Hydrophobic eutectic mixtures as volatile fatty acid extractants

Accepted Manuscript Hydrophobic Eutectic Mixtures as Volatile Fatty Acid Extractants Adriaan van den Bruinhorst, Sanne Raes, Sausan Atika Maesara, Maaike C. Kroon, A. Catarina C. Esteves, Jan Meuldijk PII: DOI: Reference:

S1383-5866(18)33408-7 https://doi.org/10.1016/j.seppur.2018.12.087 SEPPUR 15229

To appear in:

Separation and Purification Technology

Received Date: Revised Date: Accepted Date:

28 September 2018 24 December 2018 31 December 2018

Please cite this article as: A. van den Bruinhorst, S. Raes, S. Atika Maesara, M.C. Kroon, A. Catarina C. Esteves, J. Meuldijk, Hydrophobic Eutectic Mixtures as Volatile Fatty Acid Extractants, Separation and Purification Technology (2018), doi: https://doi.org/10.1016/j.seppur.2018.12.087

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Hydrophobic Eutectic Mixtures as Volatile Fatty Acid Extractants Adriaan van den Bruinhorsta*, Sanne Raesb, Sausan Atika Maesarab, Maaike C. Kroonc,1, A. Catarina C. Estevesa, Jan Meuldijkd* a

Laboratory of Physical Chemistry, Department of Chemical Engineering and Chemistry, Eindhoven

University of Technology, Het Kranenveld, Building. 14 (Helix), 5612 AZ Eindhoven, The Netherlands; b

Sub-department of Environmental Technology, Wageningen University & Research, Axis-Z, Bornse

Weilanden 9, 6708 WG Wageningen, The Netherlands c

Separation Technology Group, Department of Chemical Engineering and Chemistry, Eindhoven

University of Technology, Het Kranenveld, Building. 14 (Helix), 5612 AZ Eindhoven, The Netherlands; d

Laboratory of Chemical Reactor Engineering/Polymer Reaction Engineering, Department of Chemical

Engineering and Chemistry, Eindhoven University of Technology, Het Kranenveld, Building. 14 (Helix), 5612 AZ Eindhoven, The Netherlands; 1

Present address: Department of Chemical Engineering, Khalifa University of Science and Technology,

Petroleum Institute, P.O. Box 2533, Abu Dhabi, United Arab Emirates; *Corresponding authors:

Jan Meuldijk, +31(0)402472328, [email protected] Adriaan van den Bruinhorst, +31 (0)402478246, [email protected]

1

Abstract Organic waste streams can be converted into volatile fatty acids (VFAs) via fermentation. VFAs can be used as intermediates in the synthesis of added-value chemicals. In this work, hydrophobic eutectic mixtures were designed for the liquid-liquid extraction of VFAs from dilute aqueous solutions. The eutectic behaviour was screened for over 100 combinations of 16 hydrophobic components that were selected based on a set of predetermined criteria. Mixtures of dihexylthiourea and trioctylphosphine oxide (TOPO) showed the best extraction performance and were stable over a wide pH range. The extraction efficiency increased with increasing hydrophobicity of the VFAs, and only undissociated acids were extracted. Upon increasing the TOPO content of the eutectic mixture, the extraction performance could be improved, confirming the tuneable nature of eutectic solvents. However, the extraction performance was less than that for solutions of TOPO in hydrophobic solvents, even though mole fractions of TOPO were higher in the eutectic mixtures. It was hypothesized that the intermolecular VFATOPO interactions required for extraction are suppressed by the inter-component interactions in the eutectic mixture. The inter-component interactions are responsible for the negative deviation from ideality of the melting temperature depressions that extend the liquid window of the mixtures towards the extraction temperature. Hence, the design of novel hydrophobic extractants based on eutectic mixtures was demonstrated. Their performance might be improved by selecting counterparts that interfere less with the interactions required for VFA extraction.

Keywords Deep Eutectic Solvents, Eutectic Mixtures, Designer Solvents, Liquid-Liquid Extraction, Volatile Fatty Acid Recovery

2

1

Introduction

The conversion of organic residual waste streams to platform chemicals via fermentation is considered a viable alternative route to replace the petroleum-based production of chemicals [1]. Volatile fatty acids (VFAs) are short-chain monocarboxylic acids containing five or less carbon atoms, they are generally considered platform chemicals [2]. However, the commercialisation of VFA production via fermentation is challenging owing to the relatively low VFA concentration in the broth caused by product inhibition for the microorganisms [3,4]. A possible strategy to overcome this limitation is the continuous in-situ removal of VFAs from the fermentation broth using affinity separations, e.g. liquid-liquid extraction (LLE) [1,5]. Many extractants have been proposed for the recovery of VFAs via LLE, of which most are based on hydrophobic low-functionalised organic solvents mixed with an active component that can form complexes with the VFAs [5,6]. The solvent capacity and extraction performance is then limited by the solubility of the active component and/or its complex with the extracted VFAs [5,7]. Other VFAextractants are acid-base couples and ionic liquids [5,8], which generally include at least one component that is a liquid at the extraction temperature. The range of possible VFA extractants and diluents can be extended to components that are typically solid at the extraction temperature by using eutectic mixtures, often called deep eutectic solvents (DESs). Eutectic mixtures generally contain high mole fractions of strongly functionalised components, hence they might also overcome solubility limitations of the extracting agent in the diluent [7] or VFA–extractant complexes in the solvent. In previous work, hydrophobic eutectic mixtures have been developed and successfully applied as VFA extractants [9]. Similar hydrophobic eutectic mixtures were studied for the recovery of hydroxymethylfurfural [10], the recovery of various alkali/ transition metal salts from aqueous solutions [11,12], CO2 capture [13], as well as the extraction of components from plant leaves [14,15]. Additionally, various hydrophobic eutectic mixtures based on menthol have been explored for LLE

3

[16,17]. DESs are typically considered designer solvents [18,19]. However, most studies focus on applying known or similar DESs to the application of interest, instead of systematically designing novel hydrophobic eutectic mixtures based on the specific requirements of the application. In the present study eutectic mixtures were systematically designed for a specific purpose: the extraction of VFAs from diluted aqueous solutions. Various hydrophobic binary mixtures were explored as potential eutectic extractants. The component selection of these mixtures was rationalised via the functional groups and physical properties of its pure components. The mixtures were subjected to an experimental screening for liquid formation at the extraction temperature. The resulting eutectic mixtures were evaluated as extractant for the recovery of dilute VFAs from buffered aqueous solutions. Buffered solutions were selected because they better reflect fermentation broths, where buffering salts and acids are present as well. Buffering salts showed to have a significant effect on the extraction performance for ILs [8]. The performance of the eutectic mixtures was compared to trioctylamine (TOA) dissolved in octanol as a benchmark solvent. With this preliminary screening and VFA extraction studies it was demonstrated that DESs indeed have potential as designer solvents, since the hydrophobic eutectic mixtures were systematically designed for a specific application.

2 2.1

Materials and methods Chemicals

All chemicals were used as received from the supplier without further purification. In Table 1, the supplier and purity are listed for each chemical. Chemicals that were used for analysis purposes are specified in the description of their corresponding analysis techniques. Table 1 List of chemicals used for the eutectic mixtures and the volatile fatty acid extraction experiments. Component name

CAS

Supplier

acetic acid potassium acetate

64-19-7 127-08-2

Merck Sigma-Aldrich

Puritya wt% 100 ≥ 99

4

Component name

CAS

Supplier

propionic acid potassium propionate butyric acid sodium butyrate phosphoric acid monopotassium phosphate dipotassium phosphate 1-octanol trioctylamine decanoic acid N,N’-dihexylthiourea 1-tetradecanol dodecanoic acid 1-hexadecanol 1,2-decanediol di-n-dodecylamine 3,5-di-tertbutylcatechol cholesterol bisphenol Z N-benzyl-N-ethylaniline trioctylphosphine oxide dodecyl-methyl-sulfoxide diphenylsulfoxide tribenzylamine leucomalachite green

79-09-4 327-62-8 107-92-6 156-54-7 7664-38-2 7778-77-0 7758-11-4 111-87-5 1116-76-3 334-48-5 21071-28-3 112-72-1 143-07-7 36653-82-4 1119-86-4 3007-31-6 1020-31-1 57-88-5 843-55-0 92-59-1 78-50-2 3079-30-9 945-51-7 620-40-6 129-73-7

Sigma-Aldrich TCI Chemicals Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Frinton labs Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich ABCR GmbH Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich

a

Puritya wt% ≥ 99.5 ≥ 98 ≥ 99 ≥ 98 ≥ 85 ≥ 99 ≥ 98 ≥ 99 ≥ 98 ≥ 98 ≥ 97 ≥ 97 ≥ 99 ≥ 99 ≥ 98 ≥ 95 ≥ 98 ≥ 99 ≥ 98 ≥ 99 ≥ 99 ≥ 98 ≥ 96 ≥ 99 n.r.b

as stated by the supplier; bnot reported.

