1 Introduction

As typical protic ionic liquds (PILs), ammonium-based ILs are attracting more and more attention. On the one hand, they have general characteristics like other ILs such as low volatility, wide liquid range, good solubility with other solvents and are function adjustable according to different combinations of cations and anions. On the other hand, they also have many special properties, including low cost, high electrical conductivity, and ease of preparation through direct neutralization between corresponding acid and base, which can endow them excellent performance in many fields such as alkylation reactions [1], absorption of SO2 [2] and CO2 [3], extraction of soluble dyes [4], stability of proteins [5, 6], dissolution of cellulose [7], electrochemistry [8] and wastewater treatments [9]. Obviously, knowledge of interactions between ammonium based ILs and molecular solvents is essential for their applications in chemical and industrial processes. Some physiochemical properties of solutions containing ammonium based ILs, such as viscosity, density and conductivity, have been measured and the interactions between ILs and solvents have also been discussed [10, 11]. The molecular solvents could lead to great variations of physicochemical properties of ILs, which suggests; that the micro-structure of solutions can be changed since the properties depended on the structures. Recently it has been reported that nano-inhomogeneity structures can be generated in aqueous solutions of ILs [12, 13]. Also, a study [14] of the DMSO/trimethyl(butyl)ammonium bis((trifluoromethyl)sulfonyl)imide ([N4111] [Tf2N]) and DMSO/1-butyl-3-methylimidazolium bis((trifluoromethyl)sulfonyl)imide ([bmim] [Tf2N]) has shown that DMSO in mixtures with ILs can produce clusters, instead of homogeneous mixing. Moreover, the formation of localized, self-segregated, hydrogen-bonded nanostructures within the 1-butyl-3-methylimidazolium hexafluorophosphate ([bmim][PF6])/tetraethylene glycol (TEG) binary mixtures has also been observed [15]. Accordingly, it has been emphasized that [16] considerable efforts should be made to investigate structural heterogeneities in mixtures of ILs with molecular solvents.

NMR has been frequently utilized to study the inhomogeneities in mixtures of ILs due to its accuracy and sensitivity [17, 18]. Undoubtedly the 1H NMR chemical shifts can help us to determine the numbers and category of protons in pure substances. Furthermore, 1H NMR chemical shifts also offer us much information about microscopic structure within solutions, since the chemical environments of protons can change due to the intermolecular interactions. More interestingly, 1H NMR chemical shifts can be analyzed quantitatively by certain theoretical models. Deng [19] and co-workers have proposed a simple semi-empirical physical local composition (LC) model to correlate the 1H NMR chemical shifts of binary mixtures with satisfying results. Subsequently, relations between the thermodynamic and spectroscopic properties of mixtures have been established successfully with this LC model [20–22]. In our previous work, IR, Raman and 1H NMR measurements were made in solutions of ILs [23] and the local structure of aqueous solutions of 1-ethyl-3-methylimidazolium tetrafluoroborate (EmimBF4) and n-butylammonium nitrate (N4NO3) [24] were studied using the LC model.

Because little is known about LC of ammonium-based ILs in solutions, the triethylamine nitrate (N222NO3) ionic liquid has been synthesized as described in the present paper and the 1H NMR spectra of two binary mixtures, namely N222NO3/DMSO and N222NO3/acetone at different compositions, have been measured. DMSO is known for its special dissolving capacity and high polarity and is often used in company with ionic liquid to promote chemical reactions [25] and biological processes [26]. Acetone, as another common organic solvent, has been regarded as an efficient solvent for ILs and several properties of the acetone–IL systems have been reported [27]. However, understanding of the micro-structures of N222NO3/DMSO and N222NO3/acetone solutions remains obscure and is the subject of the present paper.

2 Experiments and Theory

2.1 Experiments

2.1.1 Synthesis of N222NO3

N222NO3 was prepared from triethylamine and nitric acid according to the procedure reported previously [28] (see Scheme 1). Nitric acid (CAS No. 7697-37-2; 65–68 mass%), triethylamine (CAS No. 121-44-8; A.R.), and DMSO (CAS No.67-68-5; A.R.) were purchased from Aladdin Reagent Co. Ltd., (Shanghai, China) and acetone (CAS No. 67-64-1, A.R.) was purchased from Tianjin Damao Chemical Reagent Factory (Tianjin, China). The water content was below 200 ppm by Karl–Fischer titration (ZSD-1, ShangHai Anting Scientific Instrument Factory). The structure of N222NO3 was identified by its 1H NMR spectra. The chemical shifts δ (ppm) are as follows: 1H NMR (Bruker, 400 MHz,CDCl3), δ (ppm): 9.49 (1H, +NH), 3.28 (6H, –CH2), 1.34 (9H, –CH3). All solutions of ILs were prepared by weight using an analytical balance with a precision of ±0.0001 g.

