Roasted Malt for Distilling: Impact on Malt Whisky New Make Spirit Production and Aroma Volatile Development

Abstract

Scotch malt whisky is typically produced using lightly kilned malted barley that imparts a relatively subtle aroma to the final product. Recently, there has been increased interest in exploring the feasibility of using roasted malts during whisky production to control congener profile. Although roasted malts are used widely within the brewing industry to develop product color and aroma in beer, applications and challenges have not yet been established for whisky production. This study investigated a role for roasted malt as a tool to impact whisky volatile composition and the consequences of such use for production efficiency of whisky new make spirit. Pot still malt was roasted at laboratory scale (0–60 min at 80–220 °C) and incorporated into a grist (≤50% w/w) for production of new make spirit. The influence of roasting conditions on malt processing characteristics and the impact on the concentration of key roasted malt volatile compounds in distillate were assessed using response surface modelling. Concentration of aroma active pyrazines and furans increased in the distillate produced using roasted malts (particularly when using malt heated >150 °C). Key indicators of process efficiency such as wort fermentability and alcohol yield reduced as intensity of malt roasting increased. Process efficiency when using low proportions of roasted malt (10% w/w) was comparable to that when using only pot still malt, but distillate volatile profile still differed significantly.

Keywords:

• Distilling
• GC-MS
• Maillard reaction
• roasting
• specialty malt
• whisky

Introduction

The majority of malt whisky produced in Scotland is derived from lightly kilned distilling malt, which is favored due to its high alcohol yield.^[1] Although green malt is occasionally used as-is, kilning is a necessary process for the production of most types of malted barley and involves the controlled drying and curing of green malt to a suitable moisture (typically 4%) and color. Kilning contributes to stability, processability, and to the aroma of the final malt product. During production of distilling malt, low kilning temperatures (65–70 °C) are chosen to promote the retention of diastatic enzymes and to maintain fermentability. Although gentle kilning promotes ethanol yield, it also limits progression of aroma generating processes such as the Maillard reaction. Consequently, the cereal raw materials selected for the production of malt whisky are often associated with a limited aroma contribution as compared with the process steps of fermentation, distillation, and maturation. Within the brewing industry, high-color specialty malts are used to promote the development of color, mouthfeel, and taste/aroma within beer,^[2] the relative contribution of a given malt is typically correlated to color and thus roasting parameters. The use of roasted malts is currently rare within the malt whisky distilling industry, but commercial examples are emerging.

Specialty malts are produced from barley, green malt or kilned pale malt under high-temperature kilning or roasting regimes (90–230 °C). Production often takes place in a dedicated roasting drum and heating conditions are typically more extensive than those applied during the production of pale malt types, because they are intended to develop the color and aroma characteristics of the final product. Flavor and color development during kilning and roasting of malted barley is derived primarily from non-enzymatic reactions such as Maillard chemistry, caramelization, and pyrolysis. Development of these reactions is dependent on a variety of parameters: temperature, duration of heat application, pH, saccharide and amino compound composition, and substrate moisture content. Previous research has explored the formation pathways of aroma active compounds in specialty malts.^[3–8] Non-enzymatic reactions during grain roasting are complex and tend to branch, resulting in a wide range of intermediate and stable reaction products. Oxygen and nitrogen containing heterocyclic compounds such as pyrazines, furans, pyridines, and pyrroles are of key importance to malt/beer/whisky flavor. Analysis of wort produced from malts of increasing color has reported a progression of aroma from sweet and husky (pale malt), to caramel and bread-like (intermediate color), and finally to chocolate, coffee, bitter, and burnt (high-color malt).^[9] Furthermore, reaction products from the malt roasting process have been observed to impact yeast cell activity during beer fermentations, resulting in increased concentrations of higher alcohols and a reduced ester content compared with the use of a pale wort.^[10,11]

The use of specialty malt during wort production is not without risk and it is well-established that high-color malts may result in a reduced attenuation of wort due to lower fermentable substrate availability during mashing and fermentation. There is also a direct impact of non-enzymatic browning reaction products on yeast metabolism and gene expression.^[10,11] To limit detrimental effects on malt processing and fermentation and to reduce the development of unpleasant acrid aroma characteristics, application rates (inclusion rates) of roasted malts within brewing grists are typically low (2–12% w/w).^[2,12,13] The production processes of beer and malt whisky new make spirit (the distillate that is matured in oak casks to produce whisky) share much in common, but there are key differences: during whisky production wort is not boiled, wash fermentation temperature is typically higher (often >30 °C), and the distillation process itself, which has no equivalent within brewing, is used to selectively recover congeners into the distilled product. It remains to be assessed whether constraints of use for roasted malts within distilleries are similar to that of the brewing industry or whether elevated use might be tolerable.

This study assesses a role for roasted malt to be used during the production of malt whisky new make spirit with a focus on consequences for distillate aroma volatile composition. The impact of malt roasting parameters and grist composition on malt processing characteristics were assessed for standard (EBC laboratory wort) and high-gravity wash production practices. The consequences for malt roasting conditions on the distillate volatile profile were assessed by response surface analysis of ten volatile compounds known to be aroma active in roasted malt. There are numerous examples of commercial distilleries in recent years opting to incorporate a proportion of roasted malt into their production processes, but a minimal amount of published literature is currently available evaluating this approach. The present research begins to elucidate some processing characteristics of roasted malt as applied to typical distillery practice and the expected contributions such malt products may provide to a new make spirit aroma volatile profile.

