This section is dedicated to the analytical methods used to investigate chemicals which typify the flavour of this vegetable. In this page the most common analytical techniques, which have been used during last fifty years, are described in summary. An historical approach helps us to point out the troubles found in guessing the structures of these VOCs and the developments contrived to do a good analytical chemistry.



At the beginning...

Colour reactions for thiosulphinates

Gas Chromatography: introduction

Gas Chromatography: mechanisms of formation of sulphide derivatives

HPLC: introduction

HPLC: reversed-phase

HPLC: normal phase

HPLC: ion-pair reversed-phase

HPLC: LC-MS



At the beginning...

The antibacterial properties of fresh garlic extract were first recognised by Pasteur in 1858 and even as early as 1892 Semmler had managed to obtain a 0.1% - 0.2% yield of volatile oil from garlic cloves by steam distillation. In 1935 Platenius developed a method of flavour estimation whereby the sulphides in oil distillates could be oxidised to sulphate in bromine water, excess bromine driven off by heating and the sulphate precipitated with barium chloride and determined by weight. Ten years later this method was improved upon by Currier who used lower reaction temperatures and converted the sulphur ultimately to methylene blue, thereby effecting a colorimetric determination. Although the use of total sulphur as a measure of garlic pungency was developed further by Kohman and Farber, it was the early work of Cavallito et al 1 in determining the chemical structure of allicin and its reaction with cysteine:

2HSCH2CHNH2COOH + RS(O)SR' → RSSCH2CHNH2COOH + R'SSCH2CHNH2COOH

and the discovery and characterisation by Stoll and Seebeck of alliinase reaction that pioneered the chemistry of the Allium species of the last fifty years.2

Back to the index



Colour reactions for thiosulphinates

Early investigations in Japan into the formation of thiosulphinates in garlic led to the development of a (paper) chromatographic technique that relied on the development of an orange-red colour formed by a reaction with alkaline sodium nitroprusside. This early colour reaction was limited by its inability to distinguish between thiosulphonates and thiosulphinates and that the colour development was not permanent. At almost the same time, during a course of experiments in the USA with homocysteine peptides it was discovered that, when the products of a reaction between N-ethylmaleimide and thiols are made alkaline, a red colour develops. Under the correct conditions, the reaction was found to be extremely sensitive and the stability of the colour made it suitable for the visualisation of thiols or thiol esters separated by paper chromatography and for colorimetric determination.

Although this colour reaction was extremely sensitive, it was found to suffer from interference from disulphides and it was subsequently modified by using N-ethylmaleimide in isopropyl alcohol followed by potassium hydroxide in the same solvent. This combination yields pink to red colours with thiosulphinates while disulphides remain colourless. Further improvements in extraction procedures (with diethylether) and colour stability (with ascorbic acid) have produced a reliable method for the colorimetric determination of thiosulphinates. 3

Back to the index



Gas Chromatography: introduction

The first reference to gas chromatography (GC) being used in the field of garlic research was in 1961 when Carson 4 used the technique to separate alkyl di- and tri- sulphides for further identification. GC appears to have been used sparingly throughout the 1960's, presumably because it was a new technique and equipment availability was limited, but as early as 1964 Saghir et al had warned that many of the compounds from allium species detected by GC may be 'artefacts of analysis'. Because of their excellent resolution and mass identification capabilities, GC and GC-MS have continued to feature prominently in the efforts to characterise allium volatiles but although these tools are of great value in the study of compounds of moderate thermal stability such as those found in distilled oils, thiosulphinates from allium species are known to decompose on heating or attempted GC analysis.

Allicin is quite reactive and unstable, hydrolysing on heating in water to give diallyl disulphide, diallyl trisulphide and the corresponding polysulphides and it was an attempt by Brodnitz et al 5 in 1971 to determine allicin by GC-MS. He also provided evidence for an unusual mode of decomposition; they indicated that GC caused diallyl thiosulphinate (allicin) to dehydrate affording a 2.4:1 mixture of two compounds (isomers 3-vinyl-3,4-dihydro-1,2-dithiin and 3-vinyl-3,6-dihydro-1,2-dithiin).

