You are on page 1of 10

Journal of Chromatography A 1643 (2021) 462034

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Automated method for volatile fatty acids determination in anaerobic


processes using in-syringe magnetic stirring assisted dispersive
liquid-liquid microextraction and gas chromatography with flame
ionization detector
M.A. Vargas-Muñoz a, Víctor Cerdà a,b, L.S. Cadavid-Rodríguez c, Edwin Palacio a,∗
a
Group of Environmental Analytical Chemistry, University of the Balearic Islands, Cra.Valldemossa km 7.5, 07122 Palma de Mallorca, Spain
b
Sciware Systems, S.L., C/Pi 37, 07193 Bunyola, Spain
c
Department of Engineering, Faculty of Engineering and Administration, Universidad Nacional de Colombia - Sede Palmira, Cra. 32 No 12-00, Palmira,
Colombia

a r t i c l e i n f o a b s t r a c t

Article history: Volatile fatty acids (VFAs) are key parameters to monitor anaerobic digestion processes. Thus, a fast, sim-
Received 12 December 2020 ple and precise determination of these analytes is necessary for a timely characterization of the biological
Revised 19 February 2021
processes present in municipal solid waste and wastewater treatment plants. In this work, an automated
Accepted 28 February 2021
method for the extraction and preconcentration of VFAs, based on dispersive liquid-liquid microextraction
Available online 10 March 2021
with magnetic stirring in syringe, and gas chromatography with flame ionization detector for the sepa-
Keywords: ration and detection, is described. The effect of parameters such as the type and volume of extraction
In-syringe magnetic-stirring-assisted solvent, pH, salting out effect and stirring time, was studied using a multivariate and univariate experi-
dispersive liquid-liquid microextraction (In mental design. Extraction and preconcentration were performed simultaneously using tert-butyl methyl
syringe-MSA-DLLME) ether (TBME) as the extraction solvent, after stirring 100 s at a constant rate. The detection limits were in
Gas chromatography the range of 0.1 - 1.3 mg L−1 and a good linearity was observed up to 10 0 0 mg L−1 of the studied VFAs,
Volatile fatty acids
with a range of R2 between 0.9997 and 0.9999. The intra and interday precision expressed as relative
Anaerobic treatment, automation
standard deviation (n= 5) varied between 0.7 and 2.4% and between 1.7 and 7.0%, respectively. Subse-
quently, the developed method was successfully applied to evaluate the presence of VFAs in wastewater
samples from anaerobic treatments and an average relative recovery of 102% was obtained.
© 2021 Elsevier B.V. All rights reserved.

1. Introduction tors present in wastewater treatment plants (WWTP), solid waste


treatment plants (SWTP) and in landfill leachates [6], because large
Volatile fatty acids are aliphatic carboxylic acids with low amounts of VFAs (>10 0 0 - 40 0 0 mg L−1 ) can inhibit the AD pro-
molecular weight and short chain lengths (C2-C7) [1]. They are in- cess [7], affecting the stability of methane production and the re-
termediate metabolites originated by microbial conversion of or- moval of nutrients (nitrogen and phosphorus compounds) from
ganic matter into biogas during anaerobic digestion (AD) and are wastewater [8]. Otherwise, a high concentration of VFAs is bene-
mainly formed from the fermentation of macromolecules (carbo- ficial to generate value-added products with different uses in the
hydrates, proteins and fats) in acidogenesis and acetogenesis steps chemical industry [9,10].
[2]. The main components of VFAs present in anaerobic digesters VFAs should be monitored at different stages of wastewater
are acetic, propionic, butyric, valeric acid and the isomeric forms treatment, as well as in the untreated wastewater, treated wastew-
of these last two [3,4]. ater and activated sludges [11]. There are traditional procedures
AD is a process used as a biological treatment of wastewater, based on titration, which only determine the total concentration
sewage sludge and organic solid waste with the aim of reducing of volatile fatty acids (TVFAs). However, the determination of in-
environmental pollution and producing renewable energy in the dividual VFAs gives more specific information about the biolog-
form of biogas [5]. VFAs are a key indicator in anaerobic reac- ical processes; thus, their separation has become more impor-
tant over time. Individual VFAs can be separated with analytical
techniques such as gas chromatography (GC), high performance

Corresponding author. liquid chromatography (HPLC), ion chromatography and capillary
E-mail address: edwin.palacio@uib.es (E. Palacio). electrophoresis. Of these techniques, GC with flame ionization de-

https://doi.org/10.1016/j.chroma.2021.462034
0021-9673/© 2021 Elsevier B.V. All rights reserved.
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