Experimental

Chemicals

All chemicals were obtained from the following commercial sources at analytical grade purity: dichloromethane (>99%), acetonitrile (≥99.9%), fructose (≥99%), sucrose (≥99.5%), maltose monohydrate (≥99.0%), maltotriose (≥95%), hexan-1-ol (≥99.9%), 2,5-dimethylpyrazine (≥98.5%), 2-methylpyrazine (≥99%), 2-ethylpyrazine (≥98%), 2,3-dimethylpyrazine (≥98.5%), 5-(hydroxymethyl)furfural (≥99%), benzaldehyde (≥99%), and 4-ethyl-2-methoxyphenol (4-ethylguiacol; ≥98.0%) from Sigma-Aldrich (Dorset, UK); glucose (≥99%), furan-2-carbaldehyde (furfural; 99%), and 5-methylfuran-2-carbaldehyde (5-methylfurfural; ≥98%) from Acros Organics (Geel, BE); 3-heptanone (98%) from Alfa Aesar (Lancashire, UK); ethanol (≥99.8%), and sodium chloride (>95%) from Fisher Chemicals (Loughborough, UK). Malted barley samples were provided by Crisp Malt (Norfolk, UK).

Malt roasting

Pot still malt (400 g) was roasted with constant agitation inside a stainless steel rotating drum (⌀ 19 cm, depth 10 cm, rotation speed 3.5 rpm) within a HP 5890 Series II GC oven (heating and cooling rate 30 °C/min; Hewlett Packard, Minneapolis, U.S.A.). Malt roasting methodology was adapted from the work of Parr et al.^[5] Design Expert 12.0 (Stat-Ease Inc., Minneapolis, MN, U.S.A.) was used for experimental design (work-flow optimization and sampling regime) and to model the impact of malt roasting parameters (two factors: roasting time and temperature) on new make spirit production and composition. Roasting variables were investigated at 25 data points (0, 15, 30, 45, 60 min at 80, 115, 150, 185, 220 °C) with replicates at each vertex of the model and three additional replicates at the central point. Roasted malt samples were stored at 8 °C prior to analysis.

Spirit production

Malt samples, produced according to the previously described experimental design, were crushed using a DLFU universal laboratory disk mill (Bühler-Miag, Uzwil, Switzerland) set to gap size 0.7 mm. Milled malt (70 g of pot still malt and 30 g of roasted pot still malt) was mixed with 310 mL of distilled water (65 °C) and mashed at 65 °C, with constant agitation for 60 min in a CM4 mashing bath (Canongate Technology, Edinburgh, UK). Following mashing, samples were cooled to 20 °C and adjusted to 450 g with room-temperature distilled water. Mash samples were filtered through pre-pleated 113 V filter paper (Whatman, Kent, UK) for 5 h, the first filtrate (50 mL) was recirculated. The wort of multiple mashes was consolidated to achieve a final volume of 1200 mL.

Wort (1100 mL) was pitched with 1 g/L of DY502 dry yeast (Anchor/Lallemand, Canada) and fermented within a 2-L glass bottle (fitted with an airlock) for 72 h at 30 °C. Samples were manually agitated every 24 h. Fermentation vessels were weighed throughout fermentation. Specific gravity of wort and wash samples was measured using an Anton-Paar DMA 4500 M density meter (Anton-Paar, Graaz, Austria).

Low wines (wash distillate) were produced using a 2-L copper pot still (Al-Ambiq, Gandra, Portugal) with a worm-tub style condenser (water flow rate 10 mL/s). The still was charged with 1000 mL of wash and 0.2 mL/L of FD20PK silicone-based antifoam (Murphy & Son Limited, Nottingham, UK) and heated using a hot plate. Samples were collected until the run-off distillate reached 1% ABV. Spirit distillation was achieved using a round bottomed glass flask (500 mL) charged with low wines (300 mL, diluted to 20% ABV with distilled water), and fitted with a Liebig condenser, containing 6 cm² of copper mesh in the adapter. The still was heated using an electric heating mantle. Foreshots (2.5 mL) were discarded, and the main cut (50 mL) was collected for analysis. Distillate alcohol by volume percentage (% ABV) was measured using an Anton-Paar DMA 35 Basic density meter.

Samples were produced to investigate the impact of roasted malt inclusion rate on distillate production and properties. Pot still malt was roasted for 30 min at temperatures 140, 180, and 220 °C (malt color: 65, 170, and 450 EBC, respectively). Grists of roasted and unroasted malt were combined at 10, 20, 30, 40, and 50% (w/w), with 100% unroasted malt used as a reference. Each grist sample was processed through to spirit distillate in triplicate, according to the previously described procedure with the following adjustments. Mash samples were filtered for 20 h at 8 °C to collect sufficient wort for fermentation. Distillation of wash (200 mL) and low wine (55 mL, diluted to 20% ABV) were both performed in round bottomed glass flasks (500 mL). Foreshots were cut at 1 mL, and the following 10 mL of distillate was collected for analysis.

