Journal of Chromatography B, 945–946 (2014) 60–67
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Relationship study of partition coefficients between ionic liquid and headspace for organic solvents by HS-GC Meiping Ni a , Ting Sun b , Lin Zhang a , Yan Liu a , Meng Xu a , Ye Jiang a,∗ a b
School of Pharmacology, Hebei Medical University, Shijiazhuang 050017, Hebei, PR China Department of Pharmacology, The fourth Hospital of Hebei Medical University, Shijiazhuang 050017, Hebei, PR China
a r t i c l e
i n f o
Article history: Received 3 July 2013 Received in revised form 24 October 2013 Accepted 9 November 2013 Available online 26 November 2013 Keywords: HS-GC ILs Partition coefficients Organic solvents
a b s t r a c t A general study was carried out to investigate the relationship between analytes (organic solvents) and matrix medium (ionic liquids, ILs) by headspace gas chromatography (HS-GC) in order to provide a guidance to choose a suitable matrix medium during the process of experiment. Thirteen ILs contained different cations or anions and two kinds of organic solvents (alkylogens and aprotic solvents which involved ability of pro-proton) performed different interactions with ILs were chosen in this study. The concentrations of analytes in headspace were determined by HS-GC and then logK (the logarithm of concentration radio between matrix medium and headspace) was calculated respectively. Factors which affect logK, such as logPO/W (the logarithm of the octanol/water partition coefficient for a solvent) for different cations (including parent nucleus and alkyl chains) and anions of ILs, were investigated. The results indicated that the longer alkyl chains, the lower polarity of parent nucleus and the higher polarity of anions performed the higher headspace efficiency for alkylogens. Meanwhile, the shorter alkyl chains and the lower polarity of parent nucleus make the higher headspace efficiency for aprotic solvents which involved ability of pro-proton. For both kinds of organic solvents, anions of ILs performed little influences to headspace efficiency. The relationship between ILs and organic solvents was primarily investigated and a helpful guidance was provided for the application of ILs as matrix medium to analyze solvents by HS-GC. The model was successfully used to determine the organic residual solvents in ketoconanzale to choose a suitable ionic liquid during the process of HS-GC. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Recently, ionic liquids (ILs) have been broadly applied in many areas, especially in analytical chemistry [1–5]. During the process of headspace gas chromatography (HS-GC), traditional matrix medium, such as water, N,N-dimethylformamide (DMF), dimethylsulfoxide (DMSO), and N,N-dimethylacetamide (DMA), exhibit poor dissolubility, unstable, and appeared large solvent peak in chromatogram, which may increase the difficulty of separation. Such as adefovir dipivoxil, a kind of nucleotide drugs, many residual solvents are introduced during the process of synthesis and these solvents cannot be headspace analyzed completely due to the reasons above. Such problems could be avoided by using ILs since ILs have shown notable properties such as great dissolubility, high thermal stability, and no vapor pressure so that organic solvents could be easily dissolved in ILs and the influence of traditional matrix medium in chromatogram could be decreased [6,7]. Thus, ILs were regarded as an ideal matrix medium for HS-GC to
∗ Corresponding author. Tel.: +86 311 86266025; fax: +86 311 86266069. E-mail addresses: [email protected], [email protected] (Y. Jiang). 1570-0232/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jchromb.2013.11.021
analyze residual organic solvents in pharmaceutical products and some applications of this had been reported in recent years [8–12]. Although ILs present many advantages for HS-GC, previous works suggested that not all of the ILs performed ideal headspace efficiency for different analytes. It is known that headspace efficiency could be characterized by logK (the logarithm of concentration radio between matrix medium and headspace). The lower value of logK represented that the higher headspace efficiency was obtained. Therefore, the ILs obtained the smaller value of logK which should be chosen during the process of experiment. However, no theoretical guidance was investigated to explain the choice of ILs during the process of HS-GC. The choice for ILs to analyze residual organic solvents in pharmaceutical products [8–12] still remained at a “trying-comparing-screening” step and these applications of ILs which will inevitably lead to the waste of time and cost. In order to avoid the time-consuming screening step, choosing suitable ILs has become a tough problem. Previously, our laboratory reported a general relationship model to explain the interactions between aliphatic alcohols and three ILs with methyl-imidazolium cations in static HS-GC [13]. The relationship between the characters of aliphatic alcohols (including polarity and boiling point) and logK established a helpful
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
61
Table 1 Instrumental conditions. Aprotic solvents Chromatograph Column Oven temperature program Split injection Column flow FID temperature
Alkylogens
SP 3420 gas chromatograph Phenomenex ZB-1 fused silica capillary column (60 m 0.53 mm I.D. phenomenex, Guangzhou, China) with 5.00 m film thickness (100% dimethylpolysiloxane) 30 ◦ C, 8 ◦ C /min to 120 ◦ C, 10 min 50 ◦ C, 2 ◦ C /min to 70 ◦ C, 5 ◦ C /min to 125 ◦ C, 5 min 150 ◦ C, splitless 2 mL/min N2 , constant flow mode 300 ◦ C
guidance of using ILs as matrix medium to analyze solvents by HSGC. It had improved that complex interactions might be existed between organic solvents and ILs, and the best model of the relationship between logK and Bp (boiling point), logPO/W (polarity) for organic solvents was identified. However, the influences by the characters (including polarity and boiling point) of organic solvents were investigated above but these structured characters of ILs were not been explored before. ILs were ionogenic compounds and also performed some characters of organic solvents. According to our prediction, logK might be influenced not only by the characters of organic solvents, but also by the characters of ILs. Thus, the influences by the characters of ILs should be further studied. In this paper, the relationship between the polarity of different groups of ILs and logK is investigated to provide a guidance to choose an ideal ionic liquid for each kind of organic solvents as the matrix medium during the process of HS-GC. Two types of organic solvents, alkylogens and aprotic solvents, were chosen to analyze the partition coefficients with ILs. Alkylogens are venomous solvents whose concentrations in pharmaceuticals should be limited [14]. Although alkylogens performed great headspace efficiency in traditional matrix medium, pharmaceutics containing these residual solvents do have poor solubility in traditional matrix medium, which increase difficulty to headspace analyze [6,7]. Consequently, ILs can be used to take place of the traditional headspace solvent. It was found that halogen atoms in organic alkylogen can absorb the cations of ILs due to the high electronegativity of halogen atoms. Similarly, anions of ILs would reject this kind of solvents because of the similar electrostatic ionic interaction. The aprotic solvents have high electronegativity and boiling point. Traditional matrix medium, such as water, is hard to analyze this kind of solvents because of their low boiling point. Meanwhile, high polarity of water might create strong interactions with organic solvents having high electronegativity. It might decrease the headspace efficiency. Thus, ILs are ideal matrix medium by HS-GC for this kind of solvents. In summary, the interactions between two kind of organic solvents mentioned above and ILs were investigated. The regression equations were obtained of ionic liquid–gas partition coefficients for these two kinds of solvents to explain the relationship between ILs and organic solvents. The results were expected to be a helpful guidance for the selection of ILs as matrix medium for analysis of residual solvents during the process of HS-GC.