tection (GC-FID) is the most used for the VFAs determination in tant volume. Finally, a comparison of the proposed method is made
wastewater because it provides a high resolution, wide linearity with other conventional methods, focusing mainly on the distilla-
and low limits of detection [12]. tion method for VFAs [33].
Recently, the beginning of using polar columns (FFAP) in GC
motivated the possibility of performing a direct aqueous injection 2. Materials and methods
(DAI) of the samples to avoid the loss of analyte during enrich-
ment step [7,13]. However, numerous studies states that DAI can 2.1. Reagents
be detrimental to the analytical performance of the GC, observ-
ing wide peaks, column contamination and detector damage. For Individual C2-C5 VFAs standards (≥99%) and 2–ethylbutyric
this reason, sample preparation is essential to protect the instru- acid (≥99%) used as internal standard (IS) were obtained from
ment [12,14] using treatment techniques such as liquid-liquid ex- Sigma-Aldrich (Madrid, Spain). Certified VFAs mix at a concen-
traction (LLE) [11,15], solid phase microextraction (SPME) [16,17] tration of 10 mM was obtained from Supelco (USA). The extrac-
and headspace (HS) techniques [18]. Among these treatment tech- tant solvents tert-butyl methyl ether – TBME (HPLC grade) (≥99%)
niques, LLE seems to be the most favorable option because it is ac- and dichloromethane were also obtained from Scharlau (Barcelona,
cepted as a practical, fast and suitable technique for a large num- Spain), as well as phosphoric acid (85% w/w). The dimethyl car-
ber of samples. LLE allows analyte preconcentration and sample bonate extractant was purchased from Sigma Aldrich. Sodium chlo-
matrix elimination. However, LLE requires extended sample prepa- ride was obtained from Scharlau. The distilled water came from the
ration times, large volumes of solvents and therefore, a costly dis- MiliQ Direct-8 purification system (Millipore Ibérica S.A.U., Madrid,
posal thereof [19]. Spain).
In this context, unconventional LLE methodologies have
emerged, gathered in the techniques of liquid phase microextrac- 2.2. Standards
tion (LPME); miniaturized LLE procedure consisting of the ex-
traction of objective compounds from an aqueous matrix using From the individual high purity VFAs standards (≥99%) a stock
reduced volumes of an organic immiscible solvent [20]. LPME solution was made at 2500 mg L−1 in distilled water. The stan-
techniques include the dispersive liquid-liquid microextraction dards for direct injection were prepared by dilution in TBME.
(DLLME), a technique based on the use of a dispersive solvent, Working solutions (5 – 10 0 0 mg L−1 ) were prepared from the stock
which has the function of dispersing the organic phase in very fine solution and 2-ethylbutyric acid was used as an internal standard
droplets in the aqueous phase, increasing the contact surface be- (IS) with a final concentration of 250 mg L−1 . Sodium chloride
tween the two phases and, therefore, improving the extraction effi- and phosphoric acid were used to adjust salt content and pH of
ciency [21]. Nonetheless, the use of dispersive solvents can change aqueous solutions in optimization studies, respectively. The inter-
the partition coefficient and decrease of the mass transfer of the nal standard chosen and the concentration of acidifying solution
analytes to the extraction solvent, reducing the enrichment effi- was prepared according to recommendation of Raposo et al, 2015
ciency [22]. To overcome this problem, alternative methods to dis- [1]. Calibration was done by using relative peak area (RPA=VFA
perse the extraction solvent have been implemented to replace the peak area / IS peak area).
dispersion solvent using different methods of mechanical agitation:
ultrasonic-assisted [23], air-assisted [24], vortex-assisted [25] and 2.3. Samples collection
magnetic stirring-assisted [26]. Among these, magnetic stirring-
assisted dispersive liquid-liquid microextraction (MSA-DLLME) is Liquid samples were obtained from wastewater from the inlet
based on the dispersion of an organic solvent in the aqueous sam- and outlet of the WWTP (Palma de Mallorca, Spain) and from an
ple using magnetic stirring to form a mild emulsification [26]. anaerobic digester treating sludge from the primary and secondary
Nevertheless, the manual methodology of a DLLME can become treatment of wastewater. Further, samples from three laboratory-
slow, laborious and imprecise. In this sense, flow analysis tech- scale fermenter reactors (FR) (located at the Universidad Nacional
niques and automation are a viable option, because they provide de Colombia – Sede Palmira) were analyzed. The reactors were
high performance, low analyst intervention, lower solvent con- designed to ferment 50 g SV L−1 of fishing waste (FW) at three
sumption, and therefore a low hazardous waste generation for en- pH values (7, 9 and 11) with inoculum-substrate ratio of 0.5. The
vironment [27]. The use of automated chromatographic systems in FW were the viscera of different species of fish (mainly snappers,
large-scale AD plants [13,28,29] and in laboratory [30,31] has al- corvines and tunas) obtained from artisanal fishing in Tumaco,
ready been reported. Thus, MSA-DLLME can be automated using Colombia. The samples were taken on days 1, 3, 5, 7, 10 and 15
the syringe laboratory technique, allowing improved reproducibil- of AD. The samples from the WWTP and FR were stored in the
ity and precision, giving place to In-syringe-MSA-DLLME [32]. dark at -20°C in 1 L and 30 mL polypropylene bottles, respectively.
In the present work, the automation of a dispersive liquid- Subsequently, they were acidified with phosphoric acid (85% w/w)
liquid microextraction method with magnetic stirring in syringe to ensure the predominance of undissociated forms of VFA species.
(In-syringe-MSA-DLLME) was developed for the VFAs determina- Once in the laboratory they were passed through a 0.45 μm filter
tion in an anaerobic digester of the WWTP from Palma of Mallorca to remove solid particles.
(Spain) and in laboratory scale biodigesters for fishing waste val-
orisation. The configuration of the analytical system is based on 2.4. In-syringe-MSA-DLLME-GC-FID system
the multi-syringe flow injection analysis (MSFIA) technique, but a
single syringe is used as the extraction chamber. A magnetic stir The main element for the treatment system (Fig. 1) is a multi-
bar is placed inside the syringe and its rotation is controlled with syringe burette module (Crison, Alella, Spain) coupled to a multi-
a motor that is activated by a computer. In this way, the agitation position valve (MPV). Multisyringe module (MSM) can be equipped
disperses the extraction solvent within the aqueous sample. Then, with four syringes that are moved simultaneously in two directions
the organic phase is separated from the aqueous phase, and the to aspirate or dispense liquids. In the proposed system it was only
organic drop enriched with the analytes is transported through the necessary to use a 5 mL glass syringe (S) (Hamilton, Switzerland),
manifold and stored until its analysis by GC-FID. Several extraction which also acts as a mixing chamber. The syringe has a three-way
solvents with different densities have been studied, in addition to solenoid valve on its head, performs the multicommutation opera-
the optimal conditions of pH, salinity, extraction time and extrac- tion.

2
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

Fig. 1. Automatic In-syringe-MSA-DLLME-GC-FID system for VFAs extraction and preconcentration. V, three-way solenoid valve of syringe; S, syringe; MSM, multisyringe
module; MPV, multiposition switching valve; SV, solenoid valve; AS, autosampler, M, motor.

The on position of the syringe is connected with the central Nitrogen (99.99%) was used as carrier gas at 1 mL min−1 . The oven
port of the MPV and the off position joins to an additional exter- temperature was planned to increase from 100°C (2 min) to 150°C
nal solenoid valve (SV). On the back, the multisyringe module has a (1 min) to 10°C min−1 and then rose to 200°C (1 min) to 20°C
panel with four ports which allows controlling and connecting the min−1 , for a time 11.5 min total analysis. The injection volume of
motor and the SV (Fig. S1A). So, these devices are first connected the sample was 1 μL and was performed in 50:1 split mode at
to an electronic circuit to prevent overheating (Fig. S1B). The mul- 250°C. The temperature of the detector was maintained at 250°C.
tiposition valve (MPV) (Sciware Systems, Bunyola, Spain) has eight
peripheral ports, as follows: extractant solvent container (6), sam-
2.6. In-syringe-MSA-DLLME-GC-FID procedure
ple (5) and air (4). The sample tube connection extends to an au-
tosampler system, which is equipped with chromatographic vials
The detailed operation protocol of the proposed method is
containing 300 μL inserts. The MSA consists of an external stirring
showed in supporting information (Table S1). Prior analysis, salt
device (Sciware Systems, Bunyola, Spain) that rotates a small stir
content and pH of samples were adjusted to 0.5 M using NaCl
magnetic bar (10 mm long and 3 mm in diameter) inside the sy-
and pH 1.0, respectively. The cleaning tubes, sample tubes, waste
ringe; a motor that causes the rotational force of the external stir-
tubes and chromatography vials were strategically located in the
ring support, connected with an elastic band (Fig. S1C); and a cir-
autosampler, thus the liquid that enters through a specific sample
cuit, which controls the motor speed through one of the rear out-
or cleaning tube goes out to 13 positions later in a vial or in the
puts of the multisyringe module (Fig. S1D). The MSA is described
waste tubes, respectively (Fig. S1E). The speed used in the motor
in detail in the authors work [26].
was always constant.
Both modules were connected to a computer, and instrumen-
tal control is done through AutoAnalysis 5.0 software (Sciware Sys-
tems). All manifold networks used in the systems have been man- 2.7. Calculations
ufactured with 0.8 mm internal diameter PTFE tube, except that
used for solvent inlet in the selection valve (1.5 mm). The screws Enrichment factor and extraction recovery (ER) were studied in
and connectors are made of PEEK, so that they are resistant to or- order to evaluate the extraction efficiency. The enrichment factor
ganic solvents. (EF) was calculated as the ratio of Corg /C0 , where Corg is the con-
centration of analytes in the organic phase after DLLME and C0 the
original concentration in the aqueous phase. The ER was defined
2.5. GC-FID instrumentation as the percentage of the total analyte which was extracted in the
C x Vorg
organic drop after DLLME (ER = org C x Vaq x 100 ) . Relative recovery
The analysis was performed on a YL-Clarity YL6500 gas chro- 0
(RR) was calculated as the percentage of the found concentration
matograph, equipped with an automatic sample injector and a
after spiking, minus real concentration in the samples before spik-
flame ionization detector (FID) of the same brand. The analytes
ing, relative to the added (spiked) concentration in the samples
were separated on a polar capillary column of fused silica with Cfound− C
polyethyleneglycol (DB-FFAP, Agilent, 30 m x 0.25 m id x 0.25 μm). (RR = Cspiked
real
x 100).