Analysis of malt and wort quality

Standard roasted malt analyses were performed according to European Brewery Convention methods with no modifications:^[14] moisture (4.2), total nitrogen (4.3.2), one thousand corn weight (4.4), cold water extract (4.6.2), wort color (4.7.1), wort fermentability (4.11), wort pH (8.17), and wort free amino nitrogen (FAN; 8.10.1). The hot water extract method (5.7) was adapted as follows: mashing duration of 1 h, mash filtered for 2 h.

HPLC analysis of sugars in wort

Fermentable carbohydrates in the wort were analyzed using a Shimadzu LC-20AD (Shimadzu, Kyoto, Japan) chromatograph, fitted with a Luna NH₂ column (5 μm, 250 × 2 mm; Phenomenex, Torrance, CA, U.S.A.) and a Shimadzu RID-20A refractive index detector. Wort (1 mL) was centrifuged at 13400 rpm for 60 sec; the supernatant was recovered and diluted with distilled water (1:1). The diluted sample (3 µL) was injected onto the column (50 °C) and eluted with 80% aqueous acetonitrile (flow rate: 0.25 mL/min) for 17 min. Glucose, fructose, sucrose, maltose, and maltotriose were calibrated between 0.5 and 50 g/L, and all curves had a coefficient of determination (R²) >0.99.

GC–MS analysis of aroma volatiles in the spirit

An internal standard (50 µL, 0.2 mg/mL of 3-heptanone solution in 20% ethanol) and sodium chloride solution (0.5 mL, 0.36 g/mL) were added to each spirit sample (15 mL, diluted to 20% ABV with distilled water). The mixture was extracted with dichloromethane (0.5 mL; DCM) by vortex mixing for 10 sec and centrifuging for 20 sec at 1000 rpm. Dichloromethane extract was recovered to vials with glass inserts (100 µL) and placed into a Shimadzu AOC5000 autosampler fitted to a Shimadzu QP2010 Ultra GC–MS. The autosampler syringe was rinsed twice with ethanol, flushed with sample five times, and then the DCM extract (1 µL) was directly injected (injector port: 120 °C, split ratio: 1:1) into an HP5MS column (30 m × 0.25 mm × 0.25 µm; Agilent J&W, CA, U.S.A.). The column temperature was held at 40 °C for 5 min, increased to 80 °C at a rate of 5 °C/min, further increased to 150 °C at a rate of 10 °C/min, and to 320 °C at a rate of 70 °C/min, holding the final temperature for 3 min. Helium carrier gas was maintained at a constant flow of 0.65 mL/min. Transfer temperature to MS was 280 °C, ion source temperature was 250 °C, ions were detected with a 4 min solvent delay, in selected ion monitoring (SIM) mode.

Ten compounds were selected for quantification in the new make spirit samples (Table 1). Selection was based on preliminary identification within the roasted malt samples using SPME GC–MS analysis and NIST08s library software. The key aroma compound groups affected by grain roasting were pyrazines and furans. The following m/z values were monitored: 2-methylpyrazine (94, 67, 53), furfural (96, 95, 67), 1-hexanol (69, 56, 43), 2,5-dimethylpyrazine (108, 81, 52), 2-ethylpyrazine (108, 81, 53), 2,3-dimethylpyrazine (108, 67), benzaldehyde (106, 105, 77), 5-methylfurfural (81, 53), 5-(hydroxymethyl)furfural (5-HMF; 97, 69, 53), and 4-ethylguaiacol (152, 77, 53).

Table 1. Summary of observed characteristics of roasted pot still malt (80–220 °C, 0–60 min) and distilled spirit (using 30% roasted malt inclusion).