Table 2 The structure properties of the ILs used in this study. Ionic liquids
logP1
logP2
logP3
[BMIM][BF4 ] [EMIM][BF4 ] [HMIM][BF4 ] [HMIM][PF6 ] [HMIM][Tf] [HMIM][T2 N] [HMIM][Cl] [HMIM][Br] [HMIM][I] [EMIM][T2 N] [EMPY][T2 N] [EMPI][T2 N] [EMPYR][T2 N]
−1.32 −1.32 −1.32 −1.32 1.18 1.18 0.32 0.65 0.65 1.18 1.18 1.18 1.18
−0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 0.70 0.60 0.18
1.33 2.17 3.00 3.00 3.00 3.00 3.00 3.00 3.00 2.17 2.17 2.17 2.17
(DMSO) (HPLC-grade) was from DIKMA. ILs obtained from Shanghai Cheng Jie Chemical Co. Ltd. (Shanghai, China). Table 2 lists Thirteen ILs and their properties. 2.2. Measurement of the partition coefficients The procedure of experiment was the same with Grant Von Wald [8]. Stock solutions were prepared by dissolving four analytes with DMSO in a 10 mL volumetric flask. Working solutions at five different concentrations were obtained by serial dilutions of stock solutions. 2 L analyte solutions was added to a 12.5 mL headspace vials and then the vials were tightly sealed using PTFE coated silicon rubber septa and aluminum crimp caps and equilibrated at the temperature of 100 ◦ C in an oven for 30 min. After homogenization, each sample was analyzed in triplicate with a 0.5 mL injection volume. For all analytes, a linear relationship was obtained. Sample vials were prepared by adding 1 mL ionic liquid to a 12.5 mL headspace vial. Figs. 1 and 2 show the headspace gas chromatogram of a 2 L standard solution prepared with [EMIM][BF4 ]
mV
3
560
2
480 400
2. Materials and methods 2.1. Reagents and materials Table 1 gives the instrumental conditions used for all measurements employed in this study. Trichloromethane, 1,1,2tichloroethylene, 1,1,1-trichloroethane, 1,2-dichloroethane, acetonitrile, acetone, acetoacetate, tetrahydrofuran (TMF), pyridine, and dioxane were analytical reagent grade and purchased from Tianjin Chemical Reagents Ltd. (Tianjin China). Dimethylsulfoxide
1
320 160
4
80 0
2
4
6
8
10
12
14
16
Fig. 1. The chromatogram of alkylogens. (1. Trichloromethane; Dichloroethane; 3. 1,1,1-Trichloroethane; 4. 1,1,2-Tichloroethylene).
min 2.
1,2-
62
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
where, C0 is the concentration of analyte in the standard solution and V0 is the volume of calibration solution added to the vial (2 L), Cg is the measured concentration in the gas phase by the linear relationship, V is the vial volume (12.5 mL), and Vl is the volume of liquid in the vial (1 mL).
5
mV 560 480
4
400
2 3
3. Results and discussions
6
320 160
1
80 0
2
4
6
8
10
12
14
16
min
Fig. 2. The chromatogram of the aprotic solvents. (1. Acetone; 2. Acetonitrile; 3. Tetrahydrofuran; 4. Dioxane; 5. Acetoacetate; 6. Pyridine).
matrix, as representative of all matrix medium on a ZB-1 column at the equilibrium temperature of 100 ◦ C. During the process of HS-GC, the partition coefficient, K, was described by the following equation: K=
Cl Cg
(1)
where, Cl is the concentration of analyte in the liquid phase, and Cg is the concentration in the gas phase. In this report, K was calculated as: K=
(C0 V0 − Cg (V − Vl ))/Vl Cl = Cg Cg
(2)
Thirteen ILs contained different cations or anions and two kinds of organic solvents were chosen in this study. The partition coefficients of different solvents between each ionic liquid and gas phase were calculated. The results of logK were listed in Tables 3 and 4. The relationship studies between ILs and organic solvents were carried out with the SPSS (Statistical Package for the Social Sciences) version 13.0 program. Previous works have revealed that because of the interaction between ILs and organic solvents, logK was influenced by the polarity and boiling point of organic solvents. In this study, the polarity of ILs which might also influence logK was investigated. The parameter which could demonstrate the whole polarity of ionic liquid was failed to be found. As we all know, ILs are composed by different organic cations and organic or inorganic anions. Cations of ILs also contained of parent nucleus and alkyl chains. The parameters which could demonstrate the anions and cations of ILs were used to investigate the relationship between logK. Thus, logP1 was used to demonstrate the polarity of anions which obtained by the polarity of natrium salt for different anions. logP2 was used to represent the polarity of nucleus obtained by the polarity of heterocyclic compounds. logP3 was used to express the polarity of alkyl chains which obtained by the polarity of alkanes.
Table 3 Values of logK for alkylogens. logK
[EMIM][BF4 ] [BMIM][BF4 ] [HMIM][BF4 ] [BMIM][T2 N] [BMPI][T2 N] [BMPY][T2 N] [BMPYR][T2 N] [HMIM][PF6 ] [HMIM][Cl] [HMIM][Br] [HMIM][I] [HMIM][Tf] [HMIM][T2 N]
Trichloromethane
Dchloroethane
Tichloroethane
Tichloroethylene
2.26 1.52 1.09 2.58 0.743 0.881 1.42 1.38 1.53 1.51 1.70 1.66 1.38
1.34 0.573 0.0750 1.88 0.390 0.490 1.05 1.64 1.75 1.78 1.90 1.92 1.64
1.81 1.77 0.465 2.36 0.531 0.665 1.20 1.53 1.63 1.63 1.79 1.82 1.53
2.02 1.50 1.07 1.89 0.115 0.240 0.810 1.69 1.76 1.73 1.92 1.99 1.69
Table 4 Values of logK for the aprotic solvents. logK
[EMIM][BF4 ] [BMIM][BF4 ] [HMIM][BF4 ] [BMIM][T2 N] [BMPI][T2 N] [BMPY][T2 N] [BMPYR][T2 N] [HMIM][PF6 ] [HMIM][Cl] [HMIM][Br] [HMIM][I] [HMIM][Tf] [HMIM][T2 N]
Ttrahydrofuran
Doxane
Pridine
Aetonitrile
Aetone
Ethyl acetate
0.938 1.33 1.72 1.70 1.38 1.44 1.52 1.25 1.26 1.30 1.31 1.31 1.31
0.401 0.862 1.21 1.01 1.69 0.750 0.810 1.95 1.93 1.97 1.93 2.00 2.01
0.612 1.01 1.44 1.95
1.38 1.82 2.22 1.06 0.750 0.770 0.840 1.24 1.27 1.27 1.27 1.29 1.31
1.42 2.87 2.30 1.10 0.780 0.790 0.880 1.41 1.45 1.45 1.45 1.47 1.47
0.710 1.15 1.54 1.95 1.63 1.66 1.76 1.17 1.18 1.20 1.21 1.25 1.24
1.67 1.75 0.914 0.910 0.960 0.960 0.960 0.950
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
63
1,2-Dichloroethane
Trichloromethane 2
2
1.5
1.5
logK
1
logK
1 0.5
0.5
0
0
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [Tf] [T2N] [Cl] [BF4] [Br] [I] [PF6]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
1,1,2-Tichloroethylene
1,1,1-Trichloroethane 2
2
1.5
1.5
logK
logK
1
1
0.5
0.5 0
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [I] [BF4] [PF6] [Tf] [T2N] [Br] [Cl]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
Fig. 3. The influence by the anions of ILs to logK.