Despite this and other contemporary studies, many authors continued to draw conclusions based on products formed by GC analysis of garlic preparations using injection port temperatures as high as 280C. Further work on the decomposition of allicin confirmed the formation of two C6H8S2 isomers, later being correctly identified as 3-vinyl-3,4-dihydro-1,2-dithiin and 2-vinyl-2,4-dihydro-1,3-dithiin (Figure 1).6

Figure 1. The thermal decomposition of allicin

Further interesting chemistry is associated with the major disulphides from garlic, i.e. diallyl disulphide and 1-propenyl disulphide. Upon heating, diallyl disulphide undergoes a sequence of complex reactions leading both to diallyl polysulphides as well as a series of acyclic and heterocyclic compounds resulting from the generation and reaction of thioacrolein and radical species CH2=CHCH2Sn with precursor diallyl polysulphides. Since most of these compounds have been identified in distilled garlic oil, they may be assumed to arise from the thermal breakdown of diallyl disulphide. A comparison of the results of HPLC and GC analyses however leads to an unusual observation: simply because a product has been distilled does not guarantee that it will survive GC analysis. As will be seen later, all of the compounds identified are breakdown products of the primary flavour compounds (thiosulphinates) of garlic and are typical of the results of GC analysis of both garlic oil and extracts which undergo thermal decomposition during analysis.

Back to the index



Gas Chromatography: mechanisms of formation of sulphide derivatives

The formation of the thermal degradation compounds found in garlic oils (and in GC analyses of garlic extracts) is explained by Block 6 by a sequence involving:

a C-S homolysis of diallyl disulphide followed by a reversible terminal and internal addition of the allyldithio radical to diallyl disulphide
an intramolecular hydrogen atom trasposition in the intermediates (formed by an internal addition of the allyl-dithio radical), giving thioacrolein and alk(en)yl radical
Diels-Alder self-condensation of thioacrolein acting as an heterodiene and its condensation to allyl mono-, di-, and trisulphides.

In Figure 3 6 the self-condensation of thioacrolein affords two vinyldithiin isomers and the condensation with a third molecule of thioacrolein affords the trimers.

Figure 3. Diels-Alder self-condensation of thioacrolein

Another interesting result is the formation (Figure 4) 6 of compounds by the thermal degradation of diallyl disulphide. Figure shows how diallyl disulphide undergoes thermal degradation to allyldithio and allyl radicals. These radicals react to form diallyl disulphide and some high molecular weight compounds with two to five sulfur atoms.

Figure 4. Thermal degradation of diallyl disulphide and addition of allyldithio radicals to diallyl disulphide

It is clear therefore that GC techniques which have commonly employed high injector port temperatures have not given a true picture of the primary flavour compounds present in garlic tissue and that many of the compounds detected by GC-MS over the last twenty-five years were 'artefacts of analysis'. By the late 1980's HPLC studies had shown the primary flavour compounds of garlic to consist almost exclusively of thiosulphinates and in 1992 a definitive study of the effect of GC conditions on thiosulphinates was undertaken. 7

Initial experiments with GC-MS were conducted with synthetic samples and used a 12 m x 0.2 mm capillary column with on-column injection but it was soon noted that narrow bore columns rapidly lost resolution, presumably due to the deposition of non-volatiles, rendering them useless. 8

Further studies showed that the use of wide bore (0.53 mm i.d.) columns did not suffer the same limitations and gave excellent resolution of most C2 - C6 thiosulphinates. As was expected, allied to the need for a wide bore column was the use of low GC temperatures, the GC injector and oven being initially cooled to 0C and the GC-MS transfer line to 100C which ensured minimum degradation of samples. 9

Back to the index



HPLC: introduction

During the late 1980's and early 1990's when GC-MS results were being subjected to increasing scrutiny, HPLC techniques capable of identifying the primary flavour compounds of garlic extracts without thermal degradation, were being developed. Although some weaknesses exist with the HPLC methods -incomplete separation of some peaks, variable retention time leading to possible misidentification of peaks and compounds having minor UV activity being overlooked - techniques are now highly developed and capable of presenting a true picture of the primary flavour compounds of allium species.

HPLC analysis of allicin in garlic extracts was first reported in 1985 by Miething, who analysed diethyl ether extracts of garlic and garlic products by normal phase (Si) HPLC. This method suffers from the fact that allicin in very unstable in ether and other organic solvents that are necessary for Si-HPLC and in 1987, Jansen et al first reported on reverse-phase (C18) HPLC analysis of allicin in aqueous garlic extract. 10

Back to the index



HPLC: reversed-phase

The separation, quantitation and variation in the amounts of all the detectable thiosulphinates present in garlic clove homogenates was first reported in 1990 together with a standard method for the quantitation of allicin using an external standard. 11 In this section all possible thiosulphinate combinations were synthesised and Figure 5 2 shows that most of them were separable from each other. The separation of the allyl methyl/methyl allyl and methyl 1-propenyl/1-propenyl methyl pairs was however poor but this was subsequently improved upon by normal phase HPLC.