Parameter	Unroasted pot still malt^a	Minimum value	Maximum value	Model fit^b	Model F-value	Model p-value	Model R^2
	Roasted malt quality characteristics
Hot water extract, dry (L°/kg)	315 ± 5	266	321	Cubic	41.86	<0.0001	0.9448
Cold water extract (% w/w)	21.1 ± 0.3	9.1	40.9	Cubic	8.35	<0.0001	0.7736
Real fermentability (%)	83.7 ± 0.6	49.8	84.7	Cubic	49.28	<0.0001	0.9527
Wort fermentable sugars (g/L)	69.5 ± 0.6	40.6	72.1	Cubic	131.60	<0.0001	0.9818
Elemental N in malt (% w/w)	1.89 ± 0.44	1.4	2.8	Quadratic	4.09	0.0072	0.4402
Wort FAN (mg/L)	154 ± 11	77	163	Cubic	35.61	<0.0001	0.9358
Wort color (EBC)	5 ± 0	4	631	Cubic	86.58	<0.0001	0.9725
Wort pH	5.79 ± 0.02	5.19	5.82	Cubic	78.61	<0.0001	0.9698
1000 corn weight (g)	42.3 ± 1.3	35.4	44.1	2FI	29.33	<0.0001	0.7586
Malt moisture (% w/w)	4.8 ± 0.1	0.2	4.8	Cubic	142.04	<0.0001	0.9831
Wort volume (mL)	289 ± 2	180	298	Cubic	29.04	<0.0001	0.9224
	Production of spirit using high gravity wort (30% of roasted malt inclusion)
	Standard analyses of wort
Wort volume (mL)	209 ± 4	150	218	Cubic	47.89	<0.0001	0.9514
Wort color (EBC)	10 ± 1	9	715	Cubic	99.01	<0.0001	0.9759
Wort FAN (mg/L)	352 ± 9	246	362	Cubic	44.29	<0.0001	0.9477
Wort fermentable sugars (g/L)	141 ± 1	106	142	Cubic	58.49	<0.0001	0.9599
Original gravity	1.0729 ± 0.0004	1.0707	1.0739	Cubic	31.71	<0.0001	0.9284
Final gravity	0.9989 ± 0.0001	0.9985	1.0127	Cubic	180.41	<0.0001	0.9866
Fermentability (%)	82.7 ± 0.1	66.8	83.1	Cubic	181.33	<0.0001	0.9867
Alcohol yield (mL AA/kg malt)	205 ± 3	112	217	Cubic	29.47	<0.0001	0.9234
	Aroma volatiles in new make spirit
2-Methylpyrazine (mg/LAA)	ND^c	ND	11.80	Cubic	39.12	<0.0001	0.9412
2,5-Dimethylpyrazine (mg/LAA)	ND	ND	2.38	Cubic	32.92	<0.0001	0.9309
2-Ethylpyrazine (mg/LAA)	ND	ND	2.28	Cubic	30.76	<0.0001	0.9264
2,3-Dimethylpyrazine (mg/LAA)	ND	ND	0.54	Cubic	99.36	<0.0001	0.9760
Furfural (mg/LAA)	7.2 ± 1.1	4.3	34.30	Quadratic	8.81	<0.0001	0.6287
5-Methylfurfural (mg/LAA)	ND	ND	1.43	Cubic	59.31	<0.0001	0.9604
5-(Hydroxymethyl)furfural (mg/LAA)	ND	ND	4.83	Cubic	12.76	<0.0001	0.8392
4-Ethylguaiacol (mg/LAA)	ND	ND	0.18	Cubic	41.54	<0.0001	0.9444
1-Hexanol (mg/LAA)	1.49 ± 0.01	1.24	3.98	Quadratic	8.83	<0.0001	0.6294
Benzaldehyde (mg/LAA)	0.54 ± 0.03	0.25	1.13	2FI	6.52	0.0017	0.4115

a mean and standard deviation of three samples.

^bModel fit was optimized, and statistical values determined using analysis of variance (ANOVA) (Design Expert 12.0).

^cND – not detected.

Compounds were identified using retention times and reference ions of standards. Calibration samples were prepared in triplicate at five dilutions. Linear calibration curves were constructed for each analyte in the range covering typical concentrations found in the distillate.

Statistical analysis

Design Expert 12.0 was used to produce statistical models using response surface methodology with two control variables (roasting time and temperature). Measured characteristics were fitted on a polynomial curve up to a cubic fit and two-way analysis of variance (ANOVA) was used to assess the significance of the response with an alpha level of 5% (p = 0.05). Significance of results from varying roasted malt grist inclusions were assessed by two-way ANOVA with an alpha level of 5% (p = 0.05). Analysis of correlation (Spearman’s rank correlation) was performed using SPSS Statistics 28.0 (IBM, New York, NY, U.S.A.).

Results and discussion

Roasted malt quality characteristics

Pot still malt was roasted under conditions of controlled time and temperature and analyzed for industry-typical measures of malt quality. Results were processed using response surface modelling to reveal trends in the development of key malt quality characteristics during malt roasting (Table 1; Figure 1).

Figure 1. Impact of roasting time (min) and temperature (°C) on development of typical malt quality characteristics in pot still malt. (A) hot water extract on dry basis (L°/kg); (B) cold water extract (% w/w); (C) wort fermentability (%); (D) wort fermentable sugars (g/L); (E) wort free amino nitrogen (mg/L); (F) wort color (EBC units); (G) wort pH; (H) 1000 corn weight (g); and (I) wort volume (mL). Roasted malt inclusion rate was 100% in analyses represented in graphs B and H, and 50% in graphs A, C-G and I. Darker tones indicate an increase in the given parameter.

