#### *3.1.4 Activity coefficient*

As discussed by Kurnia et al. [39], the lower the activity coefficient of a water-IL mixture, the higher the interaction between components in the mixture. Khan et al. also explain that for a water-IL mixture, activity coefficient below 1 signifies favorable interaction between water and ILs in the mixture [34]. When ILs interact well with water, supposedly, less water will be free to bond with each other to form hydrate. Theoretically, the activity coefficient could then reflect out the inhibition ability of IL. Therefore, validation effort was made through four sets of data [25, 30, 61] to find out if the relationship between activity coefficient and average temperature depression exists. **Figure 14** shows the graph of average temperature depression against the natural logarithm of activity coefficient.

As shown in **Figure 14**, the highest regression value is observed for BMIM-based ILs from the work of Xiao et al., which is a mere 0.6658. Meanwhile, another two sets of data record unacceptably low regression value of only 0.0045 and 0.2989. Hence, regrettably, these three sets of data could not exhibit any significant relationship between these two variables. Nevertheless, a general pattern of decreasing average temperature depression is observed when the natural logarithm of activity coefficient increases (activity coefficient increases). The incapability of the activity coefficient in reflecting the inhibition ability of IL since the calculations of activity coefficient

**161**

low temperature yet high pressure.

*3.1.5 Solubility*

**Figure 15.**

**Figure 14.**

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application…*

*Graph of average temperature depression against ln(*Y*W) for three data sets.*

*Graph of average temperature depression against solubility of IL in water.*

in COSMO-RS considers only the input of temperature, but no input of pressure is allowed. Meanwhile, in reality, hydrate occurs at low temperature (around 10°C) but high pressure. This kind of special nature of hydrate formation has thus made it hard for COSMO-RS to accurately predict out the activity coefficient water for a system of

A more soluble IL in water signifies that the IL can easily dissolve itself and interact with water molecules. Supposedly, a good IL should have high solubility in water, to bond with other water molecules and reduce the possibility of free water molecules from forming hydrate. To test the validity of the statement, four sets of data [25, 30, 61] were studied to find out if the relationship between the solubility of IL in water and average temperature depression exists. **Figure 15** shows the graph

of average temperature depression against the solubility of IL in water.

*DOI: http://dx.doi.org/10.5772/intechopen.86847*

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application… DOI: http://dx.doi.org/10.5772/intechopen.86847*

#### **Figure 14.**

*Solvents, Ionic Liquids and Solvent Effects*

*Percentage difference of predicted* ∆T *for EMIM-Cl due to temperature difference.*

**Table 9.**

**Figure 13.**

ILs using the correlation.

*3.1.4 Activity coefficient*

activity coefficient.

to 16.85°C (270–290 K) [3, 29, 30, 49, 63]. This means that the highest percentage difference is just around 3.12% (for −5°C). Hence, it can be concluded that the effect of temperature is insignificant and would not affect the screening process of

*Graph of predicted average temperature depression against simulation temperature.*

As discussed by Kurnia et al. [39], the lower the activity coefficient of a water-IL mixture, the higher the interaction between components in the mixture. Khan et al. also explain that for a water-IL mixture, activity coefficient below 1 signifies favorable interaction between water and ILs in the mixture [34]. When ILs interact well with water, supposedly, less water will be free to bond with each other to form hydrate. Theoretically, the activity coefficient could then reflect out the inhibition ability of IL. Therefore, validation effort was made through four sets of data [25, 30, 61] to find out if the relationship between activity coefficient and average temperature depression exists. **Figure 14** shows the graph of average temperature depression against the natural logarithm of

As shown in **Figure 14**, the highest regression value is observed for BMIM-based ILs from the work of Xiao et al., which is a mere 0.6658. Meanwhile, another two sets of data record unacceptably low regression value of only 0.0045 and 0.2989. Hence, regrettably, these three sets of data could not exhibit any significant relationship between these two variables. Nevertheless, a general pattern of decreasing average temperature depression is observed when the natural logarithm of activity coefficient increases (activity coefficient increases). The incapability of the activity coefficient in reflecting the inhibition ability of IL since the calculations of activity coefficient

**160**

*Graph of average temperature depression against ln(*Y*W) for three data sets.*

**Figure 15.** *Graph of average temperature depression against solubility of IL in water.*

in COSMO-RS considers only the input of temperature, but no input of pressure is allowed. Meanwhile, in reality, hydrate occurs at low temperature (around 10°C) but high pressure. This kind of special nature of hydrate formation has thus made it hard for COSMO-RS to accurately predict out the activity coefficient water for a system of low temperature yet high pressure.

