Development and Validation of Two Methods to Quantify Volatile Acids (C2-C6) by GC/FID: Headspace (Automatic and Manual) and Liquid-Liquid Extraction (LLE)

The concentration of the volatile fatty acids (VFA) is an important indicator of the status of anaerobic processes, but most of the existing methods require sample pretreatment and are labor-intensive. It was developed and validated a rapid Gas Chromatographic (GC) method to quantify seven VFA (acetic, propionic, isobutyric, butyric, isovaleric, valeric and caproic), acetone, methanol, ethanol and n-butanol by headspace (automatic and manual) and liquid-liquid extraction (LLE) with diethyl ether (only VFA). The determination was made in a Shimadzu Gas Chromatograph equipped with a Flame Ionization Detector (GC/FID), a headspace auto-sampler and an HPINNOWAX column. Isobutanol and crotonic acid were utilized as internal standards (IS). The validation parameters evaluated were: precision (coefficient of variation—C.V.% for the retention times, from 0.02 to 0.87), linearity (R2 = 0.9291 0.9997), limits of detection (from 3.97 to 36.45 mg·L−1) and instrumental precision (from 0.01 to 0.53), which provide evidence that the methods are adequate to determine these analytes in samples from anaerobic reactors and from the environment.


Introduction
The factors that affect the performance of anaerobic reactors are related to the production of the so called vola-tile fatty acids (VFA): acetic, propionic, isobutyric, butyric, isovaleric and valeric acids, among others. The concentration of the VFA gives fast information of the status of the anaerobic processes and is also an important indicator for monitoring them compared to others such as alkalinity, chemical oxygen demand (COD) and pH [1].
Some of the most common sources of carbon used by anaerobic microorganisms are acetic acid, methanol and ethanol. Therefore, in the anaerobic processes of wastewater treatment it is important to monitor the levels of such substances as well as others by-products generated, acetone and n-butanol, in order to evaluate the status of the system.
Several studies have pointed the importance of the concentration of the individual VFA as an early warning indicator for process failure [2]- [5]. These include the concentration of isobutyric and isovaleric acids [2]- [4]. According to Ahring et al. [5] the accumulation of VFA suggests an imbalance in the process.
There are lots of methods described in the literature, like those proposed for offline individual VFA measurement, that involve sample filtration or centrifugation followed by direct injection into the gas chromatograph [6]; solvent extraction followed by GC for samples from leachate [7] or high performance liquid chromatography (HPLC) of acids from marine sediments after derivatization with 2-nitrophenylhydrazides [8] and solid phase micro extraction (SPME) directly from the aqueous phase, followed by GC/mass spectrometry [9]; determination by GC/FID of its methyl esters after derivatization [10], among others. Nevertheless, these methods are time consuming because most of them require sample preparation before the GC determination or can damage the chromatographic column by injecting directly very acidic solutions. Besides being a simple method, the headspace determination is a technique that preserves the life of the chromatographic column.
Liquid-liquid extraction (LLE) is a highly sensitive and selective technique that also contributes to increasing the useful life of the chromatographic column and is important as an alternative when it is necessary to determine only the organic acids.
This study aimed to develop and validate chromatographic methods to determine some VFA (acetic, propionic, isobutyric, butyric, isovaleric, valeric and caproic) by LLE using diethyl ether as the extracting solvent due to the compatibility of the polarities of the acids with the extractant; and the same acids simultaneously with acetone, methanol, ethanol and n-butanol by headspace (automatic and manual) in environmental samples as well as in samples from anaerobic reactors using the internal standard (IS) method for quantification. The proposed methods are based on GC/FID with an HP-INNOWAX capillary column, using an automatic headspace sampler or by manual headspace and, alternatively, by LLE with ethyl ether as the solvent.
It was added an inorganic salt to the samples to promote the salting out effect in both methods. It is a useful tool that has been widely used in determinations involving phase separations. In order to decrease the solubility of the substances in the aqueous phase and increase their transition to the vapor phase an inorganic salt as potassium carbonate (K 2 CO 3 ) or sodium chloride (NaCl), for example, is added to the aqueous solution [11]. The behavior of the phase equilibrium changes significantly with the presence of a dissolved salt and it improves the efficiency of the process of extracting [12].
This paper describes an investigation of the following validation parameters for both the methods described: linearity, detection and quantification limits, precision of the method and instrumental precision, according to Ribani et al. [13], Miller & Miller [14], Duarte et al. [15] and Damasceno et al. [16] recommendations.