Under conditions of low to moderate malt roasting temperature (<150 °C), the hot water extract (HWE) of the malt grist (1:1 roasted malt and pot still malt) remained above values typical for pale malted barley (307–321 L°/kg; d.b.; Figure 1A), although a trend of decreasing extract was observed with time when using malt roasted between 115 and 150 °C. Use of malt roasted under conditions >150 °C resulted in more pronounced HWE decreases, to a minimum of 266 L°/kg (220 °C for 60 min). It is well-established that use of extensive kilning regimes and roasting is detrimental to malt amylolytic potential due to enzyme denaturation, and roasting has been demonstrated to directly impact some cereal component properties such as protein and starch solubility, and starch swelling.^[15]

Observation of cold water extract (CWE) also suggests a direct impact of heat-driven reactions on malt components. Malt cold water extract (CWE) is normally in the range of 18–21%^[16] and mostly represents soluble carbohydrate (10–12%), proteins and amino acids (∼5%), and salts (∼2%) before mashing.^[17] In the present study, CWE values were typical (19–22%) in malts roasted at temperatures up to 150 °C, but at 150–185 °C CWE gradually decreased to 9% (185 °C for 30–60 min). Whilst CWE is influenced by enzyme activity during germination, following kilning most malt enzymes are dormant until rehydration during mashing-in. Given that malts in this trial were roasted without water addition or stewing (as in crystal malt production), results suggest that non-enzymatic heat-induced reactions during roasting influenced malt component solubility. Interestingly, malt roasted at 220 °C, displayed elevated CWE values, peaking at 41%. Previous work has found soluble pyrodextrins to be formed during starch roasting,^[18,19] and this could explain the increased CWE observed in the present study. Despite their ready-solubility and contribution to extract, pyrodextrins are not fermented by distilling yeast.^[16,20]

In addition to generally reduced extract recovery in worts produced using roasted malts (1:1 roasted malt and pot still malt), analysis of wort properties indicated that use of malts roasted at an elevated temperature impacted overall fermentability of the extract. In wort produced using malt roasted between 80 and 150 °C, real fermentability was 80–85% (Figure 1C), but under elevated temperature regimes (185–220 °C) fermentability was reduced to a minimum of 50%. Real fermentability in commercial distilling malts is often >87%. Analysis of wort fermentable sugars (the sum concentration of fructose, glucose, maltose, and maltotriose) revealed that even relatively short roasting regimes resulted in a decreased fermentable sugar concentration in the wort, and sugar recovery decreased increasingly quickly alongside roasting intensity (Figure 1D). Wort produced using malt roasted at 220 °C for 60 min had just 41 g/L fermentable sugar compared with 72 g/L when using only a lightly roasted malt (80 °C for 0 min; temperature ramp only). Elemental nitrogen in roasted malt varied from 1.4 to 2.8% but showed no significant relationship with roasting time or temperature (model R² = 0.4402). Free amino nitrogen (FAN) concentration was highest in the wort produced from malt roasted at 80–115 °C (Figure 1E; 134–163 mg/L). As the malt roasting temperature increased (>115 °C), recovery of FAN into wort was substantially reduced (to a minimum value of 77 mg/L). Free amino nitrogen and fermentable sugars displayed differing patterns of development as roasting temperature and time were increased, loss of FAN mostly stabilized when roasting at >160 °C (70–90 mg/L), while sugar recovery continued to decrease (within the experimental conditions used in this study). Prior investigation of Maillard chemistry in a glucose/glycine model system, observed sugar loss exceeding that of amino acid similarly to this study.^[21] The greatest reduction of wort FAN was observed when using malt roasted between 115 and 150 °C, dropping from 144 mg/L to 95 mg/L (using malt roasted for 30 min). The differing rate of loss between sugar and FAN indicates that sugars might be interacting with other Maillard active materials within the malt or perhaps that sugars are undergoing caramelization reactions in FAN deficient, high-temperature conditions. Previously published work has indicated that caramelization reactions can lead to the development of aroma active pyrones and furans.^[22] The importance of FAN for efficient and vigorous fermentation is well-documented, playing a role in yeast cell growth during the early stages of fermentation and also impacting production of aroma-active volatiles by yeast cells.^[23–25] The grist composition (50% w/w roasted malt) used during the generation of these laboratory worts is unlikely to simulate commercial application of roasted malt within the distilling industry, but results might indicate potential issues of fermentability due to low FAN content when using elevated proportions of highly roasted malt. In addition to the contribution of wort sugar and FAN to overall fermentability, previous work has indicated that fermentation efficiency is also affected by thermal reaction products such as furfural that are known to be found in malts of high-color.^[10,11,26] In the present study, wort color negatively correlated with recovery of fermentable sugar and FAN into wort, and with overall wort fermentability (Table 2).

Table 2. Bivariate correlation analysis of roasted malt properties displayed as Spearman’s rank correlation coefficient.^a

	Wort color	Malt moisture	1000 corn weight	Elemental N in malt	Cold water extract	Hot water extract	Wort pH	Wort FAN	Wort fermentable sugars	Real fermentability
Wort color	1.000
Malt moisture	−0.973**	1.000
1000 corn weight	−0.833**	0.809**	1.000
Elemental N in malt	0.160	−0.210	−0.142	1.000
Cold water extract	−0.295	0.357*	0.168	−0.284	1.000
Hot water extract	−0.882**	0.878**	0.758**	−0.185	0.319	1.000
Wort pH	−0.941**	0.936**	0.819**	−0.184	0.410*	0.866**	1.000
Wort FAN	−0.937**	0.943**	0.793**	−0.106	0.297	0.848**	0.940**	1.000
Wort fermentable sugars	−0.948**	0.912**	0.805**	−0.104	0.304	0.922**	0.908**	0.905**	1.000
Real fermentability	−0.896**	0.865**	0.698**	−0.206	0.248	0.912**	0.856**	0.845**	0.879**	1.000

^a Wort samples produced according to EBC methodology with 50% (w/w) roasted malt within the grist.