#### *3.1.5 Solubility*

A more soluble IL in water signifies that the IL can easily dissolve itself and interact with water molecules. Supposedly, a good IL should have high solubility in water, to bond with other water molecules and reduce the possibility of free water molecules from forming hydrate. To test the validity of the statement, four sets of data [25, 30, 61] were studied to find out if the relationship between the solubility of IL in water and average temperature depression exists. **Figure 15** shows the graph of average temperature depression against the solubility of IL in water.

Here, the regression values for all three data sets are very low as well, with the lowest regression value of 0.057. Hence, similarly to the activity coefficient, no significant relationship could be deduced from this variable.

### **3.2 Prediction of inhibition ability of ammonium-based ILs**

From the validation part, it has been confirmed that sigma profile and total interaction energy of ILs can be correlated to the effectiveness of an IL as THI inhibitor. Hence, in this section, prediction work will be conducted on 20 ammonium-based ILs (refer to **Table 3**) to determine their ability as hydrate inhibitor, through the study of their sigma profile and total interaction energies.

#### *3.2.1 Sigma profile*

Although sigma profile could not directly compute a value to represent the effectiveness of an IL as a hydrate inhibitor, it does show the affinity of an IL toward the water. The higher the affinity of IL toward the water, the more hydrophilic it is, and the easier it could interact with water. This will then result in a more effective hydrate inhibitor. Hence, in this section, three sigma profile graphs will be used to determine the affinity of each ammonium-based ILs toward the water. The first figure, **Figure 16**, displays the sigma profile of the four types of cations involved here, which range from TMA to TBA.

From Section 3.1.1, it is discussed that the sigma profile graphs could be divided into three regions: hydrogen bond donor region (at the left of −1.0 e/nm<sup>2</sup> ), nonpolar region (between −1.0 and 1.0 e/nm<sup>2</sup> ), and hydrogen bond acceptor region (at the right of 1.0 e/nm<sup>2</sup> ). Here, all tetraalkylammonium-based cations have their peaks within the nonpolar region and are thus deduced to have a low affinity with water. This is because water molecules have only peaks within the hydrogen bond donor and hydrogen acceptor region. Due to this property, they do not interact well with ions that have a peak in the nonpolar region. However, when compared among themselves, TMA cation which has its highest peak at around −0.9 e/nm<sup>2</sup> performs the best because its peak is nearest to the polar region and thus has the highest affinity toward the water. This is because TMA has the lowest alkyl chain length, thus is less bulky and can easily interact with water molecules [34]. This makes TMA the most suitable cation among the four to be tuned as a hydrate inhibitor.

**163**

of interaction goes to OH<sup>−</sup>

*Sigma profile of several ammonium-based ILs.*

**Figure 17.**

**Figure 18.**

*Sigma profile of anions.*

with the water molecules. Meanwhile, BF4

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application…*

**Figure 17** shows the sigma profile of five types of different anions. From this graph, it is observed that all anions have their peaks located in the polar region at the right side, which is the hydrogen bond acceptor region [33]. This indicates that all of them are electronegative and has a lone pair ready to share with another hydrogen bond donor. Due to their readiness to interact with hydrogen bond donor, they have high affinity with water molecules and tend to bond well with water molecules. The highest tendency

ion, which has its peak at 3.6 e/nm2

lays its peak further at the right side of the sigma profile graph is effective in inhibiting as it has high affinity with water molecules. This is because the further the peak to the right, the larger the sigma value and, thus, the more electronegative an anion is. The high electronegativity then results in higher interaction energy and thus interacts better

−

Lastly, the third figure, **Figure 18**, has selectively displayed the sigma profile graph for four ILs, including TMA-OH, TEA-Cl, TPA-I, and TBA-BF4. The idea of this graph is to showcase several possible combinations of ILs by tuning the cation and anion. Here, it is easily observed that all cations show their peak in the nonpolar region. TMA cation shows its peak closest to the polar region and is thus the most suitable cation, due to its higher affinity with water. This could be explained by its short alkyl chain length as compared to others, which makes it more hydrophilic. Meanwhile, all anions

region is not an effective inhibitor anion because of its low polarized charge.

lay in the polar region on the right side. The most electronegative anion is OH<sup>−</sup>

has its peak furthest at the right. Due to its highest electronegativity and hence high

. In general, an anion that

ion that

ion that has its peak close to the nonpolar

*DOI: http://dx.doi.org/10.5772/intechopen.86847*

**Figure 16.** *Sigma profile of ammonium-based cations.*

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application… DOI: http://dx.doi.org/10.5772/intechopen.86847*

**Figure 17.** *Sigma profile of anions.*

*Solvents, Ionic Liquids and Solvent Effects*

here, which range from TMA to TBA.

region (at the right of 1.0 e/nm<sup>2</sup>

nonpolar region (between −1.0 and 1.0 e/nm<sup>2</sup>

*3.2.1 Sigma profile*

−0.9 e/nm<sup>2</sup>

hydrate inhibitor.

