A review of malting and malt processing for whisky distillation

A review of malting and malt processing for whisky distillation The science and technology of brewing and of malt and grain distilling are remarkably similar. Indeed, it has been said that malt whisky is simply a distillation of unhopped beer and that grain whisky is based on an adjunct mash containing 90% of unmalted grain instead of the 20% that is common in many styles of beer. However, there are now major differences in processing cereals in the distilling industry, which not only disprove this mantra, but also challenge some of the fundamentals underpinning brewing science. The most obvious example of this is the derivation of spirit yield. One of the most reliable statistics available to UK distillers is the historical returns made to Her Majesty's Customs and Excise declaring the actual volume of spirit produced and the ‘degree of attenuation’ achieved. For Scotch malt distilleries, the declared wort fermentability has been between 86 and 87% for at least 150 years. If the above ‘brewing’ mantra were valid, then the achieved fermentability would have been around 73%, as in a fully attenuated lager wort. Despite this obvious fact, there is no published scientific rationale that can explain why these attenuation limits are so different. However, is 87% the maximum attenuation that can be achieved from an all malt distilling wort? This was the question that scientists working in the Distillers Company research laboratories addressed over 40 years ago. They devised a simple test that involved mixing a finely ground sample of malt with water at ambient temperatures and fermenting the resulting mash directly with a distillers' yeast. The alcohol from the fully fermented mash was then distilled off and the alcohol content calculated from the specific gravity of the distillate 1. They found that the attenuation limits were generally ~90%, depending on the malt quality. This method of assessing the maximum fermentability of malt extract, without prior gelatinization of starch, was based on the fact that, if given long enough, the enzymes of malt can completely hydrolyse granular starch. The method was never published as an analytical technique, but it was the basis for a patented distilling process that was successfully scaled up to full production levels 2. Although the spirit quality was markedly different from normal, the spirit yields were declared at 430 L of alcohol per tonne of malt mashed (L alc./t) at 5% malt moisture, some 5% higher than predicted from conventional mashing and equivalent to a wort fermentability of 91%. This patented process for low‐temperature mashing and fermenting (usually termed ‘all‐grains‐in’ fermentation) has also been used for unmalted cereal with great success. By combining the mashing and fermenting processes, the malt enzymes have up to 72 h to hydrolyse granular starch compared with ~3 h by conventional mashing of gelatinized starch. Therefore the same malt can have a maximum fermentability of 73% when used in brewing, up to 87% when mashed in a conventional malt distilling process, and 91% when mashed and fermented at ambient temperatures. These values are almost constants but, to the date of writing, there is no specific scientific explanation for their distinctive end‐points. Furthermore, they clearly demonstrate that malting and mashing to achieve maximum spirit yield are markedly different from the objectives of brewing and that a better understanding of the biological and biochemical transformations that take place during malting and the subsequent processing of both malted barley and raw cereals is now required. Therefore, the purpose of this review is to highlight the anomalies between current best practice in distilling and many of the basic assumptions derived from brewing science. Many of these anomalies can be partly explained by a re‐examination of the existing research, much of which is focussed on brewing, and reinterpreting the results from a distilling viewpoint. This can reveal quite a different picture. Therefore the second objective of this article is to highlight where further research is required to answer some of the questions posed in the above examples. Since the purpose of malting barley for distilling is, by definition, to yield the maximum amount of fermentable extract, which will produce a spirit with the ‘essential flavour and aroma characteristics of malt whisky’, we will start by looking at how fermentable extract develops during the malting process and how we measure it. If the maltster has a batch of viable barley, of an approved variety with a protein content of ~10% and it has been screened over a 2.5 mm sieve to achieve a thousand corn weight of 40 g, what are the parameters that will define maximum fermentable extract? Normal practice would be to achieve an out‐of‐steep moisture of at least 45% and to germinate in a flow of fully humidified air (100% RH) at 14–16 °C for 5 days. This schedule would cater for the largest grain of the highest protein in a normal distribution. In temperate climates, such as Scotland, where the average day/night annual temperature is only 10 °C, this is not too difficult but there is still a very fine balance in choosing when to go to kiln. A typical profile of extract development for the malting barley described above is depicted in Fig. 1. If malting were to continue from 5 to 7 days, to exaggerate underlying trends, we can see that there is quite a different pattern to maximizing fermentable extract 3-5 as compared with maximizing total extract 6. The conventional laboratory analyses of measuring total achievable extract by fine grinding (setting 2 on the laboratory mill) and the readily available extract by coarse grinding (setting 7) indicates that optimum conditions are achieved after 5.5 days of germination when the fine/coarse extract difference (F/C Diff.) 6 minimizes at 0.6–0.8%. In other words, a typical distillery specification of 82% coarse grind extract and a