Scheme 1
scheme 1

Molecular structure and synthesis of N222NO3 ionic liquid

Full size image

2.1.2 Measurements of 1H NMR Chemical Shifts

The internal reference method [29] was adopted to obtain the concentration dependent chemical shifts of –CH3 (whose chemical shifts change more obviously with concentrations) in N222NO3. To avoid the influence of deuterated reagents, a 2 mm capillary tube, in which the deuterated chloroform (CHCl3-d) was sealed, was placed at the center of a 5 mm sample tube filled with the chemical shift reference tetramethylsilane (TMS). It is necessary that the inner capillary tube should be kept parallel with the exterior sample tube. The 1H NMR spectra were obtained using a Bruker AV 400 spectrometer operating at 400 MHz at different concentrations at 298.15 K and all of the experimental data are included in Table 1 and Fig. 1.

Table 1 Experimental 1H NMR chemical shifts of –CH3 of N222NO3 in N222NO3/DMSO and N222NO3/acetone systems at 25.0 °C
Full size table
Fig. 1
figure 1

Experimental and correlated 1H NMR chemical shifts of–CH3 of N222NO3 in N222NO3/DMSO and N222NO3/acetone systems at 25.0 °C: (filled circle experimental values of N222NO3/DMSO; filled triangle experimental values of N222NO3/acetone; continuous line calculated values)

Full size image

2.1.3 Measurements of Viscosity and Electric Conductivity of Binary Solutions

The viscosities of the samples were measured with an Ubbelohde viscometer (inner diameter = 1.0 mm). Each sample was placed in an external water bath, which was controlled to 25 ± 0.1 °C for 30 min. Then the measurements of viscosity were performed. Each viscosity value was reported by averaging three consecutive runs. In order to avoid moisture absorption, the ionic liquid was dried in a vacuum oven at 70 °C for 24 h before use. And, experimental data are illustrated in Fig. 2.

Fig. 2
figure 2

Viscosity η versus mole fraction of N222NO3 for binary solutions at 25.0 °C

Full size image

The electric conductivities were measured with an electric conductivity meter produced by Shanghai Precision & Scientific Instrument Co., Ltd. (DDS-11D). The cell constant was determined by using 0.1 mol·L−1 aqueous solution of KCl at 298.15 K. Each sample was placed in a water bath, which was controlled to 25 ± 0.1 °C for 30 min before measurement. Each conductivity value of the samples was obtained by averaging three consecutive runs. The electric conductivity data of binary solutions are illustrated in Fig. 3.

Fig. 3
figure 3

Specific conductivity κ versus mole fraction of N222NO3 for binary solutions at 25.0 °C

Full size image

2.1.4 The Solubility of Cellulose in Binary Solutions of N222NO3

Under the magnetic stirring and temperature control, sufficient microcrystalline cellulose (MCC) (the weight is m 1) was placed into a 100 mL three necked flask filled with N222NO3 solutions (the weight of N222NO3 is m 2). The temperature was kept at 60 °C. After 24 h dissolution, the mixtures were separated using high-speed centrifugation to obtain the precipitate (un-dissolved MCC, the weight is m 3). Then the solubility of MCC was calculated as follows and is shown in Table 2:

$$ s = \frac{{m_{1} - m_{3} }}{{m_{2} }}. $$
Table 2 The solubility of MCC in the N222NO3/DMSO and N222NO3/acetone solutions
Full size table