1,2-Dichloroethane
Trichloromethane 3
3 2.5
2.5
2
logK
1.5
2
logK
1.5
1 1 0.5 0.5
0 [BMIM] [T2N]
[BMPI] [T2N]
[BMPY] [T2N]
[BMPYR] [T2N]
0 [BMIM] [T2N]
1,1,1-Trichloroethane
[BMPI] [T2N]
[BMPY] [T2N]
[BMPYR] [T2N]
1,1,2-Tichloroethylene
3
3
2.5
2.5 2
2
logK
1.5
logK
1.5 1
1
0.5 0.5
0 0
[BMIM] [T2N] [BMIM] [T2N]
[BMPI] [T2N]
[BMPY] [T2N]
[BMPI] [T2N]
[BMPY] [T2N]
[BMPYR] [T2N]
[BMPYR] [T2N]
Fig. 4. The influence by the parent nucleus of ILs to logK.
3.1. The partitioning relationship between alkylogens and ILs Figs. 3–5 revealed that the longer alkyl chains, the lower polarity of parent nucleus, and the higher polarity of anions performed the higher headspace efficiency for alkylogens. As high electronegativity halogen atoms were contained in this kind of organic solvents, the cations of ILs could be absorbed by it. For the same reason, anions of ILs would reject the solvents because of the similar electronegativity. These two kinds of relationship might influence the headspace efficiency globally. Firstly, the influences for anions of ILs were investigated. The correlation between logP1 and logK was inspected and the result was shown in Table 5. The correlation equation represented that the headspace efficiency was increasing with the increasing of logP1 . Fig. 3 also showed that the anions of ILs performed little influence for logK though great correlation between logP1 and logK.
From Fig. 3 it could be observed that the headspace efficiency was increased with the decreasing of the polarity of anions but anions had little affect in the headspace efficiency. Fig. 4 and Fig. 5 showed that cations of ILs that included parent nucleus and alkyl chains performed an important factor which would affect the headspace efficiency. It revealed that the longer alkyl chains, the Table 5 Equations and some coefficients. Organic solvents
Equations
Multiple correlation coefficients (R)
Alkylogens
logK = 1.455 − 0.322 logP1 logK = 1.33 − 1.26 logP2 logK = 2.77 − 0.707 logP3 logK = 1.065 + 0.455 logP1 − 1.15logP2 − 0.152 logP3
0.950 0.939 0.878 0.881
64
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
1,1,1-Trichloroethane
1,2-Dichloroethane 2.5
2.5
2
2
logK
1.5
logK
1.5
1
1
0.5
0.5 0
0 [EMIM][BF4]
[BMIM][BF4]
[EMIM][BF4]
[HMIM] [BF4]
[BMIM][BF4]
[HMIM] [BF4]
Trichloromethane
1,1,2-Tichloroethylene 2.5
2.5
2
2
logK
1.5
logK
1.5 1
1 0.5
0.5
0
0
[EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
Fig. 5. The influence by the alkyl chain of ILs to logK.
Acetoacetate
Pyridine
2.5
2.5
2
2
1.5
logK
1.5
1
1
0.5
0.5
0
logK
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
TMF
Acetone
2.5
2.5
2
2
1.5
logK
1.5
1
1
0.5
0.5
0
logK
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
Acetonitrile
Acetone
2.5
2.5
2
2
1.5
1.5
logK
1
1
0.5
0.5
0
logK
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
Fig. 6. The influence by the anions of ILs to logK.
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
Acetoacetate
65
Pyridine
2
2
1.6
1.6
1.2
logK
1.2
0.8
0.8
0.4
0.4
0
logK
0 [BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
[BMIM] [T2N] [BMPI] [T2N]
TMF
[BMPY] [T2N] [BMPYR] [T2N]
Acetone
2
2
1.6
1.6
1.2
logK
1.2
0.8
0.8
0.4
0.4
0
logK
0 [BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
[BMIM] [T2N] [BMPI] [T2N]
Acetonitrile
[BMPY] [T2N] [BMPYR] [T2N]
Dioxane
2
2
1.6
1.6
1.2
logK
1.2
0.8
0.8
0.4
0.4
0
logK
0 [BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
[BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
Fig. 7. The influence by the parent nucleus of ILs to logK.
lower polarity of parent nucleus obtained the lower logK, that is, higher headspace efficiency. Finally, multivariable linear regression was carried out between logK and all the structure parameters of ILs, including logP1 , logP2 , logP3 . The results were listed in Table 5 and the correlation was great. It indicated that all the structure parameters of ILs presented influences for headspace efficiency. However, the parameters of organic solvents did not improve the correlation. It might have been caused by the similar structures of this kind of organic solvents, which performed similar interactions with ILs. All the results above implied the different groups present different effect to headspace efficiency. A whole ionic liquid was composed by these different groups. Thus, the correlation among all the characters of ILs, including logP1 , logP2 , and logP3 , with logK was researched. The result revealed that this kind of ILs involved long alkyl chain gain higher headspace efficiency and the best headspace solvents gained from this model were agreed with the choices literatures reported [8,9]. Jiang used to determine the residual solvents such as dichlormethane and trichlormethane by HS-GC–MS [9]. [BMIM][BF4 ] was used as the matrix medium in experiment. The result implied that if the ILs such as [BMIM][BF4 ] which involved longer alkyl chain was chosen as the matrix medium, this kind of organic solvents gained high headspace efficiency. Grant Von Wald [8] determined partition coefficients between a series of organic solvents and ILs. The result showed that n-heptane gained lower partition coefficients in pyridine ILs than in imidazole. The polarity of pyridine is lower than imidazole so that the point
headspace efficiency would be higher when the lower polarity for the parent nucleus of ILs could be improved.
3.2. The partitioning relationship for aprotic solvents The shorter alkyl chains and the lower polarity of parent nucleus make the higher headspace efficiency for aprotic solvents. It had been proofed that the interactions between ILs and solvents might affect the headspace efficiency above. In this part, the organic solvents which might form hydrogen bond with ILs were investigated in thirteen ILs listed in Table 2. Firstly, the influences for anions of ILs were investigated. The correlation between logP1 and logK was failed to be found. From Fig. 6 it can also be observed that anions rarely affect the headspace efficiency and it might have been caused by the little volume of anions. The correlation between logP2 and logK was also failed to be found. Fig. 7 implied that the solvents, including acetonitrile, acetone, acetoacetate, tetrahydrofuran, and pyridine, the lower polarity for parent nucleus gained higher headspace efficiency. However, dioxane was not following the role above. It was probably caused by two atoms which involved in dioxane molecule might form hydrogen bond with ILs. Consequently, the headspace efficiency was influenced. The correlation coefficient for logK and logP3 was 0.738. It implied that the correlation existed between logK and logP3 . Fig. 8 implied that the longer alkyl chains gained higher headspace efficiency. It was caused by the reason that the longer alkyl chains,
66
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
TMF
Acetone
3
3
2.5
2.5
2
2
logK
1.5 1
1
0.5
0.5
0
logK
1.5
0 [EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
Acetonitrile
[BMIM][BF4]
[HMIM] [BF4]
Dioxane
3
3
2.5
2.5
2
2
logK
1.5
logK
1.5
1
1
0.5
0.5
0 [EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
Acetoacetate
[BMIM][BF4]
[HMIM] [BF4]
Pyridine
3
3
2.5
2.5
2
2
logK
1.5 1
1
0.5
0.5
0
logK
1.5
0 [EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
Fig. 8. The influence by the alkyl chain of ILs to logK.