fermentability of 86.5% would be achieved after 5.5 days of germination. There would be no point in continuing to malt any longer since the F/C Diff. has reached its minimum and both extracts decline in parallel for the next 2 days. However, when we look at the pattern of wort fermentability over the same period, a different picture emerges. Wort fermentability peaks at 87% after 4.5 days and declines for the remainder of the malting period, therefore the maximum possible fermentable extract (FE), and hence spirit yield, can be achieved half a day earlier when the F/C Diff. is 1.8% instead of 0.8%. When a laboratory analysis is described as a measurement of ‘spirit yield’ there may be a misconception that this infers an absolute measurement of the concentration of ethyl alcohol. Except where alcohol content and distillery wash OG are measured by a direct GLC/densitometer technique 12, all of the standard methods of analysis for extract, fermentability and spirit yield are anything but ‘absolute’. They merely infer values from the measurement of the gravity of various solutions, each containing a large number of constituents with different specific gravities. The best estimate of spirit yield is therefore obtained by distillation of a measured volume of fermented wash. When the volumes of both the distillate and the residual spent wash are restored to their original values using distilled water, the specific gravity of the alcohol solution and that of the residue can be measured. Statutory tables are then used to convert the specific gravity of the alcohol fraction into equivalent degrees of gravity lost during fermentation and, when this value is added to the gravity of the residual fraction, an estimate of the original gravity of the fermented wort and the wort fermentability can be calculated using the above formula. It is also necessary to emphasize the difference between measuring real attenuation and apparent attenuation at this point. In all of the following discussion on wort fermentability, we shall consider the concentration and constituents of residual gravity as much as fermentable extract, therefore it is essential to differentiate the two ways of measuring fermentable extract. The distillation method for measuring FE and spirit yield is accurate but it is a cumbersome and time‐consuming form of analysis. The faster, but less accurate, method for measuring apparent fermentability has therefore become accepted 3 as the standard method in the UK. In this method the OG is taken from the standard method 6 for measuring coarse extract in the laboratory, and after fermentation of the extract, the final gravity (FG) is measured. In this case spirit indication tables cannot be used because they are specific only for alcohol–water solutions, therefore a correction factor (once called the solution divisor), which is itself a variable, has to be used to convert apparent fermentability to an estimate of real fermentability as follows: Since Real Fermentability(Fr) = OG–RG/OG × 100 and Apparent Fermentability(Fa) = OG–FG/OG × 100 then Solution Divisor(SD) = Fr/Fa = OG – RG/OG–FG = 0.814 to 0.820 depending on wort composition. The Solution Divisor is a conversion factor which is not constant and changes with the relative amounts of alcohol and the composition of the residual gravity 3, 4, but for a tightly specified malt of low soluble nitrogen the chosen value of 0.814 does give a good correlation with true Fr. Therefore, for brewing wort with a high residual dextrin concentration and low soluble protein 12, the factor may be 0.819. Unfortunately, this does not allow for protein with a much higher specific gravity than carbohydrate, nor that the concentration of soluble protein may vary considerably in the fermented wort. While the standard method can give a fair estimate of fermentable extract for most distilling malts 4, it can be inaccurate for highly peated malts as described above, highly modified diastatic malts and malts that may have a high sulphur content. All of these types of malt have a much higher content of solubilized protein, which will therefore distort the conversion factor as shown by the following case study from actual commercial practice. The malt in question was highly ‘peated’, with slow smouldering peat, to achieve the highest possible concentration of phenols. The malt kiln was direct fired with heavy fuel oil containing up to 3% sulphur. The peat also contained 1% sulphur. The concentration of sulphur dioxide on the finished malt was 40–50 ppm. This resulted in pH values of around 4.8 in the mash leading to increased proteolysis, since this value is the pH optimum for carboxypeptidase. The malt already contained a high concentration of soluble protein owing to the ‘stewing’ effect mentioned above, so it was not surprising that the Laboratory Fermentability was 85.9% and so deemed ‘out of specification’. These values were the result of using the FG method and a conversion factor of 0.814. However, when this malt was mashed in the distillery, the declared attenuation (Fr) was 86.4%. It only takes simple arithmetic to show that a conversion factor of 0.819 (the recommended brewing factor to allow for higher residual gravity) would have given the same degree of attenuation in the laboratory analysis. The reliability of the standard method for determining malt fermentability, and hence the predicted spirit yield (using the formula PSY = 6.06 × FE, where the correlation coefficient of 6.06 is again empirically derived) is well documented 5, but it must be emphasized that this was for one specification of malt, within one group of distilleries, and that the PSY, being only a correlation with declared spirit yields, can be variable across the whole spectrum of distilling malt. This is demonstrated in recently published data 14 (Fig. 5) and shows that the correlation between PSY and actual distillery yield, while still significant, is much more variable than might be expected (NB, the correlation coefficient of only 0.4). The other