Equipment and Chromatographic Conditions
Chromatographic analyses were performed with an HP-INNOWAX (30 m × 0.25 mm × 0.25 μm) capillary column (Agilent Technologies), in a Shimadzu GC-2010 (Kyoto, Japan) gas chromatograph equipped with FID, a COMBI-PAL auto sampler using 10 mL headspace vials and a 2.5 mL HD type Hamilton gas-tight syringe.
For the manual headspace method it was used a Fisatom hot plate and a SGE gas tight, with luer-lock-1000 µL syringe. For the LLE it was employed flasks type COD-10 mL (Figure 1), a Phoenix vortex stirrer, an Eppendorf centrifuge 5702 and the freezer of a common refrigerator; the syringe used for the injection was an Agilent 10 µL manual syringe for liquids. GC solution software was used for data treatment. After testing different possibilities, the chromatographic conditions to achieve better separation and peak resolution for the headspace method were: 35˚C (0 min) 2˚C/min 38˚C (0 min) 10˚C/min 75˚C (0 min) 35˚C/min 120˚C (1 min) 10˚C/min 170˚C (2 min), with total run time of 14.49 minutes (min). OBS: when there were many organic compounds in the sample it was necessary to increase the final temperature of the heating ramp up to 230˚C for 4 min (40˚C/min) to clean the column before the next injection. The column flow was 1.5 mL·min −1 with ultra-pure hydrogen (H 2 ) as the carrier gas. The detector used H 2 at 30 mL·min −1 with synthetic air at 300 mL·min −1 and nitrogen (N 2 ) at 30 mL·min −1 flows as flame and make-up gases, respectively. The temperatures of the detector and injector were held at 280˚C and 250˚C, respectively. After testing different temperatures for heating time and temperature of the block for the COMBI-PAL auto sampler and different volumes for the injection of the sample, the best conditions achieved were: 13 min of heating of the block at 100˚C, injection of 400 μL of sample with the syringe at 100˚C and 3 minutes of washing the syringe with N 2 after each injection.

Sample Preparation
The preservation of the samples from anaerobic reactors was tested in three ways, after centrifugation: without adding any substance, with addition of 3% H 3 PO 4 and with addition of 1 M NaOH, before to refrigerate or freeze the samples for 60 days prior to the analyses.
Because of the inability to analyze the samples from leachate directly due to saturation of the detector by the large amount of organic compounds present in the samples, the three methods tested for cleanup were: a) acidification with 2 M H 2 SO 4 followed by filtration with membrane of 1.2 µm and then freezing and thawing; centrifugation again, filtration (0.45 µm) and freezing; b) cartridge SAX eluted with 2 M H 2 SO 4 ; c) agitation of the sample in vortex with 2 M H 2 SO 4 and microcrystalline cellulose, followed by centrifugation. The influence of the salt added to the sample for the salting out effect was tested by adding 1 g of K 2 CO 3 or NaCl.

Automatic Headspace
An aliquot of 2 mL of the sample (or standard solution) was placed in a 10 mL standard vial for automatic headspace analysis, fitted with Teflon tape and Teflon-faced blue septa, containing approximately 1 g of NaCl, as follows:  1 g NaCl;  2 mL sample;  70 µL isobutanol solution 1 g·L −1 (IS for acetone and alcohols);  100 µL crotonic acid solution 700 mg·L −1 (IS for the VFA);  200 µL 2 M H 2 SO 4 solution.

Manual Headspace
It was the same preparation described in 2.3.1 and the flask with the sample was heated inside a Becker with boiling water on a hot plate for 13 min, before the injection of 400 µL of the headspace into the GC. The syringe must be kept at 90˚C before the injection to prevent condensation of the sample.
OBS: the syringe for the manual injection of the ether layer must stay in the freezer for, at least, 30 min before the injection because of the high volatility of the solvent.

Calibration Curves
The solutions for the calibration curves were prepared from stock solutions of each one of the substances at a concentration of 20 g·L −1 in ultra-purified water and in the synthetic medium [18] commonly used to feed the anaerobic reactors ( Table 2). Due to the low solubility of the VFA in water, it was necessary to add sodium hydroxide to the stock solutions to achieve its complete dissolution. These solutions were used for both methods described. From the stock solutions it was prepared the so called "mother solution" (MS), containing all the analytes to be determined in a concentration of approximately 1.0 g·L −1 . The concentrations of the calibration levels for all the substances to elaborate the calibration curves for the headspace method ranged from 1.00 to 800.00 mg·L −1 (11 levels) and, for the LLE, the concentrations ranged from 2.5 to 600.0 mg·L −1 (9 levels). The dilution was made with ultrapure water and also with a synthetic medium similar to the matrix of the samples from anaerobic reactors as the solvent using "Equation (1)" to obtain the desired concentration, in triplicate.
where C 1 is the concentration of the MS, V 1 is the volume of the MS, C 2 is the concentration of each level of the curves and V 2 is the final volume of each prepared solution.

