Determination of Volatile Carboxylic Acids (C1-C5i) and Lactic Acid in Aqueous Acid Extracts of

Silage by Liquid Chromatography

Andrea Canale, Maria E. Valente and Angelo Ciotti

A simple isocratic h.p.l.c. technique was developed for the simultaneous quantitative analysis of
lactic, formic, acetic, propionic, isobutyric, n-butyric and isovaleric acids in aqueous acid extracts of
silage. An Aminex HPX-87H strong cation exchange resin column at 41C, 0.0025M H2S04 mobile
phase and ultraviolet detector at 210 nm were utilised. Estimated recoveries of acids added to silage
ranged from 98.2 to 103.5%.

Keywords: Ion exchange h.p.l.c.; volatile carboxylic acids; lactic acid; silage

Introduction

Lactic, acetic, propionic, isobutyric, n-butyric and isovaleric acids can occur in silage as major
fermentation products, and their accurate quantitative determination is therefore important in
assessing silage quality. It is also useful to determine formic acid quantitatively, when this is used as
a silage additive.

In recent years many methods have been proposed for increasing the accuracy and speed of
determining these acids. Direct injection of the aqueous silage extract in gas chromatography avoids
esterification procedures that are very laborious and give rise to poor recoveries. With the volatile
acids from C2 to C5i, good results can be obtained, except for some ghosting problem using gas-
liquid or gas-solid separation.

Flame ionisation is unresponsive to formic acid and its detection by a katharometer is unsatisfactory.
Special problems arise with the estimation of lactic acid, whether by gas chromatography,
colorimetry or enzymic methods. With Chromosorb 101 the response, while sometimes acceptable
(recovery 90 - 5.8%), is often inconsistent or nil, depending on the manufacturer's production
batches; a similar situation occurs with PEG 6000 on TPA. Procedures involving the estimation of
lactic acid through its conversion to acetaldehyde, which is then detected either colorimetrically by
gas chromatography' have to be used with extreme caution because they can give misleading
results. It has in fact been shown that butane-2,3-diol and 3-hydroxybutan-2-one, occasionally
present in silage, are oxidised to acetaldehyde and are therefore reported as lactic acid. With poly
(ethy1ene glycol phthalic acid ester) on TPA (PEGPE 3000 on TPA) lactic acid can be separated, but
propionic and isobutyric acids are not well resolved. Enzymic methods employing L (+) and D (-)
lactate dehydrogenase (LDH) and UV determination give good results but it is necessary to
determine each stereoisomer.

H.p.l.c. resins of great efficiency are available for the separation of organic acids, as reported by
various authors for the analysis of wines, fruit juices,14 dairy product and other liquids of biological
origin. This has encouraged the authors to develop an accurate, simple and rapid procedure for the
simultaneous determination of organic acids in silage extracts

Experimental

2. Sample preparation
According to dry matter content, 60-120 g of chopped, fresh or frozen silage, was weighed in a 400
ml polyethylene bag and extracted with 250 ml of 0.05M H2S04 at 20C for 4 min in a 'Lab-Blender
Stomacher 400' (Seward Laboratory, London). The concentration of H2S04 was increased for
samples of high ash content and high buffering capacity to ensure a final pH within the range 2.0 -
3.0. The mixture, first filtered through cloth and then through Schleicher and Schull membrane filter
(BA-83 0.2 pm), was then ready for injection without further treatment.

2. Analytical method

The instruments used were: a model 8100 Liquid Chromatograph equipped with an autoinjector
incorporating a sample loop, a model 8440 variable wavelength ultraviolet visible detector and a
model 4100 computing integrator (Spectra Physics Ltd). A 300 x 7.8 mm i.d. h.p.l.c. Organic Acid
Analysis Column packed with Aminex HPX-87H and a guard column ( 40x 4 . 6m m) packed with
Aminex HPX-85H were employed (Bio-Rad Laboratories, Richmond, CA).

Analyses were performed isocratically under the following conditions: mobile phase 0.0025H~
2S04d egassed with He, flow rate 0.6 ml min-I, column temperature 41"C, injection volume 10 PI,
U.V. detection at 210 nm.

Four calibration standards were prepared by diluting a stock solution 20, 10, 5 and 3.33 times
respectively in 0.05M H2S04. This stock solution contained the following amounts (mg ml-l) of pure
acids (Hoechst AG): 18.01 lactic, 5.98 formic, 10.29 acetic, 9.85 propionic, 9.40 isobutyric, 9.50

Figure 1. Chromatogram of a standard solution containing the following acids (pg m1-l): (I) 1801
lactic; (2) 598 formic; (3) 1029 acetic; (4) 985 propionic; (5) 940 isobutyric; (6) 950 nbutyric; and (7)
911 isovaleric on an Aminex HPX-87H column. Chromatographic conditions were: mobile phase,
0.0025M H2S04; column temperature, 41C, flow rate, 0.6 ml min, injection volume, 10 PI: U.V.
detection at 210 nm

n-butyric, and 9.11 isovaleric. The external standard method was used to convert peak areas to
concentrations, and duplicate analyses were made. Standard calibration curves were defined as
y=a+bx, where y is the concentration of acid in ug ml-1 and x is the peak area in integrator units. The
slopes of these lines (b) are the response factors for each acid. Acids added in known amounts to
silage extracts were recovered, following the usual procedure, to test the overall efficiency of the
method. Duplicate analyses were performed before and after addition of the standard solution.
Both good and poor quality silages were used to cover a range of acid levels.

Results and discussion

Figure 1 shows the separation of the acids contained in one of the standard solutions used to
prepare calibration curves. The data for these lines, and the capacity factors (k') are given in Table
1. The excellent analytical precision obtained is shown by the low values of the y-intercept and by
the linear correlation coefficients (R).
The minimum detectable level (a peak height corresponding to twice the detector noise) varied from
30 ug ml-1 for isovaleric acid to 8 ug ml-1 for formic acid. This sensitivity, considering the dilution
used in preparing the sample, corresponds to as little as 0.4 and 0.1 g kg-' DM respectively.

Figure 2 shows the chromatograms of a good (a) and poor (b) quality grass silage. Under the
conditions adopted, which were optimised after several tests, the acids of interest were resolved
satisfactorily. Oxalic, malic, tartaric, citric and succinic acids were eluted before lactic acid, fumaric
acid and other unidentified components were separated without co-eluting or causing other
interference. Ultraviolet monitoring at 210 nm avoided interference from sugars and ethanol that
occur using refractometric detection.

The concentrations of volatile acids as determined by h.p.l.c. and by gas chromatography with
Chromosorb 101 were in agreement using the same two silage extracts (Table 2). The figures for
lactic acid were lower using gas chromatography, probably because of adsorption to the porous
polymer. The results of the recovery test confirmed the effectiveness of the technique (Table 3).

The Bio-Rad HPX-87H column has been used periodically to analyse about 1000 samples of silage
over a period of one year. No deterioration of separation capabilities was noticed. Although neither
increase in back-pressure nor loss of resolution were observed, the cartridge HPX-85H was replaced
every 300 injections, as a precautionary measure. In conclusion, the method combines simplicity
and minimum sample preparation with adequate speed and precision for the simultaneous analysis
of volatile carboxylic acids (C1-C5i) and of lactic acid in silages.