** Correlation significant at p < 0.01 (two-tailed).

* Correlation significant at p < 0.05 (two-tailed).

Substantial increases in wort color were found only when using malt roasted >150 °C (Figure 1F), with the greatest rate of increase observed under roasting conditions >185 °C and >15 min (554–631 EBC). Melanoidins formed during the Maillard reaction have been reported as accountable for the majority of the color change during malt roasting,^[27] and nitrogen has been identified as necessary for the formation of browning in high-temperature environments.^[28] In the present study, the majority of wort color development was measured in samples produced under roasting conditions, under which FAN content was stable (>30 min and >160 °C; Figure 1E) indicating involvement of material not contributing to utilizable nitrogen. Previous research suggests that nitrogen-rich, low molecular weight melanoidins in pale malts might be utilized to produce dark, high molecular weight melanoidins without additional free amino nitrogen in high-temperature conditions.^[27–29]

Earlier studies have demonstrated a strong relationship between wort color and pH, suggesting that an increase in acidity is attributed to development of melanoidins and their precursors, reductones.^[27,30] During the roasting of coffee beans, increasing acidity has also been linked to thermally-driven development of organic acids, such as: formic acid, acetic acid, glycolic acid, and lactic acid.^[31] Here, wort pH similarly decreased as wort color increased (r_s = -0.94, n = 32, p < 0.001; Table 2). Under roasting regimes of increasing time and temperature, the pH fell from 5.8 (80 °C for 0 min; temperature ramp only) to 5.2 (220 °C for 60 min; Figure 1G).

Grain weight (Figure 1H) and moisture decreased with increasing roasting time and temperature. The rate of weight loss (44.1 to 35.4 g) was constant, while the moisture loss (4.8 to 0.2%) was greatest in the first 15 min at all temperatures. Weight loss is mainly caused by water evaporation, gas emission during heat-induced reactions (H₂S, CO₂, CO, NH₃), and loss of dry matter as dust during roasting.^[16] Intensive roasting conditions have been reported to cause the husk of malted barley to become brittle and capable of reducing grain bed permeability due to the production of fine particles during milling,^[32] and the observations of the work presented here support this hypothesis (Figure 1I). Wort volume recovered during separation of laboratory mashes reduced from >270 mL when using malts roasted <185 °C to 180 mL when separating mash produced from malt roasted at 220 °C for 60 min.

Distillate production using high-gravity wort

High gravity wort (OG = 1.070) produced using a range of roasted malt grist application rates and malt generated under differing roasting conditions was used to evaluate the impact of roasted malt on wort production, fermentation, and distillate characteristics under conditions of relevance to distilling industry practice. Typical laboratory wort used during evaluation of high-color malt is generally of low specific gravity (OG = 1.030–1.040) and is produced using a high grist inclusion of roasted malt (50% w/w).

Response surface models assessing the impact of grain roasting parameters on high gravity wort (produced from 30% roasted malt (w/w) grist) and resulting distillate properties (Figure 2) indicated a significant response for all quality parameters assessed (p < 0.001; Table 1). Trends observed during production and processing of high-gravity wort samples were similar to those from low-gravity wort (Figure 1). Under conditions of increased roasting time and temperature, wort volume recovery, wort FAN and fermentable sugar content decreased (Figure 2A–C). Use of malt produced under increasingly extensive heating regimes also resulted in wort of reduced fermentability (Figure 2E) and reduced distillate ethanol yield (Figure 2F; determined from low wines), particularly when the grist used malt heated >185 °C.

Figure 2. Characteristics of high gravity wort (made with 30% roasted malt) as a function of roasting time (min) and temperature (°C): (A) wort volume (mL), (B) wort free amino nitrogen (mg/L), (C) wort fermentable sugars (g/L), (D) wort color (EBC units), (E) wort fermentability (%), and (F) alcohol yield (mL AA/kg malt). Darker tones indicate an increase in the given parameter.

