Quantitative Raman spectroscopy

Abstract

A method for quantifying water and / or one or more ionic liquid components in an ionic liquid (IL) / water (H2O) mixture is disclosed. The method comprises obtaining one or more Raman spectra of the IL / H2O mixture, and quantifying water and / or one or more ionic liquid components in the IL / H2O mixture using the one or more Raman spectra using a quantitative calibration model.

Description

Technical Field

[0001] The present invention relates to quantitative Raman spectroscopy. In particular, the present disclosure may relate to quantitative Raman spectroscopy for ionic liquid fiber (Ioncell)-F process. Background Art

[0002] Ex situ analytical techniques such as high performance liquid chromatography (HPLC), ion chromatography (IC), and capillary electrophoresis (CE) can be used to determine anion or cation concentrations, while nuclear magnetic resonance (NMR) spectroscopy can be used to determine the acid-base (A:B) ratio. The H2O concentration in an ionic liquid (IL) / water (H2O) mixture can be determined by Karl Fischer (KF) titration[3][5].

[0003] Determining the HO concentration in the IL / HO mixture using refractive index (RI) measurement is a simple and robust method for in-line analysis in continuous processes. However, this method is not sensitive to changes in the A:B ratio in the IL / HO mixture (results not shown). Most importantly, many of the methods mentioned above are time-consuming and cumbersome and are not suitable for in-line analysis in processes. Summary of the invention

[0004] Ionic Liquid Fiber-F is a new process for producing Lyocell-type man-made cellulose fibers based on dissolving cellulose pulp in ionic liquid (IL), dry-jet wet spinning of this solution (spinning dope) in a water (H2O) bath, and a subsequent solvent recovery step, in which IL and H2O are separated. The entire process operates in a closed-loop system [1].

[0005] The IL-F process operates on non-imidazolium based protic ILs, among which [DBNH][OAc] and [mTBDH][OAc] show high potential for producing strong cellulose fibers with minimal cellulose degradation [1][2][3].

[0006] The acid-base molar ratio (A:B ratio) in the IL and the HO content in the IL / HO mixture are key parameters for cellulose dissolution and cellulose slurry spinning. If the concentration of water in the IL / HO mixture reaches a certain limit (usually a few percent), water as a non-solvent may hinder cellulose dissolution. If acetic acid is in excess compared to the base, acetic acid also acts as a non-solvent and prevents good cellulose dissolution. In addition, the presence of water in the cellulose slurry can affect spinning by changing the rheological properties of the slurry (water acts as a softener). The A:B ratio and HO content may vary greatly within the process loop. For example, the HO concentration will gradually decrease between the spinning bath and the solvent recovery / purification step. In addition, the A:B ratio may change because of different acid and base vapor pressures or due to the formation of reactive azeotropes during recovery / purification steps using, for example, evaporation and distillation [2][4][5]. In order to ensure good process stability and uniform fiber quality, these parameters should be well adjusted in each unit operation. Therefore, rapid and quantitative monitoring of these parameters is an important part of process control.

[0007] The development of rapid, quantitative, on-line analytical methods for the simultaneous determination of H2O concentration and A:B ratio in IL / H2O mixtures could benefit an increasing number of emerging processes based on IL / H2O mixtures, such as biomass fractionation, pulp upgrading, or protein separation [6]-[9].

[0008] Vibrational spectroscopy has the great advantages of being fast, non-destructive, and requiring less work and time for sample preparation. In addition to the classical KF titration method and various electrochemical methods

[10]

[11] , vibrational spectroscopy performs well in detecting and analyzing water contained in ILs. For example, Viell et al.

[12] developed a quantitative method for determining the composition of binary mixtures of water and various ILs using mid-infrared (mid-IR) spectroscopy. In addition, near-infrared (NIR) spectroscopy has been successfully used to non-invasively and in situ determine the concentration and structure of water absorbed by room temperature ILs

[13] .

[0009] Raman spectroscopy is a well-known vibrational spectroscopic technique based on the inelastic scattering of light by materials. The proportional relationship between the intensity of the Raman scattering and the concentration of the analyte is the basis for most quantitative analyses using Raman spectroscopy in gaseous, liquid or solid samples

[14] . Raman spectroscopy is also well suited for the quantitative analysis of multiple components in a sample, as a single Raman spectrum often contains enough information to determine their concentrations simultaneously

[15] . The use of Raman probes and optical fibers also allows remote monitoring of processes in harsh environments, making the technique well suited for on-line analysis.

[0010] Raman spectroscopy has been widely used to study the molecular interactions, structure, and band assignments of pure ILs and IL mixtures

[16]

[17]

[18] . However, few studies have been conducted in the literature on methods for quantitative concentration measurements in IL / H2O mixtures.

[0011] In this application, we discuss the potential of Raman spectroscopy combined with univariate and multivariate calibration for the rapid and quantitative determination of H2O, acid, and base contents in H2O / IL mixtures over a wide, process-relevant range.

[0012] Specifically, we use 1,5-diazabicyclo[4.3.0]non-5-ene / acetate ([DBNH][OAc]) as a case study to investigate the potential of Raman spectroscopy for quantifying protic IL components (acid and base) and water in IL / water mixtures in this application. We demonstrate that the combination of Raman spectroscopy and chemometrics is successful for the quantitative analysis of IL components and water in mixtures over a wide range of concentrations. The findings of this work suggest that Raman spectroscopy should be considered more generally for the online monitoring and control of processes involving IL / H2O mixtures.

[0013] In this application, we have demonstrated that Raman spectroscopy combined with univariate or multivariate calibration can be successfully applied to the quantitative analysis of protic IL components and H2O in IL / H2O mixtures. Compared to univariate calibration, multivariate calibration using PLS regression showed superior performance for the quantitative analysis of IL components and water over a wide range of concentrations. These results suggest that the combination of Raman spectroscopy and chemometrics can be more generally applied to monitor and control a variety of processes involving the use of IL / H2O mixtures.

[0014] The disclosure includes a comparison of univariate and multivariate calibration methods for quantifying water and protic ionic liquid components. The disclosure includes keywords: ionic liquid fiber-F, ionic liquid, Raman spectroscopy, chemometrics.

[0015] According to a first aspect, a method for quantifying water and / or one or more ionic liquid components in an ionic liquid (IL) / water (H2O) mixture is disclosed. The method comprises obtaining one or more Raman spectra of the IL / H2O mixture, and quantifying water and / or one or more ionic liquid components in the IL / H2O mixture using the one or more Raman spectra using a quantitative calibration model.

[0016] In one embodiment, the ionic liquid is a protic ionic liquid.

[0017] In one embodiment, the ionic liquid is or comprises a non-imidazolium based protic ionic liquid.

[0018] In one embodiment, using a quantitative calibration model involves a univariate calibration based on the discovery of a relationship between a single spectral variable, such as peak intensity, peak area, and / or peak shift, and analyte concentration.

[0019] In one embodiment, the method includes determining the concentration of H2O in the IL / H2O mixture using a linear relationship between the H2O peak area and the H2O mass fraction in the IL / H2O mixture.

[0020] In one embodiment, the method includes utilizing a non-linear relationship between peak intensity and base concentration to determine the base concentration in the IL / H2O mixture.

[0021] In one embodiment, the nonlinear relationship is described by a power law model.

[0022] In one embodiment, using a quantitative calibration model involves using a multivariate calibration.

[0023] In one embodiment, multivariate calibration utilizes partial least squares (PLS) regression.

[0024] In one embodiment, the method includes simultaneously determining (a) acid, base, and H2O content and / or (b) H2O concentration and A:B ratio in the IL / H2O mixture.

[0025] In one embodiment, the method includes quantifying one or more ionic liquid degradation products. The degradation products can be formed in the IL / H2O mixture, for example, when water reacts with the ionic liquid. Specifically, the degradation products can be hydrolysis products, such as 1-(3-aminopropyl)-2-pyrrolidone (APP). Degradation can include degradation of the base of the ionic liquid. Quantification can include determining the concentration of the degradation products. Quantification can be performed using one or more Raman spectra to quantify one or more ionic liquid degradation products using a quantitative calibration model. Quantification can be performed substantially simultaneously with determining (a) acid, base and H2O content and / or (b) H2O concentration and A:B ratio in the IL / H2O mixture. Quantification can be performed substantially in real time, for example, for monitoring and / or controlling processes involving the use of IL / H2O mixtures. Quantification can be used for quantitative monitoring of one or more degradation products for process control, for example, simultaneously with A:B ratio and / or H2O content.

[0026] In one embodiment, quantification of one or more ionic liquid degradation products involves utilizing partial least squares (PLS) regression.A quantitative calibration model may be based on PLS for quantifying the degradation products.

[0027] In one embodiment, the quantification of one or more ionic liquid degradation products involves dividing one or more Raman spectra into two or more sub-intervals. This allows for a significant reduction in calibration error and thereby improves quantification. The number of sub-intervals can be optimized to minimize the root mean square error of calibration (RMSEC). In some embodiments, the number of sub-intervals can be less than 10.

[0028] According to a second aspect, a method for monitoring and / or controlling a process involving the use of an IL / H2O mixture is disclosed. The method comprises the method according to any one of the first aspect or embodiments thereof.

[0029] In one embodiment, the method includes quantitatively monitoring the A:B ratio and / or H2O content for process control.

[0030] According to a third aspect, a method for producing lyocell type man-made cellulose fibers is disclosed, the method being based on dissolution of cellulose pulp in IL, dry jet wet spinning of the solution in a H2O bath and a subsequent solvent recovery step, wherein IL and H2O are separated. The method comprises the method according to any one of the first aspect and / or the second aspect or their embodiments, for quantifying water and / or one or more ionic liquid components in an IL / H2O mixture.

[0031] In one embodiment, the method includes quantifying and performing solvent purification based on one or more ionic liquid degradation products. Quantifying of one or more ionic liquid degradation products allows monitoring the accumulation of the degradation products in the solvent. Therefore, harmful accumulation of degradation products that may prevent or hinder fiber production can be prevented. Quantification can be used to determine, for example, the intensity of solvent purification and / or the frequency or time of purification. Purification can include at least partially removing and / or deactivating (can include recovery and / or regeneration) one or more degradation products from the solvent.

[0032] It should be understood that the above aspects and embodiments can be used in any combination with each other. Several aspects and embodiments can be combined together to form another embodiment of the present invention.

[0033] In the following, general examples are provided to answer the questions behind the present disclosure. The answers should not be considered as the only potential alternatives related to the present invention, but rather, they can be considered as general principles that can be used to describe the present invention.