2.2

Component selection

The components that were included in the hydrophobic DES screening were selected from the chemical abstracts database via SciFinder based on the following criteria: 1. Component 1 should contain a hydroxyl, carboxyl, or a secondary amine functionality. This introduces a functional group with basicity into the eutectic mixture. Similar basic components could extract VFAs from aqueous dilutions when dissolved in volatile organic solvents;[5,6] 2. Component 2 should contain a sulfoxide, phosphine oxide, or a tertiary amine functionality. These functional groups ensure interaction with component 1, which promotes liquid miscibility and solid immiscibility, a requirement for eutectic phase behaviour; 3. Each of the components should have a (predicted) water solubility below 1 g∙L-1, this ensures the hydrophobicity of the resulting eutectic mixture;

5

4. Each of the components should have a (predicted) logarithm water-octanol partition coefficient > 104. This also ensures hydrophobicity, additionally it reduces the

above 4, i.e.

probability of selecting components that are toxic for micro-organisms;[20–23] 5. Each of the components should be solid at 298 K, i.e.

> 298 K. The liquid phase should be

formed owing to a melting temperature depression of both components; 6. None of the components should be salts, because a pH-drop was observed in the water phase for hydrophobic DESs that contain acids as well as salts, while VFA producing micro-organisms cannot survive strong acidic conditions. In the supplementary material a more detailed explanation is given for this behaviour; 7. The components should be commercially available at lab-scale. Synthesis of novel components was considered outside the scope of the present study. The selected components and their properties are listed in Table 2. The structural formulas of the components are shown in Fig. S3 of the supplementary material. Table 2 Selected components for the hydrophobic DES screening. Their molecular weight (MW), experimental melting temperature ( ), predicted (by chemical abstracts) un-buffered water solubility ( ) and logarithmic water-octanol partition coefficient ( ), and the references for the melting temperatures [ref]. Component name

MW -1 g∙mol

K

g∙kg

[ref]

Component 1 decanoic acid

172.26

304.6-305.7

0.48

N,N’-dihexylthiourea tetradecanol

244.44 214.39

307-311 310.6-311.2

64∙10 0.58∙10-3

4.81 5.93

dodecanoic acid hexadecanol 1,2-decanediol di-n-dodecylamine 3,5-di-tertbutylcatechol cholesterol bisphenol Z Component 2 N-benzyl-N-ethylaniline trioctylphosphine oxide dodecyl-methyl-sulfoxide diphenylsulfoxide tribenzylamine

200.32 242.45 174.28 353.68 222.32 386.65 268.35

317-317.5 321.6-323.3 320.0-322.1 325.0 372.8 422-422.3 457-459

0.12 27∙10-6 0.33 -3 17∙10 0.33 -6 3.3∙10 -3 14∙10

4.77 6.95 2.56b 10.85 3.72 9.62 4.87

[24– 26] [27,28] [29– 31] [32] [29,31] [33] [34] [35] [36,37] [38,39]

211.30 386.63 232.43 202.27 287.40

307-309 321-325 336-338 342.6 366-367.8

4.07 10.25 4.19 2.66b 4.31

[40] [41,42] [43] [44] [45,46]

-1

a

3.75b -3

-3

20∙10 -6 58∙10 0.35 47∙10-3 -3 19∙10

6

Component name leucomalachite green a

MW -1 g∙mol 330.47

[ref]

-1

K 372-373

g∙kg -3 0.13∙10

6.41

[47,48]

b

Chemical was supplied as a liquid at 293-298 K. Included owing to availability in the lab.

2.3

Eutectic mixture screening

The formation of hydrophobic eutectic mixtures was screened by adding mixtures (2 g) of components 1 and 2 at three molar compositions (

= 0.333, 0.5, and 0.667) to a 10 mL glass headspace vial with a

magnetic stirring bar. The vials were heated in an aluminium heating block at 250 rpm using an IKA RCT basic heating plate equipped with an IKA ETS-D5 temperature controller (

= ±0.1 K). Initially, the

mixtures were heated to 313 K. If a sample formed a clear liquid after 2 h, it was removed from the heating block and stored at ambient temperature (293-298 K). For the remaining mixtures containing solids the temperature was increased with 20 K and they were re-evaluated after 2 h. This was continued up to a maximum temperature of 373 K was reached. Samples that did not liquefy completely at this stage were discarded. Mixtures that remained liquid at ambient temperature for 24 h were selected for the initial VFA extraction experiments.

2.4

Eutectic mixture preparation

Eutectic mixtures that were used for VFA extraction were prepared on a 10 g scale. Their components were weighed into a glass flask at the aimed molar ratios using a Sartorius analytical precision balance (precision 0.1 mg). The flask was heated to the liquefaction temperature that resulted from the screening, and stirred at 250 rpm until a clear liquid was obtained (typically < 30 min). The flasks were stored after being sealed with Parafilm®.

2.5

S-L phase behaviour

The S-L phase behaviour of eutectic mixtures was studied by differential scanning calorimetry (DSC) on a TA Instruments Q100 equipped with a refrigerated cooling stage (183 K limit). The enthalpy and temperature were calibrated using high purity indium (TA Instruments, > 99.99 wt%) and cyclohexane 7

(TCI, 100 wt% batch checked by GC). Prior to analysis, the components were dried under vacuum for more than 24 h. The samples were prepared in a dry nitrogen atmosphere by transferring the two components of interest to a glass vial at the aimed mole fraction with a total mass of 0.5 g, using an analytical balance. The mixtures were subsequently stirred and heated to 5 K above the melting temperature of the highest melting component. A liquid sample of 65 μL was transferred to an aluminium Tzero pan and subsequently crimped. The sample masses typically varied between 4.0 and 5.5 mg, depending on the mixture’s density. The samples were cooled to 193 K, kept at 193 K for 5 min, heated to 328 or 343 K and kept at that temperature for 5 minutes. The heating and cooling rate was 1 K∙min-1. The cooling and heating cycle were repeated twice. The second heating cycle was used for data analysis. The thermograms were analysed and integrated using the manufacturer software Universal Analysis 2000. The solidus temperature was taken as the peak temperature of the solidus peak, the liquidus temperature as the inflection point (minimum of first derivative) after the peak maximum of the liquidus peak. Other transition temperatures, as well as the pure component melting temperatures, were taken as the onset temperature.

2.6

Sample preparation for VFA extraction

2.6.1 Water phase Buffered VFA solutions were prepared for the extraction experiments. Typically an acid concentration of 10 g∙L-1 at a pH ≈ 4.2 was used, but both were varied systematically for the VFA extraction experiments with eutectic mixtures that performed well at the initial VFA extraction. In order to maintain the desired pH for all acid concentrations, potassium phosphate-based stock buffers with a phosphate concentration of approximately 0.2 M (19-20 g∙L-1) were prepared. Since the buffer capacity may be exceeded when pure VFA is added, a mixture of the VFA and its potassium or sodium carboxylate salt

8

was added corresponding to the acid/salt ratio at the desired pH (derived from the pKa of the VFA under consideration). VFA solutions of 10 g∙L-1 and 50 g∙L-1 were prepared by the addition of the VFA and its potassium/sodium salt to a 100 mL volumetric flask, which was then filled with a phosphate buffer at the appropriate pH. For the VFA concentration series the stock solution of 50 g∙L-1 was diluted with the appropriate phosphate buffer. All stock solutions and buffers were stored in the fridge and acclimatised to ambient temperature (293-295 K) before use.

2.6.2 Solvent phase The eutectic mixtures were prepared at the desired molar ratios as described before in section 2.4. 1-Octanol with 20 % V/V of trioctylamine (TOA) was prepared as benchmark solvent, by adding 50 mL of TOA to 200 mL of 1-octanol in a measuring cylinder while mixing.

2.7

VFA extraction procedure

For the extraction experiments, 2.5 mL of the water phase of interest was added to a 12 mL (height 90 mm, inner diameter 14 mm) glass centrifuge tube with a Finn-pipette. Volumetric solvent-to-feed (S/F) ratios of 1, 0.5, 0.4, 0.3, 0.2, and 0.1 were studied by adding 2.5, 1.25, 1, 0.75, 0.5, or 0.25 mL of solvent to the water phase, respectively. The masses of the added water (bottom) and solvent (top) phase were recorded. The tubes were inserted in an insulated aluminium heating block on top of a stirring/heating plate (IKA RCT basic + ETS-D5,

= ±0.1 K). Both phases were contacted and equilibrated overnight at

298 ± 1 K by vigorous stirring at 1200 rpm, using a cylindrical magnetic stirring bar with a cross at both ends. After equilibration, the stirrer was removed and the mixture was centrifuged at 3500 rpm and 298 ± 1 K for 10 or 30 min using a Sigma 2-16 KHL centrifuge. The losses of solvent/water adherent to the stirrer were negligible (< 0.1 wt%). Approximately 1.5 mL of the water phase was sampled for pH measurements as well as for acid and salt quantification. A long and thin needle (120 x 0.7 mm) was

9

immersed in the two-phase system, by-passing the solvent phase. The solvent phase was sampled for water content analysis.