the more hydrogen atoms in ILs made the larger interaction with organic solvents. Thus, the headspace efficiency was decreased. In order to investigate the interactions between ILs and organic solvents comprehensively, the correlation for logK and all logP1 , logP2 , and logP3 was investigated. The correlation coefficient for them was 0.278. It implied that no correlation for them. Though the tendency of logK for this kind of organic solvents was observed, there was no correlation of them. The probable reasons were that different structures for this kind of organic solvents lead the difference of headspace partitioning behavior and also owing to the other interactions between organic solvents and ILs. Liu used to determine acetonitrile and DMF in pharmaceuticals and compare the headspace efficiency between [Bmim][BF4 ] and [Hmim][BF4 ] [10]. The results revealed that the headspace efficiency of [Bmim][BF4 ] was higher than [Hmim][BF4 ]. Consequently, these kinds of ILs which involves longer alkyl chains might gain higher headspace efficiency. In summary, the relationship between thirteen ILs and two kinds of organic solvents was investigated in this study, and conclusions were summarized as follows:
1. The anions of ILs had less efficiency than cations for both kinds of organic solvents. It was caused by anions small space for anions and it had little efficiency for headspace partitioning. 2. The cations for ILs influenced the headspace partitioning a lot, especially the alkyl chains of cations. There were different influences for these two kinds of solvents. The longer the alkyl
chains, the lower headspace efficiency for protic solvents which involved ability of pro-proton but the higher headspace efficiency for alkylogens. The reason was that the polarity of cations was decreased with the increase of alkyl chains which might decrease the interaction between alkylogens and ILs and then the headspace efficiency might increase the headspace efficiency. However, protic solvents which involved high electronegativity atoms, the number of hydrogen were increased with the increase of alkyl chains. Thus, the interactions between them decrease the headspace efficiency. With the decreasing of the polarity of parent nucleus, the headspace efficiency became higher for alkylogens. However, the polarity of parent nucleus performed complex influences for protic solvents which involved ability of pro-proton. For acetonitrile, acetone, acetoacetate, tetrahydrofuran, and pyridine, the headspace efficiency became higher with the decreasing of the polarity of parent nucleus. However, dioxane was not following the role above. It was probably caused by the complex structures and interactions of them and it should be further studied. 4. Application of the model for the rapid analysis of real samples After the interaction between organic solvents and ILs was explored, a sensitive method was established to determine the
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
organic residual solvents in ketoconanzale. 0.01 g of ketoconanzale were put in a 10 mL headspace vial, followed by spiking 1 mL of ILs. The sample solution was maintained at the equilibration temperature of 110 ◦ C for 30 min and a volume of 1 mL headspace gas was directly injected into GC for analysis. For ethyl, the headspace efficiency of ILs increased with the growth of cation alkyl chain. For ethyl, DMF and DMSO, the headspace efficiency of ILs decreased with the growth of cation alkyl chain. However, the anion of ILs has minor impact on headspace efficiency. Ultimately, [Bmim][PF6 ] was chosen as the best headspace solvent. An excellent separation of ethanol, dichloromethane, ethyl acetate, butyl alcohol, pyridine, DMF, and DMSO was achieved. The calibration curve was linear in the range of 1.25–200 mg L−1 for ethanol, 1.50–24.0 mg L−1 for dichloromethane, 12.5–200 mg L−1 for ethyl acetate, 12.5–200 mg L−1 for butyl alcohol, 0.500–8.00 mg L−1 for pyridine, 2.20–3.52 mg L−1 for DMF, and 12.5–200 mg L−1 for DMSO. All the average recoveries were limited in 89.8–98.2% and RSD were less than 4.0%. 5. Conclusions A relationship study of ILs–gas phase partition coefficients for two kinds of organic solvents (alkylogens and aprotic solvents) was evaluated in this paper initially. The partitioning behaviors of organic solvents were influenced by the polarity of ILs and the cations preformed more influences to headspace efficiency than anions of ILs. As the interactions between the organic solvents and ILs were initially studied, this ILs model could offer a guidance to
67
determine the organic residual solvents in pharmaceutical product by HS-GC and rapid quantitative analysis of real samples possible. Acknowledgments This work was financially supported by the National Natural Science Foundation of China (grant no. 81041087), Hebei Natural Science Foundation of China (grant no. C2009001069). References [1] L.J. He, X.L. Luo, H.X. Xie, C.J. Wang, X.M. Jiang, K. Lu, Anal. Chim. Acta 655 (2009) 52–59. [2] C.R. Contreras, C. Dominguez, J.M. Bayona, J. Chromatogr. A. 1261 (2012) 164–170. [3] L. Vallecillos, E. Pocurull, F. Borrull, Talanta 99 (2012) 824–832. [4] L. Vidal, A. Chisvert, A. Canals, A. Salvador, Talanta 81 (2010) 549–555. [5] I. Dominguez, N. Calvar, E. Gomez, A. Dominguez, Procedia Eng. 42 (2012) 1597–1605. [6] D. Wei, A. Ivaska, Anal. Chim. Acta 607 (2008) 126–135. [7] Y.X. Fan, J.Q. Qian, J. Mol. Catal. B: Enzym. 66 (2010) 1–7. [8] G.V. Wald, D. Albers, H. Cortes, T.M. Cabe, J. Chromatogr. A. 1201 (2008) 15–20. [9] H.L. Jiang, H.Y. Shen, Z.G. Le, S.C. Zhou, Chin. J. Anal. Chem. 34 (2006) 1027–1029. [10] F.H. Liu, Y. Jiang, J. Chromatogr. A. 1167 (2007) 116–119. [11] K. Urakami, A. Higashi, K. Umemoto, M. Godo, J. Chromatogr. A. 1057 (2004) 203–210. [12] G. Laus, M. Andre, G. Bentivoglio, H. Schottenberger, J. Chromatogr. A. 216 (2009) 6020–6023. [13] X.W. He, L.L. Lei, Y. Jiang, J.M. Li, M.P. Ni, Chromatographia 74 (2011) 157–161. [14] Editorial Committee of the Pharmacopoeia of the People’s Republic of China, Pharmacopoeia of the People’s Republic of China (2nd Part), Chemical Industry Press, Beijing, 2010, Appendix: 63.
