Salt Effect on the Isobaric Vapor-Liquid Equilibrium Study of Binary Mixtures H2O-NMP (N-Methyl-2-pyrrolidone).

To study the salt effect of recovering N-methyl-2-pyrrolidone (NMP) from the waste liquid produced in the polyphenylene sulfide (PPS) synthesis process, this study presents vapor-liquid equilibrium (VLE) measurement and correlation for water + NMP, water + NMP + lithium chloride, and water + NMP + sodium chloride at p = 101.3 kPa. The salt effect is discussed and the salts follow the order of lithium chloride > sodium chloride. The NRTL model was used for the correlation with binary parameters of water + NMP, water + NMP + lithium chloride, and water + NMP + sodium chloride. The correlation showed good agreement with experimental data; root-mean-square deviations are less than 0.48 K for the equilibrium temperature and 0.005 for the vapor-phase mole fraction of water.

Introduction

N-Methyl-2-pyrrolidone (NMP), a nitrogen heterocyclic compound, is a colorless liquid with slight ammonia flavor. The boiling point, flash point, and pH of NMP are 204 °C, 95 °C, and 7–9, respectively, which show its weak alkalinity. It can be mixed with water in any ratio and can be completely mixed with most solvents (ethanol, acetaldehyde, ketone, aromatic hydrocarbon, etc.). NMP is also a kind of nonproton transfer solvent, which has low viscosity, strong polarity, low volatility, little toxicity, almost no corrosion, strong biodegradation, good chemical stability, and thermal stability.[1,2] Therefore, NMP with the above excellent properties is a very widely used organic solvent.

NMP is mainly used in many industries, such as petrochemical industry, pharmaceutical industry,[2,3] pesticide,[4] dye,[5,6] and lithium-ion battery. It is widely used in extraction of aromatics, purification of olefins, etc.,[7] and also used in the production of polyphenylene sulfide, polyamide, and other polymer engineering plastics,[8,9] as well as insulation materials,[10,11] pigment and detergent,[12] etc. Polyphenylene sulfide (PPS) is a special engineering plastic with high added value and application prospect in the world,[13,14] known as the sixth largest engineering plastic, with ultrahigh cost performance. At present, the synthetic routes of PPS on the market mainly include the sodium sulfide method (Phillips method),[15−17] sulfur solution method,[18] and hydrogen sulfide method,[19] among which the sodium sulfide method is widely used. In this method, p-dichlorobenzene and sodium sulfide containing crystal water are used as raw materials, and sodium hydroxide, sodium chloride, lithium chloride, and other catalysts and additives are added in the reactor. NMP, which can promote nucleophilic reaction, is often used as a solvent to obtain a high-molecular-weight polymer. Although the yield of this method is high, the industrial wastewater often contains a lot of lithium chloride, sodium chloride, and solvent NMP, and NMP and lithium chloride are expensive as a solvent and catalyst. Therefore, the recovery of NMP and lithium chloride is an important factor restricting the economic efficiency of the industrial process of PPS. Generally, for the high-salt waste liquid produced in the PPS synthesis process, the effect of recovering NMP by the distillation method is better, but after the waste liquid is recovered by distillation, some viscous liquid remains in the tower bottom,[20,21] which greatly affects the recovery efficiency and also causes pollution to the tower kettle.

Salt effect is a phenomenon where the composition of the vapor phase in equilibrium in a binary solution usually changes by adding a salt with volatile components to the system due to interactions between the salt and solvent components.[22] To recover solvent NMP from industrial wastewater of special engineering plastics PPS, vapor–liquid equilibrium data for the systems water + NMP containing salts are necessary. Li et al.[23] described the measurement of the equilibrium liquid composition and boiling points of an NMP–water binary system at 760 mmHg with a modified Washburn ebulliometer. Gupta and Rawat[24] studied isobaric binary and ternary vapor–liquid equilibria of N-methylpyrrolidone with water and toluene at 760 mmHg. The liquid-phase splitting for a ternary system of water + NMP + 1-pentanol can be enhanced by adding the same percentage of salts (sodium chloride, potassium chloride, or potassium acetate) and the influence follows the order of sodium chloride > potassium chloride > potassium acetate.[25] As far as we know, the salt (lithium chloride and sodium chloride) effect on isobaric VLE for the system of NMP + water has not been reported.