2.2 Theory and Calculations

In 1964, Wilson first [30] proposed the LC concept to describe the microscopic solution structure; this method has been extensively applied to correlate vapor–liquid equilibrium data of binary and multicomponent mixtures. Based on the Wilson’s theory, Deng [19] assumed that the observed 1H NMR chemical shifts at a certain composition in a binary system can be simply expressed as follows:

$$ \delta_{{1,{\text{cal}}}} = \delta_{1}^{0} \varphi_{11} + \delta_{1}^{\infty } \varphi_{21} $$
(1)

where φ 11 and φ 21 are local volume fractions of molecules 1 and 2 around the central molecule 1, respectively, and δ 01 and δ ∞1 are the chemical shifts of a particular proton of the pure substance and infinitely dilute solution of molecule 1, respectively. The latter is usually obtained by extrapolating the dilute chemical shift to zero concentration. Then the local molar volume fractions can be defined as following:

$$ \varphi_{11} = \frac{{x_{1} }}{{x_{1} + \varLambda_{21} x_{2} }} $$
(2)
$$ \varphi_{21} = \frac{{\varLambda_{21} x_{2} }}{{x_{1} + \varLambda_{21} x_{2} }} $$
(3)
$$ \varLambda_{21} = \frac{{V_{2} }}{{V_{1} }}\exp [{{ - (g_{21} - g_{11} )} \mathord{\left/ {\vphantom {{ - (g_{21} - g_{11} )} {RT}}} \right. \kern-0pt} {RT}}] $$
(4)

where g 21 and g 11 are proportional to the 1–2 and 1–1 interaction energies, respectively, where V 1 and V 2 are the mole volumes of molecules 1 and 2 which are calculated from the molecular weight and density (the measured densities of N222NO3, DMSO and acetone at 298.15 K are 1.055, 1.096 and 0.788 g·cm−3 respectively, using the pycnometer method). x 1 and x 2 denote bulk mole fractions of molecules 1 and 2, respectively. (g 21 − g 11) is the mutual interaction energy parameter connected with the 1–2 and 1–1 pairs of molecules. The following objective function is used:

$$ O.F. = \left[ {\frac{1}{m - 1}\sum\limits_{k = 1}^{m} {(\delta_{{1,{\text{cal}},k}} - \delta_{1,\exp ,k} )^{2} } } \right]^{1/2} , $$
(5)

where m denotes the size of the experimental data set. δ 1,cal,k denotes kth calculated chemical shift according to Eq. 1. δ 1,expl,k denotes the kth experimental value, and (g 21 − g 11) is the only optimized parameter. According to Eqs. 1–4, the correlation of chemical shifts of binary mixtures can be achieved using one set of energy parameters and the root-mean square deviations are defined as:

$$ \Delta \delta \% = \left[ {\frac{1}{m}\sum\limits_{k = 1}^{m} {\left\{ {\frac{{(\delta_{{1,{\text{cal}},k}} - \delta_{1,\exp ,k} )}}{{\delta_{1,\exp ,k} }}} \right\}}^{2} } \right]^{1/2} \times 100\,\% $$
(6)
$$ \Delta \delta = \left[ {\frac{1}{m}\sum\limits_{k = 1}^{m} {(\delta_{{1,{\text{cal}},k}} - \delta_{1,\exp ,k} )}^{2} } \right]^{1/2} . $$
(7)

The (g 21 − g 11) and correlated deviations of two systems are listed in Table 3. According to Wilson’s definition of LC for a binary mixture, the ratio of the probability of finding the molecule 2 around the central molecule 1 can be defined in terms of bulk mole fractions and interaction energies between the 1–2 pair g 21 and 1–1 pair g 11 as shown below:

$$ \frac{{x_{21} }}{{x_{11} }} = \frac{{x_{2} \exp ( - g_{21} /RT)}}{{x_{1} \exp ( - g_{11} /RT)}} = \frac{{x_{2} }}{{x_{1} }}\exp [ - (g_{21} - g_{11} )/RT] $$
(8)

where x 11 and x 21 denote the local mole fraction of molecule 1 and 2, respectively, around the central molecule 1. Apparently the local mole fraction of every molecule in the vicinity of the central molecule 1 should be unity. Thus,

$$ x_{11} + x_{21} = 1 $$
(9)

Subsequently, the value of x 11 and x 21 can be obtained at different compositions and then the concentration dependent (x 11 − x 21) and (x 1 − x 11) can be calculated.