conclusion that can be drawn from the data in Fig. 5 is that the actual distillery yields, in the majority of the cases, were significantly higher than the PSY (by approximately 5 L alc./t and are therefore close to the values expected from the fine grind FE of a well‐modified malt with an F/C extract difference of ~1%. If there is doubt about the reliability of quoted figures for fermentable extract and PSY, then the malt soluble nitrogen ratio (or preferably the Kolbach index, KI) should be examined. The latter method is more indicative of variable RG since the results are more in line with the soluble protein values found in practice, because of thicker distillery mashes. There is also a direct method for estimating the potential RG in distillery wash and this can give a better indication of potential variations in fermentability. Measurement of the cold water extract (CWE) provides an estimate of the amount of free sugars, soluble protein, peptides and amino acids, soluble glucan and pentosan and other non‐fermentable solutes in the malt 15. Assuming that the sugars and amino acids will be utilized by yeast and that their concentration is constant, then the other water soluble constituents will form the bulk of the RG. This simple analysis is still part of the method for determining amylase activity [diastatic power (DP) and dextrinizing units] in malt but has fallen out of favour as a malt parameter in its own right. This is unfortunate, since it has been shown that there is a strong inverse relationship between CWE and fermentability for the reasons given above 8. It is therefore vital that the quality of distilling malt is depicted by all of the specified parameters and that PSY is taken only as an estimate of the minimum yield that can be achieved in practice, given the above caveats. Nevertheless, the standard method for estimating spirit yield in UK malt is probably more reliable at present 4 than at any other time over the last 30 years for the following reasons. Firstly, there are now very few directly fired kilns still in operation in which heavy fuel oil is burned. The specification for heavy fuel oil allows for a sulphur content of up to 3% and at this level, as in the example described above, the concentration of sulphur dioxide in the malt can be 30–50 ppm with a concomitant drop in wort pH. Similarly, the need to burn rock sulphur in gas fired kilns to combat the formation of nitrosamines 16, 17 is no longer a necessity since nearly all maltsters have converted to some form of indirect heating using heat exchangers 18. Therefore the ‘soluble nitrogen’ effect, brought about by enhanced proteolysis in the mash, is no longer a concern except in highly peated and enzymic malts. Secondly, in the UK there are a series of new malting varieties that have intrinsically lower soluble nitrogen ratio (SNR) values. Shortly after the standard method for measuring fermentable extract became established, the variety Triumph was the dominant malting variety grown in the UK. This cultivar, and several other subsequent crosses, all displayed a higher level of proteolysis during malting, with SNR values between 42 and 45%. The standard method had been calibrated against unsulphured malt made from Golden Promise with SNRs in the range 36–38%, thus the conversion factor of 0.814 made it extremely difficult to achieve specified fermentability with Triumph barley, especially at higher total nitrogen concentrations. After the demise of Triumph, new breeding lines resulted in varieties at the other end of the spectrum. The first of these was Chariot with typical SNRs in the range of 35–36% and all subsequent cultivars have had similar values giving high apparent fermentability figures. Such traits are obviously generic in British malting barley but the trend may be in the opposite direction in other countries. In general, attempting to achieve maximum modification from poorer varieties with higher nitrogen levels may be counter‐productive because of the higher malting loss and the ‘soluble nitrogen effect’ may also distort estimates of PSY. Therefore, distillers who wish to specify fermentable extract and PSY in their specifications should check their method of analysis to see if the ‘solution divisor’ requires re‐calibration for the type of malt required. This can be accomplished by using the distillation method for Fr and then deriving a specific solution divisor for the quick method (Fa). It was necessary to describe in detail the development of fermentable extract, and how it is measured in absolute terms, to provide the context and framework for the following discussion on the original questions posed above, viz. ‘Why does the same malt have an apparent fermentability of 73% when used for brewing, always attenuates to 87% in normal distilling practice but never yields its maximum potential of 92% and what happens in a malt mash at temperatures below the starch gelatinization temperature?’ The conclusion that can be drawn from the above observations and models of malting is that real fermentability values for unboiled distilling malt wort are dependent on (a) the total amount of fermentable carbohydrate that can be extracted and (b) the variable amount of residual gravity, especially if its soluble protein content is high. However, these variations in fermentability and fermentable extract are minor compared with the difference in the above three values of wort fermentability. We therefore have to look at the three critical differences in mashing listed below and what effect they would have on the enzymolysis of malt starch to derive an answer to our questions: for wort boiled immediately after run‐off; for wort from a conventional distilling infusion mash at say, 65 °C; for wort from an ‘all‐grains‐in’ mash/fermentation at ambient temperatures. For many years the enzymic degradation of starch in a conventional malt mash has been attributed to the action of only α‐ and β ‐amylase and the other two known diastatic enzymes, limit dextrinase and α‐glucosidase (maltase), were thought to have little or no effect on the overall fermentability of the derived wort 19, 20. An all malt brewing wort was reported to contain as much as 25% of the original starch in the form of dextrins ranging from a degree of polymerisation of 4 to over 21 21, 22. These definitive studies by Enevoldsen and Schmidt 21, 22 on the structure of brewing wort dextrins were carried out using a refined Sephadex dextran gel chromatographic system. When worts from simulated distilling mashes 8 were chromatographed, at the equivalent concentrations and on the same apparatus, a completely different pattern of dextrinization emerged. Because there was little evidence of higher dextrins in the distilling wort, it was assumed that they had been debranched by limit dextrinase, thus increasing the relative concentration of maltotriose and maltotetraose. That shift in the dextrin profile is demonstrated in Fig. 6(a) by superimposing a chromatogram of distilling wort onto the dextrin profile of a lager beer at the same column loading. The wort from this highly enzymatic malt (DP = 170) was unboiled and had been incubated for 36 h in the presence of a strong bactericide to prevent interference from the growth of natural malt microorganisms. However, re‐examination of the chromatographic data of this distilling wort, and others produced by Bathgate et al. 8, reveals a more complex pattern of starch hydrolysis. The laboratory method 23 used was one that replicated a traditional Scotch whisky distillery mash in that three separate malt mashes were made at progressively increasing temperatures (65, 77 and 88 °C). Each mash was centrifuged after a 1 h stand and the wort cooled to ambient temperature before being added to the previous centrifuged wort. [This method is still used as a laboratory standard 4 for producing distilling worts for analytical purposes.] Some of the distilling wort was inactivated by heating to boiling point after final collection (i.e. 3 h after the first mash) and was then compared with active wort incubated at 30 °C in the presence of sodium azide for a further 36 h to simulate the conditions for a standard laboratory fermentability analysis 3. The results of these HPLC profiles have now been quantified by integration of the original graphs (Fig. 6a–c and Table 2). On re‐examination of all of the HPLC analyses of these worts, and of the fermented wash derived from them, it was noted that the relative amounts of excluded unfermentable polysaccharide were constant. This provided a useful internal marker to quantify the relative amounts of fermentable sugars and the residual dextrins. Had this been done at the time of publication a different conclusion would have been drawn. Firstly, there was only a very small increase in fermentable sugars from the time final worts were collected and 36 h later (Table 2 and Fig. 6c). The only significant change was the reduction in higher dextrin to true limit dextrins with a degree of polymerisation of 4–7 glucose units (Table 2 and Fig. 6a and b). Secondly, comparison of the profiles between wort, which had been incubated for 36 h, and the fermented wash derived from the same wort clearly shows that limit dextrinase was only active in the fermenting wash (Table 2 and Fig. 6b). The third oversight was that the fractions deemed to be ‘maltotriose’ were, in fact, mixtures of true maltotriose and unfermentable trisaccharides which remained present in the wash despite being fermented with DCL ‘M’ yeast, a known maltotriose fermenter 12. A similar result had been reported in a brewing context a few years previously 24. The residual trisaccharides in the brewing wort, which had been fermented with a maltotriose fermenting strain of yeast, were identified as panose and isopanose, that is, 62‐α‐glucosyl maltose and 6‐α‐maltosyl‐glucose 25. Table 2. Chromatographic data from HPLC of distilling wort and fermented wash dextrins and equivalent concentrations (arbitrary units) of fermentable sugars (Fig. 6a–c) Glucose Maltose DP3 DP4 DP5 DP6 DP7 DP8 DP > 8 Total Wort (3 h) 660 2700 400 90 20 15 10 4 36 3935 Wort (36 h) 930 2700 375 180 102 60 18 10 80 4455 Wash 0 0 184 66 72 38 14 4 11 389 Data from Bathgate et al. 8. These misconceptions about dextrin formation and the action of limit dextrinase are only now evident because of the rigorous investigations carried out more recently by research teams at the Scotch Whisky Research Institute (SWRI). They have examined the role of distilling malt saccharifying enzymes in much more detail 26, 27 and have confirmed the importance of limit dextrinase in producing a more highly fermentable wort in distilling practice. Prior to this research, it was assumed that, as in brewing malt, limit dextrinase had little effect on the composition of wort 22 during mashing. Using a high‐performance anion exchange chromatographic system, Bringhurst et al. 27 at the SWRI not only identified dextrins with a degree of polymerisation of 4–7 glucose units in a distilling malt wort but also separated the linear from branched isomers in this range. The purified fractions with 6 and 7 glucose units were treated with pullulanase in the same manner that Enevoldsen and Bathgate 28 had used to elucidate the structure of the equivalent brewing wort dextrins. These digests, together with further characterization by sophisticated mass spectrometry, clearly showed that the branched dextrins with a degree of polymerisation of 6 and 7 in a distilling wort were nearly all isomers of either maltose or maltotriose attached by α‐1,6‐glucosidic linkages to either maltotriose or maltotetraose. However, there were very small amounts of residual 6 and 7 unit dextrins, which were not degraded by pullulanase and which may be of some significance. The postulated structures for the hydrolysed dextrins therefore complied with a proposal by Vinogradov and Bock 29 that all α‐1,6 branched dextrins of 10 glucose units or less have maltose or maltotriose as the side chain, thus making them amenable to hydrolysis