In this study, the isobaric vapor–liquid equilibrium data for the systems water + NMP, water + NMP + lithium chloride, and water + NMP + sodium chloride were determined at 101.3 kPa, and the thermodynamic consistency of the measured VLE data was checked by the van Ness test.[26] The influence of different salts on the vapor–liquid equilibrium of the system is explored. Meantime, the VLE data were correlated by the nonrandom two-liquid model (NRTL).

Experimental Section

Chemicals

N-Methylpyrrolidone, sodium chloride, and lithium chloride were all purchased from Sinopharm Chemical Reagent Co., Ltd. The water content of the salts was checked by Karl Fisher titration. NMP was used without further purification. Salts were desiccated under a vacuum for at least 24 h. The deionized water (conductivity, <1.0 μs·cm–1) was made in our laboratory by the ultrapure water machine (Nanjing Miaozhiyi Electronic Technology Co., Ltd). Ethanol, acetone, and ether were all purchased from Sinopharm Chemical Reagent Co., Ltd., and used without further purification. Helium was provided by Nanjing Tianze Gas Co., Ltd. All reagents are of analytical grade except helium (industrial grade). Specifications of the chemicals are listed in Table .

Table 1

Specifications of Experimental Reagents

component	grade	source	mass fraction	purification method	final water mass fraction	analysis method
N-methylpyrrolidone	analytical	Sinopharm Chemical Reagent Co., Ltd.	0.995	none		GC
lithium chloride	analytical	Sinopharm Chemical Reagent Co., Ltd.	≥0.99	vacuum desiccation	0.0008	KFa
sodium chloride	analytical	Sinopharm Chemical Reagent Co., Ltd.	≥0.99	vacuum desiccation	0.0008	KFa

It can detect as little as 10 μg of water and titrate 1 mg with ±0.2% accuracy in less than a minute.

Apparatus and Procedures

Accurate measurements of VLE data for (NMP + H2O), (NMP + H2O + LiCl), and (NMP + H2O + NaCl) were implemented in a modified Rose-type recirculating equilibrium still (CE-3, Tianjin Beiyang Tongchuang Distillation Equipment Co., Ltd.) under atmospheric pressure. The structure of the modified Rose-type recirculating equilibrium still is shown in Figure , and the composition of the whole of vapor–liquid equilibrium data analyzer is listed in Table . The liquid raw material added in the vapor–liquid equilibrium still was heated slowly by the thermocouple to boiling, and the return flow rate of vapor-phase condensation was controlled at about 30–40 drops/min. The evaporated gas phase rose through the vapor-phase circulating pipe and flowed downward after condensation in the spherical condenser pipe, and there is a certain amount of liquid at the gas-phase sample connection. The rest of the condensate flowed into the kettle through the liquid-phase circulating pipe, forming a vapor–liquid two-phase double circulating system. After boiling for 1.5–2 h, the temperature remained unchanged. It is considered that the vapor–liquid phase has reached equilibrium, and the equilibrium temperature was recorded, and gas and liquid sample connection shall be sampled by syringes. The closed chromatographic bottle was placed in the center of the condensed water and cooled down to the normal temperature quickly, and then the collected samples were analyzed by gas chromatography. By changing the composition of raw materials and repeating the above steps, a series of experimental data of vapor–liquid two-phase equilibrium can be obtained.

Figure 1

CE-3 vapor–liquid equilibrium still. (1) Spherical condenser, (2) condensate, (3) gas sample connection, (4) heating couple, (5) liquid sample connection, (6) thermometer, and (7) vapor–liquid riser.