Figure 5. C18-HPLC separation of thiosulphinate standards

With this method sample preparation was relatively simple and filtered, aqueous homogenates were injected directly into the HPLC. The mobile phase for C18-HPLC consisted of a 50/50 mix of methanol and water and direct injections of methanol were frequently employed to extend guard column life. It was known that allyl propyl thiosulphinate co-elutes with 1-propenyl allyl thiosulphinate and the only thiosulphinate expected, but not found, was di-1-propenyl thiosulphinate. This compound was discovered to be highly unstable and has been shown to rapidly form two dimethyl dithiabicyclohexane oxides called zwiebelanes. This method was further extended by Lawson et al to identify and quantify a number of thiosulphinate breakdown products including sulphides, vinyl dithiins and ajoene.

Back to the index



HPLC: normal phase

As a result of comparative studies, Block et al concluded that the best peak resolutions were obtainable using normal phase HPLC with hexane/isopropanol (95:5) gradients (Figure 6). 2 Aware of the instability of thiosulphinates in organic solvents Block's work concentrated on extraction and distillation procedures and comparative assessments. By careful planning of sample preparation, rapid extraction and analysis he was able to present a method comparable in stability to C18-HPLC methods but with the high resolution of Si-HPLC.

Figure 6. Normal phase HPLC of dichloromethane extract of aqueous garlic homogenate

The preparation of fresh extracts for Si-HPLC can present a number of problems particularly with emulsion formation and the presence of plant pigments and waxy materials. This work experimented with distillation and extraction procedures and was able to provide methods with excellent quantitative and qualitative agreement. Distillation was performed with high vacuum at room temperatures and aqueous condensates were collected at -196C. It was found that HPLC and NMR spectroscopic analysis of the CH2Cl2 extract of the salt-saturated condensate gave good qualitative thiosulphinate composition profiles. It is believed that this method of 'room temperature distillation' succeeds because of the stabilising effect of water, through hydrogen bonding, of the thiosulphinates. Extracts of aqueous homogenates were again undertaken with CH2Cl2 and performed quickly and at low temperatures: all analyses were undertaken within 30 minutes of extraction and reproducibility was excellent.Both Si-HPLC and Cl8-HPLC rely on the availability of external standards for accurate identification and quantitation of individual compounds. In commercial applications it is the quantitation of the principal thiosulphinate, allicin and its precursor alliin, that is important and to that end a number of methods for the preparation of pure standards of these two compounds have been published. Mayeux et al describe a simple method whereby allicin can be synthesised by oxidising diallyl disulphide with acidic hydrogen peroxide and purified using a Si-TLC plate developed in hexane/ethyl acetate immersed in ice water. Alternatively pure allicin can be isolated by Si-TLC of the ethyl acetate extract of an aqueous solution of a quality commercial garlic powder or an homogenate of fresh garlic. (If fresh garlic is used then subsequent C18-TLC is necessary to separate the 1-propenyl allyl thiosulphinate which co-elutes with allicin on Si-TLC). 12

Back to the index



HPLC: ion-pair reversed-phase

Recent work has led to the development of a method for the simultaneous determination of alliin and allicin by ion-pair reversed-phase LC. The preparation involves the drying of fresh garlic slices followed by extraction with 80% methanol. The methanol is subsequently concentrated in a rotary evaporator and then partitioned with diethyl ether. The diethyl ether layer is discarded and the aqueous layer loaded onto a pre-equilibrated ion-exchange resin and allowed to drain. The resin is washed with water until washes are neutral with litmus paper and then the alliin retained on the resin is eluted with ammonium hydroxide. After freeze drying the alliin is crystallised from 70% ethanol and repeated recrystallisation gives white needles of pure alliin. Alliinase is extracted from fresh garlic homogenates and its action on alliin yields pure allicin. Both alliin and allicin prepared in this way are used as external standards for the simultaneous quantitation of the compounds in garlic samples. 13

Fig 7. Chromatogram of methanol extract of fresh garlic

Back to the index



HPLC: LC-MS

During the past eight years great efforts have been made to combine the 'gentle' separation afforded by HPLC with the powerful identification capabilities of MS. Whilst early work gave promising results, it was some time before a successful method for use with alliums was published. Ferary et al used LC-MS to examine garlic odours to determine whether or not they contained degradation compounds of thiosulphinates. Using a cryotrapping technique they made direct injections of aqueous solutions into the HPLC which was fitted with a UV detector and coupled to a mass spectrometer. A number of different commercial coupling and ionisation systems were tested and although best results were obtained using atmospheric pressure chemical ionisation (APCI), some degradation of thiosulphinates was observed.

Calvey et al have most recently reported on an improvement to the above method using reversed phase LC-MS and LC tandem MS (LC-MS-MS) with APCI coupling. All of the major thiosulphinates were readily characterised by this technique (Figure 8) and for the first time trace amounts of propyl compounds were found in garlic (the low UV absorbance, low concentration and thermal lability of these compounds had made previous detection by other means difficult). 13

Figure 8. Total ion chromatograph of typical garlic extract

Back to the index