by pullulanase or limit dextrinase. Bringhurst et al. 27 inferred from this theory that the fractions containing 8 and 9 glucose units that they had identified must, therefore, have an analogous structure. The SWRI team also demonstrated 26 that the majority of dextrins were formed during mashing and that, during the early stages of fermentation of an unboiled wort, there was continued accumulation of true limit dextrins by further α‐amylolysis of higher dextrins. What was most revealing in this latter work was the action of limit dextrinase in hydrolysing most of the branched dextrins, but not until the fermentation was well under way. By monitoring the concentrations of 6 and 7 unit branched dextrins during the course of fermentation, they clearly demonstrated that the relative amounts of each of these fractions more than doubled during the first 10–15 h of fermentation, when α‐amylase was at its highest activity and then declined to around 40% of their original concentrations over the next 20 h, when limit dextrinase rose to its maximum activity. It is interesting to note that the concentration of the residual dextrins then remained constant until the end of a 60 h fermentation. The exact mechanism whereby limit dextrinase remains active during the first 36 h of fermentation is not entirely clear, but it would appear that there is continuous release of ‘free’ limit dextrinase from the bound form as described below. The implication is that all of the components of this fraction that can be debranched are hydrolysed during the first half of fermentation when all true limit dextrins have been formed by post‐mashing α‐amylolysis. It is now well established that limit dextrinase is present in malt, mostly as an inactive form linked to an inhibitor 30, 31. However, the theory that limit dextrinase must thereby be inactive during mashing has been disproved by the revelation that its proteinaceous inhibitor can be uncoupled by a cysteine proteinase 32 in malt and that some free limit dextrinase activity can be present in wort 33. What the definitive research at SWRI has proven is that limit dextrinase is not heat inactivated during mashing but can pass into the wort to contribute significantly to the increase in fermentability during the middle stage of fermentation. So whether free or in a bound form, what is now proved both by the SWRI work and the revision of the earlier work by Bathgate et al. 8 is that limit dextrinase only becomes fully active in a fermenting wort. The principal enzyme activity in unboiled wort, even when incubated 36 h after wort separation, is that of α‐amylase 34 since there was no appreciable increase in total fermentable sugars (Table 2 and Fig. 6b). The inference is that β‐amylase may only be active during mashing, or in the early runnings of cooled wort to the fermenter, because there was no overall increase in maltose (Table 2). However, there may have been some turnover in maltose and maltotriose by α‐glucosidase (maltase) since there was a small but measurable increase in the glucose concentration (Fig. 6c and Table 2). The initiation of debranching activity during fermentation is related to the drop in pH from around 5.0 in wort to about 4.5 in an actively fermenting wash 35. While α‐amylase is most active at pH values between 5.0–5.5, the debranching enzyme in fermenting wash has an optimum pH of 4.2–4.4. The conclusion is that, while ‘free’ limit dextrinase may be present in wort, it is the bound form, uncoupled from its inhibitor, which becomes active in the fermentation of an unboiled wort, when yeast activity reduces the pH to the latter values 36. These results can also be substantiated by now returning to the model of malt extract 37 depicted in Fig. 4. It was postulated that 59 g of the total extract of 74 g consisted of starch. If we make another simplification and estimate that the amylopectin content of malt starch is 80%, with an average chain length of 25 glucose units 38, 39, then this would result in a molecule with four α‐1,6‐glucosidic linked branch points for every 100 units in α‐1,4‐linked chains. If we now assume that the absolute limit of α‐amylolysis around a branch point is two glucose units 29, 40, that is, 63‐α‐maltosy‐maltopentaose, as postulated by Vinogradov and Bock 29 then we can envisage the simplest structure for amylopectin as being one with an equal number of A:B chains 38, 39 and six glucose units around each of the four branch points as shown in Fig. 7. In other words 7 × 4 glucose units in every 100 in the amylopectin molecule would form limit dextrins. If we also assume that all of the amylose can be hydrolysed by amylases to fermentable sugar, then 20% of the original 59 g of malt starch would yield 11.8 g of maltose and maltotriose. That would leave 80% of the starch as amylopectin of which 28% would be dextrin, assuming that there was no further hydrolysis by limit dextrinase and glucosidase. The outer chains of amylopectin could therefore yield 72% of fermentable sugars. Thus 47.2 g of amylopectin would yield 34 g of fermentable carbohydrate (72% of 47.2 g). This theoretical composition of fermentable extract is summarized in Table 3 for two scenarios: one in which all of the amylopectin branch points are intact (i.e. no de‐branching) and a second in which all of the branch points are hydrolysed by limit dextrinase. Since the model dextrin used was the smallest postulated, then the calculated fermentability of 74% represents the highest value expected for a brewing (boiled) wort, fermented with a maltotriose fermenting yeast, and in which there has been no debranching activity. The range of fermentabilities reported when DCL ‘M’ yeast was recommended for a standard brewing wort fermentability method 12 was 73.4 ± 0.9%, so matching the above theoretical calculation. Similarly, the theoretical value for fully debranched amylopectin (92.1%) matches the results reported (90–91%) for ambient temperature malt mashing/fermentations using the DCL ‘all‐grains‐in’ test for maximum