Determination of the gas-to-liquid partitioning coefficients using a new dynamic absorption method (DynAb method)

Accepted Manuscript Determination of the gas-to-liquid partitioning coefficients using a new dynamic absorption method (DynAb Method) Joren Bruneel, Christophe Walgraeve, Katrijn Van Huffel, Herman Van Langenhove PII: DOI: Reference:

S1385-8947(15)01019-0 http://dx.doi.org/10.1016/j.cej.2015.07.053 CEJ 13951

To appear in:

Chemical Engineering Journal

Received Date: Revised Date: Accepted Date:

3 June 2015 13 July 2015 16 July 2015

Please cite this article as: J. Bruneel, C. Walgraeve, K. Van Huffel, H. Van Langenhove, Determination of the gasto-liquid partitioning coefficients using a new dynamic absorption method (DynAb Method), Chemical Engineering Journal (2015), doi: http://dx.doi.org/10.1016/j.cej.2015.07.053

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.

Determination of the gas-to-liquid partitioning coefficients using a new dynamic absorption method (DynAb Method) Joren Bruneel1, Christophe Walgraeve1,*, Katrijn Van Huffel1, Herman Van Langenhove1 1

Research Group EnVOC, Faculty of Bioscience Engineering, Ghent

University, Coupure Links 653, 9000 Ghent, Belgium Tel.: +32 92 64 99 24 *Corresponding author: Christophe Walgraeve [email protected]/[email protected] URL: http://www.EnVOC.UGent.be

Abstract The determination of partitioning coefficients which describes the behavior of a pollutant between phases (gas, liquid, solid) is of fundamental importance in different scientific areas (fate and behavior of pollutants, design of abatement technologies, flavor release). A new and fast method, based on selected ion flow tube mass spectrometry, was developed to determine the gas-to-liquid partitioning coefficients (KAW) of volatile organic compounds (VOC). By using the dynamic absorption method (DynAb method), a gas stream with a known and constant compound concentration is bubbled through a liquid volume. The outlet gas concentration is continuously measured and results in a breakthrough curve. From the 1

breakthrough curve of the measured gas concentration, KAW can be calculated. In contrast to the existing dynamic methods, KAW obtained by DynAb method does not assume equilibrium of the target compound between the liquid and the air bubble leaving the liquid and therefore avoids erroneous evaluation of KAW. The developed method was applied to determine KAW for five odorous VOCs (dimethyl sulphide, dimethyl disulphide, 2-methylpropanal, 3-methylbutanal and hexanal). KAW values can be determined in short time (less than 15 minutes) with a reproducibility of less than 5 %. Measured KAW (mol m-3gas/mol m-3liquid) at 25 °C are 8.61 ± 0.20 10-2 for dimethyl sulphide, 5.90 ± 0.23 10-2 for dimethyl disulphide, 1.27 ± 0.06 10-2 for 2-methylpropanal, 1.43 ± 0.05 10-2 for 3-methylbutanal and 1.41±0.03 10-2 for hexanal. The dependence of the partitioning coefficients on temperature (4 to 25 °C) and ammonium sulphate concentration (0 to 300 g L-1) were described by Van’t Hoff and Setchenow equations, respectively. Keywords: Henry coefficient, partitioning, dynamic absorption method, SIFT-MS, VOC, odorants

Abbreviations 2-MP

2-methylpropanal

3-MB

3-methylbutanal

A

Area

AS

Ammonium sulphate

b

Asymmetric factor

Cair,eq, Cwater,eq

Concentration in the air phase or water phase at equilibrium

Ct , C0

Concentration in the gas phase at time t or time zero

CIN, COUT

Concentration in the gas phase at the inlet or outlet of bubbling vessel

 C

Normalized concentration using CIN and COUT (-)

DMDS

Dimethyl disulphide

2

DMS

Dimethyl sulphide

DynAb method

Dynamic absorption method

EPICS

Equilibrium partitioning in closed system

∆HA
W

Enthalpy change of the phase transfer from gas to liquid (kJ mol-1)

∆SA
W

Entropy change of the phase transfer from gas to liquid (kJ mol -1)

PTFE

Polytetrafluorethylene

Q

Gas flow rate (mL min-1)

KS

Salting-out constant or Setchenow constant (M-1)

KAW

Gas-to-liquid partitioning coefficient (-)

HEX

Hexanal

MFC

Mass flow controller

m/z

Mass-to-charge ratio

n

Number of samples or values

n.a.

Not available

QSAR

Quantitative structure-activity relationship

R

Universal gas constant (J mol-1 K-1)

RSD

Relative standard deviation (%)

CS

Salt concentration in water (M or g L-1)

SIFT-MS

Selected ion flow tube mass spectrometry

T

Temperature (K or °C)

t

Time (min)

tmeasure

Measuring time (min)

VOC

Volatile organic compound

V

Volume (mL)

3

1. Introduction

The availability of reliable partition coefficients is of fundamental importance in many scientific areas. For example, partitioning coefficients are used in (i) human health studies for assessing the uptake of (inhaled) organic compounds in blood and body liquids [1-3] (ii) in food industry to the describe the release of aroma’s from liquids [4-6] (iii) in waste gas treatment, to model the mass transfer of pollutants from the gas phase to liquid phase, which is invaluable for the reactor design [7].

The partition coefficient is a temperature ( ) dependent physical parameter expressing a compounds’ equilibrium between different phases. When an air/pure water two phase system is considered, the equilibrium is described by the Henry’s law coefficient, defined as the ratio of the partial pressure (or gas concentration) of the organic compound to the compound concentration in the water phase, at equilibrium. This Henry’s law is only valid if there is an infinite dilution of the compound in the solvent (water). In practice, the condition of infinite dilution is met at compound-to-solvent molar ratios less than 1:1000. The term apparent Henry’s law coefficient or gas-to-liquid partition coefficient (KAW) is used in case the liquid phase is not pure water (e.g. water containing salts) and the gas phase is air. KAW is defined as the ratio of the concentration of a compound in the air phase (  , ) and the concentration in the water phase (  , ) at equilibrium (Equation 1).

