WO2022101079A1

WO2022101079A1 - Raman spectroscopic method for the control of a liquid mixture

Info

Publication number: WO2022101079A1
Application number: PCT/EP2021/080578
Authority: WO
Inventors: Leonardo TITERICZ; Nuno FORMIGA; Javier DIAZ-MAROTO CARPINTERO; Tristan AILLET; Julien Jolly
Original assignee: Solvay Sa
Priority date: 2020-11-12
Filing date: 2021-11-04
Publication date: 2022-05-19

Abstract

A method for controlling a liquid mixture comprising at least one organic solvent and at least one quinone species, such as a working solution used in a process of manufacturing hydrogen peroxide by an AO-process, is described. The method comprises the steps of: - using a Raman spectrometer to generate a Raman spectrum of the liquid mixture; and - using a predictive model taking the Raman spectrum of the liquid mixture as input (x) and providing the concentration of the at least one quinone species as output (y); wherein said predictive model has been obtained by applying chemometrics to a training data set obtained by applying Raman spectroscopy to several liquid mixtures comprising known concentrations of the at least one quinone species.

Description

RAMAN SPECTROSCOPIC METHOD FOR THE CONTROL OF A LIQUID MIXTURE

The present invention relates to a method for the control of a liquid mixture, more particularly for the control of a working solution used for producing hydrogen peroxide by the AO-process.

Hydrogen peroxide is one of the most important inorganic chemicals to be produced worldwide. Its industrial applications include textile, pulp and paper bleaching, organic synthesis (propylene oxide), the manufacture of inorganic chemicals and detergents, environmental and other applications.

Synthesis of hydrogen peroxide is predominantly achieved by using the large scale Riedl-Pfleiderer process, also called anthraquinone loop process or AO (auto-oxidation) process.

The first step of the AO process is the reduction in an organic solvent of useful quinone(s) (alkylanthraquinone and/or tetrahydroalkylanthraquinone into the corresponding hydroquinone(s) (alkylanthrahydroquinone and/or tetrahydroalkylanthrahydroquinone) using hydrogen gas and a catalyst. The mixture of organic solvents, hydroquinone and quinone species (working solution, WS) is then separated from the catalyst and the hydroquinone species are oxidized using oxygen, air or oxygen-enriched air thus regenerating the quinone(s) with simultaneous formation of hydrogen peroxide. The organic solvent of choice is typically a mixture of two types of solvents, one being a good solvent of the quinone(s) (generally a non polar solvent, for instance a mixture of aromatic compounds) and the other being a good solvent of the hydroquinone(s) (generally a polar solvent, for instance a long chain alcohol). Hydrogen peroxide is then typically extracted with water and recovered in the form of a crude aqueous hydrogen peroxide solution, and the quinone(s) is returned to the hydrogenator to complete the loop.

Although both reduction and oxidation reactions are highly efficient in terms of yield and selectivity, the productivity of the process suffers from the appearance of undesirable products (degenerates) due to repeated operating cycles. Hence, during the life of the process, the composition of the working solution deviates considerably from its initial state and the conversion of useful, reactive quinone species as described above into non-reactive species (degenerates like Tert-Alkyl-Tetrahydro-Epoxyanthra-Quinone or Ter- Alkyl Anthrone) can greatly reduce the productivity of the process.

Monitoring the concentration of at least one, preferably of some of these quinone derivatives (useful quinones and degenerates) in the process is therefore essential to influence its parameters and to regenerate the reaction loop when required.

A known technique for this monitoring is HPLC (High Performance Liquid Chromatography), as for instance described in EP2766299 in the name of the Applicant. The drawbacks of this technique are its cost, the fact that it is destructive (and hence produces waste) and has a rather long response time (about 2 hours or more). Besides, commercially available HPLC devices are bulky and not easy to handle.

The idea behind the present invention is to use another technique for this monitoring, which has shorter response times, is non-destructive and can be used in situ in a production unit, and for which portable devices are commercially available and at a much lower price. One such technique is Raman spectrometry.