attenuation. This further corroborates the above reviews in that: There is no limit dextrinase activity during normal mashing and wort run‐off when the only active enzymes are α‐ and β‐amylase. There is considerable α‐amylase activity during mashing, wort run‐off and early fermentation, creating a turnover from higher molecular weight dextrin to true ‘limit’ dextrin. There is considerable debranching by limit dextrinase, but only when fermentation is established and the wort pH drops to 4.0–4.2. There is possibly a small contribution from α‐glucosidase in increasing the relative amount of glucose in the collected wort. Structure of outer chains of amylopectin and corresponding smallest α‐amylase limit dextrin [as proposed by Vinogradov and Bock 29]. Table 3. Theoretical fermentable extracts based on model malt (Figs 4 and 7) Component Total extract (g) Fermentable extract (no debranching) Fermentable extract (total debranching) Starch: amylose 11.8 11.8 11.8 amylopectin 47.2 34 47.2 Free sugars 9 9 9 Soluble protein 4.5 0.2 0.2 Glucans etc. 1.5 Total 74 55 68.2 Fermentability 74.3% 92.1% We now have a clearer picture of what happens during normal malt distilling practice, but this still does not explain the residual dextrin left in fully fermented wash. This only amounts to about 5% of the total carbohydrate in wort (Table 2) and is of the same order of magnitude as reported by Bringhurst et al. 27. However, the latter authors did not include non‐linear trisaccharides, which appear to be present in their chromatographic analyses of wort. The analyses carried out on Enevoldsen's apparatus 8, on the other hand, demonstrated that significant amounts of unfermented trisaccharide were present in the fully fermented wash of this high enzymatic malt. (Fig. 6b and c, Table 2) Assuming that linear maltotriose was completely fermented, those residual trisaccharides might be panose and isopanose and represent about 30% of the original wort trisaccharides. Similarly, the DP4–7 residual dextrins, which are apparently resistant to limit dextrinase or pullulanase, might be glucosyl‐maltotriose, ‐maltotetraose, ‐maltopentaose and ‐maltohexaose. The total concentration of this residual fraction is very small relative to the original concentration of maltose (Table 2) but it is still significant with respect to the final attenuation and requires further investigation. The puzzle that now presents itself is that none of these ‘pullulanase resistant dextrins’ (PRD) should exist according to the above model of Vinogradov and Bock 29. Indeed, if panose and isopanose are present, then glucosyl‐maltose and maltosyl‐glucose must be the smallest dextrins containing an α‐1,6 glucosidic linkage. The essential test would be to isolate the PRD present in the wash and treat them first with pullulanase and, if they are resistant, then with amyloglucosidase 41. If all of the remaining dextrins are hydrolysed to glucose then the above formulations are correct. This would be relatively easy to carry out, since the substrate can be sourced, quite literally by the ton, from any malt distillery's spent wash. On the assumption that PRD are residual dextrins with a degree of polymerisation of 3‐7 then they should have the structures shown in Fig. 8. Given that α‐ and β‐amylase cannot hydrolyse maltose or maltotriose on either side of the glucosyl‐α‐1,6‐linkage then the predominant isomers will have a degree of polymerisation of 4–6 as illustrated. However, the revised data from Bathgate et al. 8 in Fig. 6(c) would suggest that the trisaccharides are the most prevalent residues, therefore the core dextrin for PRD is either panose or isopanose and not maltosyl‐maltotetraose as depicted in Fig. 7. The existence of PRD in a fully fermented malt wort may also explain why attenuation maximizes at ~87% in a normal mash but never reaches the theoretical maximum. Since they are not present in an ambient‐temperature ‘all‐grains‐in’ mash and fermentation, which does go to completion at 91% fermentability, we can conclude that saccharification of starch granules follows a completely different pathway from gelatinized starch. Therefore, something must happen at the specific starch gelatinization temperature in a normal mash, causing the formation of these abnormal dextrins before amylolysis switches to the conventional mechanisms described above. It is highly unlikely that PRD is formed in the mash, since malt starch granules dextrinize at 65 °C almost as quickly as fully solubilized starch 42 and this is complete within 20 min of mashing‐in according to the conventional model. We must then investigate the degradation of semi‐crystalline starch granules in malting barley grain to understand possible PRD formation. The first research on the structure of starch in germinating barley was carried out by chemists who, in their inimitable fashion, dissolved and separated the constituent amylopectin and amylose and analysed their chemical composition. Greenwood and Thomson 43, 44 concluded that the surface of the starch granules were eroded by β‐amylase, since there was only a slight degradation of amylose and the average chain length in amylopectin was reduced by only one or two glucose units. Later work 45 on malted oats proved that the former results were statistical averages and that some amylopectin outer chains were reduced to maltose or maltotriose by the action of α‐amylase, while the vast majority of the amylopectin chains were untouched. This was only clarified when scanning electron microscopy 46 revealed that α‐amylase apparently attacked whole starch at specific surface sites by creating characteristic ‘pin‐holes’, which then widened out as the granule was degraded. Since glucose chains are arranged in parallel within the granule, it was concluded that α‐amylase must be constrained to attack the amylopectin from the outer facing, non‐reducing ends 47. Therefore for many years it was assumed that the only malt enzyme that can attack raw starch granules is α‐amylase 48, 49. This view was challenged