Here, the regression values for all three data sets are very low as well, with the lowest regression value of 0.057. Hence, similarly to the activity coefficient, no

From the validation part, it has been confirmed that sigma profile and total interaction energy of ILs can be correlated to the effectiveness of an IL as THI inhibitor. Hence, in this section, prediction work will be conducted on 20 ammonium-based ILs (refer to **Table 3**) to determine their ability as hydrate inhibitor,

Although sigma profile could not directly compute a value to represent the effectiveness of an IL as a hydrate inhibitor, it does show the affinity of an IL toward the water. The higher the affinity of IL toward the water, the more hydrophilic it is, and the easier it could interact with water. This will then result in a more effective hydrate inhibitor. Hence, in this section, three sigma profile graphs will be used to determine the affinity of each ammonium-based ILs toward the water. The first figure, **Figure 16**, displays the sigma profile of the four types of cations involved

From Section 3.1.1, it is discussed that the sigma profile graphs could be divided

performs the best because its peak is nearest to the polar region and

),

), and hydrogen bond acceptor

). Here, all tetraalkylammonium-based cations

into three regions: hydrogen bond donor region (at the left of −1.0 e/nm<sup>2</sup>

have their peaks within the nonpolar region and are thus deduced to have a low affinity with water. This is because water molecules have only peaks within the hydrogen bond donor and hydrogen acceptor region. Due to this property, they do not interact well with ions that have a peak in the nonpolar region. However, when compared among themselves, TMA cation which has its highest peak at around

thus has the highest affinity toward the water. This is because TMA has the lowest alkyl chain length, thus is less bulky and can easily interact with water molecules [34]. This makes TMA the most suitable cation among the four to be tuned as a

significant relationship could be deduced from this variable.

**3.2 Prediction of inhibition ability of ammonium-based ILs**

through the study of their sigma profile and total interaction energies.

**162**

**Figure 16.**

*Sigma profile of ammonium-based cations.*

**Figure 18.** *Sigma profile of several ammonium-based ILs.*

**Figure 17** shows the sigma profile of five types of different anions. From this graph, it is observed that all anions have their peaks located in the polar region at the right side, which is the hydrogen bond acceptor region [33]. This indicates that all of them are electronegative and has a lone pair ready to share with another hydrogen bond donor. Due to their readiness to interact with hydrogen bond donor, they have high affinity with water molecules and tend to bond well with water molecules. The highest tendency of interaction goes to OH<sup>−</sup> ion, which has its peak at 3.6 e/nm2 . In general, an anion that lays its peak further at the right side of the sigma profile graph is effective in inhibiting as it has high affinity with water molecules. This is because the further the peak to the right, the larger the sigma value and, thus, the more electronegative an anion is. The high electronegativity then results in higher interaction energy and thus interacts better with the water molecules. Meanwhile, BF4 − ion that has its peak close to the nonpolar region is not an effective inhibitor anion because of its low polarized charge.

Lastly, the third figure, **Figure 18**, has selectively displayed the sigma profile graph for four ILs, including TMA-OH, TEA-Cl, TPA-I, and TBA-BF4. The idea of this graph is to showcase several possible combinations of ILs by tuning the cation and anion. Here, it is easily observed that all cations show their peak in the nonpolar region. TMA cation shows its peak closest to the polar region and is thus the most suitable cation, due to its higher affinity with water. This could be explained by its short alkyl chain length as compared to others, which makes it more hydrophilic. Meanwhile, all anions lay in the polar region on the right side. The most electronegative anion is OH<sup>−</sup> ion that has its peak furthest at the right. Due to its highest electronegativity and hence high

interaction with water, it serves as the best anion to be used for an inhibitor. Therefore, from the graph, it is identifiable that TMA-OH is the best combination of all. This is followed by TEA-Cl, TPA-Br, and, finally, TBA-BF4. From this graph, it is inferred that to choose the right anion for the hydrate inhibitor, its peak should be located as far as possible at the right side of the graph. This indicates a highly electronegative anion that can bond well with water molecules. Meanwhile, it is reported that most of the ILs cations have their peaks located in the nonpolar region. This characteristic causes cations to behave as nonpolar molecules that are hydrophobic and does not interact well with water molecules [58]. Therefore, a cation with the lowest hydrophobicity should be chosen to be tuned as a hydrate inhibitor, so that it will not hinder interactions between IL and water molecules. This, in turn, signifies that the most recommendable cation should have its peak closest to the left polar region.