 =

 ,

 ,

(1)

Several methods are described in literature to determine the KAW. Two major groups can be distinguished. The first group consists of methods which (i) predicts the partitioning coefficients using literature data of vapour pressure and maximum water solubility or (ii) rely on quantitative structure activity relationship (QSAR) methods which estimate the partitioning coefficients from the molecular structure using statistical relationships [8].

4

It has to be noted that the measurement of the water solubility [9] and vapour pressures [10] is time consuming. The mentioned methods do not allow to assess the effect of salt concentration and the effect of complex matrices (e.g. liquid samples from biotechnologies) on the partitioning coefficients [11]. Therefore these effects and individual coefficients should be determined experimentally. Different approaches can be used.

The first approach (i) uses static equilibrium methods. The equilibrium partitioning in closed system (EPICS) is one of the most often used static methods. For the EPICS two or more closed two phase (air/water) systems are made containing the same amount (it is not necessary to know the exact amount) of the target analyte but different liquid volumes. After equilibrium, the concentration of the compound in the air phase is measured (as peak area). By applying mass balances over the two phase systems and by knowing the liquid and gas volumes, the KAW can be calculated [12, 13]. These methods were reviewed in detail by Staudinger and Roberts [8]. The main disadvantage of the static methodology are the high standard deviations (more than 10 %) on repeated measurements both for low (KAW < 0.06) and high (KAW > 8) partitioning coefficients [8]. Also between this range, high standard deviations are reported [14, 15]. A second experimental design (ii) uses a dynamic approach in which a spiked water solution (volume ) containing the target compound(s) is stripped with an inert gas stream () at constant temperature (Inert Gas Stripping method). The partition coefficient can be deduced from the slope of the linear plot of the natural logarithm of the normalised concentration (  ⁄ ) over time () (Equation 2)[16-18].

     = −  

 

(2)

With  and  the gas concentration of the pollutant at time  and at time zero. This method assumes that (i) upraising air bubbles are in equilibrium with the liquid when they leave the liquid phase and (ii) the concentration of the compound in the liquid is uniform within the

5

liquid volume, at all times. These assumptions must be checked carefully to obtain accurate results. The aim of this research is to develop a new and fast method to determine the gas-to-liquid partitioning coefficients using an absorption method, further referred to as dynamic absorption method (DynAb method). Methodologies to evaluate KAW by means of absorption are very scarce in literature [19]. The absorption method by Dumont et al. (2010) uses a circulating contaminated air flow through liquid in a closed system. In this method the gas concentration decreases to an equilibrium concentration due to absorption. From the beginning and equilibrium gas concentration the KAW is calculated. Nevertheless, this method was assessed to be non-accurate by the authors. The absorption method (EXPSAT method) developed by Dohnal and Hovorka (1999) is most related to the new absorption method described in this study. However, the absorption curve was not registered with a real time monitoring device but with a discrete concentration measurement by gas chromatography. The developed DynAb method, will be applied to determine the partitioning coefficients of important odorous compounds: dimethyl sulphide (DMS), dimethyl disulphide (DMDS), hexanal (HEX), 2-methylpropanal (2-MP) and 3-methylbutanal (3-MB). The removal of odorous compounds is important in the air treatment of livestock and bio-waste valorisation facilities. This removal of odorant compounds is not well understood. Insights in the KAW leads to a better understanding of their removal and will help to model scrubber systems. Since sulphuric acid scrubbers are widely applied in livestock air treatment to remove ammonia, the effect of temperature and ammonia sulphate on concentration KAW will be investigated. The latter concentration increases during scrubber operation and might influence the partitioning coefficients of odorous compounds and by consequence their removal efficiency.

6

2. Materials and methods In the Dynamic Absorption method (DynAb method) a gas flow with a constant concentration of the organic compounds is bubbled through a liquid phase. The compounds undergo a transfer from the gas phase into the liquid phase until equilibrium is reached. The compound concentration at the outlet is continuously monitored by SIFT-MS. From the observed outlet compound concentration profile (breakthrough curve), the partition coefficient can be determined. 2.1 Gas generation system The generation of a gas stream with a stable compound concentration is an important prerequisite to use the DynAb method. A schematic overview of the gas generation system (similar to [21]) and DynAb method set-up is given in Figure 1. A pure nitrogen gas stream of 0.2 L min-1 (") was controlled by a mass flow controller (MFC1) (Brooks Instruments, USA) and was introduced into a cylindrical chamber (length 25 cm and inner diameter 4 cm). Five PTFE tubes (outer diameter: 6.35 mm), containing the five liquid target compounds (Merck, Germany), were connected with a stainless steel capillary using a stainless steel Swagelock union and Teflon ferrules. In order to ensure a workable (and similar) compound concentration for all target compounds, a different capillary was selected for each individual compound (Internal diameter and length; DMS: 254 µm and 10 cm; DMDS, 2-MP and 3-MB: 508 µm and 5 cm). This choice for capillary diameter and length was based on the compounds’ vapour pressure. Due to the low vapour pressure of hexanal, no capillary but a PTFE tube (inner diameter of 4 mm, length 4 cm) was attached to the PTFE tube containing hexanal. These tubes were placed in the cylindrical tube. The compound diffuses at a constant rate from the PTFE tube through the capillary to the " -stream. The concentration 7

difference between the air in the vial (saturated vapour pressure) and the "-stream provides a driving force for diffusion. To ascertain a constant diffusion through the capillary, the temperature of the cylindrical tube was held constant in a thermostatic cabinet at 40 °C (Lovibond, United Kingdom). The polluted air stream (") is further diluted by a nitrogen air stream (# , MFC2) to form air stream ($ ). The latter stream is fed to a MFC4 controlling the

flow at 95±0.5 mL min-1 (%& ). Stream %& is connected to a four way valve (SS-43YFS2, Swagelok, Belgian fluid system technologies BVBA, Groot-Bijgaarden, Belgium). The surplus stream (' ) is discarded. A valve assures that the pressure is above the working pressure (2 bar) of the MFC4. A pure nitrogen air stream (% , MFC3) with the same flow as the polluted