Up till now, this technique was used to track the presence/measure the concentration of a single chemical compound, or of a limited number of compounds having very different chemical structures.

The inventors found out however that this technique can also be applied to the complex mixture which is the WS of an AO process, provided chemometric methods (chemometrics) are used to develop an adequate predictive model taking the Raman IR spectrum as input (x) and predicting the concentration of given compound(s) as output (y). This finding is surprising because actually, the compounds monitored (target molecules) have a very similar chemical structure (all of them being quinone derivatives) which implies that their peaks in the spectrum are close to each other and even overlapping.

The present invention hence concerns a method for the control of a liquid mixture comprising at least one organic solvent and at least one quinone specie, said method comprising the steps of:

- using a Raman spectrometer to generate a Raman spectrum of the liquid mixture;

- using a predictive model taking the Raman spectrum of the liquid mixture as input (x) and providing the concentration of the at least one quinone specie as output (y); wherein said predictive model has been obtained by applying chemometrics to a training data set obtained by applying Raman spectroscopy to several liquid mixtures comprising known concentrations of the at least one quinone specie.

In the frame of the present invention, by “control” is meant the measurement of the concentration (generally expressed in weight % or in g/kg related to the total weight of the mixture) of the at least one quinone specie targeted as output (y) in the liquid mixture, and this at several moments in time. This doesn’t preclude that the concentration and/or presence of other chemical species in the liquid mixture may be measured as well, neither that a continuous measurement of one or more species may be performed over a given period of time.

The liquid mixture of the present invention, is a mixture which is liquid in the conditions (p, T) in which it is controlled. It comprises at least one organic solvent and at least one quinone specie.

In one embodiment, the quinone specie(s) comprises at least one alkyl (hydro)anthraquinone which is a useful quinone (i.e. one which reacts in an AO process by being successively hydrogenated and oxidised, resulting in the production of hydrogen peroxide). In a preferred embodiment, the at least one quinone specie comprises a quinone selected from amylanthraquinone (AQ), amyltetrahydroanthraquinone (ATQ), ethylanthraquinone (EQ) and ethyltetrahydroanthraquinone (ETQ) and their mixtures; preferably it comprises both AQ and ATQ, or both EQ and ETQ.

In another embodiment, combinable with the former, the at least one quinone comprises a quinone selected from TertAmylTetrahydro- EpoxyanthraQuinone (TATEQ), Ter Amyl Anthrone (TAA), EthylTetrahydro- EpoxyanthraQuinone (ETEQ) and EthylAnthome (EA), which are in fact degenerates of the above mentioned useful quinones.

In a preferred embodiment, the method of the invention measures both the concentration of a useful quinone (like AQ and/or ATQ, or EQ and/or ETQ), and the concentration of at least one degradation product thereof (“degradate”) like epoxy or anthrone derivatives thereof, for instance TertAmylTetrahydro- EpoxyanthraQuinone (TATEQ) and/or TerAmyl Anthrone (TAA) in case of AQ and/or ATQ as useful quinone(s), or their ethyl equivalents EthylTetrahydro- EpoxyanthraQuinone (ETEQ) and Ethyl Anthorne (EA) in case of EQ and/or ETQ as useful quinones. The present invention gives good results with liquid mixtures comprising a quinone, a hydroquinone, and both epoxy and anthrone derivatives thereof; in practice, very good results are obtained when the at least one quinone specie comprises AQ, ATQ, TATEQ and TAA.

Preferably, the at least one organic solvent comprises at least one polar solvent and at least one non polar solvent in order to be able to respectively dissolve a hydroquinone and a quinone specie. Good results are obtained with at least one polar aliphatic solvent and a mixture of non polar aromatic solvents, like diiso butyl carbinol (DBC), a commercially available mixture of non polar aromatic solvents like Caromax® 20 LN, sextate (2-methyl cyclohexyl acetate), TBU (TetraButylUrea) , TOP (trioctylphosphate)... In practice, good results were obtained with DBC, Caromax® 20 LN and mixtures thereof.