Table 2

Composition of the Vapor–Liquid Equilibrium Data Analyzer

system	composition	specification
vapor–liquid equilibrium still	glass vapor–liquid equilibrium kettle	CE-3
	glass condenser
	high-temperature electric heating rod	Ferrule-type fixing
temperature display control system	temperature display control meter	accuracy: 0.1% FS
	thermocouple (temperature control)	K type
	solid-state relay	rated current: 20 A
	temperature digital display meter	accuracy: 0.1% FS
	thermocouple (temperature measurement)	Pt100
pressure display system	pressure digital display meter	0–0.1 MPa
	pressure sensor	4–20 mA

Analysis

Gas chromatography (GC-2014, Shimadzu) was adopted to measure the compositions of NMP and H2O. The GC-2014 was equipped with a capillary column of Porapak Q and a hydrogen flame ionization detector (FID), with helium as a carrier gas (>99.999% purity, 100 mL·min–1). The detection conditions for the system containing NMP and H2O were set as follows: the temperature of the injector was set as 523.15 K, the column of the GC was maintained at 483.15 K, and the detector was set at 523.15 K. At least three measurements were taken for all the samples to guarantee the reliability of the data. The mean value was adopted when the deviation of three times was not more than 0.001.

An external standard method via a GC method was established. As shown in Figure , the Pearson correlation coefficient r is 0.99998 for the corrected H2O + NMP system standard curve, which shows high accuracy. Through the standard curve, the mass relation of each component can be calculated by the relation of peak area of each component.

Figure 2

Standard curve of the H2O–NMP system.

Results and Discussion

Validation of the Apparatus

The isobaric VLE data for the binary system NMP and water at a pressure of 101.3 kPa were determined to verify the reliability of the equilibrium still, which are listed in Table . For comparison, the reference data reported by Li et al.[23] and Gupta and Rawat[24] are plotted in Figure .

Table 3

Experimental VLE Data of Binary System of H2O and NMPa

	liquid (mole fraction)		gas (mole fraction)			liquid (mole fraction)		gas (mole fraction)
T (°C)	x1	x2	y1	y2	T (°C)	x1	x2	y1	y2
100.02	1.000	0.000	1.000	0.000	112.80	0.634	0.366	0.985	0.015
100.82	0.965	0.035	0.998	0.002	116.00	0.531	0.469	0.976	0.024
101.60	0.924	0.076	0.997	0.003	122.20	0.477	0.523	0.961	0.039
101.80	0.923	0.077	0.998	0.002	126.40	0.390	0.610	0.954	0.046
101.85	0.928	0.072	0.999	0.001	133.10	0.311	0.689	0.930	0.070
102.10	0.902	0.098	0.997	0.003	140.00	0.243	0.757	0.909	0.091
102.70	0.908	0.092	0.998	0.002	144.85	0.197	0.803	0.895	0.105
104.20	0.864	0.137	0.996	0.004	150.00	0.163	0.837	0.857	0.143
104.68	0.804	0.196	0.993	0.007	156.10	0.114	0.886	0.827	0.173
104.69	0.994	0.006	0.995	0.005	163.20	0.093	0.907	0.775	0.225
105.10	0.802	0.198	0.992	0.008	170.50	0.057	0.943	0.687	0.313
105.85	0.785	0.215	0.994	0.006	175.50	0.035	0.965	0.611	0.389
106.25	0.755	0.245	0.991	0.009	180.80	0.023	0.977	0.543	0.457
106.70	0.787	0.213	0.995	0.005	189.50	0.012	0.988	0.402	0.598
107.12	0.747	0.253	0.990	0.010	196.60	0.002	0.998	0.252	0.748
107.92	0.724	0.276	0.994	0.006	198.50	0.002	0.998	0.129	0.871
109.20	0.714	0.286	0.988	0.012	199.30	0.000	1.000	0.000	1.000
110.00	0.698	0.302	0.988	0.012

Standard uncertainties u are u(x1) = u(y1) = 0.006, u(T) = 0.1 K, and u(P) = 1 kPa.

Figure 3

Comparison of H2O and NMP experimental and literature VLE data (black square, experimental; blue up-pointing triangle, literature 1;[23] pink star, literature 2[24]).

As shown in Figure , the experimental data of vapor–liquid phase equilibrium of water and NMP measured in this study are in good agreement with the data in the literature.[23,24] The difference is that the measurement range in this study is larger than that in the literature. Therefore, it can be concluded that the vapor–liquid two-phase condensation double cycle method coupled with the gas chromatography analysis method is suitable for the vapor–liquid equilibrium experiment, which also indicates that the apparatus is reliable.