2.8

pH of water phase

The pH of the water phase was determined with a Mettler Toledo SevenCompact S220 analyser equipped with a Mettler Toledo Inlab Micro pH sensor. The equipment was calibrated before each series of measurements using a 5-point calibration with the following certified buffers: pH 2.00 citrate buffer (VWR, 1.09442); pH 4.00 and 5.00 phthalate buffers (Fischer Chemical, SB98 and SB102, respectively); and pH 6.00 and 7.00 phosphate buffers (Fischer Chemical, SB104 and SB108, respectively). Aqueous samples of approximately 0.5 mL were added to a GC-vial. Subsequently, the pH sensor was immersed until a constant pH was obtained. This was repeated at least two times for each sample and the average value was recorded.

2.9

Water content

Before and after extraction, the water content of the solvent phase was determined using a coulometric Karl-Fischer titration. A Metrohm 899 Coulometer equipped with a generator electrode without diaphragm was used. 20 % V/V of chloroform was added to the titration medium (Hydranal coulomat AG) in order to improve the solubility of the hydrophobic extractants. The pure extractants were injected directly (~0.1-0.2 g). Due to their relatively high water contents, solvent phases that were sampled after extraction were diluted 10 times with a 20/80 % V/V chloroform/methanol mixture prior to injection (~0.3-0.5 g). The water contents of the diluted samples were corrected for the water content of the diluent. For each sample, the average value of a triplicate measurement was used. N,N’-dihexylthiourea (DHTU) is not compatible with the Karl-Fischer medium because it reacts with the iodine, leading to erroneous water contents. Therefore, DHTU-based samples were crimped into a vial and heated to 378 K in a Metrohm 832 KF Thermoprep sample oven to evaporate their water while 10

preserving the extractant. The headspace of the vials was continuously purged with dry nitrogen (50 mL∙min-1) and the gas phase with water vapour was then bubbled through the titration medium of a Metrohm 831 Coulometer equipped with a generator electrode without diaphragm. During preparation of the samples before (~0.2 g) and after (~0.05 g) extraction, an empty vial was capped to serve as a blank measurement by which the water contents of the samples were corrected.

2.10 Acid quantification The acid concentrations in the water phase were quantified on an Agilent 1260 Infinite HPLC equipped with a UV-Vis detector, a refractive index (RI) detector, and a Bio-Rad HPX-87H column connected to a Bio-Rad Cation-H microguard pre-column. A 5 mM aqueous H2SO4 solution was used as eluent at a flow rate of 0.6 mL∙min-1 (~70 bar). The injection volume was set at 10 µL. The column temperature was set at 338 K. The UV-Vis detector was operated at a wavelength of 210 nm and the temperature of the RI detector was set at 308 K. The run-time was 30 min per injection, see Table 3 for an overview of the detection and calibration. Calibration samples were prepared from the chemicals used during the extraction experiments, taking their purity into account. No significant differences in phosphate content could be detected between standards prepared from phosphoric acid or potassium phosphate salts. Before and after each set of samples, the calibration lines were validated with a calibration sample of 1000 mg∙L-1. All samples were injected in triplicate, always showing relative standard deviations < 1 % between the injections for concentrations within the calibration range. Table 3 Detection and calibration details of each acid for HPLC analysis, RI = refractive index and UV = ultraviolet absorption at 210 nm. Component Phosphate Acetic acid Propionic acid Butyric acid

Detector RI UV UV UV

Retention time min 9.0 14.9 17.5 21.3

Calibration 2

mg∙L 100 20 20 20

-1

-1

mg∙L 100000 20000 20000 20000

11

R

> 0.999 > 0.999 > 0.999 > 0.999

2.11 Cation quantification The Na+ and K+ contents of the water phase were evaluated before and after extraction with ion chromatography (IC). A Dionex 1100 IC equipped with an ion exchange column CS12A (2 x 250mm) was operated isocratically using a 20 mM aqueous methanesulfonic acid solution at a flow-rate of 0.25 mL∙min-1. A Dionex CSRS 500 2 mm suppressed conductivity detector was used for ion detection. The injection volume and runtime of each injection were set to 10 µL and 10 min respectively. The system was calibrated for Na+ and K+ using seven dilutions from 1000 mg∙L-1 IC standards from Sigma-Aldrich (product no. 43492 and 53337 for Na + and K+, respectively). The calibration details are listed in Table 4. Prior to analysis, the water phase was diluted 100 times with MilliQ grade water in a plastic 1.5 mL vial to a total volume of 1 mL. All samples were injected in triplicate, always showing relative standard deviations < 1 % between the injections for concentrations within the calibration range. Table 4 Detection and calibration details of each cation for IC analysis. Component Na+ K+

Retention time min 3.9-4.0 5.4-5.5

Calibration -1

mg∙L 0.1 10

R2

-1

mg∙L 50 250

> 0.999 > 0.999

2.12 Extraction performance The extraction performance was evaluated via the distribution coefficient, which is defined as:

Eq. 1

Where

is the distribution coefficient of component expressed in units

.

and

stand for the concentration of after equilibration (subscript AE) in the organic phase and aqueous phase, respectively. The concentration of acid and salt in the organic phase could not be analysed directly. Therefore, these concentrations were derived from mass balances. Considering the low water solubility of the solvent components, it was assumed that the solvent does not migrate to the aqueous

12

phase after extraction. Water does migrate to the organic phase, therefore the water content in the solvent phase was incorporated in order to close the mass balance. See the supplementary material for the derivation of an expression for VFA concentration in the organic phase after extraction.

3 3.1

Results and discussion Solvent screening and initial VFA extraction performance

3.1.1 Solvent screening The eutectic mixture screening yielded eight combinations of component 1 and component 2 that remained liquid within the temperature range 293-298 K for more than 24 h (Table 5). Two of these combinations of components involved dodecyl-methyl-sulfoxide, while the rest involved trioctylphosphine oxide (TOPO). The counterparts of dodecyl-methyl-sulfoxide were components with < 4 and were excluded from further experiments in accordance with criterion 4. Therefore, only mixtures based on TOPO were utilised for the first extraction screening. Table 5 Combinations of component 1 and 2 that resulted in a liquid after being heated to the liquefaction temperature mole fraction . The combinations that were selected for the initial VFA extraction experiments are highlighted. Component 1

Component 2

decanoic acid decanoic acid decanoic acid N,N’-dihexylthiourea N,N’-dihexylthiourea dodecanoic acid 1,2-decanediol 3,5-di-tertbutylcatechol 3,5-di-tertbutylcatechol 3,5-di-tertbutylcatechol bisphenol Z

trioctylphosphine oxide trioctylphosphine oxide dodecyl-methyl-sulfoxide trioctylphosphine oxide trioctylphosphine oxide trioctylphosphine oxide trioctylphosphine oxide trioctylphosphine oxide trioctylphosphine oxide dodecyl-methyl-sulfoxide trioctylphosphine oxide

0.333 0.500 0.333 0.500 0.667 0.500 0.500 0.333 0.500 0.500 0.667

K 313 313 313 313 333 313 313 333 313 313 333

3.1.2 Initial VFA extraction experiments In order to evaluate which of the combinations in Table 5 are suitable for VFA extraction, initial VFA extraction experiments were performed using unbuffered solutions containing 10 g∙L-1 lactic, acetic,

13

at

propionic, and butyric acid at a S/F ratio of 1. The extraction performance was evaluated for a single mixture of component 1 and 2, if multiple compositions resulted in a stable liquid

= 0.500 was

selected for the initial extraction experiments. The mixtures selected for the initial VFA experiments are highlighted in Table 5. During HPLC analysis after VFA extraction, it appeared that significant amounts of 1,2-decanediol leached to the water phase. The 1,2-decanediol-TOPO 1:1 mixture was therefore disregarded during the remainder of this study. Also the N,N’-dihexylthiourea-TOPO (DHTU-TOPO) system was not included in this initial evaluation of the extraction performance, due to a delay in the delivery of the DHTU. Fig. 1 presents the distribution coefficients of the tested VFAs for the remaining solvents, which all showed distribution coefficients that increase with the chain length/hydrophobicity of the extracted VFA (lactic acid < acetic acid < propionic acid < butyric acid). The mixtures based on bisphenol Z and 3,5-ditert-butylcatechol showed distribution coefficients close to or below 1. Therefore, these solvents were disregarded during the remainder of this study. Considering the very similar performance of the two medium-chain fatty acid-based mixtures and the low

of decanoic acid, only the dodecanoic acid-

TOPO (DodA-TOPO) system was studied in more detail. 12

Lactic acid Acetic acid

10

Propionic acid

bi / m3Aq×m-3 Org

Butyric acid 8

6

4

2

0 BisZ-TOPO xTOPO = 0.667

DTBC-TOPO xTOPO = 0.500

DodA-TOPO xTOPO = 0.500

DecA-TOPO xTOPO = 0.500

Fig. 1 Equilibrium distribution coefficients ( ) in various unbuffered DES/water systems for lactic acid (black), acetic acid (right hatch), propionic acid (grey), and butyric acid (left hatch). Solvent abbreviations: bisphenol Z (BisZ), trioctylphosphine oxide

14

(TOPO), 3,5-di-tert-butylcatechol (DTBC), dodecanoic acid (DodA), decanoic acid (DecA). -1 the mixture. Initial concentration of all acids was 10 g∙L , DES/water volume ratio was 1.

is the mole fraction of TOPO in = 1 is highlighted with a dashed line.