air stream (%& ), was also connected to the four-way valve. The streams (% , %& ) can be diverted to the bubble column or discarded (hood) by turning the four-way valve. 2.2 Dynamic absorption method (DynAb method) A volume of water was pipetted into the bubble column (max volume: 70 mL). This glass bubble column was equipped with a sintered glass plate (Porosity P1, pore size between 100 and 160 µm) to enable generation of small air bubbles when air is bubbled through the liquid. The bubble column was submerged in a temperature controlled water bath with a heating and cooling system. After temperature equilibration of the liquid (for at least 15 minutes) in the bubble column, the % -pure air stream was send to the bubble column. At time zero ( ) the four-way valve was turned whereby the gas stream entering into the bubble column changes from pure air stream (% ) to polluted air stream (%& ). Compounds are now being absorbed in the liquid, until the liquid is in equilibrium with the gas stream. At that moment the compound concentration in the outgoing air stream is equal to the inlet gas concentration. The compound concentration in the outgoing air stream is continuously measured throughout the whole experiment by SIFT-MS. This results in a breakthrough curve (Figure 2 – right curve). In order to allow for blank correction, a breakthrough experiment (Figure 2 – left curve) was conducted for which the bubble column was filled with glass beads (3 mm) representing the same volume as the water phase. The blank 8

breakthrough curve is dependent on the dead volume in the bubble column and to a minor extend to sorption effects on the constructed materials. The area ( between both breakthrough curves is proportional to the mass of absorbed compounds in the liquid phase. The calculation of the area is based on a numerical integration of the fitting curve (Equation 3) through the (blank and sample) breakthrough curve. +, +, - . /

)*  = )*, + .1− 12  /&

(3)

Factor 3 describes the asymmetric shape of the curve. From the measured absorption breakthrough curves, KAW can be calculated using Equation 4 provided that the gas flow rate () and liquid volume () are known and the area (() is determined. Equation 4 is using the  normalised gas concentration C , defined as the ratio of the outgoing compound gas

concentration ( 456 ) and inlet compound gas concentration ( 78 ). The normalised gas +, concentration, i.e. 456 ⁄ 78 10# ( )* ) is represented as a percentage value. The time

which is needed for the liquid to become in equilibrium with the gas phase (with a constant compound concentration) is theoretically infinite. But for practical reasons, the absorption experiment was stopped at the moment when the outlet concentration was equal for at least +, five minutes to the inlet concentration (i.e )* = 100%). The shape of the sample

breakthrough curve is influenced by mass transfer kinetics. It is possible that the raising air bubble may or may not be in equilibrium with the liquid when the bubble is leaving the liquid. This results into breakthrough curves of different shape. Although the shape can change, the absorbed component mass (and thus the area () will remain the same when equilibrium between the phases is obtained (applying the same compound inlet concentration). This is an important advantage of the DynAb method when compared to the dynamic stripping methods used in literature. The latter methods assume equilibrium between the raising air bubble and the liquid, which is practically difficult to verify.

9

 =

+, )*, +

+, : ; <

=

100

 +,  => D?100 − )*,*-@: ./A B E





−

 +, ./ABF > D?100 − )*,&:+C E

=

100  ( 

(4)

2.3 Selected ion flow tube mass spectrometry The compound concentration in the gas stream leaving the bubble column was continuously measured by means of SIFT-MS throughout the whole experiment (Voice 200 by Syft technologies, Interscience Louvain-La-Neuve, Belgium). SIFT-MS was introduced in the early 2000s and makes it possible to measure real time gas concentrations of multiple compounds simultaneously [22]. The technique is based on the chemical ionization of pollutants using NO+, H3O+ and O2+ precursor ions, resulting in product ions which are detected by a quadrupole mass spectrometer.

3. Results and discussion 3.1 Gas Generation system The compound concentration in the air stream should be carefully monitored. Therefore only product ions for which i) the standard deviation on the product ion signals (during a 6 hour measurement) is less than 5%, ii) branching ratio (relative abundance (%) of a product ion formed from the reaction of a certain precursor ion with the target compound) is equal or more than 50 % and iii) signal to noise ratio is higher than 20. The following product ions (based on NO+ precursor ion) were finally selected: m/z 62 of (CH3)2S+ (DMS); m/z 94 of (CH3)2S2+ (DMDS); m/z 71 of C4H7+ (2-MP); m/z 87 of C5H11O+ (3-MB) and m/z 101 C6H13O+ (HEX). The compound concentration was always kept below 2 ppmv. The compound-to-water ratio is then kept below a 1:1000 molar ratio to ensure the experiments were carried out at infinite 10

dilution (Henry’s law region). The inlet gas stream must have a constant concentration during the breakthrough experiment to gain reproducible KAW values. Long term measurements of more than 6 hours proved that pollutant concentration (0.1 to 2 ppmv) remained stable (Relative Standard Deviation (RSD) was for DMS, DMDS, 2-MB, 3-MB and HEX respectively 4.3, 4.9, 3.2, 3.8 and 3.2 %).

3.2 Henry coefficient determination of the selected compounds The developed DynAb method was used to determine the KAW partitioning coefficients for DMS, DMDS, 2-MP, 3-MB and HEX in pure water systems. KAW values were determined using the methodology described in the Material and Methods section. Therefore, 20 mL deionised water was transferred to the bubble column and inserted in the temperature controlled water bath. The air flow rate () of the compound loaded stream was set at 95 ± 0.5 mL min-1. The experiments were carried out at different temperatures (4, 11, 18 and 25°C). The experimental values are given in Table 1). In order to check the reproducibility of the DynAb method, the KAW values at 25°C were determined five times. The dimensionless KAW-values at 25 °C for DMS, DMDS, 2-MP, 3-MB and HEX are respectively 8.61 ± 0.20 10-2; 5.90 ± 0.23 10-2; 1.27 ± 0.06 10-2; 1.43 ± 0.05 10-2 and 1.41 ± 0.03 10-2. The RSD on the determined KAW ranged between 2.1 % (HEX) and 4.7% (2-MP). These low RSDs illustrate the good reproducibility of determined KAW values using the DynAb method. Table 1: Gas-to-liquid partitioning coefficient K AW (-) of DMS, DMDS, 2-MP, 3-MB and HEX as a function of -1 temperature (T). CS is the ammonium sulphate concentration in the water phase (g L )