by Sun and Henson 50, 51, who demonstrated that not only could α‐glucosidase from germinating barley degrade intact starch granules by itself, but it was also much more effective when acting synergistically with α‐amylase. For example, when one α‐amylase isomer was combined with a specific isomer of α‐glucosidase, their rate of reaction increased 11‐fold. In addition, they showed that the combined enzymes attacked starch granules in exactly the same manner as malt diastase in that characteristic ‘pin‐holes’ were observed at specific sites over the surface of the granule. Pure α‐amylase on the other hand could only hydrolyse starch at the equatorial groove of the granule and at a much slower rate than the amylase‐glucosidase complex. These authors also demonstrated that the specificity of all the diastase enzymes is markedly different when starch granules are used as substrate as compared with soluble starch. They concluded 49 that only α‐amylase and α‐glucosidase activities were relevant in the degradation of raw starch and that β‐amylase and debranching enzyme (limit dextrinase) were not significant in this role. The inference from this is that the latter enzymes play a secondary role in hydrolysing larger dextrins and amylopectin exposed by the α‐amylase‐glucosidase combination. This is a distinct possibility since starch dissolution proceeds more rapidly once the granule surface has been penetrated 50. The mechanism whereby α‐glucosidase can hydrolyse both α‐1,4 and α‐1,6 glucosidic bonds in granular amylopectin is not clear. Nevertheless, that mechanism must be dislocated precisely when gelatinization takes place and this would imply that there must be new limit dextrins created by the α‐amylase‐glucosidase complex. This theory fits with the observed pattern of starch granule degradation in that only a very small portion of the structure, at each ‘pin‐hole’, is being hydrolysed at the moment when the granule swells and ruptures, so dislocating and inactivating the enzyme–substrate conformation. Since gelatinization occurs at a precise temperature, the formation of PRD might represent a new type of residual dextrin representing those few fragments close to amylopectin branch points that are just about to be hydrolysed when the enzyme complex is suddenly destroyed. If the malt amylase–glucosidase complex behaves like some fungal glucoamylases 41, in a preferential hydrolysis of α‐1,4‐glucosidic linkages before a slower hydrolysis of the α‐1,6‐linkages 25, then any of the PRD structures in Fig. 8 are possible. However, some hydrolysis must also take place from the reducing ends, otherwise panose, isopanose and glucosyl‐maltotriose would not be formed. If this is the enzymic pathway leading to the formation of PRD, then it should occur in all malts on mashing at temperatures above the gelatinization point. In corroboration of this theory, we only need reference to Enevoldsen's original analyses of those brewing wort ‘limit’ dextrins 28 that are resistant to debranching by pullulanase. These have exactly the same structures as those shown in Fig. 8 and support the view that such dextrins are pre‐formed as a consequence of malting activity and gelatinization during mashing. However, Enevoldsen was uncertain about the creation of panose in wort, in that it might be a product of transglucosidation 52 rather than being an end product of amylolysis. We could also postulate that, if PRD are the product of an α‐amylase‐glucosidase degradation of raw starch granules, then their concentration in a consequent mash would be proportional to the number of modified starch granules in the malt. In other words, the more highly modified the malt, the greater is the possibility of creating PRD. There is some evidence in support of this in that the potential concentration of PRD in the over‐modified, high diastatic malt wort and wash reported in Table 2 is approximately double that estimated by Bringhurst et al. 27 in a normal distilling malt wort. Furthermore, a higher concentration of PRD in wort will lead to a higher residual gravity and a lower fermentability. The high enzymic malt had a fermentability of 83% compared with 86% in the reference malt 8 and the cold water extract was 24% compared with 20% in the standard malt. Therefore the evidence, limited as it is, points to a fixed amount of glucosyl‐maltosaccharides being created when partially degraded starch granules are gelatinized during mashing of malt at its optimum degree of modification. Consequently, they are ‘limit’ dextrins, not only of α‐ and β‐amylase, but also of limit dextrinase, which cannot hydrolyse a single glucose unit α‐1,6 linked to an α‐1,4‐linked maltosaccharide 28. This theory could be further ratified if residual PRD in wash were to be treated with purified malt α‐glucosidase to see if they are indeed limit dextrins of the α‐amylase/glucosidase complex. Since Sun and Henson demonstrated that malt α‐glucosidase had some activity against isomaltose 50, there is a possibility that it might also hydrolyse panose, isopanose and glucosyl maltosaccharides. We now have a thesis that answers our original questions and which can be summarized as follows: Under normal malting conditions, starch granules are partially degraded by a combination of α‐amylase and α‐glucosidase. This complex can open up the granule at specific sites on the surface and create characteristic ‘pin‐hole’ lesions, which may be widened by secondary hydrolysis by α‐ and β‐amylase, limit dextrinase and α‐glucosidase (maltase). All of these diastatic enzymes can survive mild kilning probably by forming heat stable complexes on and within the starch granules and can continue a complete degradation of starch when mashed at ambient temperatures with glucose as the end product. At normal mashing temperatures, starch granules gelatinize and dissolve with a concomitant rapid degradation to glucose, maltose, maltotriose and dextrins ranging from 4 to > 20 glucose units. If there is immediate wort boiling after run‐off, this is