After the initial extraction experiments, the DHTU-TOPO system was evaluated for buffered systems (see section 3.3). DHTU-TOPO performed sufficiently well to be included in the more detailed evaluation of the extraction performance together with DodA-TOPO.

3.2

Liquid range of novel hydrophobic eutectic mixtures

The solid-liquid (S-L) phase behaviour of the binary DodA-TOPO and DHTU-TOPO mixtures was studied in order to determine their liquid range around 298 K (the extraction temperature). The liquid range confines the composition range for mixtures of component 1 and 2 that can be applied during LLE of the VFAs. The S-L phase behaviour was studied with DSC, the thermograms are shown in the supplemental material. The resulting phase diagrams are shown in Fig. 2 and Fig. 3. The melting temperature ( enthalpy of fusion (

) and

) of pure components are collected in

Table 6. As can be seen from this table, the obtained experimental values of

and

are in good

agreement with literature values, with the exception of the DHTU melting temperature. The literature on DHTU focusses on the synthesis, resulting in a wide variety of melting temperatures. Considering the relatively low purity (~97 wt%) of DHTU used in this work, the melting temperature was considered to be in fair agreement with those reported in literature. In order to visualise the deviation from ideal eutectic phase behaviour, the ideal liquidus phase boundaries (activity coefficients

) were included in Fig. 2 and Fig. 3. The liquidus phase boundaries

were derived from the pure component data using a simplified expression (Eq. 2) for the melting temperature depression, not taking into account the heat-capacity contribution. Previous work showed that the extrapolation of liquid heat capacities to temperatures (sufficiently far) below the melting temperature of the pure component introduces a larger error than its contribution to the melting temperature depression [49]. Eq. 2 shows that

is a pure-component property that shapes the S-L 15

phase behaviour of eutectic mixtures. Hence, future design studies of eutectic mixtures could include the

in the selection criterions.

Eq. 2

Table 6 Experimental and literature values of melting temperatures ( components for which the S-L phase behaviour was studied.

) and enthalpies of fusion (

) of the pure

Component

[ref] -1 K kJ∙mol dodecanoic acid 316.8 ± 0.1 36.3 ± 0.7 this work 316.9 -a [32] 317.5 36.3 [50] N,N’-dihexylthiourea 318.7 ± 0.3 20.0 ± 0.7b this work 329-331 -a [51] 308-312 -a [28] trioctylphosphine oxide 326.4 ± 0.2 53.5 ± 0.6 this work 326.7 -a [41] 321-323 -a [42] a not reported; bpeak overlapped with the S-S transition peak, total integral of both peaks is 27 ± 1.7 kJ∙mol-1 .

The solidus peaks were not very sharp in either of the two studied systems and showed considerable fronting. The onset temperatures could therefore not be extracted reliably, while the peak temperatures were relatively constant over the composition range studied. Hence, the peak temperature was taken as the solidus transition temperature, probably leading to a slight overestimation of the solidus temperature. Based on the thermograms the difference between the onset and peak temperatures is estimated to be 2-8 K.

16

ideal liq

liquidus

solidus

eutectic

selected

330 320

T/K

310 300 290 280 270 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

xTOPO Fig. 2 T-x (mole fractions) solid-liquid phase diagram of the dodecanoic acid-trioctylphosphine oxide (DodA-TOPO) mixture, showing the liquidus and solidus transitions, and the eutectic derived from the Tammann-plot (supplementary information). The ideal liquidus phase boundary is shown as a solid line. The operating temperature of the extraction experiments is highlighted with a dashed line, five compositions were selected within the liquid range at 298 K.

The DodA-TOPO system (Fig. 2) shows clear eutectic behaviour, with negative deviations from ideality for both components (see supplementary material for activity coefficients). The eutectic temperature was estimated at 272 ± 2 K, the eutectic composition (

= 0.43 ± 0.02) was derived from the

Tammann-plot, see supplementary material. From the liquidus phase boundaries, a eutectic composition slightly richer in TOPO would be expected. This is probably still within the uncertainty of the value presented. The uncertainty of the eutectic composition is relatively large because the solidus peaks were not very sharp and uniform, hampering a reliable integration to estimate

(DSC

thermograms available in Fig. S6 of the supplementary material). Additionally, the deviations on the linear fitting parameters propagate in the eutectic composition. The slightly lower solidus temperature at

= 0.9, the extrapolated zero solidus enthalpy at

>

0.878 (Tammann plot, supplementary material), and the slight positive deviation from ideality indicate the existence of solid solutions at the hypertectic side of the phase diagram. However, the S-L phase behaviour was not further explored, because this study focusses on the exploitation of the liquid phase. The liquidus temperatures shown in Fig. 2 are sufficient to estimate the liquidus phase boundary, and to 17

ensure stable liquid formation at temperatures above the liquidus transition. Five compositions equally spaced over the liquid range at 298 K were selected for the extraction experiments, namely

=

0.38; 0.41; 0.44; 0.47; and 0.50. ideal liq

liq

S-S

trs

sol

eut

sel

330 320 310

T/K

300 290 280 270 260 250 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

xTOPO Fig. 3 T-x (mole fractions) solid-liquid phase diagram of the N,N’-dihexylthiourea-trioctylphosphine oxide (DHTU-TOPO) mixture. Two additional transitions to the solidus (sol) and liquidus (liq) transitions were observed: a transition ascribed to the solid-solid transition of DHTU (S-S), and an extra transition in the hypertectic region slightly above the solidus temperature (trs). The eutectic (eut) was derived from the Tammann-plot (supplementary information), the ideal liquidus phase boundary is shown as a solid line. The operating temperature of the extraction experiments is highlighted with a dashed line, five compositions were selected (sel) within the liquid range at 298 K.

Fig. 3 shows the phase diagram of the DHTU-TOPO system, exhibiting eutectic behaviour with a negative deviation from ideality for both components as well. The melting temperature depression of TOPO is very similar to that of the DodA-TOPO system, although the melting temperature of DHTU is slightly more depressed than for DodA. The eutectic temperature was estimated at 267 ± 2 K, the eutectic composition (

= 0.44 ± 0.05) was derived from the Tammann-plot (supplementary material). The

solidus peaks of the DHTU-TOPO system (DSC thermograms in supplementary information) were less defined than those of the DodA-TOPO system, resulting in a larger uncertainty for the eutectic composition. Also, the peak splitting, owing to solid solution formation, limited the integration of the solidus peak. DHTU shows a transition at 307.2 K that can probably be ascribed to a S-S transition, this transition could also be recognised in the mixture with 18

= 0.10. The slightly stronger melting

temperature depression of DHTU resulted in a wider liquid range at 298 K as compared to the DodATOPO system. The five selected compositions are

= 0.34; 0.38; 0.42; 0.46; and 0.50.

Concluding, the studied eutectic mixtures DodA-TOPO and DHTU-TOPO form stable liquids at 298 K around the eutectic composition as a result of intermolecular interactions between their constituents, leading to negative deviations from ideal eutectic behaviour. This was confirmed by the activity coefficients that are below unity for DodA and DHTU at the eutectic composition. The eutectic properties of both mixtures are summarised in Table 7. The difference between the ideal and real eutectic temperature,

as defined by Martins et al.,[52] is significant. Hence, there is significant

deviation from ideal behaviour and the mixtures could be classified as deep eutectic mixtures or DESs. However, the normalised melting temperature depressions, as introduced in section 2.2 of previous work,[49] and the normalised

are relatively small. Since the discussion on which definition to use is

still ongoing, the studied systems will be referred to as eutectic mixtures rather than DESs throughout the remainder of this paper. Table 7 Eutectic properties of eutectic mixtures of dodecanoic acid (DodA) and N,N’-dihexylthiourea (DHTU) with trioctylphosphine oxide (TOPO). The ideal (ID) and experimental eutectic compositions ( ) and temperatures ( ), the activity coefficient of the most ideal component ( ), the normalised melting temperature depression ( ), the difference between the ideal and real eutectic temperature ( ), and normalised to the ideal eutectic temperature are presented.