Gas-to-liquid partitioning coefficient KAW (-) -1

CS (g L ) T (°C) 0 0 0 0

4 11 18 25

n

DMS

1 1 1 5

2.79 10 -2 4.12 10 -2 6.08 10 -2 8.61±0.20 10

DMDS -2

2-MP

-2

1.70 10 -2 2.60 10 -2 4.15 10 -2 5.90±0.23 10

11

3-MB -3

1.96 10 -3 3.85 10 -3 6.91 10 -2 1.27±0.06 10

HEX -3

2.25 10 -3 4.52 10 -2 8.20 10 -2 1.43±0.05 10

-3

1.54 10 -3 3.46 10 -3 6.80 10 -2 1.41±0.03 10

The obtained KAW values were compared with literature data (Table 2). It should be noted that for 2-MB and 3-MB KAW values are only scarcely determined. When the data in Table 2 are studied, it can be seen that considerable differences exists between KAW values that were determined using different methodologies. The KAW determined with the dynamic methods correspond well with the obtained values in this study. Only 0.4 % and 3.9 % difference on average values was observed for respectively DMS and DMDS. Also, the RSD on the determined KAW values in this study are in most cases lower than the RSD on the KAW values determined by the other dynamic methods. Table 2: Overview of literature KAW-values (-) for DMS, DMDS, 2-MP, 3-MB and HEX for different methods.

Compound DMS

2

KAW 10 (-)

RSD (%)

T (°C)

Reference

8.60 4.89 5.00 5.00 6.10 6.75 7.28 7.34 7.93 8.40 12.50 8.39 8.50 8.70 8.64 8.61

n.a. n.a. < 15 0-8 n.a. 10.1 8.7 0.1 n.a. 8.3 6.1 n.a. n.a. 10.6 20 2.3

25 37 25 25 37 20 25 25 37 25 22 25 25 25 25 25

[23] [24] [14] [19] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]

4.05 4.60 4.92 4.80 4.95 6.60 9.00 5.51 6.70 7.00 5.90

1,5 0.4 0-8 < 15 9.9 n.a. 11.1 n.a. n.a. 6.8 3.9

25 25 25 25 20 25 22 25 25 25 25

[27] [30] [19] [14] [26] [36] [15] [32] [33] [34]

Absorption in horizontal flow reactor Dynamic absorption method*

1.20 1.20 0.68 1.27

10.9 10.0 16.7 4.7

25 25 25 25

[33] [17] [37]

Dynamic stripping method

1.60

3,4

25

[33]

Method Model simulations Static headspace method

Dynamic stripping method

Dynamic uptake measurements Dynamic absorption method* DMDS

Static headspace method

Dynamic stripping method

Dynamic absorption method* 2-MP

3-MB

Dynamic stripping method

12

HEX

Dynamic absorption method*

1.43

3.5

25

Static headspace method

0.70 1.10 1.16 1.43 1.29 1.68 2.36 1.26 1.41

8.6 n.a. n.a. n.a. 7.7 n.a. n.a. 13 2.1

27 25 30 37 37 37 37 25 25

Dynamic stripping method Dynamic absorption method*

[38] [13] [39] [25] [40] [24] [29] [17]

*This work, n.a. not available It can also be seen that for DMS the average KAW determined with the static methods (0.066 at 25 °C) is 23 % lower than the average of the KAW obtained by the dynamic stripping method (0.086 at 25°C). Also for DMDS it can be seen that the average KAW of DMDS determined with the static method (0.049 at 25 °C) is 23 % lower than the average of the KAW obtained by the dynamic stripping method (0.064 at 25 °C). The RSD of all values for DMS and DMDS in Table 2 is respectively 19 and 21 %. This deviation is mainly caused by static methods. The RSD on the average of the KAW determined with static methods was 23% for DMS and 21% for DMDS. In comparison, the RSD on the average of the KAW determined with dynamic methods was respectively 0.1 and 11 % for DMS and DMDS. These lower RSDs on the KAW using the dynamic methods represent a mayor advantage over the static methods. Also this study obtained better RSD on KAW than other dynamic methods. KAW were determined at different temperatures (4, 11, 17 and 25 °C). The influence of the temperature on the KAW can be expressed by the Van ’t Hoff Equation (Equation 5)

. / = −

∆H→ ∆L→ + J J

(5)

Where ∆H→ and ∆L→ are respectively the enthalpy and entropy change of the phase

transfer from the air to the water phase, is the temperature (K) and J is the universal gas constant (8.314 J mol -1 K-1). A linear plot of the natural logarithm of the dimensionless KAW in function the inverse of the temperature was made (Figure 3). From the slope and intercept, ∆H→ and ∆L→ was calculated (Table 3). The influence (based on ∆HA
W ) of the 13

temperature on the KAW is the highest for HEX (∆H→ is equal to 71.8 kJ mol-1) and the lowest for DMS (∆H→ is equal to 36.9 kJ mol-1). Different literature data are shown in Table

3. The difference in ∆H→ is maximum 27 % lower for DMS and DMDS than this study [25, 41]. The difference for the aldehydes is higher but it should be noted that high RSDs are reported (2-MP: 20% [17] and 64% [37], HEX: 23% [17]). The RSD of ∆H→ in the literature data is also much higher than this study (maximum 6%). Table 3: Enthalpy (∆HA
W) and entropy (∆SA
W) change of the phase transfer from the air phase to the water phase for DMS, DMDS, 2-MB, 3-MB and HEX deduced from Figure 3 with Equation 5. -1

∆HA
W (kJ mol )

a

∆SA
W (J mol -1) b

c

d

DMS 36.9±0.6 30.7a 29.1b,c 30.8d 104±2

e

f

DMDS 2-MP 3-MB HEX 41.8±2.0 61.5±2.0 61.0±2.0 71.8±2.7 33.2a 37.4e 54.1f d g 37.0 43.1 62.9g 117±7