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Relationship study of partition coefficients between ionic liquid and headspace for organic solvents by HS-GC Meiping Ni a , Ting Sun b , Lin Zhang a , Yan Liu a , Meng Xu a , Ye Jiang a,∗ a b
School of Pharmacology, Hebei Medical University, Shijiazhuang 050017, Hebei, PR China Department of Pharmacology, The fourth Hospital of Hebei Medical University, Shijiazhuang 050017, Hebei, PR China
a r t i c l e
i n f o
Article history: Received 3 July 2013 Received in revised form 24 October 2013 Accepted 9 November 2013 Available online 26 November 2013 Keywords: HS-GC ILs Partition coefficients Organic solvents
a b s t r a c t A general study was carried out to investigate the relationship between analytes (organic solvents) and matrix medium (ionic liquids, ILs) by headspace gas chromatography (HS-GC) in order to provide a guidance to choose a suitable matrix medium during the process of experiment. Thirteen ILs contained different cations or anions and two kinds of organic solvents (alkylogens and aprotic solvents which involved ability of pro-proton) performed different interactions with ILs were chosen in this study. The concentrations of analytes in headspace were determined by HS-GC and then logK (the logarithm of concentration radio between matrix medium and headspace) was calculated respectively. Factors which affect logK, such as logPO/W (the logarithm of the octanol/water partition coefficient for a solvent) for different cations (including parent nucleus and alkyl chains) and anions of ILs, were investigated. The results indicated that the longer alkyl chains, the lower polarity of parent nucleus and the higher polarity of anions performed the higher headspace efficiency for alkylogens. Meanwhile, the shorter alkyl chains and the lower polarity of parent nucleus make the higher headspace efficiency for aprotic solvents which involved ability of pro-proton. For both kinds of organic solvents, anions of ILs performed little influences to headspace efficiency. The relationship between ILs and organic solvents was primarily investigated and a helpful guidance was provided for the application of ILs as matrix medium to analyze solvents by HS-GC. The model was successfully used to determine the organic residual solvents in ketoconanzale to choose a suitable ionic liquid during the process of HS-GC. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Recently, ionic liquids (ILs) have been broadly applied in many areas, especially in analytical chemistry [1–5]. During the process of headspace gas chromatography (HS-GC), traditional matrix medium, such as water, N,N-dimethylformamide (DMF), dimethylsulfoxide (DMSO), and N,N-dimethylacetamide (DMA), exhibit poor dissolubility, unstable, and appeared large solvent peak in chromatogram, which may increase the difficulty of separation. Such as adefovir dipivoxil, a kind of nucleotide drugs, many residual solvents are introduced during the process of synthesis and these solvents cannot be headspace analyzed completely due to the reasons above. Such problems could be avoided by using ILs since ILs have shown notable properties such as great dissolubility, high thermal stability, and no vapor pressure so that organic solvents could be easily dissolved in ILs and the influence of traditional matrix medium in chromatogram could be decreased [6,7]. Thus, ILs were regarded as an ideal matrix medium for HS-GC to
∗ Corresponding author. Tel.: +86 311 86266025; fax: +86 311 86266069. E-mail addresses: [email protected], [email protected] (Y. Jiang). 1570-0232/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jchromb.2013.11.021
analyze residual organic solvents in pharmaceutical products and some applications of this had been reported in recent years [8–12]. Although ILs present many advantages for HS-GC, previous works suggested that not all of the ILs performed ideal headspace efficiency for different analytes. It is known that headspace efficiency could be characterized by logK (the logarithm of concentration radio between matrix medium and headspace). The lower value of logK represented that the higher headspace efficiency was obtained. Therefore, the ILs obtained the smaller value of logK which should be chosen during the process of experiment. However, no theoretical guidance was investigated to explain the choice of ILs during the process of HS-GC. The choice for ILs to analyze residual organic solvents in pharmaceutical products [8–12] still remained at a “trying-comparing-screening” step and these applications of ILs which will inevitably lead to the waste of time and cost. In order to avoid the time-consuming screening step, choosing suitable ILs has become a tough problem. Previously, our laboratory reported a general relationship model to explain the interactions between aliphatic alcohols and three ILs with methyl-imidazolium cations in static HS-GC [13]. The relationship between the characters of aliphatic alcohols (including polarity and boiling point) and logK established a helpful
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
61
Table 1 Instrumental conditions. Aprotic solvents Chromatograph Column Oven temperature program Split injection Column flow FID temperature
Alkylogens
SP 3420 gas chromatograph Phenomenex ZB-1 fused silica capillary column (60 m 0.53 mm I.D. phenomenex, Guangzhou, China) with 5.00 m film thickness (100% dimethylpolysiloxane) 30 ◦ C, 8 ◦ C /min to 120 ◦ C, 10 min 50 ◦ C, 2 ◦ C /min to 70 ◦ C, 5 ◦ C /min to 125 ◦ C, 5 min 150 ◦ C, splitless 2 mL/min N2 , constant flow mode 300 ◦ C
guidance of using ILs as matrix medium to analyze solvents by HSGC. It had improved that complex interactions might be existed between organic solvents and ILs, and the best model of the relationship between logK and Bp (boiling point), logPO/W (polarity) for organic solvents was identified. However, the influences by the characters (including polarity and boiling point) of organic solvents were investigated above but these structured characters of ILs were not been explored before. ILs were ionogenic compounds and also performed some characters of organic solvents. According to our prediction, logK might be influenced not only by the characters of organic solvents, but also by the characters of ILs. Thus, the influences by the characters of ILs should be further studied. In this paper, the relationship between the polarity of different groups of ILs and logK is investigated to provide a guidance to choose an ideal ionic liquid for each kind of organic solvents as the matrix medium during the process of HS-GC. Two types of organic solvents, alkylogens and aprotic solvents, were chosen to analyze the partition coefficients with ILs. Alkylogens are venomous solvents whose concentrations in pharmaceuticals should be limited [14]. Although alkylogens performed great headspace efficiency in traditional matrix medium, pharmaceutics containing these residual solvents do have poor solubility in traditional matrix medium, which increase difficulty to headspace analyze [6,7]. Consequently, ILs can be used to take place of the traditional headspace solvent. It was found that halogen atoms in organic alkylogen can absorb the cations of ILs due to the high electronegativity of halogen atoms. Similarly, anions of ILs would reject this kind of solvents because of the similar electrostatic ionic interaction. The aprotic solvents have high electronegativity and boiling point. Traditional matrix medium, such as water, is hard to analyze this kind of solvents because of their low boiling point. Meanwhile, high polarity of water might create strong interactions with organic solvents having high electronegativity. It might decrease the headspace efficiency. Thus, ILs are ideal matrix medium by HS-GC for this kind of solvents. In summary, the interactions between two kind of organic solvents mentioned above and ILs were investigated. The regression equations were obtained of ionic liquid–gas partition coefficients for these two kinds of solvents to explain the relationship between ILs and organic solvents. The results were expected to be a helpful guidance for the selection of ILs as matrix medium for analysis of residual solvents during the process of HS-GC.