(K) (K)

(K)

3.3

DodA-TOPO 0.314 0.429 ± 0.024 308.4 272 ± 2 0.18 0.14 36 0.12

DHTU-TOPO 0.275 0.437 ± 0.046 306.1 265 ± 2 0.36 0.17 39 0.13

Single stage VFA extraction

The VFA extraction performances of DHTU-TOPO and DodA-TOPO were evaluated in more detail via various single stage VFA extraction experiments. Each VFA was studied separately to simplify the system and avoid matrix effects or acid-acid coupling. Three acids (acetic, propionic, and butyric acid) were 19

studied, and all buffer ingredients (phosphate salts, phosphoric acid, and VFA salts) were monitored. TOA:Octanol 20:80 V/V (TOA-OCT 0.80) was analysed as benchmark for these extraction experiments. For each VFA, four parameters were examined that affect the VFA extraction: the pH of the water phase, the initial VFA concentration, the S/F ratio, and the composition of the binary eutectic mixture (see Fig. 4). pH water phase

pH < 6

Aq initial CVFA,BE

pH > 7

Solvent/Feed ratio

xTOPO in liquid region

Fig. 4 Schematic representation of the four parameters examined for the single stage extraction of volatile fatty acids (VFA), where is the VFA concentration in the aqueous phase before extraction, and the mole fraction of trioctylphosphine oxide in the eutectic mixture. The arrow indicates an increase in the studied property.

When studying the influence of one parameter on the extraction performance, all others were kept constant. The base case for the extraction experiments was: pH ≈ 4.2,

≈ 10 g∙L-1, S/F ≈ 1 (equal

volumes), and the default mole fraction of the eutectic mixture was taken as the middle value of those selected in section 3.2.

3.3.1 Effect of pH in the water phase Fig. 5 shows the results of the VFA extraction of the two studied eutectic mixtures and the benchmark. TOA-OCT 0.80 showed higher distribution coefficients than the eutectic mixtures, while the two eutectic mixtures exhibited a similar performance. All studied solvents showed an increasing affinity with increasing hydrophobicity of the VFA. The pH of the aqueous phase strongly affected the distribution coefficients of the VFAs. The extraction performance of the studied solvents increased with decreasing 20

initial pH of the aqueous phase. Under basic conditions, no extraction could be observed for any of the systems studied. This strongly implies that only protonated (undissociated) VFAs are extracted, and that the carboxylates remain in the water phase. In literature, similar extraction results were reported for ionic liquids (ILs) and other solvent systems containing TOA and TOPO [8,53–56]. 20

AA PA BA TOA-Oct DodA-TOPO DHTU-TOPO

18

bi / m3Aq×m-3 Org

16 14 12 10 8 6 4 2 0 2

3

4

5

6

7

8

pH Fig. 5 Distribution coefficient of acid ( ) after a single stage extraction versus the pH of the water phase before extraction. The studied acids are: acetic acid (AA); propionic acid (PA); and butyric acid (BA). The studied extraction solvents are: trioctylamine – 1-octanol 20/80 V/V (TOA-Oct); dodecanoic acid – trioctylphosphine oxide (DodA-TOPO) = 0.44; and N,N’dihexylthiourea – trioctylphosphine oxide (DHTU-TOPO) = 0.42. The Initial concentration of all acids was 10 g∙L-1. The dashed line highlights = 1.

The pH of the aqueous phase increased after extraction as a result of the selective removal of protonated acids. Hence, if more acid was transferred to the solvent phase, the pH of the aqueous phase showed a larger increase. Fig S8 demonstrates that for similar distribution coefficients the increase in pH was less for the eutectic mixtures than for the benchmark. In literature it was shown that TOA-Oct 0.80 co-extracts considerable amounts of the buffer’s phosphate, probably as phosphoric acid [8]. In order to evaluate the affinity towards phosphoric acid of the solvents that were studied in the present work, a single stage extraction experiment was performed using a buffer solution without VFAs as aqueous phase. The results are listed in Table 8 and confirm that phosphoric acid was extracted by TOA-Oct 0.80, leading to an increase of the pH of the aqueous phase. DodA-TOPO 0.44 extracted

21

phosphoric acid to a significantly lower extent, and no extraction of phosphoric acid was observed using DHTU-TOPO 0.42. Table 8 pH, phosphate (all degrees of protonation) concentration ( ) before and after equilibrium (BE, AE), and phosphate distribution coefficient for the different studied solvents: trioctylamine – 1-octanol 20/80 V/V (TOA-Oct 0.80); dodecanoic acid – trioctylphosphine oxide = 0.44 (DodA-TOPO 0.44); and N,N’-dihexylthiourea – trioctylphosphine oxide = 0.42 (DHTU-TOPO 0.42). Solvent

pH

TOA-Oct 0.80 DodA-TOPO 0.44 DHTU-TOPO 0.42

BE 4.29 4.29 4.29

pH g∙L-1 20.15 20.15 20.15

AE 5.98 4.34 4.30

g∙L-1 18.91 20.34 20.15

0.17 0.05 0.00

The solvents studied showed a significant difference in water uptake after equilibration. The water contents after equilibration were typically in the range of 1.5-3.5 wt% for TOA-Oct 0.80, 0.8-1.6 wt% for DodA-TOPO 0.44, and 0.3-0.5 wt% for DHTU-TOPO 0.42. Water co-extraction with the acids as observed for ILs [8,57] cannot be excluded based on our data. However, it seems to have a small effect on the total water content after equilibration, since the water content does not change significantly with the acid content of the organic phase for all acids (see Fig S9 supplementary material). Additionally, no clear pH dependency could be recognised for the water uptake. The observed differences in water uptake among the solvents studied are probably related with their hydrophobic nature. Similarly to ILs,[8,58] the eutectic mixtures and the benchmark solvent did not extract significant amounts of sodium or potassium cations. The water contents and concentrations of the metal cations before and after equilibration are presented in the data tables (supplementary information). The pH of the water phase affected the phase behaviour of DodA-TOPO 0.44. Fig. 6 (top row) shows additional phase separation at the liquid-liquid interface of the sample that was contacted to a basic aqueous phase. The phase separation became more pronounced after sampling and leaving the remaining sample overnight. The samples that were mixed with acidic VFA solutions did not show this behaviour. At the interface, DodA might dissolve in the aqueous phase with the carboxylic acid headgroup. In that case, the acid group could be deprotonated inducing phase separation via e.g. emulsion or 22

solid formation. During extraction the pH increases, hence the stability of the DodA-TOPO mixtures cannot be guaranteed throughout the process. Moreover, recovery of the VFAs and/or regeneration of the solvent phase via back-extraction with an alkaline solution would be impossible. DodA-TOPO was not further investigated.

Fig. 6 Photos of the single extraction experiments after equilibration and centrifugation for dodecanoic acid – trioctylphosphine oxide = 0.44 (top) and N,N’-dihexylthiourea – trioctylphosphine oxide = 0.42 (bottom). The extractions were performed at various initial pH of the aqueous phase: 2.9 (A); 4.1 (B); 5.6 (C); and 7.7 (D). Photo E was taken after sampling of the water phase and leaving the sample at ambient temperature overnight.

3.3.2 Effect of the VFA concentration The extraction performance of TOA-Oct 0.80 and DHTU-TOPO 0.42 was evaluated for various aqueous VFA concentrations. The results presented in Fig. 7 show a clear decrease of the distribution coefficient upon increasing butyric acid concentration at equilibrium for TOA-Oct 0.80. The results of DHTU-TOPO are less conclusive and show a weaker, similar trend. For acetic acid and propionic acid no significant influence of the VFA concentration on the distribution coefficient was observed. The concentration effect is significantly weaker for the solvents studied in this work than for phosphonium ILs described in literature [8,58].

23

12

AA PA BA TOA-OCT DHTU-TOPO

bi / m3Aq×m-3 Org

10 8 6 4 2 0 0.1

1

CAq i,AE

10

/ g×L

-1

Fig. 7 Distribution coefficient of acid ( ) after a single stage extraction versus the acid concentration in the water phase after extraction . The acids studied are: acetic acid (AA); propionic acid (PA); and butyric acid (BA). The studied extraction solvents are: trioctylamine – 1-octanol 20/80 V/V (TOA-Oct) and N,N’-dihexylthiourea – trioctylphosphine oxide (DHTU-TOPO) = 0.42. The dashed line highlights = 1, initial pH of the water phase was approximately 4.2.

The hydrophobicity of butyric acid could explain why its distribution coefficient depended strongly on its concentration in the aqueous phase. If hydrophobic solute-solvent interactions are predominant, the activity coefficient of butyric acid might be more concentration dependent than those of acetic and propionic acid. However, definitive conclusions regarding this hypothesis can only be drawn after reliable thermodynamic modelling of the extraction process, which would require a more extensive dataset. Additionally, other phenomena – such as (reverse) micelle formation [58] – were considered outside the scope of this study and were not taken into account. These could, however, greatly affect the partitioning of VFAs over the involved phases. The maximum solvent loading at a certain feed concentration for a counter-current extraction process with an infinite number of stages and a minimum solvent flow rate could be derived from the results of single-stage extraction experiments. In this specific case, the fully loaded solvent at the 1st extraction stage is then at equilibrium with the feed. Fig. 8 highlights the maximum solvent loadings at a feed concentration of 10 g∙L-1.