The respective nature of the quinone(s) and solvent(s) selected will determine the maximum amount of the quinone in the solvent, depending on the solubility of the former in the latter.

In general, good results are obtained in the following ranges:

- Concentration of quinone in the solvent(s): 15 to 45 wt%, preferably 25 to 35 wt%

- Concentration of polar solvent in the solvent mixture: 10 to 50 wt%, preferably 20 to 30wt%.

Any type of Raman spectrometer can be used in the method of the invention. However, especially if the analysis has to be performed only from time to time, it might advantageously be a portable device. This device generally uses an immersion probe the nature of which is chosen to maximize chemical compatibility. A sapphire interface gives good results in that regard. Preferably, the exposure and accumulation times are investigated in order to optimize the signal over noise ratio.

The predictive model of the invention preferably uses as input (x), the intensity of the emission of the at least one quinone specie at a given wavelength, preferably at several given wavelengths (i.e. the height of the emission peak of that specie at this/these wavelength(s).

According to the invention, this model is obtained using chemometrics. Chemometrics is defined by the International Chemometrics Societey (ICS) as the science of relating measurements made on a chemical system or process to the state of the system via application of mathematical or statistical methods. According to the invention, chemometrics are applied on a training dataset (set of spectra for known concentrations of the target molecules) with the aim of minimizing the residual error between the predicted concentration s) and the known concentration s) of the targeted quinone specie(s). Generally, at least 2 and preferably, at least 3 or even 4 quinone species are targeted, generally those as specified above.

The main methods used in quantitative chemometrics are multiple linear regression (MLR), principal components regression (PCR) and partial least squares (PLS). A preferred embodiment of the invention uses the partial least squares (PLS) method to build the predictive model.

Generally, the training data set is a set of spectra obtained on mixtures obtained by either adding given quantities of the at least one quinone specie to the liquid mixture to be controlled to increase its concentration in said at least one quinone specie, or by adding given quantities of other constituent(s) to the liquid mixture to be controlled in order to reduce its concentration in said at least one quinone specie.

By “other constituent s)” is meant one or more chemical compounds present in the liquid mixture to be controlled.

The number of mixtures required and the concentration of the at least quinone specie and/or other constituent(s) to be added to the liquid mixture to build the training data set, can easily be determined by those skilled in the art with recourse to software available on the market.

Considering the number of mixtures required especially in case of several targeted quinone species, the mixtures preparation preferably uses an automated robotic platform i.e. a device which prepares the selected mixtures automatically based on adequate input parameters including the composition of the mixtures.

Additionally or alternatively, considering the complexity of the obtained data, the model building (data set determination and/or data analysis) preferably uses a software. Most preferably, both robotics & software are used to generate the model.

Hence, in a preferred embodiment, the mixtures preparation uses an automated robotic platform and the model building (data set determination and/or data analysis) uses a software.

In the method of the invention, the experimental mixtures required to generate the training data set can advantageously be chosen thanks to a Mixture Design DOE method. This choice is possible for mixtures in which the sum of the targeted quinone species concentrations is constant meaning that if one specie concentration increases the others ones decrease. This feature can be exploited to reduce the number of experiments as it constraints the concentrations space thanks to the following relation: Ci=cste.

As to the software, the commercially available Minitab statistical software gives good results. Minitab is a statistics package developed at the Pennsylvania State University.

The present invention also concerns a process for the manufacture of hydrogen peroxide by an AO-process using a working solution comprising at least one organic solvent and at least one quinone specie, said process comprising controlling the working solution using the method as described above, wherein the working solution constitutes the liquid mixture to be controlled.