[0034] There is no reliable, on-line, quantitative and rapid analytical method for simultaneously quantifying the ionic liquid components (cations and anions) and water molecules that make up the liquid stream, such as in the ionic liquid fiber-F process. Without continuous measurement of these molecules, monitoring and control of the process would be very difficult.

[0035] We developed a rapid and quantitative method for the simultaneous quantification of ionic liquid components (cations and anions) and water molecules in liquid mixtures consisting of water and ionic liquids.

[0036] By using the developed methods, customers can monitor and control the process in real time, which is a necessary condition for good and stable process operation. This process stability is a direct economic benefit as it reduces downtime and results in more uniform product quality and less scrap.

[0037] The present invention may be included as part of a patent portfolio related to the IL-Fiber-F technology. We sell an on-line analytical method for monitoring and controlling a liquid stream consisting of ionic liquid and water in an IL-Fiber-F process. The present invention may also be used in other processes utilizing ionic liquid / water streams.

[0038] The total market for textile fibers is currently 103 metric tons per year, of which 6.3 metric tons are man-made cellulose fibers. The estimated CAGR for man-made cellulose fibers such as ionic liquid fiber-F is 7.5%. (Source: Fiber Year 2018). Customers can be pulp producers, textile fiber producers, or other industry participants interested in adding ionic liquid fiber fibers to their product portfolio. The present invention can also benefit other processes that apply mixtures of ionic liquids and water.

[0039] To the best of our knowledge, there is no equivalent on-line method for monitoring and controlling liquid flows consisting of ionic liquids and water.

[0040] One of the main issues associated with ionic liquids (ILs) is their recyclability. Only by knowing what is in IL mixtures and solutions can a viable recycling concept be developed. We have demonstrated that Raman spectroscopy can be used to quantify water and IL components in liquid mixtures. In this regard, we believe that Raman spectroscopy is a promising analytical method for online monitoring and control of ionic liquid fiber-F processes. In this application, we also extend the analytical method to more complex and more practical liquid mixtures, including degradation products such as hydrolysis products, such as 1-(3-aminopropyl)-2-pyrrolidone (APP) that can be formed after the reaction of 1.5-diazabicyclo[4.3.0]non-5-ene (DBN) with water, thereby further breaking through the limits of this analytical method. In order to measure the extent of the hydrolysis reaction and apply the correct corrective measures to reverse the reaction and keep the process within optimal working conditions, it is important to quantify degradation products such as APP. Therefore, the simultaneous quantification of four components (acetic acid, bases such as DBN, degradation products such as APP, and water) in the liquid stream was studied using Raman spectroscopy. The liquid stream can be the liquid stream of an ionic liquid fiber-F process. The sensitivity of the Raman method in quantifying degradation products such as APP is also highlighted compared to the widely used refractometry for measuring IL concentrations in aqueous mixtures. Finally, a simple modification of the multivariate partial least squares regression model based on a variable selection algorithm is proposed to improve the accuracy of the predicted calibration values.

[0041] Process analytical technology (PAT) is increasingly being used for on-line analysis and process control. According to Kuppers et al., the advantages of integrated process analysis and control include better process control, safer operation, and significant economic advantages due to better product quality and shorter troubleshooting delays (Kueppers and Haider 2003).

[0042] For process control in an industrial setting, waiting hours for an accurate and precise laboratory analysis is unacceptable. Rapid and accurate feedback mechanisms are needed to avoid poor quality and off-spec products during analytical delays. The combination of spectroscopy and chemometrics is ideal for situations where a compromise between latency and accuracy is required. Some of the accuracy and precision of laboratory methods are sacrificed in order to obtain rapid answers that can be used to continuously monitor and control processes (Geladi et al., 2004). In this regard, Raman spectroscopy is increasingly recognized as the method of choice for rapid, multi-component on-line quantitative analysis, real-time process monitoring and control (Cooper 1999).

[0043] In fact, a single rapidly acquired Raman spectrum can provide a wealth of information about a sample with its well-resolved spectral features. The proportional relationship between Raman scattering intensity and analyte concentration is the basis for most quantitative analyses using Raman spectroscopy (Smith and Dent 2004; Larkin 2011).

[0044] In multicomponent systems, quantitative Raman analysis relies on the principle of linear superposition: the Raman spectrum of a mixture is equal to the weighted sum of the Raman spectra of the components present in the mixture (Pelletier 2003). The appeal of using Raman spectroscopy for multicomponent analysis is enhanced by the absence of optical coherence between components in a sample, which means that Raman scattering from one component in a sample does not affect the Raman scattering of another component (Pelletier 2003). Interference only occurs when the absorption spectrum of one or more components significantly affects the transmission of excitation or Raman scattered light to or from the target analyte. In addition, possible changes in the interactions between analytes when changing their relative concentrations may change the shape and intensity of their respective peaks in the spectrum (Kauffmann and Fontana 2015).

[0045] From the perspective of process digitization, real-time monitoring and control, we have demonstrated that Raman spectroscopy is a very promising analytical tool for ionic liquid fiber-F processes (Guizani et al. 2020). Combining Raman spectroscopy and chemometrics will allow real-time quantification of ionic liquids (ILs) in process liquid streams, such as protic liquids, for example, 1,5-diazabicyclo[4.3.0]non-5-enium acetate [DBNH][OAc]; and water.

[0046] However, the composition of the liquid stream may change as the ionic liquid may undergo degradation such as reversible hydrolysis to become degradation products such as 1-(3-aminopropyl)-2-pyrrolidone (APP), which also forms 3-(aminopropyl)-2-pyrrolidonium acetate ([APPH][OAc]) with acetic acid.

[0047] The formation of degradation products such as APP in the liquid stream may reduce or even inhibit the cellulose solubilizing ability of the IL and ultimately lead to the irreversible formation of reaction products such as APPAc. Recently, Hyde and colleagues demonstrated that hydrolysis reactions can be avoided by controlling the pH (Hyde et al. 2019). Therefore, limiting the hydrolysis reaction is possible, but will not be discussed in detail in this application. Therefore, process feasibility depends on the continuous monitoring of the formation of degradation products and the reversal of degradation reactions such as hydrolysis when they occur. Therefore, the online Raman method can include determining the concentration of degradation products in the process stream. For example, when using an IL that is susceptible to degradation such as hydrolysis, any disclosed method can include determining the concentration of degradation products, such as the concentration of hydrolysis products. The IL can be a protic IL, such as a non-imidazolium based protic IL. Degradation products can be determined in the process stream.

[0048] Since DBN and APP display structural differences that would result in different scattering signals, we hypothesized that the Raman quantification method associated with DBN could be extended to mixtures containing APP. Therefore, we also investigated the potential of Raman spectroscopy for quantifying more challenging and complex multicomponent mixtures of DBN, APP, acetic acid (AcOH), and H2O. We also explored simple modifications to the multivariate regression model algorithm using variable selection to improve model prediction performance.

[0049] Due to the short analytical delay and acceptable accuracy in determining the expected composition of liquid streams, Raman spectroscopy shows great potential in process (specifically ionic liquid fiber-F process) monitoring and control. The concentration of water, IL components and degradation products in the liquid stream can be determined in real time using adequate Raman in situ probes. Real-time information can be used to monitor and control process operations.

[0050] Compared to the more widely used refractometry for measuring aqueous IL concentrations, Raman spectroscopy exhibits a clear advantage in detecting IL degradation products due to its improved sensitivity. This study further demonstrates that the combination of Raman spectroscopy and chemometrics opens the door to reliable monitoring and efficient control of a potentially wide range of IL-based processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings, which are included to provide a further understanding and constitute a part of this specification, illustrate examples and together with the description help to explain the principles of the present disclosure. In the drawings:

[0052] Figure 1 A sample of IL / H2O mixture according to the example is shown,

[0053] Figure 2shows NMR spectra of anhydrous ILs with different A:B ratios according to Examples (the A:B ratio of each sample is noted on the left side of the figure),

[0054] Figure 3 shows the Raman spectra of DBN, HOAc and anhydrous IL according to the examples (A:B=1.05),

[0055] Figure 4 shows the Raman spectra of anhydrous ILs with different A:B ratios according to examples,

[0056] Figure 5 shows the area normalized Raman spectra of 50 wt % water diluted IL samples according to the Examples (the color code is related to the A:B ratio),

[0057] Figure 6 shows the area normalized Raman spectra of different IL / H2O solutions according to the examples (the color code reflects the IL concentration in the IL / H2O mixture),

[0058] Figure 7 According to the example, the -1 -3800cm -1 The relationship between the integrated area of ​​Raman scattering intensity in the region and the water concentration (blue circles: calibration data set. Black stars: validation data set. Solid line: calibration model. Dashed line: 95% confidence interval limit).

[0059] Figure 8 According to the example, the -1 and 2890cm -1 Evolution of the peak intensity at as a function of the base concentration in the IL / H2O mixture (blue circles: calibration dataset. Black stars: validation dataset. Solid line: calibration model. Dashed lines: 95% confidence interval limits),

[0060] Fig. 9 shows a graph of predicted and measured acid, base, and H2O concentrations in a H2O / IL mixture according to an example,

[0061] Fig.10 Score plots based on LV1 and LV2 (top) and loading plots for four LVs (bottom) are shown according to an example.

[0062] Fig.11 The hydrolysis of [DBNH][OAc] (1) to [APPH][OAc] (2) and the condensation to APPAc (3) are shown according to the example.

[0063] Fig.12shows a comparison of RI as a function of IL weight % in a mixture with water in the presence and absence of APP according to examples,

[0064] Fig.13 Pre-processed Raman spectra of different mixtures according to the examples are shown (the color code describes the wt.% of H2O in the mixture, for clarity, the spectra are divided into the fingerprint region (lower) and the high frequency region (upper).),

[0065] Fig.14 Pre-processed Raman spectra of four APP-free IL samples with different dilution levels according to the examples are shown (for clarity, the spectra are divided into the fingerprint region (lower) and the high frequency region (upper).),

[0066] Fig.15 Pre-processed Raman spectra of a sample without APP and a sample with APP / DBN=4.47 mol / mol at two H2O contents of 0 wt% (left) and 75 wt% (right) according to the example are shown (for clarity, the spectra are divided into a fingerprint region (lower) and a high frequency region (upper).),

[0067] Fig.16 The score plot (left) and loading plot for the first three PCs according to the example are shown.

[0068] Fig.17 The cross-validated predicted and measured concentrations of AcOH, DBN, APP, and H2O according to the example are shown.