3
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

Fig. 2. (A) Results of the study of extractants with error bars. Results are expressed as relative intensity (%) giving the 100% to the solvent showing the best performance
and relating the rest to it. DMC: dimethyl carbonate, TBME: tert-butyl methyl ether, DCM: dichloromethane. (B) Standardized main Pareto chart from the fractional factorial
screening design for the calculated desirability response signal. Vertical line in the chart defines 95% of confidence level. (C) Effect of stirring time on the response of the
analytical signal (area) of the TVFAs. (D) Results of the study of salting out effect with error bars. Results are expressed as relative intensity (%) giving the 100% to the salt
concentration showing the best performance and relating the rest to it.

Linearity, sensitivity, quantification, precision and accuracy were was found that "three washes" were necessary to perform a good
calculated according to the IUPAC [34]. LOD and LOQ were calcu- cleaning of the syringe (not shown in this work). There was no sig-
lated, as three times the standard deviation of ten blanks divided nificant minimum difference between three and four washes (p=
by the slope of each calibration curve, and ten times the standard -0.033). Therefore, the procedure was worked with three washes
deviation of ten blanks divided by the slope of each calibration and took an approximate time of 4 min.
curve, respectively. Precision and accuracy was evaluated through
recovery studies [35].
3.1.2. Selection of extractant solvent
A critical step in the development of an extraction method
2.8. Statistical analysis is the selection of an appropriate solvent to extract the ana-
lytes from the sample. The optimization of the extraction condi-
Design of experiments and results evaluation were carried out tions was studied for acetic, propionic, isobutyric, butyric, isova-
using the statistical software package “STATISTICA 6.0” (Statsoft leric and valeric acids. A good solvent should have a high analyte
Inc., Tulsa, USA). All calibration solutions and samples were an- extraction capacity and a good chromatographic behavior. Solvent
alyzed in triplicate. The comparison of the proposed automated selection was carried out using tert-butyl methyl ether (TBME),
method with the 5560C distillation method was carried out with dichloromethane (DCM) and dimethyl carbonate (DMC), to evalu-
XLSTAT 2019.1 (Addinsoft SARL., Paris, France) statistical program. ate the VFAs extraction efficiency. These solvents were chosen be-
cause they are the most used and named in recent studies to ex-
3. Results and discussion tract VFAs [7,11,15]. It should be noted that TBME is a less dense
solvent than water, unlike DCM and DMC. The use of alcohols was
3.1. In-syringe-MSA-DLLME avoided due to their high miscibility in water. For this test, a 3
mL aqueous sample enriched with 500 mg L−1 was extracted with
3.1.1. System cleaning 500 μL of the solvent, at pH 1.0 and without salt. To achieve a good
The rinsing of system is necessary to minimize carry-over of mix and then a total phase separation, the stirring time and sepa-
analytes between extractions. Adsorption of VFAs can occur mainly ration time were set at 120 and 20 s, respectively. The experiment
in the port number 5 of the selection valve, inside the syringe was performed in triplicate. Results showed that DMC achieved the
and in the PTFE tubing. Two, three and four washes were tested highest extraction efficiency for all studied analytes with respect
(n= 2) in the syringe after treating the most concentrated stan- to other solvents (Fig. 2A). However, this solvent is slightly misci-
dard (10 0 0 mg L−1 ). "Two washes" are composed of a cleaning ble in water and it was not chosen because it did not provide a
performed with 2 mL of acetone (10% v/v) and then, with 2 mL good phase separation after stirring. Therefore, TBME was selected
of water. "Three washes" are composed of a cleaning with acetone as the optimal extraction solvent. TBME generally has a greater re-
and then twice with water, in the same volumes. "Four washes" in- covery for all analytes compared to DCM, due to its higher polar-
clude two cleanings with water and two cleanings with acetone. It ity [15]. Dichloromethane has a lower affinity with the more po-