Vapor–Liquid Equilibrium for Salt-Containing Systems

VLE Data of H2O (1) + NMP (2) + Lithium Chloride

The isobaric VLE data for the systems H2O (1) + NMP (2) + lithium chloride were determined at 101.3 kPa by keeping the mass fractions of lithium chloride nearly constant at 1, 3, and 5%, respectively, which are listed in Table .

Table 4

Experimental VLE Data of Different LiCl Mass Ratiosa

1% LiCl					3% LiCl					5% LiCl
T	x1	y1	ΔT	Δy1	T	x1	y1	ΔT	Δy1	T	x1	y1	ΔT	Δy1
100.02	1.000	1.000	0.34	0.000	100.02	1.000	1.000	0.34	0.002	100.02	1.000	1.000	0.33	0.000
102.00	0.952	0.998	0.21	0.002	103.00	0.945	0.997	0.05	0.002	104.10	0.945	0.996	0.23	0.003
103.20	0.907	0.995	0.22	0.001	104.20	0.91	0.996	0.08	0.002	105.90	0.906	0.994	0.42	0.001
104.90	0.863	0.994	0.33	0.002	106.80	0.852	0.994	0.32	0.005	108.40	0.846	0.992	0.32	0.005
107.50	0.793	0.993	0.15	0.004	109.50	0.778	0.992	0.16	0.002	112.00	0.771	0.987	0.18	0.003
111.60	0.676	0.986	0.12	0.007	115.00	0.673	0.982	0.45	0.007	118.90	0.667	0.980	0.23	0.008
121.00	0.504	0.965	0.16	0.012	120.30	0.592	0.97	0.23	0.004	126.40	0.552	0.964	0.16	0.002
127.30	0.408	0.949	0.13	0.003	126.80	0.505	0.955	0.35	0.004	132.90	0.487	0.949	0.28	0.003
136.30	0.311	0.924	0.44	0.010	136.20	0.382	0.932	0.76	0.001	142.00	0.391	0.919	0.12	0.005
142.10	0.256	0.893	0.15	0.002	144.30	0.296	0.906	0.43	0.004	151.80	0.282	0.871	0.11	0.002
144.90	0.243	0.882	0.15	0.004	150.00	0.266	0.879	0.88	0.004	157.00	0.253	0.827	0.31	0.001
154.30	0.163	0.831	0.21	0.008	160.80	0.198	0.834	0.56	0.007	162.50	0.212	0.797	0.33	0.007
161.20	0.129	0.783	0.38	0.006	168.00	0.139	0.779	0.62	0.003	163.10	0.201	0.778	0.24	0.009
167.80	0.098	0.725	0.25	0.009	172.90	0.118	0.733	0.51	0.004	164.70	0.200	0.777	0.11	0.003
173.00	0.082	0.698	0.45	0.004	177.80	0.103	0.674	0.65	0.002	168.60	0.157	0.722	0.16	0.007
180.20	0.057	0.614	0.20	0.001	180.50	0.078	0.622	0.34	0.005	173.10	0.128	0.681	0.32	0.003
186.30	0.043	0.527	0.41	0.002	186.10	0.06	0.555	0.65	0.002	179.10	0.100	0.598	0.15	0.002
192.80	0.026	0.432	0.31	0.006	190.00	0.047	0.497	0.52	0.002	185.50	0.068	0.529	0.14	0.004
197.00	0.006	0.321	0.25	0.005	193.00	0.035	0.457	0.16	0.002	189.80	0.041	0.469	0.31	0.003
199.00	0.002	0.252	0.32	0.004	197.20	0.02	0.393	0.67	0.004	193.00	0.031	0.405	0.23	0.004
200.90	0.003	0.181	0.14	0.006	201.10	0.005	0.307	0.32	0.004	196.00	0.022	0.380	0.22	0.005
202.30	0.002	0.114	0.18	0.002	202.90	0.004	0.224	0.19	0.004	197.30	0.019	0.350	0.21	0.003
204.70	0.002	0.027	0.23	0.004	204.10	0.002	0.132	0.78	0.001	199.50	0.015	0.289	0.15	0.004
204.90	0.000	0.000	0.35	0.000	205.00	0.002	0.051	0.55	0.000	202.60	0.008	0.217	0.16	0.003
					205.80	0.000	0.000	1.67	0.002	204.10	0.001	0.142	0.08	0.001
										207.30	0.001	0.057	0.44	0.007
										209.00	0.000	0.000	0.15	0.012

Standard uncertainties u are u(x1) = u(y1) = 0.006, u(T) = 0.1 K, and u(P) = 1 kPa.