the final composition of starch derived carbohydrates according to the conventional paradigm. All malt worts also contain a small amount of glucosyl maltodextrins, based on a core of 62‐α‐glucosyl maltose (panose) or 6‐α‐maltosyl glucose (isopanose), which are remnants of the α‐amylase/glucosidase degradation of granular starch. These dextrins are resistant to the action of debranching enzymes and their concentration may vary between 4 and 8% of the malt extract, depending on the degree of modification of the host starch granules. They are created at the active sites of this enzyme complex when the granule is gelatinized. In a conventional mash of unboiled distilling wort, the spectrum of wort dextrins produced from gelatinized starch is reduced to true ‘limit’ dextrins of 4–8 glucose units by continued α‐amylolysis. These dextrins will contain side chains of either maltose or maltotriose residues surrounding the α‐1,6‐glucosidic linkage and can be debranched by limit dextrinase during fermentation, leaving only the above core dextrins in the spent wash. If the purpose of mashing was only to achieve the maximum spirit yield, then every malt distillery would mash at ambient temperature and immediately add yeast without wort extraction, so that the entire process could take place in a single vessel. With virtually no energy input and minimum plant utilization, the absolute maximum spirit yield would be achieved. However, would that spirit have the ‘essential flavour and aroma characteristics’ of malt whisky? The answer is an emphatic ‘no’. A trained flavour profile panel assessed 53 the ‘new make spirit’, made from the first industrial trials of a low‐temperature malt mash/fermentation, as being excessively ‘grainy’ and ‘sulphury – vegetable’ – these being the more polite comments. This is not entirely surprising since almost the entire endosperm is hydrolysed during the ‘all‐grains‐in’ fermentation. Consequently, the excessive concentration of amino acids and hydrolysed lipids 54 must lead to a completely different spectrum of higher alcohols, esters and fatty acid residues. It is also important to note that some of these fatty acids and esters may be derived from lipids associated with raw starch granules, both internally, complexed with the amylose fraction, and externally as part of the surrounding amyloplast membrane 55. Similarly some amino acids must be formed from degradation of protein also associated with the amyloplast and the starch granule. Therefore a significant quantity of triglycerides, glycolipids, phospholipids and free amino acids is released into the fermenting wort during the digestion of raw starch. The extreme degradation of malt starch, protein and lipids during low‐temperature mashing also highlights several other factors that affect the quality of malt spirit. First of all the predominant fermentable sugar produced is glucose 47 and not maltose, thus the pattern of fermentation is quite different. Secondly, the potential for undesirable ‘grainy’ character is much higher since the fermented wash is distilled in the presence of finely ground husk. Thirdly, excess amino acids and lipids are ideal substrates for the much higher concentration of bacteria in wort not separated from spent grains and not partially sterilized by normal mashing temperatures 56. It must be remembered that malt wort only requires a free amino acid (FAN) concentration of 150–200 mg per litre to achieve maximum attenuation and any wort with concentrations higher than that will leave substantial amounts of unassimilated amino acids in the finished wash 57. Consequently, all highly modified malts with high FAN are susceptible to Lactobacillus and Acetobacter acidification during late fermentation with inevitable effects on spirit quality. Low‐temperature malt mashing and fermentation have therefore not produced acceptable ‘new make spirit’ for the above reasons despite the very high spirit yields. This is not the case in grain distilling however, when any unacceptable congeners can be removed by adjustment of the distillation column. Therefore to produce acceptable Scotch malt whisky within the guidelines of the legal definition, barley must be malted and mashed in the traditional way and the loss of 4 or 5% of fermentable extract must be accepted as the norm. Similarly, mashing at 65 °C exercises some control over the concentration of commensular bacteria, especially Lactobacillus and Acetobacter species, passing to the fermenter. It is recognized that limited bacterial activity is an essential part of the craft of malt distilling, but heavy contamination of competing bacteria 58 can reduce spirit yield and produce undesirable off‐flavours. On the other hand, completely sterile fermentations may result in a reduction in the desirable esters and other congeners characterizing the distilled spirit. Now that nearly every Scotch malt whisky distillery has a state‐of‐the‐art mashing and lautering plant and stainless steel wash backs, all fitted with best practice in‐place cleaning and sterilization equipment, the industry may be approaching the latter conditions rather than preserving the former partial sterilizations. This may be another case of slavish adherence to brewing technology, which is not in the best interests of the malt whisky distiller. This review has attempted to show that malt distilling has many similarities to, and has benefitted from, brewing science and technology, but there are still significant areas that are germane only to distilling practice and which require further research and development. In particular, the enzymic degradation of cereal starch at temperatures below gelatinization is not well understood, nor is the detail of secondary saccharification in fermenting wort. Mashing at lower temperatures also has a major effect on proteolysis and FAN concentrations in wort; therefore research in this area, together with optimum levels of desirable versus undesirable microflora, may reveal opportunities to develop new flavour profiles.