4
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

lar analytes (acetic and propionic acids); important acids to de- lower volumes of TBME and more acidic pH improves the extrac-
termine the biodigestion efficiency [36]. Moreover, a solvent less tion. On the other hand, salt concentration and stirring time has
dense than water such as TBME is a more environmentally friendly a significant positive effect, so adding salt and stirring longer in-
solvent compared to DCM [30]. creases the extraction. The interaction between the extractant vol-
ume and the stirring time was the most significant factor showing
a negative effect.
3.1.3. Sample volume selection
As can be seen in the desirability profile of the preliminary
The effect of sample volume on extraction was also investi-
study (Fig. S2A), the optimum values within the studied range
gated. As expected, when the volume of the enriched aqueous so-
were: 200 μL, an extraction time of 120 s, a pH of 1.0 and a salt
lution was increased from 1 to 3 mL, the analytical signal increased
concentration of 1.0 M. However, variables with significant effect
linearly due to the greater availability of the analyte mass for ex-
on a central composite design (CCD) were evaluated to find the
traction. An average increase of 2.2 times was obtained for all an-
optimum values (Fig. S2B). Based on the above, it was decided to
alytes. Volumes greater than 3 mL were not analyzed because it
study again the volume of the extractant between 150 and 300
was necessary to save space for the volume of the extraction sol-
μL, and the extraction time between 60 and 180 s in a two-factor
vent (studied from 0.2 to 0.5 mL) and to aspirate the minimum
CCD with three central points (n= 3) showing that the optimum
volume of air (1.2 mL) necessary to introduce the sample and the
values are a minimum volume for the extractant (120 μL) and a
solvent into the 5 mL syringe. Therefore, 3 mL was chosen for the
maximum time for agitation (205 s). However, despite these re-
following experiments in view of high sensitivity.
sults, there were many problems with the recovery in the vial of
such small volumes of organic extract and reproducibility problems
3.1.4. Multivariate and univariate optimization of the were presented when trying to extract with volumes smaller than
in-syringe-MSA-DLLME 200 μL. When very low volumes of extractant are used, there is
Once the extraction solvent was selected, the optimization pa- a risk of dilution of the organic drop along the pipe, resulting in
rameters affecting the DLLME was developed using the protocol the observation of aqueous phase microdroplets in the VFAs ex-
detailed in Table S1. The objective of optimizing the analytical tract, which may change the recovered concentration and affect
method is to find the best system conditions and is measured in the performance of the chromatograph column. Therefore, taking
general, in terms of sensitivity of the analytical signal. This can be into account the small difference between the ranges of the vol-
done as a univariate analysis, studying a single variable while the umes studied, it was decided to take 200 μL as the most appro-
other factors remain constant, or as a multivariate analysis, design- priate volume.
ing experiments where all the variables and their interactions are On the other hand, stirring times up to 10 min were tested in
studied at the same time. In the present study a combination of a univariate analysis and 100 s was chosen as the most suitable
the two methodologies was used to show greater precision of the value for extraction (Fig. 2C). Although pH and saline concentra-
results. Statistical software Statistical 6.0 (Statsoft, Tulsa, USA) was tion were significant factors in the Pareto chart, it was not decided
used for all multivariate optimization. to evaluate these parameters in the CCD and they were set as 1.0
First, a preliminary study was carried out to find the variables and 0.5 M, respectively. The pH of the sample had a negative effect,
with a significant effect on the response, through a fractional factor therefore, a pH 1.0 is safe to have the VFA in its molecular form,
evaluation (2k-1 ), including three central points. For the screening favoring even better the undissociated form of VFA with a lower
test, 11 experiments were performed in triplicate and four vari- pKa such as acetic acid (pKa= 4.75). A deeper univariate analysis
ables were covered with wide evaluation ranges: extractant vol- was performed for the salting out effect (Fig. 2D). Finally, a robust-
ume (20 0-50 0 μL), salt content (0.0 - 1.0 M), pH (1.0 - 2.0) and ness validation test was performed in a two-factor complete facto-
stirring time (40 - 120 s). The peak area of each VFA was used as rial design (stirring time and extraction volume) with three central
the response signal. The sum of normalized multiple responses of points (Fig. S2C). The data obtained show a robust system toler-
the analyte area was used to transform the individual responses of ant to an oscillation of 5% over the optimal value of the variables,
each acid into one. For this, the area of individuals VFAs in each working with a 95% confidence (p= 0.05).
of the experiments was divided over the largest area obtained for
said analyte under the optimization conditions developed. For this
research, 3 mL of sample enriched with 500 mg L−1 was used.
The use of dispersants was evaluated, however it was not used 3.1.5. Stirring time
because previous tests with TBME solvent and dispersants such as The effect of shaking for longer times on the analytical response
acetone, methanol and acetonitrile showed a dispersion of micro- has been evaluated (Fig. 2C), showing that equilibrium is reached
droplets that did not separate into two phases. This fact is possi- for all objective analytes after 7 min. It should be noted that the
bly attributed to the similar densities between the chosen extrac- extraction for acetic at 500 mg L−1 did not increase after 100 s,
tant and the dispersants. However, not using an additional solvent but the extraction of longer chain acids than acetic acid contin-
brings the benefit of having a lower reagent expense and avoid ued to increase over time, contributing to a rapid increase in the
dilution of the analytes in the medium. In addition, conventional TVFAs concentration. In a previous experiment (not shown here)
liquid-liquid extraction procedures for VFAs usually only use a sin- it was observed that 100 s was adequate to perfectly observe 5
gle organic solvent. Likewise, thanks to the small vortex generated mg L−1 of acetic acid and even lower concentrations of the same
by motor it is possible to form the dispersion. The fastest speed al- acid as was shown later in the method validation. Likewise, per-
lowed by the MSA system was fixed (equivalent to 1.4 V). A rapid forming extractions with long times can lead to greater miscibility
separation between the organic phase and the aqueous phase was between the aqueous and organic phases, increasing the time for
observed and, therefore, the separation time was set at 20 s. phase separation. Thus, based on the previous discussion and the
The results obtained in the Pareto chart (Fig. 2B) and in ANOVA results obtained with the preliminary analysis and the CCD, where
test (Table S2) showed that the curvature, extractant volume, pH, it was observed a small difference between 100 and 120 s, it was
salt concentration and extraction time were significant with a 95% decided to choose 100 s to perform the extraction. This, since the
confidence level, indicating that optimum values could be found sensitivity of the proposed method is in accordance with the con-
within the studied range. As expected, the Pareto chart displays a centrations present in the analyzed matrices, and at the same time
negative effect on the extractant volume and pH, showing that at the requirements of longer stirring times are reduced.

5
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

Table 1
Figures of merit of the developed In-syringe-MSA-LLME method for VFAs determination.

Acetic Propionic Isobutyric Butyric Isovaleric Valeric

Retention time (min) 5.16 ± 0.05 6.11 ± 0,05 6.42 ± 0.05 7.12 ± 0.05 7.63 ± 0.05 8.51 ± 0.05
LOD (mg L−1 ) 1.3 1.0 0.8 0.5 0.3 0.1
LOQ (mg L−1 ) 4.3 3.3 2.5 1.8 1.1 0.2
Linear range (mg L−1 ) 4.3 - 1000 3.3 - 1000 2.5 - 1000 1.8 - 1000 1.1 - 1000 0.2 - 1000
Slope ± SD 0.0555 ± 0.002 0.2712 ± 0.003 0.5716 ± 0.002 0.5426 ±0.002 0.7681 ± 0.003 0.7966 ± 0.001
Intercept ± SD 0.0003 ± 0.00006 0.0007 ± 0.00003 0.0001± 0.00002 0.0034 ± 0.0002 0.0034 ± 0.0002 0.0039 ± 0.0006
Determination coefficient (R 2 ) 0.9998 0.9999 0.9997 0.9998 0.9999 0.9999
Intra-day precision (%) (n= 5) 1.3 2.4 2.3 2.3 1.0 0.7
Inter-day precision (%) (n= 5) 6.9 3.6 3.0 1.7 1.9 2.1
Enrichment factor 1.2 3.2 4.4 4.3 5.3 5.5
ER to 10 mg L−1 (%) (n= 5) 5.7 20.0 31.6 30.1 32.4 38.5
ER to 50 mg L−1 (%) (n= 5) 6.5 22.7 28.5 29.4 36.1 37.0
ER to 100 mg L−1 (%) (n= 5) 7.7 22.1 27.8 26.4 37.5 35.1

SD: standard deviation, ER: extraction recovery.

