RESULTS AND DISCUSSION
Direct gas chromatographic determination of carboxylic acids
As it was mentioned in the introduction, for better chromatographic behaviour prior to GC analysis short-chain carboxylic acids should be derivatized. For derivatization, we have chosen BSTFA, as the reaction can be expected to be complete at room temperature, as BSTFA and its by-products (trimethylsilyltrifluoroacetamide and trifluoroacetamide) are more volatile than many other products, causing less chromatographic interference, and, finally, as hydrogen fluoride, a byproduct of silylation with BSTFA, reduces flame ionisation detector fouling [25]. However, BSTFA as other silyl reagents is moisture sensitive, and therefore the solvents and the samples should be dry.
Three solvents – acetone, DEE and DMF – were tested as a derivatization medium. For this, to 1 ml of succinic acid solutions (1 g l–1) 50 μl of BSTFA was added and after 20 min direct GC analysis was carried out. The analysis revealed that derivatization in acetone was not complete. In the chromatogram obtained after the derivatization a relatively small peak of a succinic acid derivatization product emerges at 12.25 min. However, a huge asymmetric peak of underivatized succinic acid at 10.70 min is still present as in the chromatogram obtained before derivatization (Fig. 1a). When the solvent acetone was replaced with DEE or DMF, the derivatization was complete (chromatograms of succinic acid in DEE before and after derivatization are presented in Fig. 1b). The boiling point of DEE is 34.6°C, thus it is difficult to handle, concentrations of a solution may change and thus lead to analytical errors. Because of that, even in the case of direct injection of solutions into a gas chromatograph, diethyl ether is not the first choice solvent. In the case of HS-GC, DEE would not fit at all, since the analytes must be transferred to the headspace by heating and the heating temperature must be lower than the boiling point of the solvent. Obviously, at temperature of less than 34.6°C, only a very small fraction of derivatized acids would be transferred to the headspace resulting in very high detection limits. Based on the above, we decided in favour of less volatile solvent DMF (boiling point 152°C).
To determine the amount of BSTFA required for derivatization, 5–50 μl of BSTFA was added to 1 ml of 1 g l–1 solutions of oxalic, lactic, succinic, malic and citric acids in DMF, and derivatization was carried out for 20 min. The results of the GC analysis demonstrated that for all the acids 30 μl of BSTFA was sufficient.
Further, the effect of derivatization time on the derivatization yield was studied. The acids were subjected to the GC analysis immediately after the addition of BSTFA and after 5 to 60 min of the addition of BSTFA. The results showed that the duration of derivatization had no effect on the peak areas. Therefore gas chromatographic analysis of the analytes of interest can be performed immediately after the derivatization reagent is added.
In order to correct the loss of analyte during sample handling, an internal standard should be added to the analyte solutions. An internal standard should be a compound that is not normally found in foods (our test items), should be well separated in the chromatogram, but contemporaneously should be eluted at a reasonable time. For four of our analytes, n-tridecane was selected as an internal standard. Only in the case of succinic acid, the peak of n-tridecane overlapped with the peak of the analyte, so n-tetradecane was chosen (chromatograms are presented in Fig. 2).
For all the analytes, quality parameters such as linearity, limits of detection (LOD) and repeatabilities were determined under the optimized conditions. The calibration curves were drawn with 7 calibration points with three replicate injections and were linear up to 1 g l–1 for all the analytes. The limits of detection were defined as three times the baseline noise. The repeatabilities were determined by five repetitions analysis. Concentration of an internal standard in the final solutions was 0.01 g l–1. The retention times, correlation coefficients, limits of detection and relative standard deviations are presented in Table 1.
| Carboxylic acid | RT, min | RSD% (c = 10 mg l–1, n = 5) | R2 | LOD, mg l–1 |
|---|---|---|---|---|
| Lactic acid | 8.48 | 2.2 | 0.999 | 0.27 |
| Oxalic acid | 9.61 | 4.9 | 0.996 | 0.55 |
| Succinic acid | 12.25 | 4.5 | 0.997 | 0.59 |
| Malic acid | 14.65 | 4.2 | 0.998 | 0.51 |
| Citric acid | 18.58 | 3.1 | 0.999 | 0.33 |
Headspace gas chromatographic determination of carboxylic acids
The next step of the work was to examine the possibility to determine the analytes of interest by headspace gas chromatography. The derivatization conditions were the same as in the case of direct GC determination.