[0069] Fig.18 shows the lowest RMSEC values ​​of the 10 best models using increasing number of intervals according to the example,

[0070] Fig.19 shows the sample composition according to the example,

[0071] Fig. 20 The ratio of DBN to DBN+APP (left) and DBN weight % (right) in the mixture according to the example are shown,

[0072] Fig.21 Pre-treated Raman spectra of samples with APP / DBN molar ratios of 0 and 4.47 at two H2O contents of 25 wt% (left) and 50 wt% (right) according to the example are shown (for clarity, the spectra are divided into the fingerprint region (lower) and the high frequency region (upper)),

[0073] Fig. 22 The raw pre-processed Raman spectra (top) and the averaged centered pre-processed Raman spectra of different samples according to the example are shown (the color code indicates the wt. % of H2O in the mixture).

[0074] Fig.23 shows the variance and cumulative variance explained by the first ten PCs according to the example,

[0075] Fig.24 The RMSE and variance according to the example are shown as a function of the number of LVs.

[0076] Fig.25 shows the model residuals according to the example,

[0077] Fig.26 shows the model scores and loadings according to the examples,

[0078] Fig. 27 shows the model regression vector according to the example,

[0079] Fig.28 The RMSE and variance according to the example are shown as a function of the number of LVs.

[0080] Fig.29 shows the model residuals according to the example,

[0081] Fig.30 shows the model scores and loadings according to the examples,

[0082] Fig.31 shows the model regression vector according to the example,

[0083] Fig.32 The RMSE and variance according to the example are shown as a function of the number of LVs.

[0084] Fig.33 shows the model residuals according to the example,

[0085] Fig.34 shows the model scores and loadings according to the examples,

[0086] Fig.35 shows the model regression vector according to the example,

[0087] Fig.36 The RMSE and variance according to the example are shown as a function of the number of LVs.

[0088] Fig.37 shows the model residuals according to the example,

[0089] Fig.38 shows the model scores and loadings according to the examples, and

[0090] Fig.39 The model regression vector according to the example is shown.

[0091] In the figures, similar reference numerals may be used to designate equivalent or at least functionally equivalent parts. DETAILED DESCRIPTION

[0092] The detailed description provided below in conjunction with the drawings is intended as a description of examples and is not intended to represent the only form in which the examples can be constructed or utilized. However, the same or equivalent functions and structures can be implemented by different examples.

[0093] The disclosed methods can be used for ionic liquids. In one embodiment, the ionic liquid is or comprises a protic ionic liquid. In one embodiment, the ionic liquid is or comprises an ionic liquid based on a non-imidazolium and / or superbase. In one embodiment, the ionic liquid is or comprises an amidine or guanidine based protic ionic liquid. In one embodiment, the ionic liquid is or comprises a DBN based and / or TBD based ionic liquid.

[0094] The DBN-based ionic liquid may comprise a 1,5-diazabicyclo[4.3.0]non-5-enium cation having the formula (I),

[0095]

[0096] in

[0097] R1 is selected from the group consisting of hydrogen, straight and branched C1-C6 alkyl, C1-C6 alkoxy, C1-C 10 Alkoxyalkyl and C 6-18 aryl groups, which are optionally substituted with one or more substituents selected from hydroxy and halogen;

[0098] as well as

[0099] Anions are selected from the group consisting of halides such as fluoride, chloride, bromide and iodide; pseudohalides such as cyanide, thiocyanate and cyanate; carboxylates, preferably formate, acetate, propionate or butyrate; alkyl sulfites, alkyl sulfates, dialkyl phosphites, dialkyl phosphates, dialkyl phosphites and dialkyl phosphonates.

[0100] The DBN-based ionic liquid may also comprise the above 1,5-diazabicyclo[4.3.0]non-5-enium cation having formula (I), wherein R1 is H and the anion is a carboxylate anion, preferably formate, acetate, propionate or butyrate.

[0101] In one embodiment, the DBN-based ionic liquid is or comprises [DBNH][CO2Et] and / or [DBNH][OAc].

[0102] The TBD-based ionic liquid may comprise a cationic 1,5,7-triazabicyclo[4.4.0]dodec-5-enium [TBDH]+ moiety and an anion selected from the group according to formula a), formula b) and formula c),

[0103] a)

[0104] b)

[0105]

[0106] c)

[0107]

[0108] wherein each of R, R2, R3, R4, R5, R7, R8, R9 and R10 is H or an organic group, and X- is selected from the group consisting of halides, pseudohalides, carboxylates, alkyl sulfites, alkyl sulfates, dialkyl phosphites, dialkyl phosphates, dialkyl phosphites and dialkyl phosphonates.

[0109] All embodiments can be used in any combination with one another.

[0110] 1. Materials and Methods

[0111] 1.1. Materials

[0112] Samples of 1,5-diazabicyclo[4.3.0]non-5-ene (DBN) (CAS No. 3001-72-7; purity ≥ 99.0% by mass) and acetic acid (HOAc) (CAS No. 64-19-7; purity ≥ 99.8% by mass) were purchased from Fluorochem and Sigma-Aldrich, respectively, and used without further purification.

[0113] Sample preparation

[0114] 1.2.1. Anhydrous IL

[0115] [DBNH][OAc] was prepared by slow and controlled addition of equimolar amounts of HOAc to DBN. The mixture was initially stirred and cooled at 25°C to transfer the exothermic reaction enthalpy. When the mixture was close to equimolar amounts, the system was heated at 70°C to avoid crystallization of the IL. The system was kept at this temperature for another hour under mixing to ensure that the reaction proceeded to completion. The water content of the synthesized IL was less than 0.5 wt%.

[0116] pass 1This IL sample was analyzed by H NMR to determine its A:B ratio. Additional samples were prepared by mixing the initial IL with known amounts of HOAc or DBN to achieve the target A:B ratio. These samples were then mixed for homogenization and purified by 1 H NMR was reanalyzed to finally determine the A:B ratio. The samples prepared were in the A:B molar ratio range of 0.4-1.98.

[0117] 1.2.2.IL / H2O mixture

[0118] IL / H2O mixtures were prepared by diluting anhydrous IL samples with water. The IL / H2O mixtures were adjusted to the target concentrations using a 1 mg precision electronic scale during the dilution process. The IL concentration range covered was large enough to simulate concentrated and diluted samples (1.25 wt% - 90 wt% IL in IL / H2O mixtures) with different molar A:B ratios (0.4-1.98). The IL / H2O mixtures were split into a calibration set and a validation set for the model calibration and validation procedures. Because each sample can be represented by its A:B ratio and IL concentration, the sample subsets were Figure 1 Shown in.

[0119] Figure 1 A sample of IL / H2O mixture according to the example is shown.

[0120] 1.3. 1H NMR measurements

[0121] Anhydrous IL samples were passed 1 H NMR characterization. The samples were loaded into standard 5 mm NMR tubes and dissolved in DMSO-d6. Spectra were collected using a Bruker 400 MHz Ultra Shield NMR instrument (Bruker, Billerica, Massachusetts, USA) at 23 °C. NMR spectra were collected following a standard zg30 sequence with 8 transients and an acquisition time of 2.56 s. All spectra were referenced to tetramethylsilane. ACD / 1D NMR processor software was used to process raw data. The A:B ratio was calculated by considering the ratio between the peak areas associated with the acid and the base, respectively.

[0122] 1.4. Raman spectrum acquisition

[0123] Samples were analyzed with an Alpha 300R confocal Raman microscope (WITec GmbH, Germany) under ambient conditions. Nearly 100 μL of IL / H2O mixture was spread on a microscope concave slide and covered with a cover glass. Raman spectra were obtained by using a frequency-doubled Nd:YAG laser (532.35 nm) and a Nikon 20x (NA=0.95) air objective lens with constant power. The Raman system was equipped with a DU970 N-BV EMCCD camera behind a 600 line / mm grating. The excitation laser was horizontally polarized. After fixing the focus using the microscope mode, each single spectrum was collected, averaging 32 scans with an integration time of 0.1 s / scan. The baseline of the spectrum was corrected using WITec Project 1.94 (WITec GmbH, Germany) using a second-order polynomial equation.

[0124] A total of 75 spectra were collected with at least two replicates per sample. The calibration set consisted of 54 spectra and the validation set consisted of 21. The samples and spectra of these two sets were independently prepared and collected on different days.

[0125] 1.5. Development of calibration methods for determination of acid, base and H2O concentrations in IL / H2O mixtures

[0126] Univariate calibration

[0127] Univariate calibration is the simplest way to build a quantitative calibration model. It is based on finding the relationship between a single Raman spectroscopy variable (peak intensity, peak area, peak shift) and the analyte concentration. (The Mathworks, Inc., Natick, MA, USA) routines were used for Raman spectral analysis, correlation identification, and calculation of calibration model parameters.

[0128] The coefficient of determination (R 2 ) and the root mean square error of calibration (RMSEC) and the root mean square error of prediction (RMSEP) expressed as follows to evaluate the quality of the fit:

[0129]

[0130] in and denote the measured and predicted values, respectively, and N denotes the number of samples in the calibration dataset, and

[0131]

[0132] in and denote the measured value and the predicted value, respectively, and K denotes the number of samples in the validation dataset.

[0133] Although univariate calibration can work successfully in some cases, it sometimes fails to give good results when dealing with complex Raman spectra that exhibit overlapping bands or unexpected shifts within the calibration range. In these cases, multivariate methods are more suitable for calibration model building

[19]

[20] .

[0134] 1.5.2. Multivariate Calibration: Partial Least Squares Regression (PLS)

[0135] When building calibration models using spectral data and multivariate analysis, two main problems are often encountered: there are typically more spectral variables than samples, and high variable collinearity is encountered in most spectral data. This makes classical multiple linear regression (MLR) based on the raw spectral variables either impossible or highly unstable.

[0136] A robust alternative to MLR is Partial Least Squares (PLS) regression. PLS projects the data (in our case, the spectral intensities) into a new space with a smaller number of new variables. These new variables are called components or latent variables (LVs). PLS calculates the LV in a way that maximizes the covariance between the X-blocks (spectral intensities) and the Y-blocks (concentrations of different substances).

[0137] LVs are mutually orthogonal, which suppresses the problems associated with matrix inversion when computing the model coefficients. In addition to the mutual orthogonality of LVs, the reduction in spatial dimensionality also makes PLS particularly suitable for building prediction models from spectral data. A very good explanation of the principle of PLS ​​is given in

[21] .

[0138] During the calibration procedure, the spectra collected are placed as rows in an X matrix (independent block) with n rows and k columns. Each row represents a spectrum, and each column represents a single wavelength. The variables we aim to predict are placed in a Y matrix (dependent block) with n rows and m columns. Each row represents a sample, and each column represents a single variable.