170±7

170±7

205±9

g

[27] [25] [41] [32] [37] [41] [17]

3.3 Effect of salt concentration on the KAW partitioning coefficients Sulphuric acid chemical scrubbers are frequently used for industries with NH3 sources (livestock, bio-waste valorisation, etc.). The ammonia is absorbed in the form of ammonium sulphate (AS). During scrubber operation the AS concentration increases continuously in the washing liquid. When the AS-concentration reaches a limit value, a fraction of the washing water must be drained and renewed by fresh water [42]. According to the Belgian regulation, the AS concentration cannot exceed 278 g L-1. Next to NH3, also odorous compounds are emitted, which can cause nuisance in neighbourhood. In order to evaluate the scrubbers for the odorous gas removal, it is important that the dependence of the AS concentration on the KAW value of typical odorous compounds is quantified. It should be mentioned that the odour detection threshold of ammonia is factor 4200 (2-MP) to 6800 (DMDS) higher than for the odorous compounds studied, so the target compounds represent important odour compounds [43]. The effect of the AS salt concentration (0 to 300 g L-1) on the KAW values was determined. The obtained KAW partitioning coefficients are given in Table 4. Table 4: Gas-to-liquid partitioning coefficient KAW (-) of DMS, DMDS, 2-MP, 3-MB and HEX in function of ammonium sulphate concentration in water (CS).

14

Gas-to-liquid partitioning coefficient KAW (-) -1

CS (g L ) T (°C) 100 200 300 300

25 25 25 4

n 3 3 3 3

DMS

DMDS -1

1.85±0.03 10 -1 3.91±0.10 10 7.18±0.08 10-1 -1 2.55±0.04 10

2-MP -1

1.31±0.03 10 -1 3.08±0.14 10 6.36±0.18 10-1 -1 1.83±0,1 10

3-MB -2

2.22±0.07 10 -2 4.08±0.21 10 7.38±0.52 10-2 -2 1.54±0.02 10

HEX -2

2.95±0.06 10 -2 6.14±0.06 10 1.26±0.04 10-1 -2 2.56±0.08 10

-2

2.68±0.06 10 -2 6.11±0.06 10 1.41±0.08 10-1 -2 2.7±0.09 10

Increasing the AS-concentration from 0 to 300 g L-1, increases the KAW drastically (factor 8.3 for DMS, 10.8 for DMDS, 10.0 for HEX, 5.8 for 2-MP and 8.8 for 3-MB)(Table 4, Figure A.1 and Figure A.2). The dependence of the AS concentration on the KAW can be quantified by the Setchenow Equation (Equation 6):

MNO P

,Q R = Q * ,

(6)

The KAW,0 and KAW,S represent the gas-to-liquid partitioning coefficient respectively in a pure water system and AS containing water system. The slope of a linear plot of the log (KAW,0/KAW,S) versus the salt concentration * (M) in water is equal to the Setchenow saltingout constant Q (M-1)(Figure 4). High regression coefficients were observed for all

compounds (R2 > 0.99). Q constant at 25°C is for DMS, DMDS, HEX, 2-MP, 3-MB respectively 0.41±0.03, 0.46±0.02, 0.45±0.04, 0.34±0.01 and 0.41±0.01. To the best of our knowledge, only one study has evaluated the influence of ammonium sulphate on the KAW (0.332 M-1) of DMS at 25°C [35]. The Q (M-1) values in AS for the five studied compounds are in the same order as benzene (0.366), trichloromethane (0.309), chloro-benzene (0.374), anisole (0.415) at 30 °C [44]. In comparison with e.g. NaCl, the Q of the sulphide compounds is roughly the half in comparison with AS [32]. This is mainly due to the monovalent ions of NaCl. For the aldehydes, a linear relationship between the KS and number of carbon atoms (from C4 to C6) was observed (R2 = 0.98). Since the KAW was also determined at a temperature of 4 °C and AS concentration of 300 g L-1 it was possible to calculate also the salting out constant at 4°C (Table 1). The Q at 4°C is for respectively DMS, DMDS, HEX, 2-MP and 3-MB 0.42, 0.45, 0.55, 0.39 and 0.46. For 15

DMS and DMDS the Q at 4°C and 25°C does not differ significantly (α=0.05). However, for

the aldehydes, a significant difference (α=0.05) was observed. The Q at 4°C were 22.8 % (HEX), 16.6 % (2-MP) and 12.9 % (3-MB) higher at 4°C when compared to 25°C.

3.4 Method optimization The KAW-values of the compounds were determined using a liquid volume of 20 mL, 15 mL, 10 mL and 5 mL. The determined KAW-values were not significantly different from each other which shows that the determination of the KAW by the DynAb method is independent of the liquid volume . However, the time needed to determine the KAW values (-*;  ) is dependent on i) the magnitude of the KAW , ii) the liquid volume used, iii) temperature at which the KAW should be determined. The lower the liquid volume, the less mass needs to be absorbed and the smaller the time needed to obtain equilibrium. From the experimental data, a linear relationship was found between the measuring time (-*;  ) and the product of the area between the breakthrough curves and the air flow rate (which is proportional with the absorbed mass in the liquid phase)(Q.A)(Figure 5). The lower the KAW of the compound, the larger -*;  and the higher the resulting area between the breakthrough curves (A). For compounds with high KAW -values, the equilibrium time will be shorter and the resulting area between the breakthrough curves will be lower. For compounds with a low KAW , the liquid volume can be reduced in order to avoid an excessive analysis time. Nevertheless, the minimal liquid volume is limited by the blank breakthrough curve. When the product of the area ( (min) and

the air flow rate  (mL min -1) becomes smaller than 5000 mL (when gas concentration is

16

+, normalised to 100 ( )* ) and  is in mL min-1), a higher relative standard deviation (>5%)

on the determined KAW was observed (Figure 5). Figure 6 provides a nomogram (based on Equation 4 and Figure 5) that can be used to determine the optimal liquid volume to perform the experiment (acceptable RSD and low -*;  ). The dotted line represents the lowest recommended Q.A which result into KAW with a RSD below 5%. For example, the -*;  for 2-MB (1.27 10-2) was 1 hour using a volume of 20 mL. This -*;  could be reduced to only 15 minutes by lowering the liquid volume to

5 mL (Figure 5 and Figure 6). It can also be seen for example that the highest KAW that can be reliable measured is 0.4 when a liquid volume of 20 mL is chosen (Figure 6). The DynAb method was evaluated for KAW values in the range of 10-3 to 101, but it is theoretically possible make use of this method out of this range.