3.1.6. Salting out effect analytes to the organic phase lies in the fact of avoiding the prob-
The effect of salt content was evaluated, since it sometimes lems derived from the injection of water into the GC. Therefore,
produces an improvement in extraction efficiency [7,11]. Therefore, the method developed has good performance and great applicabil-
NaCl concentrations were added up to 2.0 M. According to Fig. 2D, ity for wastewater samples.
the extraction efficiency after adding 0.5 M of NaCl increased for
acetic 38%, for propionic 37% and for the rest of the fatty acids an 3.3. Application to wastewater samples
average increase of about 20% was obtained. There were no signif-
icant differences between 0.5 M and 1.0 M, but a concentration of In order to evaluate the applicability of the method developed
2.0 M slightly decreased VFAs extraction. Therefore, a salt concen- in syringe, 4 types of wastewater samples with different matri-
tration of 0.5 M was chosen. The addition of salt to a water sam- ces were analyzed (Table 2). Sample 1 comes from the inlet (I) of
ple increases the extraction of hydrophobic and hydrophilic ana- wastewater in the WWTP of Palma de Mallorca, sample 2 corre-
lytes. This reduces hydrogen bonds between hydrophilic analytes sponds to the outlet (O) of the purified water after the complete
and water, which reduces their solubility in water and increases treatment, sample 3 was obtained from anaerobic digester super-
their mass transfer to the organic phase. However, when the con- natant (AD) which treats the residual sludge from the process of
centration of sodium chloride increases, the extraction efficiency of primary and secondary treatment of the wastewater, and sample
fatty acids decreases slightly, due to an increase in viscosity caused 4 comes from the laboratory-operated (FR) fermenter reactor (day
by the interactions between the ionic form of the salt and these 10, pH 7.0). The results showed that only sample 4 (FR) contained
polar analytes. appreciable amounts of all the VFAs analyzed (Fig. 3).
On the other hand, no VFAs were found in the WWTP samples
3.2. Figures of merit of the in-syringe-MSA-DLLME-GC-FID method above the LOD. The possible absence or low content of VFAs in the
AD sample can be attributed to the rapid transformation of VFAs
To evaluate the performance of the proposed method, its fig- into biogas, indicating the efficiency of the process and an absence
ures of merit were studied and shown in Table 1. The linear work of inhibition in methane production [37]. Therefore, to evaluate the
range was established from the LOQ to the highest standard, show- effects of the matrix, all wastewater samples were spiked with the
ing correlation coefficients (R2 ) greater than 0.9997. As can be seen target analytes at 10, 50 and 100 mg L−1 and recovery studies were
in this table, a wide range of concentrations were quantified, i.e., performed. Samples were also spiked with a certified VFAs stan-
4.3-10 0 0 mg L−1 of acetic acid, 3.3-10 0 0 mg L−1 of propionic, 2.51- dard at a concentration of 1 mM for method validation.
10 0 0 mg L−1 of isobutyric acid, 1.8 - 10 0 0 mg L−1 of isobutyric As can be seen in Table 2, a good recovery of the standards
acid, 1.1 - 10 0 0 mg L−1 of isovaleric and 0.2 - 10 0 0 mg L−1 of va- added to the four samples of wastewater was obtained, with a
leric. LODs were lower than 1.3 mg L−1 and LOQs were lower than general average of 102.5 ± 8.2 %. However, the relative recovery
4.3 mg L−1 . In this case, the LOD are only estimates to verify the (RR) of the longer chain VFAs (100 mg L−1 of isovaleric and va-
absence of interference during the VFAs analysis, since typical con- leric acids) of the inlet sample was slightly lower with respect to
centrations in real samples do not require the detection of trace the RR percentages found for the other samples. This is attributed
quantities. The intra-day and inter-day precision were determined to the wastewater matrix in the inlet is much more complex be-
by analyzing five spiked samples from anaerobic digester of the cause it has not undergone any treatment as of the other sam-
WWTP of Palma de Mallorca (AD) at 50 mg L−1 within same day ples analyzed did, and possibly a large variety of higher molecular
and during five different days, respectively. As can be seen, RSD weight compounds (carboxylic acid of longer aliphatic chain, car-
were found from 0.7 to 2.4% in one day and from 1.7 to 6.9% in boxylic acids having an aromatic cycle, carbohydrates, amino acids,
five days (n= 5). alcohols, others [38]) in the sampled aliquot generated some type
Accuracy of the proposed method was evaluated in five repeti- of interference at the analysis time. Finally, these results demon-
tions by spiking 10, 50 and 100 mg L−1 to the anaerobic digester strate the suitability of the proposed method to selectively deter-
sample. Enrichment factors were between 1.2 to 5.5. Recovery rates mine VFAs in wastewater in terms of recovery, sensitivity and ac-
varied from 5.7 to 38.5%. An increase in extraction efficiency for curacy. Likewise, as mentioned earlier, the AD sample was used to
acids having a longer chain was observed. Although these values test the accuracy of the method, showing good precision in mea-
were low, the preconcentration carried out is more than enough to surements within the same day and on 5 different days. Therefore,
detect low concentrations of VFAs, since in this case it works with the applicability of the proposed method to this type of samples is
concentrations in the range of mg L−1 , due to the characteristics validated.
of the studied matrix, a high preconcentration of the analytes is The performance of the proposed method is comparable with
not required. Likewise, the most important reason to transfer the other extraction methods for VFAs determination in wastewater

6
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

Fig. 3. Chromatogram of the VFAs found in the fermenter reactor sample at pH 7.0 on day 10. (1) Acetic acid, (2) Propionic acid, (3) Isobutyric acid, (4) Butyric acid, (5)
Isovaleric acid, (6) Valeric acid, (7) Isocaproic acid, (8) Caproic acid and (IS) Internal standard: 2 ethylbutyric acid.

(Table S3). Respect to manual LLE methods [11,15,30,39,40], the ex- to a liquid-liquid extraction in batches, without the high volume
tractant solvent consumption is reduced from 5 to 50 times, mak- demands and the time it requires.
ing this process more economical and environmentally friendly.
Likewise, the extraction of the target analytes into few microliters 3.3.1. Determination of VFAs in the laboratory biodigester scale
of extracting solvent favors the preconcentration of those VFAs To determine the presence of VFAs in different concentration
present at low concentrations. Other extraction methods achieved ranges, samples of VFAs production were taken of anaerobic fer-
lower LOD; however, this involved the use of sequential extrac- mentation reactors with fish waste at three pH units (7.0; 9.0 and
tions accompanied by mass spectrometry detection [11], or long 11.0). The pH of a biodigester not only influences methane produc-
HS-SPME methods carried out manually [41,42]. Moreover, SPME tion but also has an important inhibition effect on acid producing
requires careful handling due to the fiber fragility; otherwise its microorganisms. Fig. S3 of support information shows the lowest
durability and extraction efficiency are affected. [12]. The method VFAs production at pH 11.0, otherwise the neutral-slightly alkaline
also is versatile because it can be applied to samples with high region (pH 7.0-9.0) favors the production of organic acids in quan-
concentration, making a dilution of organic drop in TBME, if is nec- tity and variety. These results are consistent with other works in
essary. In-syringe-MSA-DLLME provides precise control of the mea- which it seems that strong alkaline conditions inhibit the activity
surement of the sample volume, TBME volume and organic drop of acidogenic bacteria, as reflected at a pH above 9.0 [43].
collected, due to the use of a 16,0 0 0 step motor in the burette. In
the same way, extraction in a closed system in a syringe was able 3.4. Comparison of proposed automated method vs distillation
to minimize the loss of acid by volatilization. Thus, the proposed method
method is also competitive in terms of precision and reproducibil-
ity. The most frequently method used for the routine analysis of
On the other hand, the fact of carrying out in just 7 min the en- VFAs in wastewater samples from anaerobic treatment processes
tire treatment process (sampling, extraction, phase separation, in- is distillation (with subsequent titration) [37,44]; found as method
jection of the extract (in vials) and cleaning) means a great time 5560 C in standard APHA protocols [33]. In order to make a com-
saver, considering that batch extraction can take up more than 30 parison of the proposed method with the distillation method, sam-
min per sample according to the chosen method. In-syringe-MSA- ples from fermenter reactor at pH 7.0 (days 1, 3, 5, 7, 10 and
DLLME system treats up to 8 samples h−1 . The extraction process 15) were selected and analyzed. Although the total time of pro-
(agitation and separation) takes at least 3 min and therefore, it posed method (extraction + separation and detection by GC-FID) is
is possible to simultaneously analyze up to 5 samples h−1 with longer than distillation time; method 5560C only determines TV-
the chromatograph, which could make possible an online analysis. FAs, expressed as milligrams of acetic acid per liter, and consumes
Likewise, it was not necessary to include the centrifugation steps a large volume of reagents and water for its preparation, generat-
and a salt drying (Na2 SO4 anhydride), because a good separation ing a large amount of liquid waste.
of the organic phase was achieved and the collection of organic The methods were compared quantitatively using different sta-
extract in the vial was precise. In contrast with manual SPME, tistical approaches: the Bland-Altman, Deming and Passing-Bablok
LLE and vortex-assisted LLE (VALLE), the method analysis time was tests (Fig. S4). These procedures are adequate to compare the to-
significantly shorter. Accordingly, all of the aforementioned allows tal concentrations of VFAs, because they are arbitrary to select
this automated method to be positioned as the first to be so close which is the dependent and independent variable and consider