VLE Data of H2O (1) + NMP (2) + Sodium Chloride

The isobaric VLE data for the systems H2O (1) + NMP (2) + sodium chloride were determined at 101.3 kPa by keeping the mass fractions of sodium chloride nearly constant at 1, 3, and 5%, respectively, which are listed in Table .

Table 5

Experimental VLE Data of Different NaCl Mass Ratiosa

1% NaCl					3% NaCl					5% NaCl
T	x1	y1	ΔT	Δy1	T	x1	y1	ΔT	Δy1	T	x1	y1	ΔT	Δy1
100.12	1.000	1.000	0.33	0.001	100.12	1.000	1.000	0.32	0.000	100.12	1.000	1.000	0.34	0.000
100.62	0.969	0.999	0.21	0.003	103.60	0.944	0.997	0.23	0.001	103.80	0.924	0.994	0.42	0.001
101.84	0.925	0.998	0.12	0.002	104.90	0.902	0.995	0.16	0.002	104.40	0.902	0.994	0.21	0.002
102.90	0.906	0.997	0.22	0.004	106.90	0.843	0.993	0.19	0.001	107.00	0.853	0.993	0.31	0.001
104.90	0.868	0.994	0.31	0.005	110.00	0.772	0.991	0.31	0.003	111.00	0.752	0.990	0.22	0.002
105.10	0.858	0.993	0.03	0.005	115.30	0.649	0.984	0.43	0.002	115.60	0.639	0.986	0.14	0.005
107.90	0.770	0.990	0.11	0.002	118.20	0.563	0.980	0.21	0.001	118.20	0.563	0.980	0.19	0.002
108.20	0.760	0.991	0.26	0.004	122.50	0.468	0.971	0.36	0.004	123.00	0.458	0.972	0.23	0.004
113.20	0.632	0.982	0.21	0.002	129.60	0.377	0.942	0.24	0.002	130.00	0.356	0.952	0.41	0.006
122.40	0.461	0.972	0.15	0.004	137.30	0.282	0.924	0.43	0.004	137.60	0.282	0.924	0.13	0.003
128.90	0.371	0.954	0.32	0.001	145.00	0.236	0.902	0.47	0.002	143.60	0.252	0.905	0.21	0.003
137.50	0.272	0.927	0.41	0.003	148.80	0.201	0.876	0.13	0.003	144.60	0.237	0.902	0.16	0.005
143.90	0.218	0.906	0.13	0.002	161.50	0.152	0.816	0.52	0.005	150.90	0.190	0.866	0.23	0.002
147.00	0.204	0.896	0.45	0.006	173.00	0.106	0.726	0.42	0.002	162.60	0.142	0.796	0.26	0.005
155.00	0.162	0.836	0.23	0.004	182.50	0.054	0.656	0.13	0.004	170.00	0.116	0.746	0.32	0.004
166.70	0.109	0.748	0.31	0.005	188.80	0.021	0.574	0.34	0.003	180.20	0.052	0.646	0.31	0.003
177.00	0.085	0.683	0.12	0.003	191.90	0.013	0.495	0.12	0.002	187.30	0.021	0.574	0.41	0.002
179.20	0.081	0.665	0.18	0.004	193.00	0.012	0.463	0.38	0.003	192.30	0.014	0.475	0.13	0.004
185.20	0.059	0.574	0.24	0.005	196.40	0.009	0.308	0.21	0.005	193.00	0.013	0.452	0.24	0.006
191.10	0.054	0.477	0.26	0.003	198.60	0.007	0.258	0.32	0.006	196.40	0.009	0.328	0.31	0.002
193.50	0.041	0.427	0.31	0.002	199.30	0.002	0.185	0.13	0.002	198.60	0.007	0.258	0.33	0.001
196.00	0.029	0.371	0.25	0.004	200.50	0.002	0.124	0.24	0.004	200.00	0.002	0.156	0.13	0.004
198.10	0.026	0.294	0.31	0.003	202.10	0.001	0.050	0.35	0.003	202.10	0.001	0.047	0.43	0.005
199.20	0.025	0.222	0.41	0.004	203.90	0.000	0.000	0.39	0.001	203.90	0.000	0.000	0.26	0.001
201.40	0.014	0.137	0.23	0.002
203.00	0.001	0.031	0.41	0.001
203.90	0.000	0.000	0.35	0.001

Standard uncertainties u are u(x1) = u(y1) = 0.006, u(T) = 0.1 K, and u(P) = 1 kPa.