### *3.2.2 Total interaction energies*

In Section 3.1.2, a correlation has been developed to describe the relationship between the average depression temperature of IL-hydrate system and the total interaction energies. It is found out that both cation and anion interaction energy have a different effect on IL inhibition ability. High anion interaction energy is preferable, while high cation interaction energy will reduce an IL inhibition ability.

$$
\Delta T = 1.758 + 0.0643 E\_{\text{INT}, \text{C}} - 0.00589 E\_{\text{INT}, \text{A}} \tag{5}
$$

Using the above correlation, the ability of ammonium-based ILs has been predicted through the calculation of average temperature depression. **Table 10** shows the list of ammonium-based ILs together with their total interaction energies and predicted inhibition ability measured in terms of average temperature depression.

**Table 10** shows a list of tetraalkylammonium-based ILs, which range from cation tetramethylammonium to tetrabutylammonium paired with five types of different anions that are hydroxide ion, a tetrafluoroborate ion, chloride ion, bromide ion, and iodide ion. From this table, it is observed that when the anion is fixed, an increase in cation interaction energy, which is caused by the increase in alkyl chain length, will reduce average temperature depression. This again agrees to the earlier statement which explained that the longer alkyl chain length of cation, the bulkier it is and thus harder for it to interact with water molecules [49, 60]. This, as a result, increases its hydrophobicity, reduces its ability to bond with water, and is thus a less effective thermodynamic hydrate inhibitor [62]. In fact, among the five TBA ionic liquids (ILs), three of them show negative temperature depression. This is because of the poor combination of the bulky cation (TBA) and weak electronegativity anion (Br<sup>−</sup>, I<sup>−</sup>, BF4 <sup>−</sup>), resulting in a super ineffective inhibitor. A negative temperature depression signifies that instead of serving as hydrate inhibitor, they have now become hydrate promoter that favors the formation of the hydrate phase.

In terms of the effect of anion, we can see that the higher the interaction energy of anion, the higher the average temperature depression is. Here, the rank of EINT is as OH<sup>−</sup> > Cl<sup>−</sup> > Br<sup>−</sup> > l<sup>−</sup> > BF4 − . This resulted in the average temperature depression to follow the same pattern. For instance, looking at tetramethylammonium ILs, inhibition ability rank is as TMA-OH > TMA-Cl > TMA-Br > TMA-I > TMA-BF4. Hence, this again proves that the interaction energy provided by anion plays a crucial role in determining its inhibition ability. Also, this prediction agrees well with work reported by Tariq et al. [62]. In his review work, he reported that for a methylimidazoliumbased IL, the order of efficiency follows as such C2C1im-Cl > C2C1im-Br > C2C1im-I > C2C1im-BF4. Regrettably, OH<sup>−</sup> ILs are not studied in Tariq's work; yet, the whole ranking ranging from Cl<sup>−</sup> to BF4 − is similar to the predicted ranking. This proves that the

**165**

electronegative OH<sup>−</sup>

*Predicted average temperature depression of AILs.*

**Table 10.**

**4. Conclusions**

highly electronegative OH<sup>−</sup>

*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application…*

developed correlation is performing outstandingly in predicting the inhibition ability of ILs. Lastly, from this model, TMA-OH is identified to show the strongest ability as THI, with the highest depression temperature of 1.97°C. This is due to the highly

chain length TMA cation that does not hinder the IL interaction with water molecules.

In conclusion, among the four identified fundamental properties, sigma profile and hydrogen bonding energy have been successfully correlated to the inhibition ability of IL. Sigma profile provides a qualitative understanding of each IL in the sense of their affinity toward water molecules. Meanwhile, hydrogen bonding energy, or later upgraded to total interaction energy, has been able to satisfactorily predict out a quantitative value of average temperature depression provided by each IL. This value will then tell us the effectiveness of each IL as a thermodynamic hydrate inhibitor. The correlation developed is validated with open literature and is found out to have an average error of 20.49%. From the predicted data, it is observed that TMA-OH depresses the temperature of IL-hydrate system by 1.97°C, whereas the widely studied EMIM-Cl can only experimentally depress the system by 1.22°C. TMA-OH has shown the highest inhibition ability due to the combination of its short alkyl chain length cation and a

suitable to be used for substituted cations, as the introduced functional group such as hydroxyl group will provide extra H bonding with water molecules. COSMO-RS simulation, on the other hand, has been proven to be applicable in computing fundamental

anion that bonds well with water molecules and a short alkyl

anion. Findings, however, show that this correlation is not

*DOI: http://dx.doi.org/10.5772/intechopen.86847*


*Pre-Screening of Ionic Liquids as Gas Hydrate Inhibitor via Application… DOI: http://dx.doi.org/10.5772/intechopen.86847*

#### **Table 10.**

*Solvents, Ionic Liquids and Solvent Effects*

should have its peak closest to the left polar region.