The factors informing use of roasted malts within the brewing industry are process-based (e.g., wort filtration and extract yield) and product-based (e.g., color development, mouthfeel, foam stability, volatile profile), and while many considerations will be shared by distillers, some are of limited relevance to typical processes (e.g., foam stability and wort color). The grist specifications used by brewers tend to include only small proportions of colored malt in order to achieve the required product specification and to prevent issues of brewhouse processing. Given the substantial processing differences of malt whisky production, it remains to be assessed whether the proportions of roasted malt that may be used by distillers is similar to established practice in the brewing industry. The yield of wort from grist using malt roasted up to 180 °C (30 min roast time) was not significantly different (p> 0.05) to that from use of only kilned pot still malt (245 ± 5 mL), even at high inclusion rate (50%; Figure 3A). Supplementing grist with increasing proportions of malt roasted at 220 °C, resulted in significantly (p < 0.001) reduced wort volume recovery, perhaps due to lack of intact husk and reduced malt component solubility impeding grain bed permeability. At 50% inclusion of malt roasted at 220 °C (30 min), wort yield was reduced by 25% compared with a lightly kilned pot still malt control (from 245 ± 5 mL to 183 ± 7 mL). Wort original gravity was similar regardless of roasting temperature or grist inclusion rate (1.072–1.073), although trends during fermentation indicated differences in fermentability and subsequent alcohol yield. Increased use of roasted malt was linearly coupled with elevated final gravity values. Wort produced using only pot still malt had a fermentability of 83.4 ± 0.2% (Figure 3B) and resulted in a final wash of gravity 0.998 ± 0.000 following fermentation with distilling yeast. Fermentability of wort reduced alongside increasing use of roasted malt (final gravity of fermented wash increased), particularly when using malt heated at >180 °C (fermentability: 63.9–76.3%, final gravity: 1.005–1.015). Use of roasted malt at application rates typical of brewing practice (10%) resulted in filtered wort yield (236–252 mL), fermentability (80.7–83.1%), and ethanol yield (225–246 mL AA/kg malt; in low wines) similar to a pot still malt control (wort volume: 245 ± 5 mL, fermentability: 83.4 ± 0.2%, ethanol yield: 241 ± 9 mL AA/kg malt; Figure 3). Within the grist proportions measured during this study, use of malt roasted at 140 °C resulted in ethanol yield (Figure 3C; 235–244 mL AA/kg malt) similar to using a grist of only kilned pot still malt (241 mL AA/kg malt). Use of malt roasted at 220 °C resulted in reduced ethanol yield, particularly at high grist inclusion rates (139 ± 4 mL AA/kg malt at 50% use).

Figure 3. Characteristics of wort prepared with 0–50% roasted malt (140, 180, and 220 °C temperature roasts for 30 min): (A) collected wort volume (mL); (B) fermentability (%); and (C) alcohol yield (mL AA/kg malt). Error bars indicate the standard deviation of triplicate independent samples.

Aroma volatiles in spirit produced using roasted malt

All response surface models (Figure 4; assessing new make spirit produced using 30% (w/w) roasted malt) except benzaldehyde were statistically significant (Table 1; p < 0.05), indicating that volatile development during grain roasting can impact aroma compound recovery into distillate.

Figure 4. Concentrations of aroma volatiles in spirit (mg/LAA; made with 30% roasted malt, high gravity wort), as a function of roasting time (min) and temperature (°C): (A) 2-methylpyrazine; (B) 2,5-dimethylpyrazine; (C) 2-ethylpyrazine; (D) 2,3-dimethylpyrazine; (E) furfural; (F) 5-methylfurfural; (G) 5-(hydroxymethyl)furfural; (H) 4-ethylguaiacol; and (I) 1-hexanol. Darker tones indicate an increase in the given parameter.

Four pyrazines were quantified in new make spirit: 2-methylpyrazine (nutty),^[33] 2-ethylpyrazine (nutty), 2,3-dimethylpyrazine (cocoa, coffee), and 2,5-dimethylpyrazine (chocolate, roasted nut). Pyrazines are nitrogen heterocycles, formed from complex interactions between FAN and reducing sugars during the Maillard reaction, and they are found in a variety of roasted foods, including barley, cocoa, coffee, and nuts. All four pyrazines were only detected in spirits made with ≥115 °C roasts (Figure 4A–D), and previous work has indicated pyrazine generation during grain roasting requires a greater input of thermal energy compared with other Maillard reaction products such as oxygen-containing heterocycles.^[9] The concentration of pyrazines detected in new make spirit generally increased alongside malt roasting temperature and development was most evident during the first 15 min of roasting. These results are in accordance with previous reports on barley and malt roasted under conditions of high-temperature and low moisture.^[4,6,9] The most abundant pyrazine detected in the new make spirit samples was 2-methylpyrazine (≤11.8 mg/LAA), followed by 2,5-dimethylpyrazine (≤2.38 mg/LAA), 2-ethylpyrazine (≤2.28 mg/LAA), and 2,3-dimethylpyrazine (≤0.54 mg/LAA). The pyrazine concentration in the new make spirit increased linearly with increased use of roasted malt within the grist (Figure 5A–D).

Figure 5. Concentrations of aroma volatiles in spirit (mg/LAA), made including 0–50% of roasted malt: (A) 2-methylpyrazine; (B) 2,5-dimethylpyrazine; (C) 2-ethylpyrazine; (D) 2,3-dimethylpyrazine; (E) furfural; (F) 5-methylfurfural; (G) benzaldehyde; (H) 4-ethylguaiacol; and (I) 1-hexanol. Error bars indicate the standard deviation of three independent samples.