Table 3 The energy parameter (g 21 − g 11) and the deviations of correlation of 1H NMR chemical shifts
Full size table

3 Results and Discussion

3.1 Correlation of 1H NMR Chemical Shifts

Taking N222NO3/DMSO for example, the N222NO3 could be regarded as the central substance 1 and then 2 is assigned to DMSO. Thus, according to Eqs. 1–4, the chemical shifts of –CH3 of N222NO3 can be correlated with the only optimized parameter (g 21 − g 11). Obviously, as is shown in Fig. 1, the –CH3 chemical shifts of N222NO3 in solutions can be well correlated by this semi-empirical LC model, just as its excellent performance in the aqueous solutions of n-butylammonium nitrate (N4NO3) [23] and EmimBF4 [24] ionic liquids.

3.2 Local Compostion in the Solutions

According to the LC theory, it is easy to understand that the binary solutions of ILs should not be completely homogeneous in the local area. Therefore the number of IL or molecular solvent molecules in the local region should be not always be the same as in the bulk, which is the so-called the LC behavior. In N222NO3/DMSO systems, for example, since N222NO3 is considered as the substance 1, x 11 denotes the local mole fractions of N222NO3 around the N222NO3, and x 21 denotes the local mole fractions of DMSO around the N222NO3. With the help of the LC model, the local mole fractions x 11 and x 21 can be calculated using Eqs. 8–9 since the (g 21 − g 11) can be obtained by the correlation of δ(–CH3) of N222NO3. In order to illustrate the LC more clearly, the concentration-dependent (x 11 − x 21) of the two systems is shown graphically in Fig. 4.

Fig. 4
figure 4

Concentration-dependent (x 11 − x 21) for binary solutions of N222NO3

Full size image

Clearly (x 11 − x 21), which denotes the difference between the local mole fractions of N222NO3 and DMSO (or acetone) in the vicinity of N222NO3, behaves similarly in the two systems and decreases from 1 to −1 with x 2 (the bulk mole fraction of solvent) over the whole concentration range. Apparently, there are only N222NO3 in the pure ionic liquid (when x 2 = 0) and here x 11 = 1 and x 21 = 0. With x 2 increasing, (x 11 − x 21) becomes smaller and the amount of solvent gathered around N222NO3 keeps increasing, indicating solvent molecules can gradually break into the self-associated organization of N222NO3 and destroy the network of N222NO3. Taking N222NO3/DMSO for example, when x(DMSO) < 0.3, the value of (x 11 − x 21) remains positive, indicated that there are more N222NO3 but fewer DMSO distributed around N222NO3. However, with the amount of DMSO increasing continuously (0.3 < x(DMSO) < 1), the negative value of (x 11 − x 21) demonstrates predominance of the interactions between N222NO3 and DMSO instead of self-association of N222NO3. Within the N222NO3-rich region, self-association among N222NO3 molecules should dominate, indicating that a small quantity of DMSO does not cause significant interference with the structure of N222NO3. When the amount of DMSO is further increased, much more DMSO can move into the local area of N222NO3 and the original network of N222NO3 can be gradually broken down due to the interactions between the DMSO and N222NO3. In the N222NO3/acetone system, the local structure of solutions shows similar variations. Accordingly, (x 11 − x 21) can offers us a picture of the change of micro-structure in binary solutions. Actually, it has been reported [14] that DMSO is not be completely mixed with ionic liquid in DMSO/[N4111][Tf2N] mixtures. Venkatesu and co-workers [31, 32] have studied mixtures of trimethylammonium-based ILs with DMSO and detected that DMSO molecules can reduce hydrogen bonding between the cation and anion due to ion–dipole interactions between DMSO and ILs. However at higher IL concentrations, the interactions between DMSO and IL became weaker and extensive self-association in the IL has been detected, which is consistent with our results.