Initially, pure carboxylic acids were examined. For this, 10–20 mg of the acids were weighed into a headspace vial, 30 μl of BSTFA was added, the vial was heated for 5 min at 200°C and analysed by headspace gas chromatography under the conditions described in the Experimental section. In the resulting chromatograms the peaks of the derivatized acids were observed. The retention times of them are given in Table 2.
| Carboxylic acid | RT, min | LOD, mg l–1 |
|---|---|---|
| Lactic acid | 10.64 | 22 |
| Oxalic acid | 12.22 | 19 |
| Succinic acid | 14.10 | 45 |
| Malic acid | 16.05 | 87 |
| Citric acid | 19.75 | – |
Further, it was investigated whether the derivatized and dissolved in DMF acids were transferred to the headspace. Since the temperature of the analyte transfer to the headspace must be lower than the boiling point of the solvent (152°C), the samples were heated at 140°C. In order to transfer the maximum amount of the analyte to the headspace, the heating time was increased to 20 min. Unfortunately, the concentrations of the derivates in the headspace were very low, thus the limits of detection were high and not of practical use. Citric acid was not detected at all (Table 2).
In order to obtain lower LOD, the solvent with a higher boiling point should be chosen. On the other hand, headspace gas chromatography is suitable for identifying the acids in dry samples.
Real sample analysis
The possibility to apply headspace gas chromatography for identifying of carboxylic acids is illustrated by the determination of succinic acid in amber. Amber is a fossil resin derived from coniferous trees. It is found all over the world, but because of different plants from which it originates, it is different in the composition. Succinic acid is found only in Baltic amber [26]. Thus identification of the amber from the Baltic Sea region can be based on the presence of succinic acid in it.
We have analysed the amber collected at the seaside of Palanga, excavated in the Kaliningrad region and originating from Japan. 20–30 mg of the crushed amber was weighed into a headspace vial, 30 μl of BSTFA was added, the vial was closed and heated for 15 min at 200°C. In Fig. 3 the chromatograms of the derivatized amber from Palanga (a), Kaliningrad (b) and Japan (c) are presented. In the chromatograms (a, b) a peak of succinic acid derivate is evidenced, but it is absent in the chromatogram of Japan amber (C).
Headspace gas chromatography was also applied to detect citric acid in boiled smoked chicken sausages Berlinki (Poland). Citric acid is added to the sausages to regulate the acidity. As BSTFA is moisture sensitive, before derivatization the sausages were crushed and dried at 70°C for 5 h. 50 mg of dried and grinded sausages was weighed into the headspace vial, 30 μl of BSTFA was added, the vial was closed and heated for 15 min at 200°C. A peak of derivatized citric acid is evidenced in the chromatogram (Fig. 4).
Unfortunately, for citric acid quantification, the sensitivity of the headspace gas chromatographic method was not sufficient. Thus the direct GC method was applied. To 0.5 g of grinded sausages 5 ml of DEE was added, the sausages were extracted for 5 min and centrifuged for 1 min at 5000 rpm. 3 ml of the extraction solution was evaporated to dryness, 1 ml of DMF containing an internal standard n-tridecane and 30 μl of BSTFA were added and the solution was injected into the gas chromatograph for analysis. A chromatogram of the extract is presented in Fig. 5. It was determined that the sausages contained 0.12 mg g–1 of citric acid. However, differently from the headspace chromatogram (Fig. 4), many peaks were observed in the chromatogram of the extract (Fig. 5). It resulted in a bad separation of the peaks of interest, consequently the repeatability of the results was not good (RSD for three replicate measurements was 21.3%).