[0139] In short, in PLS, there is an outer relationship where the X-block and Y-block are decomposed into a score matrix and a loading matrix:

[0140] X=TP′+E

[0141] Y=UQ′+F *

[0142] in:

[0143] T and U represent the score matrices of X-block and Y-block respectively.

[0144] P′ and Q′ represent the score matrices of X block and Y block respectively.

[0145] E and F * Represent the residual matrices of block X and block Y after being projected onto a limited number of LVs, respectively.

[0146] There is also an inner relationship between the scores that relates the two blocks.

[0147] U=BT,

[0148] where B is the matrix containing the regression vectors.

[0149] The blend relation can be written as:

[0150] Y=TBQ+F,

[0151] where ||F|| is to be minimized.

[0152] This hybrid relationship ensures the ability to use the model parameters to make future predictions from the test set.

[0153] In this study, the X block contains the Raman spectrum of each sample, and the Y block contains the weight fractions of acid, base, and H2O in each sample.

[0154] Through R 2 Evaluation of coefficients, RMSEC, and RMSEP investigated the number of LVs, quality of fit, and model validation.

[0155] use and PLS Toolbox (Eigenvector Research, Inc., Manson, USA) software package was used for data analysis and model building.

[0156] 2. Results and Discussion

[0157] 2.1. Anhydrous IL samples 1 H NMR

[0158] The NMR spectra of the anhydrous IL and the corresponding A:B ratios are Figure 2 The DMSO-d6 singlet state is seen at ≈2.5 ppm. The singlet states observed in the range of ≈1.6 ppm-1.75 ppm correspond to the protons of the methyl groups in the acetic acid / acetate molecules. The remaining multiplets correspond to the different protons of the base and its protonated forms. Detailed information on the chemical shifts of the acid and base protons is given in the Supplementary Material.

[0159] As expected, only mixed peaks are observed, not peaks representing molecular and ionic species separately. For a fast exchange system, assuming that the observed1 The H chemical shift is a weighted average of the molecular and ionic species.

[0160] When increasing the A:B ratio, it can be noted that the different peaks associated with HOAc, DBN and their ionized forms shift towards higher ppm values. These shifts can be directly related to the different shielding / deshielding effects experienced by the protons when changing the A:B ratio.

[0161] Figure 2 Shown are NMR spectra of anhydrous ILs with different A:B ratios according to Examples (the A:B ratio of each sample is noted on the left side of the figure).

[0162] For example, as the A:B ratio increases, the protons belonging to the acid methyl groups (1.6ppm-1.8ppm) shift downfield. This is probably due to the higher hydrogen bonding network between these protons and the oxygen atoms of the carboxylate groups in the adjacent free acetic acid molecules. This reduces the shielding around these protons and causes them to resonate at a higher frequency.

[0163] The position of the methyl group singlet state is linearly related to the A:B ratio, with a clear increase in slope starting from A:B = 0.85 indicating higher proton deshielding above this value. + Similar linear relationships and slope changes were found for the equivalent protons bonded to C3 and C9 in . However, the slope changes were different. The slope was significant below A:B = 1.1 and decreased significantly above this value, which means that proton deshielding decreased significantly above this value. These results are illustrated in the Supplementary Materials. As explained in

[22] , different chemical shifts can be used to calculate IL ionicity.

[0164] The A:B ratio is calculated by taking the ratio of the integrated area of ​​the methyl group singlet state (acid) to the sum of the integrated areas of the different multiplet states (base). The area of ​​the triplet state observed in the range of ≈3.27ppm-3.95ppm (corresponding to two hydrogens bonded to C9) is used as a reference. The estimated relative error for calculating the A:B ratio is less than 5%.

[0165] Raman spectroscopy

[0166] 2.2.1. Raman spectra of acids, bases and ILs

[0167] The area-normalized Raman spectra of DBN, HOAc and IL (A:B = 1.05) are shown in Figure 3 As shown. In the HOAc spectrum, ≈895cm -1 and ≈2950cm -1 The two main peaks at ≈1665cm are attributed to CC and CH stretching vibrations. -1The medium intensity band at can be attributed to C=O stretching vibration

[23] .

[0168] Figure 3 Raman spectra of DBN, HOAc, and anhydrous IL according to Examples are shown (A:B=1.05).

[0169] To the best of our knowledge, the assignment of specific Raman bands for DBNs has not been proposed in the literature. The band assignments in this paper are based on Raman spectra tables

[24]

[25] . -1 and 732cm -1 The peak at 518cm may be due to CC vibration. -1 The peak at 1650cm can be attributed to CN vibration. -1 The peak at 1452 cm -1 and 2850cm -1 The peak at 313 cm -1 、518cm -1 and 732cm -1 The peak at does not overlap with the peak of HOAc and may be a good candidate for building a calibration model.

[0170] The Raman spectrum of IL (A:B = 1.05) shows a complex pattern in which we can distinguish some features from the spectra of HOAc and DBN, respectively. For example, it can be noted that the 313 cm -1 、518cm -1 and 732cm -1 The peaks at 800 cm-1 show a decrease in intensity compared to pure alkali without any significant changes in shape or peak position, again confirming their potential use for calibration. -1 -1000cm -1 or 2800cm -1 -3200cm -1 The overlapping peaks between HOAc and DBN are more complicated to explain and are caused by the overlap / displacement of the HOAc and DBN peaks. The DBN peak associated with the -CH2 vibration is observed to shift from 2850 cm-1 in the soda ash to 2850 cm-1 in the soda ash. -1 to 2880cm in IL -1 Ionization and the interaction between the molecule and ionized HOAc and DBN may be responsible for these shifts.

[0171] 2.2.2. Raman spectra of anhydrous A / B mixtures in excess of base or acid

[0172] Figure 4The Raman spectra of the ILs with nearly equimolar amounts of acid and base and with excess base or acid are shown. We have already mentioned that the 1650 cm -1 The peak intensity near 1650cm-1 decreases sharply in nearly equimolar ILs. When the composition deviates from equimolar, we can observe that when DBN is in excess, -1 The peak intensity increases significantly when the acid is in excess. We can notice that the peak at 1700 cm -1 Based on these observations, we can conclude with greater certainty that these peaks are associated with C=N and C=O vibrations in the non-ionized DBN (base excess) and HOAc (acid excess) forms.

[0173] In fact, for nearly equimolar compositions, once proton exchange occurs when acid and base are mixed, most of the molecules are ionized, the two double bonds are delocalized, and no "real" C=O or C=N vibrations are seen in the Raman spectrum of the IL. When there is an excess of acid or base, these bands are more prominent due to the presence of neutral molecules.

[0174] It is also worth mentioning here that in the case of protic ILs such as [DBNH][OAc], a chemical equilibrium occurs between ionized and neutral species (the equilibrium constant is directly related to the so-called ionicity of the IL). This then explains why the band intensity of the double bond does not completely disappear even in near equimolar compositions (in the presence of neutral species) [4].

[0175] Figure 4 Raman spectra of anhydrous ILs with different A:B ratios according to examples are shown.

[0176] Excess base or acid in the mixture also affects the high wavenumber region (2200cm -1 -3200cm -1 ). At low A:B ratios (excess base), the -1 -2900cm -1 The area in the range increases with the appearance of a shoulder at low wavenumbers most likely associated with molecular vibrations in the free base, while at high A:B ratios, the peak at 3000 cm -1 -3100cm -1 The area within the range is increased mainly due to the presence of molecular acid.

[0177] 2.2.3. Effect of water on the Raman spectra of ILs as a function of the A:B ratio

[0178] So far, we have demonstrated that excess acid or base induces distinct changes in the Raman spectra of A / B mixtures. In this section, we discuss the effect of water dilution of the IL. The spectra of water-diluted IL samples (50 wt% water) with different A:B ratios are shown in Figure 2. Figure 5 Shown in.

[0179] Dilution with water does induce a clear change in the Raman spectrum compared to anhydrous IL. -1 -3700cm -1 A broad band attributed to different OH vibration modes in water molecules appears in the wavenumber range of 1595 cm -1 Another peak at 1650 cm-1, which is attributed to the bending vibration mode of water, appears nearby

[26] . -1 The peak at 1600 cm-1 is dramatically reduced in the presence of water (probably due to ionization) and is only visible in samples with low A:B ratios. -1 and 1700cm -1 The peak at is also visible in dilute samples with high A:B ratios. Here, too, the chemical equilibrium defines the degree of ionization of the different molecules.

[0180] Figure 5 Shown are area normalized Raman spectra of 50 wt % water-diluted IL samples according to the Examples (color code is related to the A:B ratio).

[0181] It was also reported that upon the introduction of water, the IR and NMR spectra of imidazolium-based ILs with different anions underwent drastic changes due to the interaction between IL and H2O molecules

[27] .

[0182] 2.2.4. Raman spectra of IL / water mixtures

[0183] Raman spectra of different IL / H2O samples Figure 6 As shown in the figure, at 3000cm -1 -3700cm -1 In the wavenumber range of , the intensity of the broad peaks attributed to different OH vibration modes in water molecules increases with the dilution factor, while the intensity of scattering from acid and base molecules / ions in the IL (almost all the remaining spectral bands) decreases. This is expected because the Raman scattering intensity is proportional to the concentration of the scattering molecules.

[0184] Although the presence of water causes a significant change in the Raman spectrum of the IL, its scattering does not significantly interfere with the scattering produced by the IL molecules.

[0185] Figure 6Shown are area-normalized Raman spectra of different IL / H 2 O solutions according to the examples (the color code reflects the IL concentration in the IL / H 2 O mixture).

[0186] 2.2.5. Univariate calibration method for determining H2O concentration in IL / H2O mixtures

[0187] To predict the H2O concentration in H2O / IL mixtures, the 2200 cm -1 -3800cm -1 Although at 3000cm -1 -3200cm -1 There is a small overlap between the peaks attributed to IL and H2O in the region, but the respective signals are still relatively well separated.

[0188] We have developed a very simple method for calibration. In our spectrum processing procedure, the spectral range is first narrowed to 2200 cm -1 -3800cm -1 The original spectra were then area normalized. The peak areas associated with IL and H2O were defined as follows:

[0189]

[0190]

[0191] H2O peak area H2O 峰面积 The graphical representation as a function of H2O concentration is Figure 7 Shown in.

[0192] Figure 7 According to the example, the -1 -3800cm -1 The integrated area of ​​the Raman scattering intensity in the region versus the water concentration (blue circles: calibration dataset. Black stars: validation dataset. Solid line: calibration model. Dashed lines: 95% confidence interval limits).