7
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

both methods with an experimental error [7]. Deming regression

RR (%)
and the Passing-Bablok regression indicates that there is no signif-

106.2

107.4

102.5
103.5
112.5
102.6
88.3
95.5

98.1

97.3
98.3

97.3

93.3

97.6
96.7
80.1
icant difference between the two methods (α >0.05). The difference
found by Bland-Altman (194 mg L−1 ) can be attributed to various
sources of error that underestimate or overestimate TVFAs concen-

106.2 ± 2.4
Cf (mg L−1 )

125.8± 4.4
trations. Particularly, total concentrations determined by the distil-
47.8 ± 2.2
80.1 ± 0.1

49.2 ± 4.0

53.7 ± 0.8
97.6 ± 1.0

13.3 ± 0.7
23.5 ± 0.3

118.0±7.5
100.0±0.9

65.0± 0.2
9.7 ± 1.0

9.3 ± 0.6

98.7±1.2
96.3±6.0
8.8 ±1.0

lation method were higher than concentrations found with auto-


Valeric

mated method. Factors in the distillation process such as sample


ND

ND

ND
pretreatment, titrant addition rate and magnetic stirring can affect
the accuracy of VFAs monitoring. On the other hand, ionic interfer-
ents (bicarbonate, VFAs, sulphide, phosphate, ammonium and oth-
RR (%)

100.6

108.4

107.6

103.9
94.1
97.5
82.3

96.7

98.2

93.1

98.8

96.3
94.8
92.4
97.0

99,0
ers) and solid interferents (i.e. cellulose, hemicellulose, and other
carbon and phosphorus salts) can react with the titrant, affecting
the titrant consumption and therefore the results [44]. So, distil-

535.9 ± 3.0
545.5 ± 2.1
108.4 ± 2.8
Cf (mg L−1 )

lation method can be empirical and provide an incomplete and


48.8 ± 5.5

48.5 ± 4.2
82.3 ± 0.1

53.8 ± 0.8
98.8 ± 0.2

583.3±1.9
637.9±3.9
100.3±1.1

101.1±0.3

642.0±2.1
9.7 ± 1.8

9.3 ± 0.7
Isovaleric

99.5±5.1
9.4 ±1.8

somewhat variable recovery, requiring a recovery factor for each


apparatus [12,33]. Likewise, the recoveries of this method can de-
ND

ND

ND

crease if individual VFAs are present in very low concentrations.


Finally, the fact of not considering the concentrations of isocaproic
acid in the TVFAs could also influence the total quantification of
RR (%)

112.9
110.5

103.5

100.1
105.9
108.7

109.9
105.5
102.9
93.5

94.4
95.0

94.9
96.3
96.8
99.6

organic acids. In another study carried out in large-scale anaerobic


digestion plants where the titration method was compared with a
GC method, a difference of 873 mg L−1 of TVFAs was found [44].
Thus, compared with In-syringe-MSA-DLLME-GC-FID, lower aver-
887.2± 14.6
839.1 ± 9.2
848.6 ± 3.3
108.7 ± 7.9

105.5 ± 1.6
Cf (mg L−1 )

54.9 ± 4.5
55.0 ± 5.3
93.5 ± 1.0

10.0 ± 2.1
53.0 ± 2.6

945.4±4.2
926.9±4.6
11.3 ±2.2

Ca: concentration added; Cf: concentration found (average ± SD, n= 3), SD: standard desviation (n= 3), RR: relative recovery, ND: No detected.

age differences were provided.


9.4 ± 3.3
91.1±5.1

83.7±1.7

90.7±1.2
Butyric

4. Conclusions
ND

ND

ND

Concentration added of 1 mM is equivalent to 60.1 mg L−1 of Ac, 74.1 mg L−1 of Pr, 88.1 mg L−1 of Ib y Bu, 102.1 mg L−1 of Iv and Va.

A simple, fast and selective method based on in-syringe-MSA-


DLLME was used as sample preparation procedure followed by GC-
RR (%)

110.8
108.2

103.7

120.3

107.7
103.6

116.3
105.1
109.3

111.4
100.4
110.3
91.2

91.5

94.0

99.3

FID for the extraction and determination of volatile fatty acids (C2-
C5) in anaerobic digestion processes. To our knowledge, this is
the first automated method for individual quantification of VFAs
that allows extracting, preconcentration and cleaning at the same
471.7 ± 2.5
483.3 ± 1.6
103.6 ± 1.8

527.4 ± 2.0
Cf (mg L−1 )

109.3 ±6.9
54.1 ± 2.5

52.5 ± 3.1

53.8 ± 4.6
91.2 ± 0.3

12.0 ± 0.7

572.1±3.4
568.7±6.0
11.1 ±2.6
Isobutyric

9.4 ± 1.3

time, avoiding the need to perform multiple extractions in a batch


91.4±3.9

80.6±5.4

87.5±0.9
Application of the proposed In-syringe-MSA-DLLME-GC-FID method to determine VFAs in wastewater samples.

method and the problems derived from the direct aqueous injec-
ND

ND

ND

tion in column. A multivariate and univariate optimization was ap-


plied to find the best analytical conditions in the system. Results
showed tert-butyl methyl ether as the most suitable solvent for
RR (%)

111.1
103.2

112.7

112.4
113.4

111.2
109.5
111.0

110.1

106.2

107.1
108.1
101.8
110.0

extraction. The characteristics of the designed system and the se-


91.3

96.4

lected extraction solvent avoided the centrifugation stage and the


drying of the extract, having a great time saving. The method was
successfully applied for the analysis of complex samples coming
1697.2 ±3.5
110.1 ±11.7

1686.1± 7.0

1739.6 ±3.4
1794.4± 4.2
113.4 ± 2.6
109.5 ± 1.1
Cf (mg L−1 )

1761.5±5.9
11.3 ± 2.0

56.2 ± 1.3
55.0 ± 2.8

from WWTP and anaerobic processes. Sample matrix interference


10.3 ±1.1

9.6 ± 3.2
Propionic

55.6±5.9

67.6±1.3
82.2±9.0

78.7±0.7

was no observed, so the approach presented may be useful for


the routine analysis of VFAs in the control of anaerobic processes
ND

ND

ND

of municipal or industrial SWTPs and WWTPs, being competi-


tive in terms of sensitivity, precision, frequency and solvent sav-
RR (%)

ing, respect to previously informed manual procedures. Likewise,


113.7
111.8

102.9

113.8

113.8
108.4
103.8

109.6

109.5
112.3

104.3
107.9
85.5

91.7
84.5
98.0

the method can be extended to micro scale studies in the labo-


ratory and other complex aqueous matrices (leachate, wetlands).
Finally, the statistical comparison between the distillation method
8055.1 ± 88.5
8064.3 ± 53.4
8097.4 ± 11.6

and the automated method did not have a significant difference in


8159,4 ±18.7
109.1 ± 11.0

8119.9±27.4
108.4 ± 2.0
Cf (mg L−1 )

112.3± 0.9
55.9 ± 0.7

11.4 ± 0.3
54.7 ± 2.2
10.3 ± 0.3
49.0 ± 2.3

the quantification of TVFAs. Therefore, the proposed methodology


11.4 ±1.6

62.3±7.7

51.3±3.1

68.3±2.7

has good accuracy and is reliable for the analysis of organic acids.
Acetic

ND

ND

ND

Credit Author Statement

M.A. Vargas-Muñoz: Research, writing


Ca (mg L−1 )

Víctor Cerdà: Methodology


L.S. Cadavid-Rodríguez: Reviewing, methodology
1 mM∗

1 mM∗

1 mM∗

1 mM∗
100

100

100

100

Edwin Palacio: Conceptualization, methodology, supervision


10
50

10
50

10
50

10
50
0

Declaration of Competing Interest


Sample

The authors declare that they have no known competing finan-


Table 2

AD

cial interests or personal relationships that could have appeared to


FR
O
I

influence the work reported in this paper.