Three furan derivatives, previously identified as key volatiles in roasted malt,^[5] were monitored in new make spirit: furfural (bready, caramel), 5-methylfurfural (caramel, maple), and 5-HMF (buttery, caramel). In agreement with previous literature, furan content of new make generally increased with use of malt produced with increasingly extensive roasting regimes. Both 5-methylfurfural and 5-HMF were only identified in new make spirit produced using malt roasted >115 °C and >150 °C respectively (Figure 4F and G). By contrast, furfural was the sole furan monitored that could be identified within new make produced using only unroasted pot still malt (Figure 4E). Furfural is commonly found in pale malt and was likely produced during kilning. All three furan compounds were detected in greatest concentration in samples produced using malt roasted at 220 °C, furfural and 5-methylfurfural when using malt heated for 60 min (34 mg/LAA and 1.4 mg/LAA), and 5-HMF when malt was heated for 30 min (4.8 mg/LAA). Furans are oxygen heterocycles, formed via the Maillard reaction or via caramelization.^[22] Furan production through the caramelization pathway may explain the patterns observed in this work, in which development continued under conditions of decreasing sugar concentration (Figure 2C) and stable FAN content (Figure 2B; >160 °C).

Incorporation of malt produced under more extensive heating regimes into the new make spirit production process allowed 4-ethylguaiacol (smoky, spicy) to be found within the distillate. It was detected primarily in spirit produced using malt roasted at 220 °C (Figure 4H; 0.03–0.16 mg/LAA); however, some minor development was observed when using malt heated at 185 °C for ≥30 min (Figure 5H; 0.01 mg/LAA). The concentration of 4-ethylguaiacol exhibited a positive quadratic relationship (R² = 0.999) with inclusion of malt roasted at 220 °C (Figure 5H). In whisky, 4-ethylguaiacol content is often linked to cask charring or malt peating processes.^[34,35] In this study, 4-ethylguaiacol is likely to be a thermal degradation product of barley lignin and hemicellulose.^[36] This result may indicate a possible alternative route to incorporating typical characteristics of peated malt into distillates given issues of peat harvest sustainability.

Benzaldehyde (bitter almond, maraschino cherry aromas) was detected in all spirit samples (0.2–1.1 mg/LAA). Although modelling did not indicate a strong relationship between roasting condition and benzaldehyde development (R² = 0.412), assessment of grist composition (using malt heated to 140, 180, and 220 °C) did indicate that benzaldehyde content in distillate increased alongside greater heat application to grist components (Figure 5G). Benzaldehyde may be generated through a variety of mechanisms during distillate production, but thermally driven Strecker degradation of phenylalanine explains the present trends.^[37] All spirit samples were also found to contain 1-hexanol (herbaceous, woody aroma), with highest concentrations observed when using malt heated under the most extreme temperature conditions (Figure 4I; 1.2–4.0 mg/LAA) and when grist inclusion of roasted malt was increased (Figure 5I; 1.0–2.6 mg/LAA). Hexanal, the corresponding aldehyde, is a lipid-derived aroma volatile, which can be produced enzymatically in pale malts^[9] or by thermal oxidation in roasted malts.^[38] Hexanal can be reduced to 1-hexanol during the fermentation process.^[39] Previous research has highlighted that malt hexanal content increases during roasting^[5,40] and yeast cell conversion of such a 1-hexanol precursor during wash production may explain the trends observed in this work.

Conclusions

The principal reason that a distiller might use roasted malt is due to the potential for the material to impart an aroma or taste to the distillate that may be difficult or impossible to achieve using a typical lightly kilned malted barley. The present research confirms that use of a roasted malt within a distillery grist can significantly impact the aroma volatile profile in new make spirit compared with the use of only pot still malt and also that the volatile compound profile can be further controlled through manipulation of malted barley roasting parameters. Modelling indicated that recovery of pyrazines (2-methylpyrazine, 2,3-dimethylpyrazine, 2,5-dimethylpyrazine, and 2-ethylpyrazine) and furans (furfural, 5-methylfurfural, and 5-HMF) into distillate was optimally achieved by incorporation of malt roasted >150 °C into the grist. The phenolic compound, 4-ethylguaiacol was primarily found in new make spirit derived from malt roasted at 220 °C. Given that many of the aroma compounds a distiller may seek to recover from roasted malts are derived from sugars and amino acids within barley, it is likely unavoidable that use of a grist incorporating roasted malt results in reduced ethanol yield. However, this work indicates that use of only low proportions of roasted malt within a distillery grist may still significantly impact the distillate volatile profile whilst minimizing the negative processing impacts associated with malts of high color. Should impact on yield prove tolerable within process specifications, the use of roasted malt during the production of malt whisky new make spirit may provide a tool for distillers seeking greater control over product volatile compound composition.

Declaration of interest statement

The authors have no competing interests to report.

Abbreviations
ABV	=	alcohol by volume
CWE	=	cold water extract
DCM	=	dichloromethane
EBC	=	European Brewery Convention
FAN	=	free-amino nitrogen
HWE	=	hot water extract
GC–MS	=	gas chromatography-mass spectrometry
LAA	=	liters of absolute alcohol
OG	=	original gravity
SIM	=	selected ion monitoring
SPME	=	solid-phase microextraction.

Acknowledgments

The authors thank Maarten Gorseling and Alasdair Brown of the International Centre for Brewing and Distilling for their technical contribution to this study.

Additional information

Funding

The authors gratefully acknowledge Crisp Malt (Norfolk, UK), Holyrood Distillery (Edinburgh, UK), Scottish Research Partnership in Engineering (Glasgow, UK), Heriot-Watt University (Edinburgh, UK), and The Macpherson Trust (Newark, UK) for their financial support of this work.