It is also interesting to focus on the differences of the two systems. As shown in Fig. 4, when (x 11 − x 21) = 0, x(DMSO) ≈ 0.3 and x(acetone) ≈ 0.5. In other words, when x(DMSO) exceeds 0.3, there are more DMSO than N222NO3 molecules distributed around N222NO3. However in the case of N222NO3/acetone, the interactions between acetone and N222NO3 do not dominate before x(acetone) reaches 0.5. That is to say, DMSO molecules can break into the self-association structure of N222NO3 at lower concentrations than can acetone. In addition, the difference between the local and bulk composition (x 1 − x 11) has also been calculated and is shown in Fig. 5. Since x 1 and x 11 denote the bulk and local concentrations of N222NO3, respectively, the absolute values of (x 1 − x 11) should represent the extent of LC in the binary solutions. It can be seen that the absolute values of (x 1 − x 11) change in the order: N222NO3/DMSO > N222NO3/acetone, indicating that the LC behavior should be more remarkable in N222NO3/DMSO systems and be weaker in the N222NO3/acetone system, which is consistent with the (x 11 − x 21) analysis. So far as we know, the effect of the molecular solvents on the network of ILs mainly depends on the solvent dielectric constant [33]; other studies [34] have pointed out that molecular solvents with high dielectric constant have greater impact in disrupting the electrostatic attraction between the ions of the ILs. Accordingly it can be deduced that molecular solvents with higher dielectric constant, like DMSO, can give rise to more conspicuous LC behavior, the so-called structural heterogeneities, in solutions of N222NO3 ionic liquid.

Fig. 5
figure 5

Concentration-dependent (x 1 − x 11) for binary solutions of N222NO3

Full size image

In order to demonstrate our results, the physicochemical properties including viscosity and electric conductivity of N222NO3/DMSO and N222NO3/acetone have been measured. As shown in Fig. 2, the viscosity η of all solutions decreases greatly with increasing concentration of the molecular solvents. In addition, since the viscosity of acetone is smaller than that of DMSO, the viscosity of N222NO3/DMSO is larger than that of N222NO3/acetone. However with increasing solvent concentration, the slope of the curve corresponding to the N222NO3/DMSO system is more negative than that of N222NO3/acetone, especially within the range 0 < x(N222NO3) < 0.6, indicating that DMSO more efficiently reduces the viscosity of N222NO3.

In the case of electric conductivity κ shown in Fig. 3, with mole fraction of N222NO3 increasing, κ sharply increases in the molecular solvent-rich region and then progressively became smaller in the N222NO3-rich region. Also, the maximum value of κ of N222NO3/DMSO is much larger than in the N222NO3/acetone system, indicating DMSO molecules more effectively promote the dissociation of N222NO3.

In order to further elucidate this, the solubility of MCC in N222NO3/DMSO and N222NO3/acetone at different concentrations were measured. As show in Table 2, with increasing DMSO concentration, the solubility of MCC gradually increases in the N222NO3/DMSO system. However the solubility of MCC decreases with increasing acetone concentration in the N222NO3/acetone system. In other words, DMSO is an efficient co-solvent to enhance the ability of N222NO3 to dissolve the MCC. Actually, it has been detected [35] that the dissociation of cellulose in ionic liquids is mainly determined by the hydrogen bonding interactions of hydroxyl protons of cellulose with cations and anions. Therefore it can be deduced that DMSO should more readily break down the network of N222NO3 and promote the dissociation of ionic liquid. Accordingly, the experimental data of viscosity, electric conductivity and MCC solubility demonstrate that DMSO molecules easily interact with N222NO3 ionic liquid, which obviously is consistent with the LC analysis.

4 Conclusions

With the help of the LC model, the correlations of 1H NMR chemical shifts of two systems, N222NO3/DMSO and N222NO3/acetone, have been performed. The changes of local mole fractions have revealed that solvents do not mix homogeneously with N222NO3 in the local region. Furthermore, the micro-structure of systems change with the concentration of solutions. Specifically, N222NO3 mainly self-associates with itself in N222NO3-rich region and a small quantity of solvent does not induce great variations of the original network of N222NO3. Nevertheless, in solvent rich region, N222NO3 chiefly interacts with molecular solvents leaving fewer N222NO3 molecular self-associated. In addition, compared to acetone, DMSO molecules appear more able to insert into the network of N222NO3. In other words, the ability of DMSO to break down the self-associated network of N222NO3 is greater than that of acetone, which is in accordance with the order of their dielectric constants. Consequently, the molecular solvents with higher dielectric constant should lead to more impact on the structure and properties of N222NO3 ionic liquid. Also, the experimental data of viscosity, electrical conductivity and MCC solubility show consistent results with the LC analysis.

Evidently it is interesting and practical to study the local heterogeneity of ILs in solutions which can dramatically affect the properties of ILs/solvent systems. This work offered us a clear insight about the LC behavior in N222NO3 solutions due to addition of solvents. More kinds of molecular solvents will be involved in further studies to verify this simple and convenient approach and final analysis results.