*3.2.2 Total interaction energies*

interaction with water, it serves as the best anion to be used for an inhibitor. Therefore, from the graph, it is identifiable that TMA-OH is the best combination of all. This is followed by TEA-Cl, TPA-Br, and, finally, TBA-BF4. From this graph, it is inferred that to choose the right anion for the hydrate inhibitor, its peak should be located as far as possible at the right side of the graph. This indicates a highly electronegative anion that can bond well with water molecules. Meanwhile, it is reported that most of the ILs cations have their peaks located in the nonpolar region. This characteristic causes cations to behave as nonpolar molecules that are hydrophobic and does not interact well with water molecules [58]. Therefore, a cation with the lowest hydrophobicity should be chosen to be tuned as a hydrate inhibitor, so that it will not hinder interactions between IL and water molecules. This, in turn, signifies that the most recommendable cation

In Section 3.1.2, a correlation has been developed to describe the relationship between the average depression temperature of IL-hydrate system and the total interaction energies. It is found out that both cation and anion interaction energy have a different effect on IL inhibition ability. High anion interaction energy is preferable, while high cation interaction energy will reduce an IL inhibition ability.

∆*T* = 1.758 + 0.0643*E*INT,C − 0.00559*E*INT,A (5)

Using the above correlation, the ability of ammonium-based ILs has been predicted through the calculation of average temperature depression. **Table 10** shows the list of ammonium-based ILs together with their total interaction energies and predicted inhibition ability measured in terms of average temperature depression. **Table 10** shows a list of tetraalkylammonium-based ILs, which range from cation tetramethylammonium to tetrabutylammonium paired with five types of different anions that are hydroxide ion, a tetrafluoroborate ion, chloride ion, bromide ion, and iodide ion. From this table, it is observed that when the anion is fixed, an increase in cation interaction energy, which is caused by the increase in alkyl chain length, will reduce average temperature depression. This again agrees to the earlier statement which explained that the longer alkyl chain length of cation, the bulkier it is and thus harder for it to interact with water molecules [49, 60]. This, as a result, increases its hydrophobicity, reduces its ability to bond with water, and is thus a less effective thermodynamic hydrate inhibitor [62]. In fact, among the five TBA ionic liquids (ILs), three of them show negative temperature depression. This is because of the poor combination of the bulky cation (TBA) and weak electronegativity

<sup>−</sup>), resulting in a super ineffective inhibitor. A negative tempera-

. This resulted in the average temperature depression to

ILs are not studied in Tariq's work; yet, the whole rank-

is similar to the predicted ranking. This proves that the

ture depression signifies that instead of serving as hydrate inhibitor, they have now

In terms of the effect of anion, we can see that the higher the interaction energy of anion, the higher the average temperature depression is. Here, the rank of EINT is

follow the same pattern. For instance, looking at tetramethylammonium ILs, inhibition ability rank is as TMA-OH > TMA-Cl > TMA-Br > TMA-I > TMA-BF4. Hence, this again proves that the interaction energy provided by anion plays a crucial role in determining its inhibition ability. Also, this prediction agrees well with work reported by Tariq et al. [62]. In his review work, he reported that for a methylimidazoliumbased IL, the order of efficiency follows as such C2C1im-Cl > C2C1im-Br > C2C1im-I >

become hydrate promoter that favors the formation of the hydrate phase.

**164**

as OH<sup>−</sup>

anion (Br<sup>−</sup>, I<sup>−</sup>, BF4

> Cl<sup>−</sup>

ing ranging from Cl<sup>−</sup>

> Br<sup>−</sup>

C2C1im-BF4. Regrettably, OH<sup>−</sup>

> l<sup>−</sup>

 > BF4 −

 to BF4 − *Predicted average temperature depression of AILs.*

developed correlation is performing outstandingly in predicting the inhibition ability of ILs. Lastly, from this model, TMA-OH is identified to show the strongest ability as THI, with the highest depression temperature of 1.97°C. This is due to the highly electronegative OH<sup>−</sup> anion that bonds well with water molecules and a short alkyl chain length TMA cation that does not hinder the IL interaction with water molecules.