[0193] The H2O in the IL / H2O mixture was obtained. 峰面积 A significant linear relationship with the H2O mass fraction was observed for samples with different A:B ratios at equivalent H2O mass fractions. In fact, some spectral features at 2200 cm -1 -3010cm -1 Changes occur within the region, especially at low and high A:B ratios as described previously. These changes can be particularly observed for samples containing 50 wt% water, if the A:B range is the largest.

[0194] Overall, these changes do not significantly affect the model. The RMSEC and RMSEP are 2.234 wt% and 2.257 wt%, respectively. These values ​​are reasonable considering the simple procedure and the large concentration range of H2O content from 10 wt% to 98.75 wt%. The prediction errors are comparable to those reported by Viell et al.

[12] , who used advanced spectral processing methods to quantify H2O in H2O / IL mixtures using mid-infrared (mid-IR) spectroscopy. The prediction errors reported in their study were less than 2.3 wt% over the entire concentration range. Table 1 gives the model parameters and fit quality metrics for predicting water concentration in IL / H2O mixtures.

[0195] Table 1: Model parameters and fit quality metrics for predicting water concentration in IL / H2O mixtures.

[0196]

[0197] 2.2.6. Univariate calibration method for determining the base concentration in IL / H2O mixtures

[0198] The Raman scattering intensity of the IL components (acid and base) increases with the increase of IL concentration in the IL / H2O mixture, e.g. Figure 6 Some peak intensities show a very good correlation with the base concentration, especially at 743 cm -1 and 2890cm -1 The intensities of these peaks as a function of the base concentration in the IL / H2O mixture are Figure 8 Shown in.

[0199] However, the relationship between the peak intensity and the base concentration is not linear. The reason for this nonlinearity over the range of base concentrations is unclear. Polynomial models can be used to approximate complex nonlinear relationships because they are simply Taylor series expansions of the unknown nonlinear function. However, we found that the observed relationship can be well described by simple power law models, which we used for the calibration equations.

[0200] Figure 8 According to the example, the -1 and 2890cm -1 Evolution of the peak intensity at as a function of the base concentration in the IL / H2O mixture (blue circles: calibration dataset. Black stars: validation dataset. Solid line: calibration model. Dashed lines: 95% confidence interval limits).

[0201] like Figure 8 As shown in Table 2, the fitting quality is relatively good within the range of base concentration covered by the two peaks. Table 2 gives the model parameters and fitting quality indicators (R2 , RMSEC and RMSEP).

[0202] For both peaks, RMSEC and RMSEP are below 2 wt%, which is also a fairly reasonable value for the coverage. However, using 2890 cm -1 A peak at will yield better model sensitivity because of the greater intensity range over the same covered base concentration range.

[0203] Table 2: Model parameters and fit quality metrics for predicting base concentrations in IL / H2O mixtures.

[0204]

[0205] So far, univariate calibration has shown good potential for quantifying water and base in IL / H2O mixtures. However, univariate calibration also shows limitations because we cannot predict the acid concentration in IL / H2O mixtures using a single spectral feature. Assuming that only water, acid, and base are present in the solution, the acid concentration can still be obtained by the difference to 100%.

[0206] Using multivariate calibration can greatly improve the quality and performance of the calibration model because it exploits all the information in the spectrum rather than focusing on just a single variable. In the next section, we will discuss the potential of one of the multivariate calibration methods, namely PLS regression.

[0207] 2.2.7. PLS regression for simultaneous determination of IL composition and H2O content in IL / H2O mixtures.

[0208] Preprocessing is a very important step when performing multivariate data analysis. When trying to build a PLS calibration model, we tested several spectral preprocessing methods and evaluated them based on the quality of model fit. The preprocessing strategies retained included baseline correction, smoothing (10-point moving window), area normalization, and mean centering of the raw spectra (X blocks). The Y block data was mean centered.

[0209] The number of model LVs was chosen based on minimization of RMSECV and visual inspection of the LVs. The model with four LVs produced a very good fit, with more than 99% explained variance in both the X and Y blocks.

[0210] The measured and predicted values ​​of acid, base and H2O concentrations for the calibration and validation datasets are presented in Fig. 9 The reader can see that the different data points of the two sets are well distributed around the y=x line, which reflects a good model fit quality.

[0211] Fig. 9Shown are predicted and measured graphs of acid, base, and H2O concentrations in H2O / IL mixtures according to examples.

[0212] The PLS calibration method is clearly successful in simultaneously predicting the acid, base, and H2O content in IL / H2O mixtures. The PLS model statistics are shown in Table 3.

[0213] Table 3: PLS model statistics for predicting acid, base, and HO contents in HO / IL mixtures.

[0214]

[0215] The RMSE is reasonable with respect to the relatively wide coverage of the three variables. It must be noted that some validation set samples have acid, base, and H2O concentrations from the calibration set samples, which also reflects the good model predictive ability.

[0216] To the best of the present inventors' knowledge, these results are the first to demonstrate in the literature the potential of Raman spectroscopy combined with chemometrics for the rapid and quantitative determination of Il components and HO content in IL / HO mixtures. The applicability of this approach can be extended to a wider range of processes involving IL / HO mixtures [6]-[9].

[0217] 2.2.8. Discussion of PLS ​​calibration model

[0218] Table 4 gives the percentage variance captured by the PLS regression model for Block X and Block Y. LV1 and LV2 have explained close to 99% of the variance in the data. The latent variable LV1 explains the largest portion of the variance, 96.53% and 93.14% in Block X and Block Y, respectively.

[0219] Table 4: Percentage variance captured by the PLS regression model

[0220] Latent variables X-block explained variance, % Y-block explained variance, % 1 96.53 93.14 2 2.31 2.86 3 0.53 3.24 4 0.38 0.40

[0221] The score plot based on the first two LVs and the loading plots for the four LVs are shown in Fig.10 The scores represent the coordinates of the samples in the new LV space. Samples with similar scores on the same LV are similar with respect to that LV. The corresponding loadings are used to explain what causes the changes in the scores and can help to clarify the physical meaning of a particular LV. Specific loadings show the spectral features responsible for the patterns in the score plot. A high loading value for a certain wavelength (or wavelength range) means that the former contributes greatly in establishing the latent variable.

[0222] The positive peak in the LV1 loading spectrum shows an almost complete water peak signal. The negative peak constitutes an almost opposite anhydrous IL spectrum. From our previous discussion, we can infer that LV1 mainly explains the changes in IL (correspondingly, water) concentration in the mixture.

[0223] This observation is confirmed when looking at the LV1-LV2 score plot, where the scores are colored according to the water concentration in the mixture. Samples with IL concentrations close to the average concentration (about 40 wt%) have a score value of zero because the data is mean centered. We can see that samples with negative LV1 scores have IL concentrations below this average, while samples with positive LV1 scores have IL concentrations below this average.

[0224] From the loading plot of LV2, it can be seen that this latent variable is independent of the H2O concentration in the sample, because the loading value of LV2 is close to zero in the wavenumber region corresponding to OH vibration.

[0225] The LV2 loadings show both positive and negative peaks. A closer look reveals that the negative peaks correspond to the vibrational frequencies of the chemical bonds found in bases, and the positive peaks occur at the vibrational frequencies of acids (see previous discussion on peak assignments). This is confirmed on the score plot, where the samples are found to separate along LV2 based on the A:B ratio.

[0226] The same applies to LV3 and LV4, which will explain the changes in the A:B ratio rather than the concentration, since the respective loadings show features related to molecular vibrations in A and B. Both LVs will also partially explain the presence of molecular acids and / or bases in the sample, since some spectral features related to them appear on the loading plots.

[0227] In summary, LV2, LV3, and LV4 were associated with the variation in the A:B ratio in the mixture, whereas LV1 explained the variation in the IL concentration in the mixture.

[0228] Fig.10 Score plots based on LV1 and LV2 (top) and loading plots for four LVs (bottom) are shown according to an example.

[0229] Part 2: Quantification of ionic liquid degradation products and improved prediction performance by interval PLS.

[0230] The composition of the liquid stream may change as ionic liquids such as 1,5-diazabicyclo[4.3.0]non-5-ene (DBN) may undergo degradation such as reversible hydrolysis to degradation products such as 1-(3-aminopropyl)-2-pyrrolidone (APP) which also forms 1-(3-aminopropyl)-2-pyrrolidonium acetate ([APPH][OAc]) with acetic acid ( Fig.11 ). In addition, according to Fig.11In the second reaction shown, [APPH][OAc] may undergo condensation or decomposition to become 1-(3-acetamidopropyl)-2-pyrrolidone (APPAc).

[0231] Fig.11 The hydrolysis of [DBNH][OAc] (1) to [APPH][OAc] (2) and the condensation to APPAc (3) are shown according to the examples.

[0232] Some examples are provided for quantifying ionic liquid degradation products. In one embodiment, the degradation product may be a hydrolysis product, such as APP, and while some examples are provided for this particular degradation product, any other degradation product may also be applicable.

[0233] 3. Materials and Methods

[0234] 3.1. Raw materials

[0235] Samples of DBN (CAS No. 3001-72-7; purity ≥ 99.0% by mass) and HOAc (CAS No. 64-19-7; purity ≥ 99.8% by mass) were purchased from Fluorochemicals and Sigma-Aldrich, respectively, and used without further purification.

[0236] Experimental procedures

[0237] Sample preparation

[0238] [DBNH][OAc] stock solutions were synthesized using the procedure described in (Guizani et al. 2020). 1-(3-Aminopropyl)-2-pyrrolidonium acetate ([APPH][OAc]) was synthesized by the University of Helsinki. The stock solution of [APPH][OAc] contained DBN in part. The two stock solutions used for sample preparation were characterized using capillary electrophoresis, NMR, and Karl Fischer titration to determine the initial AcOH, DBN, APP, and H2O concentrations. Their specifications are given in Table 8 in the Electronic Supplementary Information (ESI).

[0239] From these stock solutions and distilled H2O, forty samples of defined composition (each weighing more than 3 g) were prepared by gravimetric method using an electronic weighing scale with an accuracy of 0.1 mg. The forty samples can be divided into four categories according to the water content (about 0 wt %, about 25 wt %, about 50 wt % and about 75 wt %). Table 5 gives the concentration ranges of the four individual molecular components. The compositions of the forty samples are Fig.19The training samples were prepared so that they spanned a wide range of the four analyte concentrations and covered as wide a range of sample compositions as possible that might be encountered in a process liquid stream. Table 5 gives the concentration ranges of the four molecules.