24

AA PA BA

-1 COrg i,AE / g×L

40

30

30

20

20

10

10

0

AA PA BA

40

0 0

10

20

30

40

-1 CAq i,AE / g×L

50

0

5

10

15

20

25

-1 CAq i,AE / g×L

Fig. 8 Acid concentration after extraction ( ) in the organic phase (Org) versus that in the aqueous phase (Aq). The studied extraction solvents are: N,N’-dihexylthiourea – trioctylphosphine oxide = 0.42 (left) and trioctylamine – 1-octanol 20/80 -1 V/V (right). The maximum solvent loadings at a feed acid concentration of 10 g∙L are highlighted.

Although the benchmark achieved higher VFA loadings for all acids studied, DHTU-TOPO could extract butyric acid more selectively. As this was a preliminary study, only solutions of single VFAs were explored. Similar single-stage extraction experiments on a (synthetic) fermentation broth should be performed to give conclusive insight in the solvent’s performance in terms of extraction capacity and selectivity.

3.3.3 Effect of S/F ratio Since the aim of the extraction was to concentrate the VFAs inside the solvent, only S/F ratios below 1 were considered; higher ratios would result in a more diluted organic phase. The results shown in Fig. 9 demonstrate that TOA-Oct 0.80 as well as DHTU-TOPO 0.42 showed an increase in the distribution coefficients upon decreasing the S/F ratio. This effect was most pronounced for butyric acid, followed by propionic acid. Only a minor effect of S/F ratio has been observed for acetic acid. The extraction performance of TOA-Oct 0.80 showed a stronger S/F ratio dependency than DHTU-TOPO 0.42. Low S/F ratios were more effective for the extraction of VFAs from dilute aqueous solutions using the solvents studied. This is economically advantageous, since less solvent (investment) is required to achieve higher VFA concentrations from the same feed.

25

16

AA PA BA TOA-OCT DHTU-TOPO

14

bi / m3Aq×m-3 Org

12 10 8 6 4 2 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

S/F Fig. 9 Distribution coefficient of acid ( ) after a single stage extraction versus the solvent-to-feed ratio (S/F). The studied acids are: acetic acid (AA); propionic acid (PA); and butyric acid (BA). The studied extraction solvents are: trioctylamine – 1-octanol 20/80 V/V (TOA-Oct) and N,N’-dihexylthiourea – trioctylphosphine oxide (DHTU-TOPO) = 0.42. The Initial concentration of all acids was 10 g∙L-1. The dashed line highlights = 1.

3.3.4 Effect of the mole fraction of TOPO One of the advantages of DESs and eutectic mixtures is that their properties can be tuned by varying the mole fraction of their components, i.e. their composition. Within the liquid range of DHTU-TOPO, five mole fractions were selected and applied as VFA extractant. The resulting distribution coefficients are collected in Fig. 10. The extraction performance improved upon increasing TOPO content for all VFAs studied. This improvement was more pronounced with increasing hydrophobicity of the VFA, in the order acetic acid < propionic acid < butyric acid.

26

5

bi / m3Aq×m-3 Org

4

AA PA BA

3

2

1

0 0.32

0.36

0.40

0.44

0.48

0.52

xTOPO Fig. 10 Distribution coefficient of acid ( ) after a single stage extraction versus the mole fraction of trioctylphosphine oxide ( ) in N,N’-dihexylthiourea – trioctylphosphine oxide mixtures (solvent). The acids studied are: acetic acid (AA); propionic -1 acid (PA); and butyric acid (BA). The Initial concentration of all acids was 10 g∙L . Arrows serve as a guide to the eye.

In literature, a similar behaviour has been reported for the extraction of propionic acid with TOPO dissolved in various non-protic hydrophobic solvents such as hexane, kerosene, and methyl isobutyl ketone [7,59]. It was found that TOPO was the key active component and that the solvent contribution was negligible. In DHTU-TOPO, DHTU could in principle also act as an active component, since it is capable to form hydrogen bonds with the VFAs. However, with decreasing DHTU content the extraction performance increases, also indicating that TOPO was indeed the major active component. The propionic acid distribution coefficients exhibited for TOPO-based extractants in literature are generally higher than those presented in Fig. 10. Several research groups studied the VFA distribution coefficient as a function of the molar concentration of TOPO in a solvent [7,59]. The results collected in Fig. 11 demonstrate that distribution coefficients between 4 and 25 would be expected when these trends are extrapolated to the TOPO concentrations in the eutectic mixture. A possible explanation for the much lower distribution coefficients obtained in this study, could be the competition between DHTU and VFAs for interaction with TOPO. Additionally, the non-protic diluents studied in literature hardly interact with the VFAs, increasing the strength of the interactions of TOPO with the VFAs. Strong

27

interactions between TOPO and VFAs considerably decrease the activity of the VFAs, resulting in higher distribution coefficients. As opposed to the eutectic DHTU-TOPO mixtures, where DHTU can interact with both TOPO and the VFAs.

bPA / m3Aq×m-3 Org

25

DHTU hexane kerosene MIBK

20 15 10 5 0 0.0

0.2

0.4

0.6

0.8

CTOPO / mol×L

1.0

1.2

1.4

-1

Fig. 11 Distribution coefficient of propionic acid ( ) versus the trioctylphosphine oxide (TOPO) concentration ( ) in the extractant with four different counterparts: N,N’-dihexylthiourea (DHTU); hexane (Keshav et al.[7]); kerosene (Bilgin et al.[59]); and methyl isobutyl ketone (MIBK, Bilgin et al.[59]). Dashed lines are second order polynomial extrapolations of the literature data.

The competition between VFA and DHTU is supported by a paper of Gilmore et al. on TOPO-based DESs which was submitted simultaneously with this work [60] as well as by the extraction of VFAs using ILs [5,8,57,61]. For the TOPO-based DESs high distribution coefficients were obtained for uranyl salt extraction since there is less competition between the coulombic salt complexation and the hydrogen bonding interactions leading to DES formation. For ILs, the coulombic attractions responsible for their formation typically do not block hydrogen bonding sites required for VFA extraction. Hence, the interactions involved in solvent formation and solvent extraction performance should be dissimilar if competition is expected. Whether the hydrogen bonding nature of DHTU leads to competition of VFAs with DHTU for the accepting oxide group of TOPO, might be verified by studying similar eutectic mixtures. The secondary amine groups could for instance be replaced with a non-hydrogen bond donating tertiary amine functionality, e.g. replacing DHTU with N,N’-dimethyldihexylthiourea. It would also be interesting to study the intermolecular interactions in more detail – e.g. with 15N and 31P-NMR or light/neutron 28

scattering – to obtain insight in the molecular structure of the eutectic mixture and the complexation of the VFA within this structure.

4

Conclusions

Novel eutectic mixtures were systematically designed for a specific application: the extraction of VFAs from dilute aqueous solutions. The S-L phase behaviour of trioctylphosphine oxide (TOPO) mixed with dodecanoic acid or dihexylthiourea (DHTU) allowed to reach TOPO concentrations well above the solubility limit of common diluents. The eutectic mixtures exhibited low water uptake and a minor extraction of phosphoric acid originating from the pH buffer, improving the selectivity as compared to 20% V/V trioctylamine in n-octanol. Exclusively undissociated VFAs were extracted. The extraction performance of DHTU-TOPO increased at low acid concentrations and small S/F ratios, which emphasizes its potential for VFA extraction from fermentation broths. The extraction performance could be tuned by changing the composition of the eutectic mixtures. However, the addition of TOPO led to a significantly smaller increase of the distribution coefficient than expected from previously reported TOPO-diluent solutions. This was ascribed to the competition between the VFA and DHTU to form intermolecular hydrogen bonds with TOPO. Future designs of eutectic mixtures and DESs for VFA extraction processes should focus on combining an active extractant (e.g. a phosphine oxide) with a hydrophobic component being not a hydrogen bond donor, and taking the enthalpy of fusion into account. Ultimately, this could result in the development of a high performance VFA extractant with a high (molar) concentration of the base functionality, and a high hydrophobicity.

Acknowledgements

29

Financial support from Netherlands Organization for Scientific Research (NWO) and the company Paques B.V. is gratefully acknowledged. This work is part of the research program “MES meets DES”, with project number STW-Paques 12999.

Supplementary material The supplementary information includes more details about: pH drop of acid-quaternary ammonium halide salt-based eutectic mixtures; mass balance determination of VFA concentration in solvent phase; structural formulas of screened components; Tammann-plots, DSC thermograms, and activity coefficients; pH difference before and after equilibrium; water uptake organic phase; and data tables.

References [1]

M.T. Agler, B.A. Wrenn, S.H. Zinder, L.T. Angenent, Waste to bioproduct conversion with undefined mixed cultures: the carboxylate platform, Trends Biotechnol. 29 (2011) 70–78. doi:10.1016/j.tibtech.2010.11.006.

[2]

W.S. Lee, A.S.M. Chua, H.K. Yeoh, G.C. Ngoh, A review of the production and applications of waste-derived volatile fatty acids, Chem. Eng. J. 235 (2014) 83–99. doi:10.1016/j.cej.2013.09.002.