Table 2 The structure properties of the ILs used in this study. Ionic liquids
logP1
logP2
logP3
[BMIM][BF4 ] [EMIM][BF4 ] [HMIM][BF4 ] [HMIM][PF6 ] [HMIM][Tf] [HMIM][T2 N] [HMIM][Cl] [HMIM][Br] [HMIM][I] [EMIM][T2 N] [EMPY][T2 N] [EMPI][T2 N] [EMPYR][T2 N]
−1.32 −1.32 −1.32 −1.32 1.18 1.18 0.32 0.65 0.65 1.18 1.18 1.18 1.18
−0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 −0.67 0.70 0.60 0.18
1.33 2.17 3.00 3.00 3.00 3.00 3.00 3.00 3.00 2.17 2.17 2.17 2.17
(DMSO) (HPLC-grade) was from DIKMA. ILs obtained from Shanghai Cheng Jie Chemical Co. Ltd. (Shanghai, China). Table 2 lists Thirteen ILs and their properties. 2.2. Measurement of the partition coefficients The procedure of experiment was the same with Grant Von Wald [8]. Stock solutions were prepared by dissolving four analytes with DMSO in a 10 mL volumetric flask. Working solutions at five different concentrations were obtained by serial dilutions of stock solutions. 2 L analyte solutions was added to a 12.5 mL headspace vials and then the vials were tightly sealed using PTFE coated silicon rubber septa and aluminum crimp caps and equilibrated at the temperature of 100 ◦ C in an oven for 30 min. After homogenization, each sample was analyzed in triplicate with a 0.5 mL injection volume. For all analytes, a linear relationship was obtained. Sample vials were prepared by adding 1 mL ionic liquid to a 12.5 mL headspace vial. Figs. 1 and 2 show the headspace gas chromatogram of a 2 L standard solution prepared with [EMIM][BF4 ]
mV
3
560
2
480 400
2. Materials and methods 2.1. Reagents and materials Table 1 gives the instrumental conditions used for all measurements employed in this study. Trichloromethane, 1,1,2tichloroethylene, 1,1,1-trichloroethane, 1,2-dichloroethane, acetonitrile, acetone, acetoacetate, tetrahydrofuran (TMF), pyridine, and dioxane were analytical reagent grade and purchased from Tianjin Chemical Reagents Ltd. (Tianjin China). Dimethylsulfoxide
1
320 160
4
80 0
2
4
6
8
10
12
14
16
Fig. 1. The chromatogram of alkylogens. (1. Trichloromethane; Dichloroethane; 3. 1,1,1-Trichloroethane; 4. 1,1,2-Tichloroethylene).
min 2.
1,2-
62
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
where, C0 is the concentration of analyte in the standard solution and V0 is the volume of calibration solution added to the vial (2 L), Cg is the measured concentration in the gas phase by the linear relationship, V is the vial volume (12.5 mL), and Vl is the volume of liquid in the vial (1 mL).
5
mV 560 480
4
400
2 3
3. Results and discussions
6
320 160
1
80 0
2
4
6
8
10
12
14
16
min
Fig. 2. The chromatogram of the aprotic solvents. (1. Acetone; 2. Acetonitrile; 3. Tetrahydrofuran; 4. Dioxane; 5. Acetoacetate; 6. Pyridine).
matrix, as representative of all matrix medium on a ZB-1 column at the equilibrium temperature of 100 ◦ C. During the process of HS-GC, the partition coefficient, K, was described by the following equation: K=
Cl Cg
(1)
where, Cl is the concentration of analyte in the liquid phase, and Cg is the concentration in the gas phase. In this report, K was calculated as: K=
(C0 V0 − Cg (V − Vl ))/Vl Cl = Cg Cg
(2)
Thirteen ILs contained different cations or anions and two kinds of organic solvents were chosen in this study. The partition coefficients of different solvents between each ionic liquid and gas phase were calculated. The results of logK were listed in Tables 3 and 4. The relationship studies between ILs and organic solvents were carried out with the SPSS (Statistical Package for the Social Sciences) version 13.0 program. Previous works have revealed that because of the interaction between ILs and organic solvents, logK was influenced by the polarity and boiling point of organic solvents. In this study, the polarity of ILs which might also influence logK was investigated. The parameter which could demonstrate the whole polarity of ionic liquid was failed to be found. As we all know, ILs are composed by different organic cations and organic or inorganic anions. Cations of ILs also contained of parent nucleus and alkyl chains. The parameters which could demonstrate the anions and cations of ILs were used to investigate the relationship between logK. Thus, logP1 was used to demonstrate the polarity of anions which obtained by the polarity of natrium salt for different anions. logP2 was used to represent the polarity of nucleus obtained by the polarity of heterocyclic compounds. logP3 was used to express the polarity of alkyl chains which obtained by the polarity of alkanes.
Table 3 Values of logK for alkylogens. logK
[EMIM][BF4 ] [BMIM][BF4 ] [HMIM][BF4 ] [BMIM][T2 N] [BMPI][T2 N] [BMPY][T2 N] [BMPYR][T2 N] [HMIM][PF6 ] [HMIM][Cl] [HMIM][Br] [HMIM][I] [HMIM][Tf] [HMIM][T2 N]
Trichloromethane
Dchloroethane
Tichloroethane
Tichloroethylene
2.26 1.52 1.09 2.58 0.743 0.881 1.42 1.38 1.53 1.51 1.70 1.66 1.38
1.34 0.573 0.0750 1.88 0.390 0.490 1.05 1.64 1.75 1.78 1.90 1.92 1.64
1.81 1.77 0.465 2.36 0.531 0.665 1.20 1.53 1.63 1.63 1.79 1.82 1.53
2.02 1.50 1.07 1.89 0.115 0.240 0.810 1.69 1.76 1.73 1.92 1.99 1.69
Table 4 Values of logK for the aprotic solvents. logK
[EMIM][BF4 ] [BMIM][BF4 ] [HMIM][BF4 ] [BMIM][T2 N] [BMPI][T2 N] [BMPY][T2 N] [BMPYR][T2 N] [HMIM][PF6 ] [HMIM][Cl] [HMIM][Br] [HMIM][I] [HMIM][Tf] [HMIM][T2 N]
Ttrahydrofuran
Doxane
Pridine
Aetonitrile
Aetone
Ethyl acetate
0.938 1.33 1.72 1.70 1.38 1.44 1.52 1.25 1.26 1.30 1.31 1.31 1.31
0.401 0.862 1.21 1.01 1.69 0.750 0.810 1.95 1.93 1.97 1.93 2.00 2.01
0.612 1.01 1.44 1.95
1.38 1.82 2.22 1.06 0.750 0.770 0.840 1.24 1.27 1.27 1.27 1.29 1.31
1.42 2.87 2.30 1.10 0.780 0.790 0.880 1.41 1.45 1.45 1.45 1.47 1.47
0.710 1.15 1.54 1.95 1.63 1.66 1.76 1.17 1.18 1.20 1.21 1.25 1.24
1.67 1.75 0.914 0.910 0.960 0.960 0.960 0.950
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
63
1,2-Dichloroethane
Trichloromethane 2
2
1.5
1.5
logK
1
logK
1 0.5
0.5
0
0
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [Tf] [T2N] [Cl] [BF4] [Br] [I] [PF6]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
1,1,2-Tichloroethylene
1,1,1-Trichloroethane 2
2
1.5
1.5
logK
logK
1
1
0.5
0.5 0
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [I] [BF4] [PF6] [Tf] [T2N] [Br] [Cl]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
Fig. 3. The influence by the anions of ILs to logK.