[0240] Table 5: Concentration ranges of AcOH, DBN, APP and H2O in the prepared mixtures

[0241] molecular AcOH DBN APP <![CDATA[H2O]]> Concentration range, weight % 7.5-32.6 2.7-67.4 0-58.4 0.1-75.2

[0242] 3.2.2. Refractometry

[0243] The reader might reasonably wonder whether other simpler online methods could be considered instead of Raman spectroscopy. We asked ourselves similar questions when screening alternative analytical methods. Refractometry is widely used for online process monitoring and control and is suitable for quantifying IL concentrations in aqueous mixtures (Liu et al. 2008)(Kaneko et al. 2018). Therefore, we considered refractometry to be an important alternative that should be investigated and performed refractive index (RI) measurements on a set of forty samples to evaluate the potential of refractometry. RI was measured at 293.15 K using a Peltier heated Abberefractometer (Abbemat 300, Anton Paar, Austria).

[0244] Raman spectroscopy

[0245] Samples were analyzed with an Alpha 300R confocal Raman microscope (Witec GmbH, Germany) under ambient conditions. Samples approaching 100 μL were spread on a microscope concave slide and covered with a cover glass. Raman spectra were obtained by using a frequency-doubled Nd:YAG laser (532.35 nm) with a 30 mW constant power and a Nikon 20x (NA = 0.4) air objective. The Raman system was equipped with a DU970 N-BV EMCCD camera behind a 600 line / mm grating. The excitation laser was horizontally polarized. After fixing the focus using the microscope mode, each single spectrum was collected, averaging 32 scans with an integration time of 0.5 s / scan. Forty spectra were collected for a total of forty mixtures.

[0246] Data analysis

[0247] use (MathWorks Inc.) performed data analysis and graphics

[0248] 3.3.1. Exploratory Data Analysis: Principal Component Analysis (PCA)

[0249] The spectra were first baseline corrected using a second-order polynomial and then area normalized. PCA was performed on the preprocessed mean-centered spectra. For more detailed information on PCA, the reader is referred to the professional literature (Brereton 2003)(Geladi 2003)(Geladi et al. 2004).

[0250] 3.3.2. Partial Least Squares Regression (PLS)

[0251] The PLS1 algorithm was used in this study to generate models for each component in the sample set. PLS modeling employed the same preprocessing methods as described for PCA. Decomposition into latent structures was accomplished by maximizing the covariance between the sample preprocessed spectra and their analyte-specific mean centered concentrations using the NIPALS algorithm (Geladi and Kowalski 1986).

[0252] Model validation and selection of a sufficient number of latent variables for the PLS model were performed using a model cross-validation procedure based on the leave-one-out method. The cross-validation root mean square error RMSECV was used as a quantitative measure of the selected model LV. RMSECV was calculated using the following formula:

[0253]

[0254] where y i and denote the measured value and the predicted value, respectively, and n denotes the number of samples in the data set.

[0255] 4. Results and Discussion

[0256] 4.1. Limitations of Refractometry

[0257] Refractometry was first considered due to its simplicity and demonstrated applicability in the analysis of mixtures of IL and water (Liu et al. 2008)(Kaneko et al. 2018). Therefore, we would like to discuss the options for further development of Raman analysis based on the results we obtained from refractometry. The RI evolution of the forty samples was Fig.12 In addition to the RI measurements of the forty samples, we also measured the RI of aqueous solutions of [DBNH][OAc] / H2O mixtures without APP to evaluate the RI sensitivity for detecting APP formation.

[0258] The RI evolved linearly with the IL mass fraction and the trend was very similar in the presence or absence of APP. Samples measured at similar water contents but with larger differences in APP content did not show any significant differences in the RI values, although some scatter in the RI values ​​was visible at low dilution levels. Overall, these results suggest that refractometry is very limited in probing the extent of DBN hydrolysis to APP and that alternative more sensitive analytical methods are needed.

[0259] Fig.12 Shown is a comparison of RI as a function of IL weight % in a mixture with water in the presence and absence of APP according to examples.

[0260] 4.2. Raman spectroscopy of liquid mixtures

[0261] Pre-processed Raman spectra of different mixtures Fig.13 The spectra are colored according to the wt% concentration of H2O in the mixture. Samples with the same water content are clearly divided into four different categories corresponding to the four dilution levels. 3000 cm -1 -3700cm -1 The scattering intensity in the region is caused by the OH stretching vibration of water molecules and increases with the water content (Sun 2009). -1 -2000cm -1 The scattering intensity in this region decreases with decreasing dilution level because it is mainly associated with other molecules. The bending mode of water (~1640 cm -1 )( et al. 1991) is less affected due to the low intensity compared to the scattering intensity of other molecules, as discussed in our previous article (Guizani et al. 2020).

[0262] Fig.13 Pre-processed Raman spectra of different mixtures according to the examples are shown (the color code describes the wt. % H2O in the mixture, for clarity, the spectra are divided into the fingerprint region (lower) and the high frequency region (upper).).

[0263] Effect of H2O addition

[0264] The spectra of four APP-free samples with different water contents are shown in Fig.14 The band assignments were made according to the existing literature on Raman spectroscopy-structure correlations and characteristic group frequencies (Larkin 2011).

[0265] In the sample without APP, 464 cm -1 and 518cm -1The peak at 740cm originates from the CNC bending / deformation mode in DBN. -1 The peak near 920cm is most likely derived from CC vibration in DBN. -1 and about 2930cm -1 The two main peaks at about 2890 cm-1 are attributed to the CC and CH stretching bands in AcOH, respectively. -1 and about 2980cm -1 The peaks at about 1650 cm -1 The moderate intensity band at can be safely attributed to the C=O stretching band from AcOH.

[0266] After adding water, due to the dilution effect, 280cm -1 -3000cm -1 In contrast, when the water content increases, the scattering intensity in the region around 3100 cm- 1 -3700cm- 1 The broad peak associated with the OH vibration of water molecules increases significantly at 240 nm. It is well known that Raman scattering of analytes can show a strong dependence on the molecular environment (Kauffmann and Fontana 2015). Therefore, in addition to the intensity changes due to the analyte concentration, changes in the molecular environment can also lead to band shifts and shape changes.

[0267] Upon addition of water, the analyte is expected to interact with the water molecules via hydrogen bonds. These would partially explain some of the changes in the spectra of the diluted samples other than the decrease in intensity. For example, the C=O stretching band shifts downward to about 1600 cm -1 , which is most likely due to the structural changes of the solution in the presence of water (Nakabayashi et al. 1999; Gofurov et al. 2019). In addition, after the addition of H2O, the -1 The band at 2800 cm-1 disappears, which may indicate the absence of a specific AcOH structure (dimer or trimer) that is only present in the anhydrous IL. At higher frequencies, the reader can note that the band at 2800 cm-1 disappears. 1 -2870cm- 1 The significant intensity reduction in the region reflects the modification of the DBN vibration mode in the presence of water.

[0268] Fig.14 Pre-processed Raman spectra of four APP-free IL samples with different dilution levels according to the examples are shown (for clarity, the spectra are divided into the fingerprint region (lower) and the high frequency region (upper).)

[0269] Impact of adding apps

[0270] For both the almost anhydrous mixture and the mixture containing 75 wt% water, the Raman spectra of the sample without APP and the IL sample with APP / DBN = 4.47 mol / mol are Fig.15 The reader will notice that, regardless of the water content, the spectra change dramatically as the APP / DBN ratio changes. For example, around 464 cm -1 and about 516cm -1 The bands at 2890 cm-1 are significantly reduced after adding APP. These may originate from the CNC bending / deformation modes in DBN, which are absent in APP, even though there is still a CNC bond system left in APP. In the high-frequency region, the bands at about 2890 cm-1 attributed to the in-phase and out-of-phase stretching of -CH2 in DBN -1 and about 2980cm -1 The band at was significantly reduced after adding APP.

[0271] After adding APP, about 1640cm -1 -1675cm -1 The peak near 332 cm is broadened and is thought to represent overlapping contributions from the C=O of the keto group in APP and the C=O of the carboxylic acid group in AcOH. As mentioned previously, a clear shift in the wavenumber range is seen when water is added to the organic mixture. The scattering intensity at 332 cm -1 -340cm -1 The region was increased and was assigned to the δC-C vibration mode present in the aliphatic amino-propyl chain of APP.

[0272] Fig.15 Pre-treated Raman spectra of a sample without APP and a sample with APP / DBN = 4.47 mol / mol at two H2O contents of 0 wt% (left) and 75 wt% (right) according to the example are shown (for clarity, the spectra are divided into a fingerprint region (bottom) and a high frequency region (top).).

[0273] The observed changes in spectral features due to changes in APP / DBN ratio and H2O content encouraged the development of quantitative analysis using Raman spectroscopy.

[0274] 4.3. Principal Component Analysis (PCA)

[0275] PCA is a method for data reduction and visualization. PCA is central to chemometrics and is often used for exploratory multivariate data analysis and unsupervised pattern recognition. In PCA, the dimensionality of a dataset is reduced by transforming the original spectral dataset into a smaller dataset consisting of a small number of uncorrelated variables (PCs) that retain most of the variation present in all the original variables. The goal is to identify the direction of greatest variability in the data and interpret it in terms of the underlying chemistry.

[0276] PCA was performed on the preprocessed (background corrected, area normalized and mean centered) data (see Fig. 22 The results show that the first three PCs capture more than 99% of the data variance (see Fig.23 ). The first PC explains 89% of the variance, while the next three PCs explain 5.1%, 4.4%, and 0.52% of the variance, respectively. Therefore, the pseudo-rank of the data matrix should be three or four, which is physically reasonable because four different molecules are present in the mixture and contribute to the scattering. Sample N35 showed difficulty during pre-processing (background subtraction) and has a very high Hoteling T 2 and Q residual values. Therefore, sample N35 was considered an outlier and was excluded from the decomposition procedure.

[0277] The scores and loadings of the first three PCs are Fig.16 For PC1, the scores are colored according to the weight % of H2O in the mixture. The PC1 scores are related to the water content in the different mixtures. Samples with the same water content have similar scores on PC1, which is around 280 cm -1 -1800cm -1 The region shows positive contributions (mainly corresponding to the scattering of DBN, APP, and AcOH), and at about 3020 cm -1 -3720cm -1 The region shows a negative contribution (mainly corresponding to the scattering of H2O).

[0278] PC2 and PC3 separate the samples within each group based on the proportion of DBN in the sum of DBN and APP. PC2 and PC3 also show some interesting features. In PC2, the anhydrous samples and the samples with the highest water content have negative scores and are separated from the less extreme samples with 25 wt% and 50 wt% H2O and positive scores, respectively. In PC3, the spread of scores becomes narrower due to increasing dilution, and the shape of the scores for the different samples is very similar to the shape of the DBN weight % in the mixture, as shown in Figure 2. Fig. 20 In summary, the PCA decomposition shows that the variance in the data can be almost completely captured in the first four components. Depending on the sample composition, these four PCs reflect most of the chemical information.