The process of the invention is particularly suitable for the manufacture of hydrogen peroxide by the AO-process wherein said process has a production capacity of hydrogen peroxide of up to 100 kilo tons per year (ktpa). Preferably said process is a small to medium scale AO-process operated with a production capacity of hydrogen peroxide of up to 50 kilo tons per year (ktpa), and more preferably with a production capacity of hydrogen peroxide of up to 35 kilo tons per year (ktpa), and in particular a production capacity of hydrogen peroxide of up to 20 kilo tons per year (ktpa). The dimension ktpa (kilo tons per annum) relates to metric tons.

The working solution of said AO-process is controlled by a method as described above meaning that its concentration in at least one quinone specie, generally a useful quinone (hydroquinone or quinone specie) and/or a degenerate thereof as described above, is measured by said method. This measurement can be performed on samples withdrawn from time to time from the stream of working solution. However, advantageously, it is performed by immerging a measuring probe/device in a pipe, reservoir or other device wherein the working solution circulates to perform the AO process.

Although of course the working solution must be “sampled” on line i.e. in the stream of working solution of the AO-process, the control per se i.e. the application of the method of the invention to said samples or directly in the continuous stream, can be applied remotely i.e. the samples and/or the Raman spectrum measured on them can be sent to another location, for instance a specialized lab or central large-scale .mother plant. Alternatively and preferably, the control is realized using a portable Raman spectrometer or a Raman spectrometer installed on or close to a pipe, reservoir or other device where the working solution circulates

A particular advantage of the AO-process of the present invention is that the hydrogen peroxide can be manufactured in a plant which may be located at any, even remote, industrial or end user site provided the Raman spectrometer is correctly calibrated and equipped with/connected to adequate software including the predictive model specific to the working solution of said AO-process.

In a preferred embodiment of the invention, the working solution is regenerated either continuously or intermittently, based on the results of its control. Regeneration means conversion of certain degradates, like epoxy or anthrone derivatives as described above, back into useful quinones.

In case of an intermittent reversion, this reversion of the working solution may be performed, for instance, at a different site in the equipment of another hydrogen peroxide production plant, e.g. in the respective regeneration equipment of a similar or preferably a larger scale hydrogen peroxide production plant. Alternatively, the working solution may be regenerated in separate mobile regeneration equipment for the reversion of the working compounds contained in the working solution, e.g. in a mobile regeneration unit that is used on demand or as appropriate in a number of different locations where a small to medium hydrogen peroxide manufacturing process according to the AO-process is performed. Another option is to intermittently or periodically perform the regeneration of the working solution under particular conditions in the main equipment of the small to medium hydrogen peroxide manufacturing process according to the AO-process itself.

In any case, the efficiency of the regeneration of the working solution can be controlled by applying the control method of the invention to the working solution before and after regeneration.

While the invention has been described with reference to details of the illustrated embodiment, these details are not intended to limit the scope of the invention as defined in the appended claims.

Also, the invention will be further described based on the Examples below, which again are not intended to limit the scope of the invention as defined in the appended claims.

Experimental conditions The Raman spectra have been collected with the Raman Rxn2 spectrometer from Kaiser Optical Systems. The excitation line is 785 nm with a spectral coverage from 150 to 3425 cm-1. The power of the excitation line was fixed to 150 mW.

The exposure and accumulation times have been investigated to optimize the signal over noise ratio. An immersion optic (IO) with 1/8” diameter was used for spectra collection. This immersion probe relies on sapphire interface to maximize chemical compatibility.

The samples were prepared using a Zinsser Analytic robotic platform, and the data were computed thanks to Minitab software.

Example 1

A first model was built by considering only useful compounds AQ and ATQ diluted into a Caromax/DBC solvent. For this model, samples were prepared by mixing pure solutions of AQ (200 g/kg), ATQ (200 g/kg) and the DBC solvent. AQ (6 levels) and ATQ (6 levels) with two concentration levels for the solvent lead to a dataset of 72 samples. The concentration ranges are indicated in Table 1 below. The PLS model was applied with success on this dataset (R²=0.99).