Consistency Check

In this study, the reliability of experimental data was judged by the method of van Ness tests, which is based on the Gibbs–Duhem theory. To verify whether the experimental data can pass the van Ness thermodynamic test, the predicted values of mole fraction of the vapor phase and the equilibrium pressure for the binary NRTL model were compared with the experimental data.

A point-by-point test of van Ness consistency test was applied in this study to guarantee the elimination of unreliable experimental point, which may be neglected in the area test method. The van Ness test is expressed as follows:where N is the number of experimental data points, ycal represents the mole fraction of component i in the vapor phase calculated by the NRTL model, and yexp stands for the experimental mole fraction of component i in the vapor phase. The criterion of this test is that Δy must be less than 1.

By the van Ness test, the values of Δy are 0.433, 0.308, and 0.407 for the system of (NMP + water + lithium chloride) and 0.315, 0.271, and 0.304 for the system of (NMP + water + sodium chloride) at salt concentrations of 1, 3, and 5% (mass fraction), respectively. All the values of Δy are less than 1, confirming that the measured data passed the thermodynamic consistency of the point-by-point test. From the checking by the above method, we can conclude that the experimental VLE data are thermodynamically consistent.

Salt Effect on Vapor–Liquid Equilibria

The trend of VLE data of the water and NMP system with different LiCl mass ratios was studied and is shown in Figure .

Figure 4

VLE data of different LiCl mass ratios (circle, 0% LiCl; up-pointing triangle, 1% LiCl; down-pointing triangle, 3% LiCl; square, 5% LiCl).

The vapor–liquid equilibrium T–y–x diagram is shown in Figure . It can be found that the addition of LiCl has a certain impact on the T–y and T–x curves in the vapor–liquid equilibrium of water and NMP; that is, the T–y and T–x curves show an upward shift phenomenon, and the upward shift of both increases with the increase in the addition amount of LiCl. When the water content of vapor phase or liquid phase is fixed, the vapor–liquid equilibrium temperature increases with the increase in LiCl content. In terms of the separation degree of the system, the addition of LiCl will reduce the two-phase area of the vapor–liquid equilibrium of the water and NMP system, making the separation of the two components of the system more difficult. The reason is probably that the addition of LiCl as a salt to the system will have a certain impact on the vapor–liquid equilibrium of the system and change the relative volatility between components, that is, the so-called vapor–liquid equilibrium salt effect; second, LiCl will form complexes with NMP components in the system. On the one hand, it will take away part of NMP, making NMP reduced in the vapor–liquid equilibrium system. On the other hand, because the complex formed in the process has specific structural characteristics, that is, its thermal stability is strong, it may also have a certain impact on the vapor–liquid equilibrium of the system.

The trend of VLE data of the water and NMP system with different NaCl mass ratios was studied and is shown in Figure .

Figure 5

VLE data of different NaCl mass ratio (circle, 0% NaCl; up-pointing triangle, 1% NaCl; down-pointing triangle, 3% NaCl; square, 5% NaCl).

As can be seen from Figure , the addition of NaCl has a certain impact on the vapor–liquid equilibrium of water and NMP system, and it is more obvious when the equilibrium temperature is between 140 and 200 °C. Comparing Figures and 5, it can be found that the effect of NaCl addition on the system is less obvious than that of LiCl, and according to Figure , the vapor–liquid equilibrium curve of the system does not show a trend of obvious change with the increase in NaCl addition; that is, the vapor–liquid equilibrium of the system is almost not affected by the amount of NaCl addition. This also reflects the particularity of the influence of LiCl on the vapor–liquid equilibrium of NMP + H2O system and further explains the effect of the formation of complexes by adding salt on the vapor–liquid equilibrium of the system. Therefore, to recover solvent NMP from industrial wastewater containing lithium chloride of special engineering plastics PPS, it is necessary to change the vapor–liquid equilibrium of water + NMP containing lithium chloride by the decomplexation of the complexes caused by NMP and lithium chloride.