4. Conclusions A new dynamic absorption method (DynAb) for measuring gas-to-liquid partition coefficients (KAW ) was developed. The method is based on the absorption of target compounds from a constant gas concentration stream in the liquid and monitoring the outlet target concentration by SIFT-MS. From the breakthrough curve of the outgoing target concentration, KAW can be calculated. An important advantage is that the method is applicable without reaching equilibrium concentrations in the gas bubbles leaving the liquid phase. In this study, DynAb was applied for measuring gas-to-liquid partitioning coefficients of odour compounds (dimethyl sulphide, dimethyl disulphide, 2-methylpropanal, 3-methylbutanal and hexanal). By using the developed method, KAW can be measured in short time (less than 15 minutes) and with high reproducibility (RSD less than 5 %). The effect of temperature (4 to 25 °C) and ammonium sulphate (AS) concentration (0 to 300 g L-1) was investigated to get more insight in the scrubber performance. When the temperature increases from 4 to 25°C, the KAW increases with a factor 5. Raising the AS concentration (0 to 300 g L-1), increases this factor to 10. This implies that a lower removal efficiency will be obtained in scrubbers, 17

due to a lower mass transfer. The determination of KAW values of odorous compounds in this study is important in scrubber reactor design for livestock and bio-waste valorisation applications.

5. Acknowledgement This research has been financially supported by the agency for Innovation and Technology (IWT, project number B-13856).

6. References [1] W. Lindinger, A. Hansel, Analysis of trace gases at ppb levels by proton transfer reaction mass spectrometry (PTR-MS), Plasma Sources Sci. Technol., 6 (1997) 111-117. [2] T. Karl, P. Prazeller, D. Mayr, A. Jordan, J. Rieder, R. Fall, W. Lindinger, Human breath isoprene and its relation to blood cholesterol levels: new measurements and modeling, J. Appl. Physiol., 91 (2001) 762-770. [3] T. Esper, M. Wehner, C.D. Meinecke, H. Rueffert, Blood/Gas Partition Coefficients for Isoflurane, Sevoflurane, and Desflurane in a Clinically Relevant Patient Population, Anesthesia and Analgesia, 120 (2015) 45-50. [4] M. Marin, I. Baek, A.J. Taylor, Volatile release from aqueous solutions under dynamic headspace dilution conditions, J. Agric. Food Chem., 47 (1999) 4750-4755. [5] M. Martuscelli, G. Savary, P. Pittia, N. Cayot, Vapour partition of aroma compounds in strawberry flavoured custard cream and effect of fat content, Food Chemistry, 108 (2008) 1200-1207. [6] L.K. Mao, Y.H. Roos, S. Miao, Flavour Release from Monoglyceride Structured Oil-inWater Emulsions through Static Headspace Analysis, Food Biophysics, 9 (2014) 359-367. [7] N.J.R. Kraakman, J. Rocha-Rios, M.C.M. van Loosdrecht, Review of mass transfer aspects for biological gas treatment, Appl. Microbiol. Biotechnol., 91 (2011) 873-886. [8] J. Staudinger, P.V. Roberts, A critical review of Henry's law constants for environmental applications, Critical Reviews in Environmental Science and Technology, 26 (1996) 205-297. [9] F. Etzweiler, E. Senn, H.W.H. Schmidt, Method for measuring aqueous solubilites of organic compounds, Analytical Chemistry, 67 (1995) 655-658. [10] J. Parshintsev, C.K. Lai, K. Hartonen, M. Kulmala, M.L. Riekkola, A new approach to determine vapor pressures of compounds in multicomponent systems by comprehensive two-dimensional gas chromatography coupled to time-of-flight mass spectrometry, Talanta, 124 (2014) 21-26. [11] N.N. Nirmalakhandan, R.E. Speece, QSAR model for predicting Henry's constant, Environmental Science & Technology, 22 (1988) 1349-1357. [12] J. Dewulf, D. Drijvers, H. Vanlangenhove, Measurement of Henry's law constant as function of temperature and salinity for the low-temperature range, Atmospheric Environment, 29 (1995) 323-331.