1,2-Dichloroethane
Trichloromethane 3
3 2.5
2.5
2
logK
1.5
2
logK
1.5
1 1 0.5 0.5
0 [BMIM] [T2N]
[BMPI] [T2N]
[BMPY] [T2N]
[BMPYR] [T2N]
0 [BMIM] [T2N]
1,1,1-Trichloroethane
[BMPI] [T2N]
[BMPY] [T2N]
[BMPYR] [T2N]
1,1,2-Tichloroethylene
3
3
2.5
2.5 2
2
logK
1.5
logK
1.5 1
1
0.5 0.5
0 0
[BMIM] [T2N] [BMIM] [T2N]
[BMPI] [T2N]
[BMPY] [T2N]
[BMPI] [T2N]
[BMPY] [T2N]
[BMPYR] [T2N]
[BMPYR] [T2N]
Fig. 4. The influence by the parent nucleus of ILs to logK.
3.1. The partitioning relationship between alkylogens and ILs Figs. 3–5 revealed that the longer alkyl chains, the lower polarity of parent nucleus, and the higher polarity of anions performed the higher headspace efficiency for alkylogens. As high electronegativity halogen atoms were contained in this kind of organic solvents, the cations of ILs could be absorbed by it. For the same reason, anions of ILs would reject the solvents because of the similar electronegativity. These two kinds of relationship might influence the headspace efficiency globally. Firstly, the influences for anions of ILs were investigated. The correlation between logP1 and logK was inspected and the result was shown in Table 5. The correlation equation represented that the headspace efficiency was increasing with the increasing of logP1 . Fig. 3 also showed that the anions of ILs performed little influence for logK though great correlation between logP1 and logK.
From Fig. 3 it could be observed that the headspace efficiency was increased with the decreasing of the polarity of anions but anions had little affect in the headspace efficiency. Fig. 4 and Fig. 5 showed that cations of ILs that included parent nucleus and alkyl chains performed an important factor which would affect the headspace efficiency. It revealed that the longer alkyl chains, the Table 5 Equations and some coefficients. Organic solvents
Equations
Multiple correlation coefficients (R)
Alkylogens
logK = 1.455 − 0.322 logP1 logK = 1.33 − 1.26 logP2 logK = 2.77 − 0.707 logP3 logK = 1.065 + 0.455 logP1 − 1.15logP2 − 0.152 logP3
0.950 0.939 0.878 0.881
64
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
1,1,1-Trichloroethane
1,2-Dichloroethane 2.5
2.5
2
2
logK
1.5
logK
1.5
1
1
0.5
0.5 0
0 [EMIM][BF4]
[BMIM][BF4]
[EMIM][BF4]
[HMIM] [BF4]
[BMIM][BF4]
[HMIM] [BF4]
Trichloromethane
1,1,2-Tichloroethylene 2.5
2.5
2
2
logK
1.5
logK
1.5 1
1 0.5
0.5
0
0
[EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
Fig. 5. The influence by the alkyl chain of ILs to logK.
Acetoacetate
Pyridine
2.5
2.5
2
2
1.5
logK
1.5
1
1
0.5
0.5
0
logK
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
TMF
Acetone
2.5
2.5
2
2
1.5
logK
1.5
1
1
0.5
0.5
0
logK
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
Acetonitrile
Acetone
2.5
2.5
2
2
1.5
1.5
logK
1
1
0.5
0.5
0
logK
0 [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
[HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [HMIM] [BF4] [PF6] [Tf] [T2N] [Cl] [Br] [I]
Fig. 6. The influence by the anions of ILs to logK.
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
Acetoacetate
65
Pyridine
2
2
1.6
1.6
1.2
logK
1.2
0.8
0.8
0.4
0.4
0
logK
0 [BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
[BMIM] [T2N] [BMPI] [T2N]
TMF
[BMPY] [T2N] [BMPYR] [T2N]
Acetone
2
2
1.6
1.6
1.2
logK
1.2
0.8
0.8
0.4
0.4
0
logK
0 [BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
[BMIM] [T2N] [BMPI] [T2N]
Acetonitrile
[BMPY] [T2N] [BMPYR] [T2N]
Dioxane
2
2
1.6
1.6
1.2
logK
1.2
0.8
0.8
0.4
0.4
0
logK
0 [BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
[BMIM] [T2N] [BMPI] [T2N]
[BMPY] [T2N] [BMPYR] [T2N]
Fig. 7. The influence by the parent nucleus of ILs to logK.
lower polarity of parent nucleus obtained the lower logK, that is, higher headspace efficiency. Finally, multivariable linear regression was carried out between logK and all the structure parameters of ILs, including logP1 , logP2 , logP3 . The results were listed in Table 5 and the correlation was great. It indicated that all the structure parameters of ILs presented influences for headspace efficiency. However, the parameters of organic solvents did not improve the correlation. It might have been caused by the similar structures of this kind of organic solvents, which performed similar interactions with ILs. All the results above implied the different groups present different effect to headspace efficiency. A whole ionic liquid was composed by these different groups. Thus, the correlation among all the characters of ILs, including logP1 , logP2 , and logP3 , with logK was researched. The result revealed that this kind of ILs involved long alkyl chain gain higher headspace efficiency and the best headspace solvents gained from this model were agreed with the choices literatures reported [8,9]. Jiang used to determine the residual solvents such as dichlormethane and trichlormethane by HS-GC–MS [9]. [BMIM][BF4 ] was used as the matrix medium in experiment. The result implied that if the ILs such as [BMIM][BF4 ] which involved longer alkyl chain was chosen as the matrix medium, this kind of organic solvents gained high headspace efficiency. Grant Von Wald [8] determined partition coefficients between a series of organic solvents and ILs. The result showed that n-heptane gained lower partition coefficients in pyridine ILs than in imidazole. The polarity of pyridine is lower than imidazole so that the point
headspace efficiency would be higher when the lower polarity for the parent nucleus of ILs could be improved.