8
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

Acknowledgments [17] M. Ábalos, J.M. Bayona, Application of gas chromatography coupled to chemical
ionisation mass spectrometry following headspace solid-phase microextraction
for the determination of free volatile fatty acids in aqueous samples, J. Chro-
The authors acknowledge financial support from the Comuni- matogr. A. 891 (20 0 0) 287–294, doi:10.1016/S0 021-9673(0 0)0 0655-5.
tat Autonoma de les Illes Balears through the Direcció General de [18] J.A. Cruwys, R.M. Dinsdale, F.R. Hawkes, D.L. Hawkes, Development of a static
Política Universitaria i Recerca with funds from the Tourist Stay Tax headspace gas chromatographic procedure for the routine analysis of volatile
fatty acids in wastewaters, J. Chromatogr. A. 945 (2002) 195–209, doi:10.1016/
Law (PRD2018/45) and also to the Govern de les Illes Balears and S0021-9673(01)01514-X.
the Oficina de Cooperació al Desenvolupament i Solidaritat of the [19] C.I.C. Silvestre, J.L.M. Santos, J.L.F.C. Lima, E.A.G. Zagatto, Liquid-liquid extrac-
University of the Balearic Islands (OCDS-CUD2018/1). M.A.V-M. ac- tion in flow analysis: A critical review, Anal. Chim. Acta. 652 (2009) 54–65,
doi:10.1016/j.aca.2009.05.042.
knowledges the support from the Spanish Ministry of Science, In-
[20] S. Hu, X. Chen, R. qin Wang, L. Yang, X. hong Bai, Natural product applications
novation and Universities (MCIU) for the pre-doctoral research fel- of liquid-phase microextraction, TrAC - Trends Anal. Chem. 113 (2019) 340–
lowship (FPU19/06082). E.P. thanks to Emaya for the provision of 350, doi:10.1016/j.trac.2018.11.006.
[21] M. Rezaee, Y. Yamini, M. Faraji, Evolution of dispersive liquid-liquid mi-
wastewater samples and Dr. Sabrina Clavijo for her contribution.
croextraction method, J. Chromatogr. A. 1217 (2010) 2342–2357, doi:10.1016/
j.chroma.2009.11.088.
[22] P.P. Zhang, Z.G. Shi, Q.W. Yu, Y.Q. Feng, A new device for magnetic stirring-
Supplementary materials assisted dispersive liquid-liquid microextraction of UV filters in environmental
water samples, Talanta 83 (2011) 1711–1715, doi:10.1016/j.talanta.2010.11.076.
Supplementary material associated with this article can be [23] A. Saleh, Y. Yamini, M. Faraji, M. Rezaee, M. Ghambarian, Ultrasound-assisted
emulsification microextraction method based on applying low density organic
found, in the online version, at doi:10.1016/j.chroma.2021.462034. solvents followed by gas chromatography analysis for the determination of
polycyclic aromatic hydrocarbons in water samples, J. Chromatogr. A. 1216
(2009) 6673–6679, doi:10.1016/j.chroma.20 09.08.0 01.
References [24] M.A. Farajzadeh, M.R.A. Mogaddam, Air-assisted liquid-liquid microextraction
method as a novel microextraction technique; Application in extraction and
[1] F. Raposo, R. Borja, J.A. Cacho, J. Mumme, Á.F. Mohedano, A. Battimelli, D. Bol- preconcentration of phthalate esters in aqueous sample followed by gas
zonella, A.D. Schuit, J. Noguerol-Arias, J.C. Frigon, G.A. Peñuela, J. Muehlenberg, chromatography-flame ionization detection, Anal. Chim. Acta. 728 (2012) 31–
C. Sambusiti, Harmonization of the quantitative determination of volatile fatty 38, doi:10.1016/j.aca.2012.03.031.
acids profile in aqueous matrix samples by direct injection using gas chro- [25] E. Yiantzi, E. Psillakis, K. Tyrovola, N. Kalogerakis, Vortex-assisted liquid-
matography and high-performance liquid chromatography techniques: Multi- liquid microextraction of octylphenol, nonylphenol and bisphenol-A, Talanta
laboratory validation study, J. Chromatogr. A. 1413 (2015) 94–106, doi:10.1016/ 80 (2010) 2057–2062, doi:10.1016/j.talanta.20 09.11.0 05.
j.chroma.2015.08.008. [26] B. Horstkotte, R. Suárez, P. Solich, V. Cerdà, In-syringe-stirring: A novel ap-
[2] I.H. Franke-Whittle, A. Walter, C. Ebner, H. Insam, Investigation into the ef- proach for magnetic stirring-assisted dispersive liquid-liquid microextraction,
fect of high concentrations of volatile fatty acids in anaerobic digestion on Anal. Chim. Acta. 788 (2013) 52–60, doi:10.1016/j.aca.2013.05.049.
methanogenic communities, Waste Manag 34 (2014) 2080–2089, doi:10.1016/ [27] M. Alexovič, B. Horstkotte, I. Šrámková, P. Solich, J. Sabo, Automation of disper-
j.wasman.2014.07.020. sive liquid–liquid microextraction and related techniques, TrAC - Trends Anal.
[3] M.A. Khan, H.H. Ngo, W. Guo, S.W. Chang, D.D. Nguyen, S. Varjani, Y. Liu, Chem. 86 (2017) 39–55, doi:10.1016/j.trac.2016.10.003.
L. Deng, C. Cheng, Selective production of volatile fatty acids at different pH [28] P.F. Pind, I. Angelidaki, B.K. Ahring, A new VFA sensor technique for anaerobic
in an anaerobic membrane bioreactor, Bioresour. Technol. 283 (2019) 120–128, reactor systems, Biotechnol. Bioeng. 82 (2003) 54–61, doi:10.1002/bit.10537.
doi:10.1016/j.biortech.2019.03.073. [29] K. Boe, D.J. Batstone, I. Angelidaki, An innovative online VFA monitoring system
[4] Y. Li, Y. Chen, J. Wu, Enhancement of methane production in anaerobic di- for the anaerobic process, based on headspace gas chromatography, Biotechnol.
gestion process: A review, Appl. Energy. 240 (2019) 120–137, doi:10.1016/j. Bioeng. 96 (2006) 712–721, doi:10.1002/bit.
apenergy.2019.01.243. [30] T. Pruksatrakul, P. Phoopraintra, P. Wilairat, P. Chaiyen, R. Chantiwas, Develop-
[5] V. Stazi, M.C. Tomei, Enhancing anaerobic treatment of domestic wastewater: ment of a sequential injection-liquid microextraction procedure with GC-FID
State of the art, innovative technologies and future perspectives, Sci. Total En- for analysis of short-chain fatty acids in palm oil mill effluent, Talanta 165
viron. 635 (2018) 78–91, doi:10.1016/j.scitotenv.2018.04.071. (2017) 612–618, doi:10.1016/j.talanta.2017.01.010.
[6] B.K. Ahring, M. Sandberg, I. Angelidaki, Volatile fatty acids as indicators of pro- [31] F. Robert-Peillard, J.L. Boudenne, B. Coulomb, Individual volatile fatty acids
cess imbalance in anaerobic digestors, Appl. Microbiol. Biotechnol. 43 (1995) determination by chromogenic derivatization coupled to multi-syringe chro-
559–565, doi:10.10 07/BF0 0218466. matography, Talanta 115 (2013) 737–743, doi:10.1016/j.talanta.2013.06.019.
[7] M. Ghidotti, D. Fabbri, C. Torri, S. Piccinini, Determination of volatile fatty [32] F. Maya, B. Horstkotte, J.M. Estela, V. Cerdà, Lab in a syringe: Fully auto-
acids in digestate by solvent extraction with dimethyl carbonate and gas mated dispersive liquid-liquid microextraction with integrated spectropho-
chromatography-mass spectrometry, Anal. Chim. Acta. 1034 (2018) 92–101, tometric detection, Anal. Bioanal. Chem. 404 (2012) 909–917, doi:10.1007/
doi:10.1016/j.aca.2018.06.082. s00216- 012- 6159- 4.
[8] W. Janczukowicz, J. Rodziewicz, K. Czaplicka, I. Kłodowska, A. Mielcarek, The [33] American Public Health Association - APHAStandard Methods for the Exami-
effect of volatile fatty acids (VFAs) on nutrient removal in SBR with biomass nation of Water and Wastewater, 1998 20th Ed, Washington DC.
adapted to dairy wastewater, J. Environ. Sci. Heal. - Part A. 48 (2013) 809–816, [34] International Union of Pure and Applied Chemistry - IUPAC, Harmonized
doi:10.1080/10934529.2013.744658. guidelines for single-laboratory validation of methods of analysis, Budapest
[9] G. Strazzera, F. Battista, N.H. Garcia, N. Frison, D. Bolzonella, Volatile fatty acids (20 02), doi:10.1351/pac20 0274050835.
production from food wastes for biorefinery platforms: A review, J. Environ. [35] F. Raposo, R. Borja, J.A. Cacho, J. Mumme, K. Orupõld, S. Esteves, J. Noguerol-
Manage. 226 (2018) 278–288, doi:10.1016/j.jenvman.2018.08.039. Arias, S. Picard, A. Nielfa, P. Scherer, I. Wierinck, E. Aymerich, C. Cavinato,
[10] A. Gherghel, C. Teodosiu, S. De Gisi, A review on wastewater sludge valorisa- D.C. Rodriguez, N. García-Mancha, P.N.T. Lens, V. Fernández-Cegrí, First inter-
tion and its challenges in the context of circular economy, J. Clean. Prod. 228 national comparative study of volatile fatty acids in aqueous samples by chro-
(2019) 244–263, doi:10.1016/j.jclepro.2019.04.240. matographic techniques: Evaluating sources of error, TrAC - Trends Anal. Chem
[11] A. Banel, B. Zygmunt, Application of gas chromatography-mass spectrometry 51 (2013) 127–143, doi:10.1016/j.trac.2013.07.007.
preceded by solvent extraction to determine volatile fatty acids in wastewater [36] D.T. Hill, R.D. Holmberg, Long chain volatile fatty acid relationships in anaer-
of municipal, animal farm and landfill origin, Water Sci. Technol. 63 (2011) obic digestion of swine waste, Biol. Wastes. 23 (1988) 195–214, doi:10.1016/
590–597, doi:10.2166/wst.2011.204. 0269- 7483(88)90034- 1.
[12] R. Fernández, R.M. Dinsdale, A.J. Guwy, G.C. Premier, Critical analysis of meth- [37] L.S. Cadavid-Rodríguez, M.A. Vargas-Muñoz, J. Plácido, Biomethane from fish
ods for the measurement of volatile fatty acids, Crit. Rev. Environ. Sci. Technol. waste as a source of renewable energy for artisanal fishing communities, Sus-
46 (2016) 209–234, doi:10.1080/10643389.2015.1073493. tain. Energy Technol. Assessments. 34 (2019) 110–115, doi:10.1016/j.seta.2019.
[13] V. Diamantis, P. Melidis, A. Aivasidis, Continuous determination of volatile 05.006.
products in anaerobic fermenters by on-line capillary gas chromatography, [38] E. Palacio-Barco, F. Robert-Peillard, J.L. Boudenne, B. Coulomb, On-line analysis
Anal. Chim. Acta. 573–574 (2006) 189–194, doi:10.1016/j.aca.2006.05.036. of volatile fatty acids in anaerobic treatment processes, Anal. Chim. Acta. 668
[14] F.R. Mansour, N.D. Danielson, Solvent-terminated dispersive liquid-liquid mi- (2010) 74–79, doi:10.1016/j.aca.2009.12.019.
croextraction: a tutorial, Anal. Chim. Acta. 1016 (2018) 1–11, doi:10.1016/j.aca. [39] R.T. Rosmalina, U.Hamidah Widyarani, N. Sintawardani, Determination of
2018.02.005. volatile fatty acids in tofu wastewater by capillary gas chromatography with
[15] N.T. Mkhize, T.A.M. Msagati, B.B. Mamba, M. Momba, Determination of volatile flame ionization detection: A Comparison of extraction methods, IOP Conf. Ser.
fatty acids in wastewater by solvent extraction and gas chromatography, Phys. Earth Environ. Sci. (2020) 483, doi:10.1088/1755-1315/483/1/012038.
Chem. Earth. 67–69 (2014) 86–92, doi:10.1016/j.pce.2013.10.008. [40] E.M. Siedlecka, J. Kumirska, T. Ossowski, P. Glamowski, M. Gołebiowski, J. Gaj-
[16] S.-P. Yo, Analysis of volatile fatty acids in wastewater collected from a pig farm, dus, Z. Kaczyński, P. Stepnowski, Determination of volatile fatty acids in envi-
Chemosphere 38 (1999) 823–834. ronmental aqueous samples, Polish J. Environ. Stud. 17 (2008) 351–356.