Validation
The quantification of the analytes was made by the internal standard method, with the comparison of the response factor-RF, "Equation (2)" of each substance with the calibration curve under the same conditions.
i IS where A i is the peak area of each substance and A IS is the peak area of the internal standard considered. Linearity, precision of the method (C.V.% of the retention times) and limits of detection and quantification (LD and LQ, respectively) were assessed based on Ribani et al. [13], Duarte et al. [15] and Damasceno et al. [16] recommendations. Three replicates of each standard level were used to build the analytical curves, with the aid of the GC Solution software. The precision of the method was investigated through the C.V.% of the retention times (RT) of the replicates of the standard solution at the different concentrations, using "Equation (3)". The instrumental precision was evaluated by 10 successive injections of a unique standard solution with all the analytes at 50 mg·L −1 . Finally, LD and LQ were determined from the calibration curves with the "Equations (4)" and "(5)", suggested by Ribani et al. [13].
( ) where C.V.% is the coefficient of variation, SD is the standard deviation of RT and m is the average of RT of the analyte.

( )
where S is the standard deviation of the intercept and s is the slope of the calibration curve.

Results and Discussion
The smaller differences between the standard deviation (from 0.1 to 34 mg·L −1 ) of the responses were observed for the samples centrifuged and analyzed without addition of any preservative and frozen or refrigerated (for this last option, in flasks without headspace), after 60 days. The better cleanup for samples from landfill leachate was extraction from SAX cartridge with 2 M H 2 SO 4 (C.V.% from 1.0 to 20.8 for the concentrations obtained).
Both of the salts tested for the salting out effect: K 2 CO 3 and NaCl were appropriate to transfer the substances to the vapor phase. However, the former also promoted an increase of the pH, which is not suitable for this kind of determination, once the pH for the VFA determination must be ≤2.0. So, it was established the NaCl as the substance to promote the salting out effect.

Headspace
A typical chromatogram of the headspace analysis (that is similar for manual and automatic ways) is in Figure 2, Table 3 (peaks between 6.0 and 7.0 min may be related with some substances present in previous injections, probably retained due to the high polarity of this column). In Figure 3 ( Table 4) there is a chromatogram of a standard solution of the acids (100 mg·L −1 ) by GC/FID after extraction with ethyl ether. The chromatograms exhibit good separation with well-resolved peaks for the analytes in a short time.

Calibration Curves
No considerable changes have been observed in the equations obtained from the standards diluted with water and with synthetic medium for both methods (headspace and ethyl ether extraction), which indicates that these methods were not affected by this kind of changes in the matrix of the sample.

Validation
The methods studied were linear at the considered range of concentrations (1.0 to 800.0 mg·L −1 ). The coefficient of correlation (R 2 ) for the headspace analysis of the standards diluted in synthetic medium ranged from 0.9291 (for caproic acid, which has the higher boiling point among the considered acids and, therefore, the lower volatility- Table 1) to 0.9997 (automatic, Table 5) and from 0.9614 to 0.9997 for the manual technique. It was observed the carry-over effect mainly for the automatic headspace, which is avoided with the introduction of "whites" every 3 samples (containing only water and sulfuric acid). The best results were obtained for the correlation coefficients by the method of extraction with ethylic ether (from 0.9951 to 0.9997, Table 6). These results show the higher extraction efficiency of this method; however, it does not allow the determination of acetone and alcohols, due to the fact that the retention times are identical to the ethylic ether, besides being more labor-intensive.
The LD and LQ for the headspace (from 5.8 to 28.6 mg·L −1 ) and for the LLE (from 3.97 to 36.45 mg·L −1 ) demonstrate the reliability of these methods.
The C.V.% of the retention times were very low (from 0.02 to 0.87) for the headspace and from 0.07 to 0.34 for the LLE methods, respectively; and from 0.01 to 0.53 for the 10 successive injections of the 50 mg·L −1 standard by all the techniques. These values reflect the good precision of the method and the instrumental precision, respectively.
The LD and LQ for the headspace (from 5.8 to 28.6 mg·L −1 ) and for the LLE (from 3.97 to 36.45 mg·L −1 ) demonstrate the reliability of these methods.
The tabulated values for the Ftest (ANOVA) are 3.0204 and 3.5004 for the headspace and the LLE methods, respectively and the calculated values are in Table 7. These values (F calculated > F tabulated ) are further evidence of the linearity of the methods studied [19].

Conclusion
The gas chromatographic determination (GC/FID) by headspace (automatic and manual) described is a useful and rapid analytical method for monitoring eleven volatile substances: acetone, some alcohols and VFA in samples from anaerobic processes and from the environment. The method described for LLE with diethyl ether is an efficient analytical procedure for the determination of VFA. The validation parameters analyzed for the methods developed indicated good linearity, precision and both low detection and quantification limits, allowing its application to monitor the status of wastewater anaerobic treatment systems in a short time of analysis.   Table 4).