3.2. The partitioning relationship for aprotic solvents The shorter alkyl chains and the lower polarity of parent nucleus make the higher headspace efficiency for aprotic solvents. It had been proofed that the interactions between ILs and solvents might affect the headspace efficiency above. In this part, the organic solvents which might form hydrogen bond with ILs were investigated in thirteen ILs listed in Table 2. Firstly, the influences for anions of ILs were investigated. The correlation between logP1 and logK was failed to be found. From Fig. 6 it can also be observed that anions rarely affect the headspace efficiency and it might have been caused by the little volume of anions. The correlation between logP2 and logK was also failed to be found. Fig. 7 implied that the solvents, including acetonitrile, acetone, acetoacetate, tetrahydrofuran, and pyridine, the lower polarity for parent nucleus gained higher headspace efficiency. However, dioxane was not following the role above. It was probably caused by two atoms which involved in dioxane molecule might form hydrogen bond with ILs. Consequently, the headspace efficiency was influenced. The correlation coefficient for logK and logP3 was 0.738. It implied that the correlation existed between logK and logP3 . Fig. 8 implied that the longer alkyl chains gained higher headspace efficiency. It was caused by the reason that the longer alkyl chains,
66
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
TMF
Acetone
3
3
2.5
2.5
2
2
logK
1.5 1
1
0.5
0.5
0
logK
1.5
0 [EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
Acetonitrile
[BMIM][BF4]
[HMIM] [BF4]
Dioxane
3
3
2.5
2.5
2
2
logK
1.5
logK
1.5
1
1
0.5
0.5
0 [EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
Acetoacetate
[BMIM][BF4]
[HMIM] [BF4]
Pyridine
3
3
2.5
2.5
2
2
logK
1.5 1
1
0.5
0.5
0
logK
1.5
0 [EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
[EMIM][BF4]
[BMIM][BF4]
[HMIM] [BF4]
Fig. 8. The influence by the alkyl chain of ILs to logK.
the more hydrogen atoms in ILs made the larger interaction with organic solvents. Thus, the headspace efficiency was decreased. In order to investigate the interactions between ILs and organic solvents comprehensively, the correlation for logK and all logP1 , logP2 , and logP3 was investigated. The correlation coefficient for them was 0.278. It implied that no correlation for them. Though the tendency of logK for this kind of organic solvents was observed, there was no correlation of them. The probable reasons were that different structures for this kind of organic solvents lead the difference of headspace partitioning behavior and also owing to the other interactions between organic solvents and ILs. Liu used to determine acetonitrile and DMF in pharmaceuticals and compare the headspace efficiency between [Bmim][BF4 ] and [Hmim][BF4 ] [10]. The results revealed that the headspace efficiency of [Bmim][BF4 ] was higher than [Hmim][BF4 ]. Consequently, these kinds of ILs which involves longer alkyl chains might gain higher headspace efficiency. In summary, the relationship between thirteen ILs and two kinds of organic solvents was investigated in this study, and conclusions were summarized as follows:
1. The anions of ILs had less efficiency than cations for both kinds of organic solvents. It was caused by anions small space for anions and it had little efficiency for headspace partitioning. 2. The cations for ILs influenced the headspace partitioning a lot, especially the alkyl chains of cations. There were different influences for these two kinds of solvents. The longer the alkyl
chains, the lower headspace efficiency for protic solvents which involved ability of pro-proton but the higher headspace efficiency for alkylogens. The reason was that the polarity of cations was decreased with the increase of alkyl chains which might decrease the interaction between alkylogens and ILs and then the headspace efficiency might increase the headspace efficiency. However, protic solvents which involved high electronegativity atoms, the number of hydrogen were increased with the increase of alkyl chains. Thus, the interactions between them decrease the headspace efficiency. With the decreasing of the polarity of parent nucleus, the headspace efficiency became higher for alkylogens. However, the polarity of parent nucleus performed complex influences for protic solvents which involved ability of pro-proton. For acetonitrile, acetone, acetoacetate, tetrahydrofuran, and pyridine, the headspace efficiency became higher with the decreasing of the polarity of parent nucleus. However, dioxane was not following the role above. It was probably caused by the complex structures and interactions of them and it should be further studied. 4. Application of the model for the rapid analysis of real samples After the interaction between organic solvents and ILs was explored, a sensitive method was established to determine the
M. Ni et al. / J. Chromatogr. B 945–946 (2014) 60–67
organic residual solvents in ketoconanzale. 0.01 g of ketoconanzale were put in a 10 mL headspace vial, followed by spiking 1 mL of ILs. The sample solution was maintained at the equilibration temperature of 110 ◦ C for 30 min and a volume of 1 mL headspace gas was directly injected into GC for analysis. For ethyl, the headspace efficiency of ILs increased with the growth of cation alkyl chain. For ethyl, DMF and DMSO, the headspace efficiency of ILs decreased with the growth of cation alkyl chain. However, the anion of ILs has minor impact on headspace efficiency. Ultimately, [Bmim][PF6 ] was chosen as the best headspace solvent. An excellent separation of ethanol, dichloromethane, ethyl acetate, butyl alcohol, pyridine, DMF, and DMSO was achieved. The calibration curve was linear in the range of 1.25–200 mg L−1 for ethanol, 1.50–24.0 mg L−1 for dichloromethane, 12.5–200 mg L−1 for ethyl acetate, 12.5–200 mg L−1 for butyl alcohol, 0.500–8.00 mg L−1 for pyridine, 2.20–3.52 mg L−1 for DMF, and 12.5–200 mg L−1 for DMSO. All the average recoveries were limited in 89.8–98.2% and RSD were less than 4.0%. 5. Conclusions A relationship study of ILs–gas phase partition coefficients for two kinds of organic solvents (alkylogens and aprotic solvents) was evaluated in this paper initially. The partitioning behaviors of organic solvents were influenced by the polarity of ILs and the cations preformed more influences to headspace efficiency than anions of ILs. As the interactions between the organic solvents and ILs were initially studied, this ILs model could offer a guidance to
67
determine the organic residual solvents in pharmaceutical product by HS-GC and rapid quantitative analysis of real samples possible. Acknowledgments This work was financially supported by the National Natural Science Foundation of China (grant no. 81041087), Hebei Natural Science Foundation of China (grant no. C2009001069). References [1] L.J. He, X.L. Luo, H.X. Xie, C.J. Wang, X.M. Jiang, K. Lu, Anal. Chim. Acta 655 (2009) 52–59. [2] C.R. Contreras, C. Dominguez, J.M. Bayona, J. Chromatogr. A. 1261 (2012) 164–170. [3] L. Vallecillos, E. Pocurull, F. Borrull, Talanta 99 (2012) 824–832. [4] L. Vidal, A. Chisvert, A. Canals, A. Salvador, Talanta 81 (2010) 549–555. [5] I. Dominguez, N. Calvar, E. Gomez, A. Dominguez, Procedia Eng. 42 (2012) 1597–1605. [6] D. Wei, A. Ivaska, Anal. Chim. Acta 607 (2008) 126–135. [7] Y.X. Fan, J.Q. Qian, J. Mol. Catal. B: Enzym. 66 (2010) 1–7. [8] G.V. Wald, D. Albers, H. Cortes, T.M. Cabe, J. Chromatogr. A. 1201 (2008) 15–20. [9] H.L. Jiang, H.Y. Shen, Z.G. Le, S.C. Zhou, Chin. J. Anal. Chem. 34 (2006) 1027–1029. [10] F.H. Liu, Y. Jiang, J. Chromatogr. A. 1167 (2007) 116–119. [11] K. Urakami, A. Higashi, K. Umemoto, M. Godo, J. Chromatogr. A. 1057 (2004) 203–210. [12] G. Laus, M. Andre, G. Bentivoglio, H. Schottenberger, J. Chromatogr. A. 216 (2009) 6020–6023. [13] X.W. He, L.L. Lei, Y. Jiang, J.M. Li, M.P. Ni, Chromatographia 74 (2011) 157–161. [14] Editorial Committee of the Pharmacopoeia of the People’s Republic of China, Pharmacopoeia of the People’s Republic of China (2nd Part), Chemical Industry Press, Beijing, 2010, Appendix: 63.