18

[13] C. Jouquand, V. Ducruet, P. Giampaoli, Partition coefficients of aroma compounds in polysaccharide solutions by the phase ratio variation method, Food Chemistry, 85 (2004) 467-474. [14] G. Darracq, A. Couvert, C. Couriol, A. Amrane, D. Thomas, E. Dumont, Y. Andres, P. Le Cloirec, Silicone oil: An effective absorbent for the removal of hydrophobic volatile organic compounds, Journal of Chemical Technology and Biotechnology, 85 (2010) 309-313. [15] A. Godayol, M. Alonso, J.M. Sanchez, E. Antico, Odour-causing compounds in air samples: Gas-liquid partition coefficients and determination using solid-phase microextraction and GC with mass spectrometric detection, Journal of Separation Science, 36 (2013) 1045-1053. [16] D. Mackay, W.Y. Shiu, R.P. Sutherland, Determination of air-water Henry's law constants for hydrophobic pollutants, Environmental Science & Technology, 13 (1979) 333337. [17] T. Karl, C. Yeretzian, A. Jordan, W. Lindinger, Dynamic measurements of partition coefficients using proton-transfer-reaction mass spectrometry (PTR-MS), International Journal of Mass Spectrometry, 223 (2003) 383-395. [18] P.M. Heynderickx, S. De Clercq, P. Saveyn, J. Dewulf, H. Van Langenhove, Determination of the sorption and desorption kinetics of perfume raw materials in the liquid phase with vesicular dispersion: Application of SIFT-MS, Chemical Engineering Journal, 217 (2013) 281-288. [19] E. Dumont, G. Darracq, A. Couvert, C. Couriol, A. Amrane, D. Thomas, Y. Andres, P. Le Cloirec, Determination of partition coefficients of three volatile organic compounds (dimethylsulphide, dimethyldisulphide and toluene) in water/silicone oil mixtures, Chemical Engineering Journal, 162 (2010) 927-934. [20] V. Dohnal, S. Hovorka, Exponential saturator: A novel gas-liquid partitioning technique for measurement of large limiting activity coefficients, Ind. Eng. Chem. Res., 38 (1999) 20362043. [21] C. Walgraeve, J. Bruneel, K. Van Huffel, K. Demeestere, L. Vincze, B. De Meulenaer, H. Van Langenhove, Sorption behaviour of targeted volatile organic compounds on airborne particulate matter using selected ion flow tube mass spectrometry, Biosyst. Eng., 131 (2015) 84-94. [22] D. Smith, P. Spanel, Selected ion flow tube mass spectrometry (SIFT-MS) for on-line trace gas analysis, Mass Spectrometry Reviews, 24 (2005) 661-700. [23] V.P. Aneja, J.H. Overton, The emission rate of dimethyl sulfide at the atmosphericoceanic interface, Chem. Eng. Commun., 98 (1990) 199-209. [24] S.M. van Ruth, E. Villeneuve, Influence of beta-lactoglobulin, pH and presence of other aroma compounds on the air/liquid partition coefficients of 20 aroma compounds varying in functional group and chain length, Food Chemistry, 79 (2002) 157-164. [25] S.M. van Ruth, G. de Vries, M. Geary, P. Giannouli, Influence of composition and structure of oil-in-water emulsions on retention of aroma compounds, Journal of the Science of Food and Agriculture, 82 (2002) 1028-1035. [26] A.G. Vitenberg, B.V. Ioffe, Z.S. Dimitrova, I.L. Butaeva, Determination of gas-liquid partition-coefficient by means of gas-chromatographic analysis, Journal of Chromatography, 112 (1975) 319-327. [27] A. Przyjazny, W. Janicki, W. Chrzanowski, R. Staszewski, Headspace gaschromatographic determination of distribution coefficient of selected organosulfur compounds and their dependence on some parameters, Journal of Chromatography, 280 (1983) 249260. [28] M.C. Iliuta, F. Larachi, Gas-liquid partition coefficients and Henry's Law constants of DMS in aqueous solutions of Fe(II) chelate complexes using the static headspace method, Journal of Chemical and Engineering Data, 50 (2005) 1700-1705. [29] S.M. van Ruth, I. Grossmann, M. Geary, C.M. Delahunty, Interactions between artificial saliva and 20 aroma compounds in water and oil model systems, Journal of Agricultural and Food Chemistry, 49 (2001) 2409-2413.

19

[30] W.W. Du, W. Parker, Liquid-gas partitioning of selected volatile organic sulfur compounds in anaerobically digested sludges, Water Science and Technology, 66 (2012) 573-579. [31] A.W. Lantz, S.M. Wetterer, D.W. Armstrong, Use of the three-phase model and headspace analysis for the facile determination of all partition/association constants for highly volatile solute-cyclodextrin-water systems, Analytical and Bioanalytical Chemistry, 383 (2005) 160-166. [32] D. Liu, A. Feilberg, A.M. Nielsen, A.P.S. Adamsen, PTR-MS measurement of partition coefficients of reduced volatile sulfur compounds in liquids from biotrickling filters, Chemosphere, 90 (2013) 1396-1403. [33] P. Pollien, A. Jordan, W. Lindinger, C. Yeretzian, Liquid-air partitioning of volatile compounds in coffee: dynamic measurements using proton-transfer-reaction mass spectrometry, International Journal of Mass Spectrometry, 228 (2003) 69-80. [34] E. Schuhfried, F. Biasioli, E. Aprea, L. Cappellin, C. Soukoulis, A. Ferrigno, T.D. Mark, F. Gasperi, PTR-MS measurements and analysis of models for the calculation of Henry's law constants of monosulfides and disulfides, Chemosphere, 83 (2011) 311-317. [35] W.J. Debruyn, E. Swartz, J.H. Hu, J.A. Shorter, P. Davidovits, D.R. Worsnop, M.S. Zahniser, C.E. Kolb, Henrys law solubilities and setchenow coefficient for biogenic reduced sulfur species obtained from gas-liquid uptake measurements, Journal of Geophysical Research-Atmospheres, 100 (1995) 7245-7251. [36] L. Arellano-Garcia, A. Gonzalez-Sanchez, H. Van Langenhove, A. Kumar, S. Revah, Removal of odorant dimethyl disulfide under alkaline and neutral conditions in biotrickling filters, Water Science and Technology, 66 (2012) 1641-1646. [37] R.S. Strekowski, C. George, Measurement of henry's law constants for acetone, 2butanone, 2,3-butanedione, and isobutyraldehyde using a horizontal flow reactor, Journal of Chemical and Engineering Data, 50 (2005) 804-810. [38] D.M. Jung, S.E. Ebeler, Headspace solid-phase microextraction method for the study of the volatility of selected flavor compounds, Journal of Agricultural and Food Chemistry, 51 (2003) 200-205. [39] G. Savary, E. Guichard, J.L. Doublier, N. Cayot, Mixture of aroma compounds: Determination of partition coefficients in complex semi-solid matrices, Food Research International, 39 (2006) 372-379. [40] D.F. Nahon, M. Harrison, J.P. Roozen, Modeling flavor release from aqueous sucrose solutions, using mass transfer and partition coefficients, Journal of Agricultural and Food Chemistry, 48 (2000) 1278-1284. [41] X.L. Zhou, K. Mopper, Apparent partition-coefficient of 15 carbonyl-compounds between air and seawater and between air and fresh-water - implications for air see exchange, Environmental Science & Technology, 24 (1990) 1864-1869. [42] L.J.S. Hadlocon, R.B. Manuzon, L.Y. Zhao, Development and evaluation of a full-scale spray scrubber for ammonia recovery and production of nitrogen fertilizer at poultry facilities, Environmental Technology, 36 (2015) 405-416. [43] Y. Nagata, Measurement of Odor Threshold by Triangle Odor Bag Method, in: M. Review (Ed.), Government of Japan, Japan Ministry of the Environment, 2003, pp. 118-127. [44] M. Gorgenyi, J. Dewulf, H. Van Langenhove, K. Heberger, Aqueous salting-out effect of inorganic cations and anions on non-electrolytes, Chemosphere, 65 (2006) 802-810.