[3]

S. Pratt, D. Liew, D.J. Batstone, A.G. Werker, F. Morgan-Sagastume, P.A. Lant, Inhibition by fatty acids during fermentation of pre-treated waste activated sludge, J. Biotechnol. 159 (2012) 38–43. doi:10.1016/j.jbiotec.2012.02.001.

[4]

L.A. Royce, P. Liu, M.J. Stebbins, B.C. Hanson, L.R. Jarboe, The damaging effects of short chain fatty acids on Escherichia coli membranes, Appl. Microbiol. Biotechnol. 97 (2013) 8317–8327. doi:10.1007/s00253-013-5113-5.

[5]

C.S. López-Garzón, A.J.J. Straathof, Recovery of carboxylic acids produced by fermentation,

30

Biotechnol. Adv. 32 (2014) 873–904. doi:10.1016/j.biotechadv.2014.04.002. [6]

A.S. Kertes, C.J. King, Extraction chemistry of fermentation product carboxylic acids, Biotechnol. Bioeng. 28 (1986) 269–282. doi:10.1002/bit.260280217.

[7]

A. Keshav, K.L. Wasewar, S. Chand, Equilibrium and Kinetics of the Extraction of Propionic Acid Using Tri-n-Octylphosphineoxide, Chem. Eng. Technol. 31 (2008) 1290–1295. doi:10.1002/ceat.200700490.

[8]

E. Reyhanitash, B. Zaalberg, S.R.A. Kersten, B. Schuur, Extraction of volatile fatty acids from fermented wastewater, Sep. Purif. Technol. 161 (2016) 61–68. doi:10.1016/j.seppur.2016.01.037.

[9]

D.J.G.P. van Osch, L.F. Zubeir, A. van den Bruinhorst, M.A.A. Rocha, M.C. Kroon, Hydrophobic deep eutectic solvents as water-immiscible extractants, Green Chem. 17 (2015) 4518–4521. doi:10.1039/C5GC01451D.

[10]

C.H.J.T. Dietz, M.C. Kroon, M. Di Stefano, M. van Sint Annaland, F. Gallucci, Selective separation of furfural and hydroxymethylfurfural from an aqueous solution using a supported hydrophobic deep eutectic solvent liquid membrane, Faraday Discuss. 206 (2018) 77–92. doi:10.1039/C7FD00152E.

[11]

D.J.G.P. van Osch, D. Parmentier, C.H.J.T. Dietz, A. van den Bruinhorst, R. Tuinier, M.C. Kroon, Removal of alkali and transition metal ions from water with hydrophobic deep eutectic solvents, Chem. Commun. 52 (2016) 11987–11990. doi:10.1039/C6CC06105B.

[12]

E.E. Tereshatov, M.Y. Boltoeva, C.M. Folden, First evidence of metal transfer into hydrophobic deep eutectic and low-transition-temperature mixtures: indium extraction from hydrochloric and oxalic acids, Green Chem. 18 (2016) 4616–4622. doi:10.1039/C5GC03080C.

[13]

C.H.J.T. Dietz, D.J.G.P. van Osch, M.C. Kroon, G. Sadowski, M. van Sint Annaland, F. Gallucci, L.F. 31

Zubeir, C. Held, PC-SAFT modeling of CO2 solubilities in hydrophobic deep eutectic solvents, Fluid Phase Equilib. 448 (2017) 94–98. doi:10.1016/j.fluid.2017.03.028. [14]

J. Cao, M. Yang, F. Cao, J. Wang, E. Su, Tailor-made hydrophobic deep eutectic solvents for cleaner extraction of polyprenyl acetates from Ginkgo biloba leaves, J. Clean. Prod. 152 (2017) 399–405. doi:10.1016/j.jclepro.2017.03.140.

[15]

J. Cao, M. Yang, F. Cao, J. Wang, E. Su, Well-Designed Hydrophobic Deep Eutectic Solvents As Green and Efficient Media for the Extraction of Artemisinin from Artemisia annua Leaves, ACS Sustain. Chem. Eng. 5 (2017) 3270–3278. doi:10.1021/acssuschemeng.6b03092.

[16]

B.D. Ribeiro, C. Florindo, L.C. Iff, M.A.Z. Coelho, I.M. Marrucho, Menthol-based eutectic mixtures: Hydrophobic low viscosity solvents, ACS Sustain. Chem. Eng. 3 (2015) 2469–2477. doi:10.1021/acssuschemeng.5b00532.

[17]

C. Florindo, L.C. Branco, I.M. Marrucho, Development of hydrophobic deep eutectic solvents for extraction of pesticides from aqueous environments, Fluid Phase Equilib. 448 (2017) 135–142. doi:10.1016/j.fluid.2017.04.002.

[18]

M. Francisco, A. van den Bruinhorst, M.C. Kroon, Low-Transition-Temperature Mixtures (LTTMs): A New Generation of Designer Solvents, Angew. Chemie Int. Ed. 52 (2013) 3074–3085. doi:10.1002/anie.201207548.

[19]

C. Florindo, F.S. Oliveira, L.P.N. Rebelo, A.M. Fernandes, I.M. Marrucho, Insights into the Synthesis and Properties of Deep Eutectic Solvents Based on Cholinium Chloride and Carboxylic Acids, ACS Sustain. Chem. Eng. 2 (2014) 2416–2425. doi:10.1021/sc500439w.

[20]

M.J. Playne, B.R. Smith, Toxicity of organic extraction reagents to anaerobic bacteria, Biotechnol. Bioeng. 25 (1983) 1251–1265. doi:10.1002/bit.260250508. 32

[21]

M. Vermuë, J. Sikkema, A. Verheul, R. Bakker, J. Tramper, Toxicity of homologous series of organic solvents for the gram-positive bacteria Arthrobacter and Nocardia Sp. and the gramnegative bacteria Acinetobacter and Pseudomonas Sp., Biotechnol. Bioeng. 42 (1993) 747–758. doi:10.1002/bit.260420610.

[22]

J. Sikkema, J.A. de Bont, B. Poolman, Mechanisms of membrane toxicity of hydrocarbons., Microbiol. Rev. 59 (1995) 201–22. doi:0146-0749/95/$04.00+0.

[23]

H.J. Heipieper, G. Neumann, S. Cornelissen, F. Meinhardt, Solvent-tolerant bacteria for biotransformations in two-phase fermentation systems, Appl. Microbiol. Biotechnol. 74 (2007) 961–973. doi:10.1007/s00253-006-0833-4.

[24]

K. Bica, J. Shamshina, W.L. Hough, D.R. MacFarlane, R.D. Rogers, Liquid forms of pharmaceutical co-crystals: exploring the boundaries of salt formation, Chem. Commun. 47 (2011) 2267–2269. doi:10.1039/C0CC04485G.

[25]

M. Ramos Riesco, F.J. Martínez-Casado, J.A.R. Cheda, M.I. Redondo Yélamos, A. FernándezMartínez, S. López-Andrés, Study of the Polymorphism in Copper(II) Decanoate through Its Phase Diagram with Decanoic Acid, and Texture of the Columnar Thermotropic Liquid Crystal Developable Domains in This and Similar Systems, Cryst. Growth Des. 15 (2015) 497–509. doi:10.1021/cg501572s.

[26]

Y. Cai, M. Liu, X. Song, J. Zhang, Q. Wei, L. Zhang, A form-stable phase change material made with a cellulose acetate nanofibrous mat from bicomponent electrospinning and incorporated capric– myristic–stearic acid ternary eutectic mixture for thermal energy storage/retrieval, RSC Adv. 5 (2015) 84245–84251. doi:10.1039/C5RA14876F.

[27]

M.R. Sambrook, P.D. Beer, J.A. Wisner, R.L. Paul, A.R. Cowley, F. Szemes, M.G.B. Drew, Anion-

33

Templated Assembly of Pseudorotaxanes: Importance of Anion Template, Strength of Ion-Pair Thread Association, and Macrocycle Ring Size, J. Am. Chem. Soc. 127 (2005) 2292–2302. doi:10.1021/ja046278z. [28]

M.R. Maddani, K.R. Prabhu, A Concise Synthesis of Substituted Thiourea Derivatives in Aqueous Medium, J. Org. Chem. 75 (2010) 2327–2332. doi:10.1021/jo1001593.

[29]

C. Mosselman, J. Mourik, H. Dekker, Enthalpies of phase change and heat capacities of some long-chain alcohols. Adiabatic semi-microcalorimeter for studies of polymorphism, J. Chem. Thermodyn. 6 (1974) 477–487. doi:10.1016/0021-9614(74)90009-3.

[30]

N.D.D. Carareto, A.O. Dos Santos, M.P. Rolemberg, L.P. Cardoso, M.C. Costa, A.J.A. Meirelles, On the solid-liquid phase diagrams of binary mixtures of even saturated fatty alcohols: Systems exhibiting peritectic reaction, Thermochim. Acta. 589 (2014) 137–147. doi:10.1016/j.tca.2014.05.022.