Table 1

Calibration

AQ (g/kg)

0 - 20 - 35 - 50 - 65 - 80 - 100 - 120 - 140 - 160 - 180 - 200

ATQ (g/kg)

0 - 50 - 70 - 90 - 150 - 180 - 200

DBC (g/kg)

100 - 135 - 170 - 200 - 250

This model was evaluated on both a synthetic sample and on an industrial working solution having the same AQ and ATQ concentrations, namely 95 g/kg AQ and 139 g/kg ATQ. The relative errors obtained were respectively of 9% for the synthetic model, and of 30% for the industrial working solution.

Example 2

A second model was built starting from the industrial working solution and considering 2 additional components present in the industrial working solution, namely degenerates TATEQ and TAA which are present in a total amount of a few grams per kg. A training dataset of 40 samples was generated by adding given amounts of AQ, ATQ, TATEQ and TAA to said industrial solution and a PLS model was built based thereon. The average relative errors were respectively of about 6.5% for the useful quinones (AQ & ATQ) and of about 50% for the degenerates (TATEQ and TAA).

Claims

C L A I M S

1. A method for the control of a liquid mixture comprising at least one organic solvent and at least one quinone specie, said method comprising the steps of:

2. The method according to claim 1, wherein the at least one quinone specie comprises a quinone selected from amylanthraquinone (AQ), amyltetrahydroanthraquinone (ATQ), ethylanthraquinone (EQ) and ethyltetrahydroanthraquinone (ETQ) and their mixtures, preferably it comprises both AQ and ATQ, or both EQ and ETQ.

3. The method according to claim 1 or 2, wherein the at least one quinone comprises a quinone selected from TertAmylTetrahydro-EpoxyanthraQuinone (TATEQ), Ter Amyl Anthrone (TAA), EthylTetrahydro-EpoxyanthraQuinone (ETEQ) and EthylAnthorne (EA).

4. The method according to any of claims 1 to 3, wherein the at least one quinone specie comprises AQ, ATQ, TATEQ and TAA.

5. The method according to any of claims 1 to 4, wherein the at least one organic solvent comprises at least one polar solvent and at least one non polar solvent.

6. The method according to any of claims 1 to 5, wherein the predictive model uses as input (x), the intensity of emission of the at least one quinone specie at a given wavelength, preferably at several given wavelengths.

7. The method according to any of claims 1 to 6, using the partial least squares (PLS) method to build the predictive model.

8. The method according to any of claims 1 to 7, wherein the training data set is a set of spectra obtained on mixtures prepared by either adding given quantities of the at least one quinone specie to the liquid mixture to increase its concentration in said at least one quinone specie, or by adding given quantities of other constituent(s) to the liquid mixture in order to reduce its concentration in said at least one quinone specie.

9. The method according to claim 8, wherein the mixtures preparation uses an automated robotic platform and wherein the model is built using a software.

10. A process for the manufacture of hydrogen peroxide by an AO-process using a working solution comprising at least one organic solvent and at least one quinone specie, said process comprising controlling the working solution using the method according to any of claims 1 to 9, wherein the working solution constitutes the liquid mixture to be controlled.

11. The process according to claim 10, which is a small to medium scale AO- process operated with a production capacity of hydrogen peroxide of up to 50 kilo tons per year (ktpa), preferably of up to 35 kilo tons per year (ktpa), more preferably of up to 20 kilo tons per year (ktpa).

12. The process according to claim 10 or 11, which is performed by immerging a measuring probe/device in a pipe, reservoir or other device wherein the working solution circulates to perform the AO process.

13. The process according to claim 12, wherein the control is realized using a portable Raman spectrometer or a Raman spectrometer installed on or close to a pipe, reservoir or other device where the working solution circulates.

14. The process according to any of claims 10 to 13, wherein the working solution is regenerated either continuously or intermittently, based on the results of its control.

15. The process according to claim 14, wherein the efficiency of the regeneration of the working solution is controlled by applying the method of any of claims 1 to 9 to the working solution before and after regeneration.