[0279] Fig.16 Score plots (left) and loading plots for the first three PCs according to the example are shown.

[0280] 4.4. Partial least squares regression (PLS) for quantification of DBN, APP, AcOH, and H2O in liquid mixtures

[0281] 4.4.1. PLS model based on full spectrum

[0282] The PLS model was established on thirty-nine samples after discarding sample N35, which showed some signal anomalies when included in the model, and background correction was unsuccessful, resulting in high Q residuals and obvious abnormal behavior. The results of the PLS regression are shown in Table 6. Since the regression uses the PLS1 algorithm, each component is modeled separately. The motivation for this choice is that the error produced by the PLS1 regression is lower than that of PLS2, in which all components are modeled simultaneously (Brereton 2003).

[0283] For different models, the number of selected LVs varied between 3 and 5. Each model captured more than 96% of the prediction variability (spectra) and more than 99% of the response variability (concentrations).

[0284] Table 6: PLS regression results: explained variance and RMSECV

[0285]

[0286] The RMSECVs for AcOH, DBN, APP, and H2O are 0.23 wt%, 2.09 wt%, 1.15 wt%, and 0.64 wt%, respectively. The model shows that AcOH and H2O are better predictable than DBN and APP, although the results for the latter two molecules are still in a good range. Fig.17 Cross-validated predicted versus measured concentrations are shown for AcOH, DBN, APP, and H2O. The reader can notice that the data points for the four molecules are very close to the 1:1 identity line. The Pearson correlation coefficients for all four examples are above 0.99. Overall, this analytical approach shows very good results even for more complex mixtures containing APP. Additional data on the PLS model are given in the ESI. Finally, it is important to emphasize the universality of this analytical approach based on spectroscopy and multivariate analysis. The applications are almost endless, and the reader is encouraged to try to apply them to their own systems where changes in signal intensity may be correlated with analyte concentration.

[0287] Fig.17 Cross-validated predicted and measured concentrations of AcOH, DBN, APP, and H2O according to examples are shown.

[0288] 4.4.2. Improving the prediction performance of PLS ​​models through variable selection

[0289] The goal of variable selection is to obtain a model that is easier to understand and has better predictive performance. In the search for the best variable selection procedure, one might be tempted to try all possible combinations of predictor variables to select the best one. However, this proves to be prohibitive due to the large number of variables and induces a high risk of overfitting when the number of variables is higher than the number of samples (Andersen and Bro 2010). Both situations are encountered when dealing with spectral data. Therefore, the goal here is not to find the best model, but to find a model that is better in terms of prediction and understanding.

[0290] In this regard, we adopted a simple approach in which the spectral range was divided into 10 subintervals and the PLS model was determined based on all possible combinations of intervals. The algorithm calculated the RMSEC for all combinations and the combination that yielded the lowest RMSEC was selected. The results are presented in Fig.18 For example, the reader can see that using the entire spectrum to predict DBN or APP yields the worst case prediction in terms of the lowest possible RMSEC. For AcOH and H2O, the worst case in terms of the lowest RMSEC is obtained using a subinterval.

[0291] Fig.18 The lowest RMSEC values ​​of the 10 best models using increasing numbers of bins according to the example are shown.

[0292] The best case with the lowest RMSEC was found between these two extremes. Their optimal number of subintervals and the corresponding lowest RMSEC are summarized in Table 7. The reader can notice that the model calibration error can be further reduced by this simple procedure.

[0293] Table 7: PLS regression results: explained variance and RMSECV

[0294]

[0295] More Examples

[0296] A. Sample Preparation

[0297] Table 8: Specifications of original [APPH][OAc] and [DBNH][OAc] stock solutions

[0298] Weight fraction [APPH][OAc] stock solution [DBNH][OAc] stock solution <![CDATA[H2O (KFT), wt%]]> 0.97％ 0.05％ APP+, dry weight % 63.0％ 0.0％ DBN+, dry weight % 12.1％ 75.4％ HOAc, dry weight % 24.9％ 24.6％ [APPH][OAc], dry weight % 83.9％ 0.0％ [DBNH][OAc], dry weight % 16.1％ 100％

[0299] Fig.19 Sample compositions according to examples are shown.

[0300] B. Raman spectroscopy analysis

[0301] Fig. 20The ratio of DBN to DBN+APP (left) and the DBN weight % (right) in the mixture according to the examples are shown.

[0302] Fig.21 Pre-treated Raman spectra of samples with APP / DBN molar ratios of 0 and 4.47 at two H2O contents of 25 wt% (left) and 50 wt% (right) according to the example are shown (for clarity, the spectra are divided into the fingerprint region (lower) and the high frequency region (upper)).

[0303] C.PCA

[0304] Fig. 22 The raw pre-processed Raman spectra (top) and the averaged centered pre-processed Raman spectra of different samples according to the examples are shown (the color code indicates the wt. % of H2O in the mixture).

[0305] Fig.23 The variance and cumulative variance explained by the first ten PCs according to the example are shown.

[0306] D.PLS model

[0307] a.AcOH

[0308] Fig.24 The RMSE and variance according to the examples are shown as a function of the number of LVs.

[0309] Fig.25 Model residuals according to examples are shown.

[0310] Fig.26 Model scores and loadings according to the examples are shown.

[0311] Fig. 27 The model regression vector according to the example is shown.

[0312] b.DBN

[0313] Fig.28 The RMSE and variance according to the examples are shown as a function of the number of LVs.

[0314] Fig.29 Model residuals according to examples are shown.

[0315] Fig.30 Model scores and loadings according to the examples are shown.

[0316] Fig.31 The model regression vector according to the example is shown.

[0317] c.APP

[0318] Fig.32The RMSE and variance according to the examples are shown as a function of the number of LVs.

[0319] Fig.33 Model residuals according to examples are shown.

[0320] Fig.34 Model scores and loadings according to the examples are shown.

[0321] Fig.35 The model regression vector according to the example is shown.

[0322] d.H2O

[0323] Fig.36 The RMSE and variance according to the examples are shown as a function of the number of LVs.

[0324] Fig.37 Model residuals according to examples are shown.

[0325] Fig.38 Model scores and loadings according to the examples are shown.

[0326] Fig.39 The model regression vector according to the example is shown.

[0327] The different functions discussed herein may be performed in a different order and / or concurrently with each other.

[0328] Unless otherwise indicated, any range or device value given herein may be expanded or altered without losing the effect sought. In addition, any example may be combined with another example unless expressly prohibited.

[0329] Although the subject matter has been described in language specific to structural features and / or acts, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as examples of implementing the claims, and other equivalent features and acts are intended to fall within the scope of the claims.

[0330] It should be understood that the benefits and advantages described above may relate to one embodiment or may relate to several embodiments. The embodiments are not limited to embodiments that solve any or all of the stated problems or embodiments that have any or all of the stated benefits and advantages. It will be further understood that reference to 'an' item may refer to one or more of those items.

[0331] The term 'comprising' is used herein to mean including the identified methods, blocks or elements, but such blocks or elements do not comprise an exclusive list and the method or apparatus may contain additional blocks or elements.

[0332] Numerical descriptors such as 'first', 'second', etc. are used herein merely as a means of distinguishing otherwise similarly named parts. Numerical descriptors should not be construed as indicating any particular order, such as priority, order of manufacture, or order of appearance in any particular structure.

[0333] Although the present invention has been described in conjunction with specific types of devices and / or methods, it should be understood that the present invention is not limited to any specific type of devices and / or methods. Although the present invention has been described in conjunction with a plurality of examples, embodiments and implementations, the present invention is not limited thereto, but rather encompasses various modifications and equivalent arrangements falling within the scope of the claims. Although various examples have been described above with some degree of specificity or with reference to one or more individual embodiments, a person skilled in the art may make many changes to the disclosed examples without departing from the scope of this specification.

[0334] References:

[0335] [1] A. Michud et al., "Ioncell-F: ionic liquid-based cellulosic textilefibers as an alternative to viscose and Lyocell," Text.Res.J., vol.86, no.5, pp.543–552, 2016.

[0336] [2] T. Kakko, AWT King, and I. Kil- "Homogenous esterification of cellulose pulp in [DBNH][OAc]," Cellulose, vol.24, no.12, pp.5341–5354, 2017.

[0337] [3] A. Parviainen et al., "Sustainability of cellulose dissolution and regeneration in1,5-diazabicyclo[4.3.0]non-5-enium acetate: A batch simula-tion of the IONCELL-F process," RSC Adv., vol.5, no.85, pp.69728–69737, 2015.

[0338] [4]J.Canongia Lopes and L.P.Rebelo,“Ionic liquids and reactiveazeotropes:the continuity of the aprotic and protic classes,”Phys.Chem.Chem.Phys.,vol.12,no.8,pp.1648–1648,2010.

[0339] [5]W.Ahmad et al.,“Feasibility of thermal separation in recycling ofthe distillable ionic liquid[DBNH][OAc]in cellulose fiber production,”Chem.Eng.Res.Des.,vol.114,pp.287–298,2016.

[0340] [6]L.S.Sciarini,A.Rolland-Sabaté,S.Guilois,P.Decaen,E.Leroy,and P.LeBail,“Under-standing the destructuration of starch in water-ionic liquidmixtures,”Green Chem.,vol.17,no.1,pp.291–299,2015.

[0341] [7]Y.Kohno and H.Ohno,“Ionic liq-uid / water mixtures:From hostility toconciliation,”Chem.Commun.,vol.48,no.57,pp.7119–7130,2012.

[0342] [8]A.M.Stepan,A.Monshizadeh,M.Hum-mel,A.Roselli,and H.Sixta,“Cellulose fractionation with IONCELL-P,”Carbohydr.Polym.,vol.150,pp.99–106,2016.

[0343] [9]A.M.Stepan,A.Michud,S.Hellstén,M.Hummel,and H.Sixta,“IONCELL-P&F:Pulp Fractiona-tion and Fiber Spinning with Ionic Liquids,”Ind.Eng.Chem.Res.,vol.55,no.29,pp.8225–8233,2016.

[0344]

[10] K.Wippermann,J.Giffin,and C.Korte,“In Situ Determination of theWater Content of Ionic Liquids,”J.Electrochem.Soc.,vol.165,no.5,pp.H263–H270,2018.