9
M.A. Vargas-Muñoz, V. Cerdà, L.S. Cadavid-Rodríguez et al. Journal of Chromatography A 1643 (2021) 462034

[41] L. Feng, Y. Huang, H. Wang, Solid-phase microextraction in combination with [43] N. Bermúdez-Penabad, C. Kennes, M.C. Veiga, Anaerobic digestion of tuna
GC-FID for quantification of the volatile free fatty acids in wastewater from waste for the production of volatile fatty acids, Waste Manag 68 (2017) 96–
constructed wetlands, J. Chromatogr. Sci. 46 (2008) 577–584, doi:10.1093/ 102, doi:10.1016/j.wasman.2017.06.010.
chromsci/46.7.577. [44] H. Sun, J. Guo, S. Wu, F. Liu, R. Dong, Development and validation of a simpli-
[42] M. Ábalos, J.M. Bayona, J. Pawliszyn, Development of a headspace solid- fied titration method for monitoring volatile fatty acids in anaerobic digestion,
phase microextraction procedure for the determination of free volatile fatty Waste Manag 67 (2017) 43–50, doi:10.1016/j.wasman.2017.05.015.
acids in waste waters, J. Chromatogr. A. 873 (20 0 0) 107–115, doi:10.1016/
S0021- 9673(99)01263- 7.

10

You might also like