Calculation

The vapor–liquid phase equilibrium can be expressed as followswhere the Poynting factor , , and φ̂ associated with nonideality were all close to 1 since the pressure was low. x and y represent the mole fraction of component i in the liquid phase and vapor phase, respectively. is the saturation vapor pressure of pure component i, which was estimated by the Antoine expression.[27] Considering the nonideality of the liquid phase, eq can be simplified as

The saturation vapor pressure of pure component is calculated by the Antoine equation, which is given aswhere A, B, and C are the parameters for each component i, and Tmin and Tmax are the limits of the temperature range, which are listed in Table .

Table 6

Parameters of the Antoine Equation

component	A	B	C	Tmin	Tmax	literature
water	7.19621	1730.630	–39.724	log		(28)
NMP	14.65738	4112.28	–66.866	380.73	475.72	(29)

Correlations of the VLE Data

For the description of vapor–liquid equilibrium state of the nonideal system in this study, excess Gibbs function should be mentioned. This function is closely related to the nonideality of liquid, which can be expressed by eq .where G represents excess Gibbs function, and x and γ represent the mole fraction and activity coefficient of component i in the liquid phase, respectively.

On the basis of Wilson’s model, Renon and Prausnitz introduced Scott’s two-liquid model to treat the mixed solution as a two fluid and then established the NRTL (nonrandom two-liquid) model, namely, the ordered two-fluid model. Derivation of formula can be found anywhere in the literature,[30] and the expression of activity coefficient of binary system is shown as followswhereSince the NRTL model is frequently used to correlate the VLE data of the salt-containing systems,[31] in this work, the NRTL model is adopted to correlate the VLE data. The parameters are regressed from the VLE data by minimizing the following objective functionwhere and are the calculated and experimental activity coefficient of component i in the salt-containing system. The regressed parameters for all the systems are listed in Table .

Table 7

Regressed Parameters of NRTL

system	τ12	τ21
H2O(1)–NMP(2)	–5519.074	1128.327
H2O–NMP (1% LiCl)	–9075.977	3187.593
H2O–NMP (3% LiCl)	–6325.647	2161.272
H2O–NMP (5% LiCl)	–3985.238	2015.275
H2O–NMP (1% NaCl)	–9489.609	4145.941
H2O–NMP (3% NaCl)	–10089.067	3371.964
H2O–NMP (5% NaCl)	–9768.230	3309.599

The root-mean-square deviation (RMSD) for the temperature (T) and the mole fraction of the vapor phase (y1) are expressed as follows

The values of RMSD(y1) and RMSD(T) are listed in Table , which are less than 0.005 and 0.48 K, respectively. According to the calculated values of RMSD(y1) and RMSD(T), the NRTL model is suitable for the VLE calculation for the systems of water + NMP + salts.

Table 8

RMSD for the Equilibrium Temperature (T) and Mole Fractions of the Vapor Phase (y1) of the NRTL Model

		RMSD
salt	w (%)	y1	T (K)
lithium chloride	1	0.005	0.27
	3	0.004	0.48
	5	0.005	0.22
sodium chloride	1	0.003	0.26
	3	0.003	0.32
	5	0.003	0.28

Conclusions

The isobaric vapor–liquid equilibrium data for the systems NMP + water + lithium chloride and NMP + water + sodium chloride were determined at a pressure of 101.3 kPa. The consistency of the measured VLE data was checked by the van Ness test. Meanwhile, the NRTL model was adopted to correlate the VLE experimental data of the systems, and the interaction parameters of the NRTL model were also regressed. The correlated results were in agreement with the measured data. The salting-out effect of the salts follow the order of lithium chloride > sodium chloride, Therefore, to recover solvent NMP from industrial wastewater containing lithium chloride of special engineering plastics PPS, it is necessary to change the vapor–liquid equilibrium of water + NMP containing lithium chloride by the decomplexation of the complexes caused by NMP and lithium chloride.