20

Figure 1: Schematic overview of the experimental set-up of the DynAb method.  Figure 2: Breakthrough curve. Normalised gas concentration (C = CSTU ⁄CVW 10#) in function of time (min). The dotted area (A) is proportional to the absorbed mass of the compound in the liquid phase.

Figure 3: Logarithmic plot of KAW (-)(y) in function of the reciprocal of the temperature 10 3 T-1 (K-1)(x) for DMS ( ), DMDS ( ), HEX ( ), 2-MP ( ) and 3-MB ( )(n = 1). According to Equation 4, the following linear regression was obtained for DMS: y = -4,4425x + 12,459 (R2 0.999); DMDS: y = -5,0318x + 14,087(R2 0.999); HEX: y = -8,637x + 24,715 (R2 0.999); 3-MB: y = -7,3328x + 20,394 (R2 0.999) and 2-MP: y = -7,4023x + 20,49 (R2 0.999). Figure 4: Logarithmic plot of the ratio (log(r)) of the gas-to-liquid partitioning coefficient of a certain ammonium sulphate concentration (KAW,S) to the gas-to-liquid partitioning coefficient in pure water (KAW,0) in function of the concentration ammonium sulphate (CS)(M) for DMS ( ) and DMDS ( ), HEX ( ), 2-MP ( ) and 3-MB ( )(n = 3). The linear fitting according to Equation 5 for DMS, DMDS, 2-MP, 3-MB and HEX are respectively: log(r)=0.42 Cs (R2 0.997), log(r)=0.46 Cs (R2 0.999), log(r)=0.33 Cs (R2 0.999), log(r)=0.41 Cs (R2 0.999) and log(r)=0.43 Cs (R2 0.995). Figure 5: A) Measuring time (tmeasure)(min) in function of the area between the breakthrough curves times air flow rate (Q.A 10-3)(mL) for V is equal to 20 mL and air flow rate 95 ± 0.5 mL min-1 for DMS and DMDS. B) Relative standard deviation on KAW based on n = 3 (RSD)(%) in function of the product of the area (A) between blank and sample breakthrough curve and the air flow rate (Q.A 10-3)(mL) for 20 mL of liquid and air flow rate 95 ± 0.5 mL min-1 for DMDS. Figure 6: Logarithm of the calculated area A (min) between breakthrough curve of blank and liquid sample times flow rate Q (mL min-1) (Log(Q.A))(log(mL)) in function of gas-to-liquid partitioning coefficient (KAW)(-) for (left) lower coefficients and (right) higher coefficients. The areas are based on normalised concentrations to 100 (-). The dotted line represents the minimum Q.A (log(5000)=3.7 log(mL)) that should be used to calculate KAW with a RSD lower than 5 %. Figure A.1: Gas-to-liquid partitioning coefficient (KAW )(-) in function of ammonium sulphate concentration (CS)(g L-1) for DMS ( ) and DMDS ( ) using the DynAb method. Figure A.2: Gas-to-liquid partitioning coefficient (KAW )(-) in function of ammonium sulphate concentration (CS)(g L-1) for HEX ( ), 2-MP ( ) and 3-MB ( ) using the DynAb method.

21

Figure 1

Thermostatic cabinet N2-reservoir Q1

MFC 1

Q2

MFC 2

Capillary sytem

MFC 3

Q4a

Waste

Pressure Q4b

Four-way valve

Q3

MFC 4

Q5 Waste Heater

Cooler

Waste

Bubble column

Liquid reservoir

SIFT-MS

Figure 2

Normalised Gas Concentration (-)

110

100 90 80 70

A

60 50 40 30 20 10 0 0

5

10

15

Time (min)

20

25

30

Figure 3

-2 -2.5 -3

Ln (KAW) (-)

-3.5 -4

-4.5 -5 -5.5 -6 -6.5 -7 3.30

3.35

3.40

3.45

3.50

103 T-1 (K-1)

3.55

3.60

3.65

Figure 4

1.2

Log (KAW,S/KAW,0) (-)

1

0.8

0.6

0.4

0.2

0

0

0.5

1

1.5

CS (M)

2

2.5

Figure 5

A)

B) 25

70

tmeasure = 0,3563 Q.A+ 3,6529 R² = 0,999

60

20

tmeasure (min)

50

RSD (%)

40 30

15

10

20 5 10 0

0 0

50

100

Q.A

150

10-3 (mL)

200

0

10

20

Q.A

10-3

30

(mL)

40

Figure 6

Log (Q.A) (log(mL))

5.5 5 4.5 4

5.5 5 4.5 4

3.5

3.5

3

3 0

0.01 0.02 0.03 0.04 0.05 0.06

KAW (-)

V = 20 mL V = 50 mL V = 100 mL V = 150 mL V = 200 mL

6

Log (Q.A) (log(mL))

V =1 mL V = 5 mL V = 10 mL V = 15 mL V = 20 mL

6

0

0.2

0.4

0.6

KAW (-)

0.8

1

100

COUT

80

COUT / CIN 100 (-)

KAW 60

40

CIN

20

VOC air

0 0

5

10

15

Time (min)

20

25

30

A new method to measure gas-to-liquid partitioning coefficients (KAW) was developed

The DynAb method does not assume equilibrium between the gas bubble and liquid

KAW is determined in less than 15 min with high reproducibility (RSD < 5%)

KAW values for five odorous sulphides and aldehydes were determined

Effect of temperature and ammonium sulphate concentration on KAW was investigated

22