[0345]

[11] A.García-Mendoza and J.C.Aguilar,“Analysis of water in roomtemperature ionic liquids by linear sweep,differential pulse and square wavecathod-ic stripping voltammetries,”Electrochim.Acta,vol.182,pp.238–246,2015.

[0346]

[12] J.Viell and W.Marquardt,“Concen-tration measurements in ionicliquid-water mixtures by mid-infrared spectroscopy and indirect hardmodeling,”Appl.Spectrosc.,vol.66,no.2,pp.208–217,2012.

[0347]

[13] C.D.TRAN,H.L.D.P.SILVIA,and D.OLIVEIRA,“Absorption of Water byRoom-Temperature Ionic Liquids:Effect of Anions on Concentration and State ofWater,”Appl.Spectrosc.,vol.57,no.2,pp.152–157,2015.

[0348]

[14] M.J.PELLETIER,“Quantitative Analy-sis Using Raman Spectrometry,”Appl.Spectrosc.,vol.57,no.1,p.20A–42A,2003.

[0349]

[15] O.Svensson,M.Josefson,and F.W.Langkilde,“Reaction monitoringusing Raman spectroscopy and chemometrics,”Chemom.Intell.Lab.Syst.,vol.49,no.1,pp.49–66,1999.

[0350]

[16] R.W.Berg,Raman spectroscopy and ab-initio model calculations onionic liquids,vol.138,no.11.2007.

[0351]

[17] V.H.Paschoal,L.F.O.Faria,and M.C.C.Ribeiro,“VibrationalSpectroscopy of Ionic Liq-uids,”Chem.Rev.,vol.117,no.10,pp.7053–7112,2017.

[0352]

[18] S.SATYEN,T.HIROI,K.IWATA,and H.HAMAGUCHI,“Raman Spectroscopy andthe Heterogeneous Liquid Structure in Ionic Liquids,”in Ionic LiquidsCompletely UnCOILed:Critical Expert Overviews,N.V.Plechkova and K.R.Seddon,Eds.2015,pp.165–187.

[0353]

[19] P.Geladi,“Chemometrics in spectros-copy.Part 1.Classicalchemometrics,”Spectrochim.Acta-Part B At.Spectrosc.,vol.58,no.5,pp.767–782,2003.

[0354]

[20] P.Geladi,B.Sethson,J. T.Lillhonga,T.Lestander,andJ.Burger,“Chemometrics in spectroscopy:Part 2.Examples,”Spectrochim.Acta-PartB At.Spectrosc.,vol.59,no.9,pp.1347–1357,2004.

[0355]

[21] P.Geladi and B.R.Kowalski,“Par-tial least-squares regression:atutorial,”Anal.Chim.Acta,vol.185,no.C,pp.1–17,1986.

[0356]

[22] M.Shen,Y.Zhang,K.Chen,S.Che,J.Yao,and H.Li,“Ionicity of ProticIonic Liquid:Quan-titative Measurement by Spectroscopic Methods,”J.Phys.Chem.B,vol.121,no.6,pp.1372–1376,2017.

[0357]

[23] T.Nakabayashi,K.Kosugi,and N.Ni-shi,“Liquid structure of aceticacid studied by Raman spectroscopy and Ab initio molecular orbital calcula-tions,”J.Phys.Chem.A,vol.103,no.43,pp.8595–8603,1999.

[0358]

[24] E.Smith and G.Dent,Modern Raman Spectroscopy-A PracticalApproach.Chichester,UK:John Wiley&Sons,Ltd,2004.

[0359]

[25] P.Larkin,Infrared and Raman spec-troscopy:principles and spectralinterpretation.2011.

[0360]

[26] Q.Sun,“The Raman OH stretching bands of liquid water,”Vib.Spectrosc.,vol.51,no.2,pp.213–217,2009.

[0361]

[27] S.Cha et al.,“Structures of ionic liquid-water mixturesinvestigated by IR and NMR spec-troscopy,”Phys.Chem.Chem.Phys.,vol.16,no.20,pp.9591–9601,2014.

[0362] Andersen CM,Bro R(2010)Variable selection in regression-a tutorial.JChemom 24:728–737.https: / / doi.org / 10.1002 / cem.1360

[0363] Brereton RG(2003)Chemometrics:Data Analysis for the Laboratory andChemical Plant.John Wiley&Sons,Ltd

[0364] Cooper JB(1999)Chemometric analysis of Raman spectroscopic data forprocess control applications.Chemom Intell Lab Syst 46:231–247.https: / / doi.org / 10.1016 / S0169-7439(98)00174-9

[0365] Geladi P(2003)Chemometrics in spectroscopy.Part 1.Classicalchemometrics.Spectrochim Acta-Part B At Spectrosc 58:767–782.https: / / doi.org / 10.1016 / S0584-8547(03)00037-5

[0366] Geladi P,Kowalski BR(1986)Partial least-squares regression:atutorial.Anal Chim Acta 185:1–17.https: / / doi.org / 10.1016 / 0003-2670(86)80028-9

[0367] Geladi P,Sethson B, J,et al(2004)Chemometrics in spectroscopy:Part 2.Examples.Spectro-chim Acta-Part B At Spectrosc 59:1347–1357.https: / / doi.org / 10.1016 / j.sab.2004.06.009

[0368] Gofurov S,Makhmanov U,Kokhkharov A,Is-mailova OB(2019)Structural andOptical Characteristics of Aqueous Solutions of Acetic Acid.Appl Spectrosc73:503–510.https: / / doi.org / 10.1177 / 0003702819831325

[0369] Guizani C,Hellstén S,Witos J,Sixta H(2020)Quantitative Ramanspectroscopy for the IoncellTM pro-cess.Part 1:comparison of univariate andmultivariate calibration methods for the quantification of water and proticionic liquid components.Cellulose 27:157–170.https: / / doi.org / 10.1007 / s10570-019-02809-y

[0370] Hyde AM,Calabria R,Arvary R,et al(2019)Investigating theUnderappreciated Hydrolytic Instabil-ity of 1,8-Diazabicyclo[5.4.0]undec-7-ene and Related Unsaturated Nitrogenous Bases.Org Process Res Dev 23:1860–1871.https: / / doi.org / 10.1021 / acs.oprd.9b00187

[0371] Kaneko K,Yoshimura Y,Shimizu A(2018)Water concentration dependence ofthe refractive index of var-ious ionic liquid-water mixtures.J Mol Liq 250:283–286.https: / / doi.org / 10.1016 / j.molliq.2017.12.009

[0372] Kauffmann T,Fontana M(2015)Simultaneous quantification of ionicsolutions by Raman spectrometry and chemometric analysis.IMEKO XXI WorldCongr

[0373] Kueppers S,Haider M(2003)Process analytical chemistry-future trendsin industry.Anal Bioanal Chem 376:313–315.https: / / doi.org / 10.1007 / s00216-003-1907-0

[0374] Larkin P(2011)Infrared and Raman spectrosco-py:principles andspectral interpretation

[0375] Liu W,Cheng L,Zhang Y,et al(2008)The physical properties of aqueoussolution of room-temperature ionic liquids based on imidazolium:Database andevaluation.J Mol Liq 140:68–72.https: / / doi.org / 10.1016 / j.molliq.2008.01.008

[0376] Nakabayashi T,Kosugi K,Nishi N(1999)Liquid structure of acetic acidstudied by Raman spectroscopy and Ab initio molecular orbital calculations.JPhys Chem A 103:8595–8603.https: / / doi.org / 10.1021 / jp991501d

[0377] M, G, D(1991)Raman study of the bendingband of water.Spectrochim Acta Part A Mol Spectrosc 47:897–906.https: / / doi.org / 10.1016 / 0584-8539(91)80277-P

[0378] Pelletier MJ(2003)Quantitative analysis us-ing Ramanspectrometry.Appl Spectrosc 57:20A-42A

[0379] Smith E,Dent G(2004)Modern Raman spectros-copy-a practicalapproach.John Wiley&Sons,Ltd,Chichester,UK

[0380] Sun Q(2009)The Raman OH stretching bands of liquid water.VibSpectrosc 51:213–217.https: / / doi.org / 10.1016 / j.vibspec.2009.05.002

Claims

1. A method for producing lyocell type man-made cellulose fibers, the method being based on dissolution of cellulose pulp in an ionic liquid, dry-jet wet spinning of the solution in a H2O bath and a subsequent solvent recovery step, wherein the ionic liquid and the H2O are separated, the method comprising a sub-method for quantifying water and / or one or more ionic liquid components in an ionic liquid / H2O mixture, the sub-method comprising - obtaining one or more Raman spectra of the ionic liquid / H2O mixture, and - quantifying water and / or one or more ionic liquid components in the ionic liquid / H2O mixture using the one or more Raman spectra using a quantitative calibration model; The sub-method includes simultaneously determining the acid, base and H2O content and / or the H2O concentration and acid-base ratio in the ionic liquid / H2O mixture.

2. The method according to claim 1, wherein: The ionic liquid is a protic ionic liquid.

3. The method according to claim 1, wherein: The ionic liquid is or comprises a non-imidazolium based protic ionic liquid.

4. The method according to claim 1, wherein: Using a quantitative calibration model involves univariate calibration, which is based on finding the relationship between a single spectral variable such as peak intensity, peak area and / or peak shift and analyte concentration.

5. The method according to claim 4, comprising determining the H2O concentration in the ionic liquid / H2O mixture by using a linear relationship between the H2O peak area and the H2O mass fraction in the ionic liquid / H2O mixture.

6. A method according to claim 4 or 5, comprising determining the base concentration in the ionic liquid / H2O mixture using a non-linear relationship between peak intensity and base concentration.

7. The method according to claim 6, wherein: The power law model is used to describe this nonlinear relationship.

8. The method according to claim 1, wherein: Using a quantitative calibration model involves using multivariate calibration.

9. The method according to claim 8, wherein: The multivariate calibration utilized partial least squares regression.

10. The method of claim 1, comprising quantifying one or more ionic liquid degradation products.

11. The method according to claim 10, wherein: Quantification of the one or more ionic liquid degradation products involves utilizing partial least squares regression.

12. The method according to claim 10 or 11, wherein: Quantification of the one or more ionic liquid degradation products involves dividing the one or more Raman spectra into two or more sub-intervals.

13. The method of claim 10, comprising performing solvent purification based on the quantification of the one or more ionic liquid degradation products.

14. A method for monitoring and / or controlling a process involving the use of an ionic liquid / H2O mixture, the method comprising a method according to any one of the preceding claims.

15. The method according to claim 14, comprising quantitatively monitoring the acid-base ratio and / or H2O content for process control.

Images