[31]

N.D. Dorighello Carareto, M.C. Costa, A.J.A. Meirelles, J. Pauly, High Pressure Solid-Liquid Equilibrium of Fatty Alcohols Binary Systems from 1-Dodecanol, 1-Tetradecanol, 1-Hexadecanol, and 1-Octadecanol, J. Chem. Eng. Data. 60 (2015) 2966–2973. doi:10.1021/acs.jced.5b00330.

[32]

M.C. Costa, M.P. Rolemberg, L.A.D. Boros, M.A. Krähenbühl, M.G. de Oliveira, A.J.A. Meirelles, Solid−Liquid Equilibrium of Binary Fatty Acid Mixtures, J. Chem. Eng. Data. 52 (2007) 30–36. doi:10.1021/je060146z.

[33]

C.-C. Kwan, M.J. Rosen, Relationship of structure to properties in surfactants. 9. Syntheses and properties of 1,2- and 1,3-alkanediols, J. Phys. Chem. 84 (1980) 547–551. doi:10.1021/j100442a021.

[34]

C.W. Hoerr, H.J. Harwood, A.W. Ralston, Solubilities of High Molecular Weight Symmetrical 34

Normal Aliphatic Secondary Amines, J. Org. Chem. 09 (1944) 201–210. doi:10.1021/jo01184a011. [35]

S.P. Verevkin, C. Schick, Substituent Effects on the Benzene Ring. Determination of the Intramolecular Interactions of Substituents in tert-Alkyl-Substituted Catechols from Thermochemical Measurements, J. Chem. Eng. Data. 45 (2000) 946–952. doi:10.1021/je0001126.

[36]

J.C. va. Miltenburg, A.C.G. va. Genderen, G.J.K. va. den Berg, Design improvements in adiabatic calorimetry, Thermochim. Acta. 319 (1998) 151–162. doi:10.1016/S0040-6031(98)00402-X.

[37]

J. Kaloustian, A.-M. Pauli, P. Lechene de la Porte, H. Lafont, H. Portugal, Thermal Analysis of anhydrous and Hydrated Cholesterol, J. Therm. Anal. Calorim. 71 (2003) 341–351. doi:10.1023/A:1022818902212.

[38]

B. Dannels, A.F. Shepard, Reductive coupling of aromatic compounds, J. Catal. 15 (1969) 106– 110. doi:10.1016/0021-9517(69)90014-1.

[39]

M.E. McGreal, V. Niederl, J.B. Niederl, Condensations of Ketones with Phenols, J. Am. Chem. Soc. 61 (1939) 345–348. doi:10.1021/ja01871a032.

[40]

G. Trapani, A. Reho, A. Latrofa, Trimethylamine-Borane as Useful Reagent in the N-Acylation or N-Alkylation of Amines by Carboxylic Acids, Synthesis (Stuttg). (1983) 1013–1014. doi:10.1055/s1983-30605.

[41]

R.C. Castells, A.M. Nardillo, Thermodynamics of the molecular association of tri-n-octylphosphine oxide and haloalkanes using gas-liquid chromatography, J. Solution Chem. 14 (1985) 87–100. doi:10.1007/BF00648898.

[42]

C.R. Hilliard, N. Bhuvanesh, J.A. Gladysz, J. Blümel, Synthesis, purification, and characterization of phosphine oxides and their hydrogen peroxide adducts, Dalt. Trans. 41 (2012) 1742–1754. doi:10.1039/C1DT11863C. 35

[43]

B.M. Choudary, B. Bharathi, C.V. Reddy, M.L. Kantam, Tungstate-exchanged Mg-Al-LDH catalyst: an eco-compatible route for the oxidation of sulfides in aqueous medium, J. Chem. Soc. Perkin Trans. 1. (2002) 2069–2074. doi:10.1039/b205292j.

[44]

R.H. Smith, D.H. Andrews, Thermal Energy Studies. II. Phenyl Derivatives of Metals, J. Am. Chem. Soc. 53 (1931) 3661–3667. doi:10.1021/ja01361a008.

[45]

M.M. Kreevoy, Y. Wang, Kinetic and equilibrium acid-base behavior of tertiary amines in anhydrous and moist dimethyl sulfoxide, J. Phys. Chem. 81 (1977) 1924–1928. doi:10.1021/j100535a008.

[46]

G.O. Dudek, F.H. Westheimer, The Solvolysis of Tetrabenzylpyrophosphate, J. Am. Chem. Soc. 81 (1959) 2641–2646. doi:10.1021/ja01520a011.

[47]

C.D. Ritchie, W.F. Sager, E.S. Lewis, Application of Linear Free Energy Relationships to Some Reactions of Triarylmethane Derivatives, J. Am. Chem. Soc. 84 (1962) 2349–2356. doi:10.1021/ja00871a016.

[48]

M. Barbero, S. Cadamuro, S. Dughera, M. Rucci, G. Spano, P. Venturello, Solvent-free Brønsted acid catalysed alkylation of arenes and heteroarenes with benzylic alcohols, Tetrahedron. 70 (2014) 1818–1826. doi:10.1016/j.tet.2014.01.011.

[49]

A. van den Bruinhorst, A Systematic Evaluation of Deep Eutectic Solvent Formation, Stablity, and Physicochemical Properties, in: Deep Eutectic Solvents a New Gener. Des. Solvents, Technische Universiteit Eindhoven, 2018: pp. 224–321.

[50]

T. Inoue, Y. Hisatsugu, R. Ishikawa, M. Suzuki, Solid-liquid phase behavior of binary fatty acid mixtures: 2. Mixtures of oleic acid with lauric acid, myristic acid, and palmitic acid, Chem. Phys. Lipids. 127 (2004) 161–173. doi:10.1016/j.chemphyslip.2003.10.013. 36

[51]

N. Azizi, E. Farhadi, Rapid and highly efficient synthesis of thioureas in biocompatible basic choline hydroxide, J. Sulfur Chem. 38 (2017) 548–554. doi:10.1080/17415993.2017.1327591.

[52]

M.A.R. Martins, S.P. Pinho, J.A.P. Coutinho, Insights into the Nature of Eutectic and Deep Eutectic Mixtures, J. Solution Chem. (2018). doi:10.1007/s10953-018-0793-1.

[53]

S.T. Yang, S.A. White, S.T. Hsu, Extraction of Carboxylic Acids with Tertiary and Quaternary Amines: Effect of pH, Ind. Eng. Chem. Res. 30 (1991) 1335–1342. doi:10.1021/ie00054a040.

[54]

R. Canari, A.M. Eyal, Selectivity in Monocarboxylic Acids Extraction from Their Mixture Solutions Using an Amine-Based Extractant: Effect of pH, Ind. Eng. Chem. Res. 42 (2003) 1301–1307. doi:10.1021/ie020898w.

[55]

E. Alkaya, S. Kaptan, L. Ozkan, S. Uludag-Demirer, G.N. Demirer, Recovery of acids from anaerobic acidification broth by liquid-liquid extraction., Chemosphere. 77 (2009) 1137–42. doi:10.1016/j.chemosphere.2009.08.027.

[56]

B.-H. Um, B. Friedman, G.P. van Walsum, Conditioning hardwood-derived pre-pulping extracts for use in fermentation through removal and recovery of acetic acid using trioctylphosphine oxide (TOPO), Holzforschung. 65 (2011) 51–58. doi:10.1515/hf.2010.115.

[57]

Š. Schlosser, J. Marták, M. Blahušiak, Specific phenomena in carboxylic acids extraction by selected types of hydrophobic ionic liquids, Chem. Pap. 72 (2018) 567–584. doi:10.1007/s11696017-0365-7.

[58]

J. Marták, Š. Schlosser, Extraction of lactic acid by phosphonium ionic liquids, Sep. Purif. Technol. 57 (2007) 483–494. doi:10.1016/j.seppur.2006.09.013.

[59]

M. Bilgin, C. Ar soy, S.I. K rba lar, Extraction Equilibria of Propionic and Butyric Acids with Tri-noctylphosphine Oxide/Diluent Systems, J. Chem. Eng. Data. 54 (2009) 3008–3013. 37

doi:10.1021/je900063p. [60]

M. Gilmore, É.N. McCourt, F. Connolly, P. Nockemann, M. Swadźba-Kwaśny, J.D. Holbrey, Hydrophobic Deep Eutectic Solvents Incorporating Trioctylphosphine Oxide: Advanced Liquid Extractants, ACS Sustain. Chem. Eng. 6 (2018) 17323–17332. doi:10.1021/acssuschemeng.8b04843.

[61]

J. Marták, S. Schlosser, Phosphonium ionic liquids as new, reactive extractants of lactic acid, Chem. Pap. 60 (2006) 395–398. doi:10.2478/s11696-006-0072-2.

38

Highlights

   

Two of the designed novel solvents deviate from ideal eutectic phase behavior Trioctylphosphine oxide-based eutectic mixtures only extract undissociated acids Eutectic mixtures show distribution coefficients: acetic < propionic < butyric acid Acid distribution coefficients increase with mole fraction trioctylphosphine oxide

39