Tuesday, November 18, 2014

Chill filtration

Same whisky, chilled with a drop of water on the left
Most of the Scotch whisky in the market is filtered in some way before bottling. The methods vary from crude physical filtering (ie spirit is run through a cloth or a cellulose sheet) to temperature and pressure controlled nanofiltering systems. 

The main reasons for filtering the spirits are removing any solid components, improving the flavour and avoiding the haze formation later in the bottle or glass. 

Spirits have been chill filtered at least from the late 14th century. Russian vodka was traditionally rapidly cooled after distillation with ice. The excess water transformed into ice and most of the oils hardened on top of that. Filtering became common before the 17th century. The early filters were usually made of cloth (felt, cotton), but also of paper, sand and charcoal. Chemical purification was also used quite early on, just as with wine, mead or beer. Spirit cleansing additives of the 17th and 18th century included ash, potash, burnt wormwood, and the gentry also used milk, eggs, fresh black bread, soda and isinglass to produce top quality vodka. 

Charcoal in Jack Daniel's distillery (alcademics.com)
Coal filtering was also widely used in USA and Canada in the early 19th century. In 1810's Kentucky "layers of white flannel, clean white sand and pulverised charcoal (about 20 inches thick layer) from good green wood such as sugar tree hickory" were used as a filter. The now famous Lincoln county process consisting of maple wood coal filtration at a thickness of 10 feet was invented in 1825 by Alfred Eaton and is still used by Jack Daniel's and George Dickel. Active charcoal was widely used from the start of the 20th century. The first written evidences of active charcoal filters are from 1785 from St Petersburg, Russia. Farmacist T.Y. Lowitz found that raw distillates became clearer and less harsh tasting after treatment with "raw coal".  The activated coal filtering was patented in 1901 in Austria and in 1907 in Russia and used widely ever since, especially in vodka production.

Sand and coal filter tanks in Zhitomir vodka distillery
The distillers filter their spirits for various reasons, most of which are based on traditional practices and empirical evidence, not much has been released as scientific papers. In Tennesee, the Lincoln County process is believed to reduce fusel oils and harsh flavours, resulting in a "less grainy" spirit, according to John Lunn, the master distiller of George Dickel. Most bourbons are filtered before and after aging with active charcoal, usually said to "mellow" the taste. In China, Japan and Taiwan the rice spirits are filtered to remove higher alcohols (fusels) to enhance the flavour. The vodka producers usually just try to remove any colour or taste to render the spirit "clean". Scotch whisky producers claim that their filtering does not affect the flavour and only prevents the haze formation in the bottle or glass. Actually, charcoal is not very efficient in removing oils from the spirit, at least at higher alcoholic strengths and the normal filtration of rice spirits removes only a small fraction of higher alcohols, although an extremely efficient nanofiltration can reduce higher alcohols up to 44%. Interestingly, the amounts of higher alcohols has quite dramatically dropped during the 20th century, which probably diminishes the clouding, but also must affect the flavour, both directly and through lesser high chain ester formation.

EU limit [g/l]Modern spirit, est.Analysis 1905Analysis 1966
Scotch malt10~ 11,5-3,52,7-3,9
Scotch blend10<10,5-1,91,7-2,2
Scotch grain10<10,5-3,00,7-1,3
Bourbon10~ 1,5
Rice spirits102-5


Fruit brandies2


Canadian whisky10

Irish whisky10

Amount of higher alcohols in different spirits [g/l]

The haze formation occurs when the long chain fatty esters become insoluble in lower temperatures or at lower alcoholic strengths. The most crucial esters in haze formation in whiskies are ethyl laureate (ethyl-dodecanoate), ethyl-palmitate (ethyl-hexadecanoate) and ethyl-9-hexadeanoate (ethyl-palmitoleate). Alcocol content of 45% abv is believed to be a crucial limit for these long chain esters to percipitate in room temperature. Other compounds responsible for the cloudiness are high molecular weight lipids, especially sitosterol beta-D-glucoside (probably cask-derived) and slightly ethanol-soluble lignins from the cask.

The haze-forming long chain esters are not very aromatic, ethyl laureate gives some floral, fruity and waxy aromas, as ethyl palmitate and ethyl palmitoleate mostly contribute to the mouthfeel (waxy, oily), although they are reported to give some aromas of coconut and fruits. Probably the more important factor is their ability to act as a surfactant and to enhance/suppress other aromas.

(Coal) filtering increases the formation of short esters and the conversion of aldehydes and ketones to alcohols, probably because of the slight oxiditation caused by the process. On the other hand, some (7%) ethyl acetate (pear-drops) was removed in a typical Scotch filtration, so the net effect remains unclear. The highly efficient nanofiltration and active coal filtration techniques used by brandy and vodka industries remove great deal of long esters and some terpenes. Even a light filtration removed all of the nerol (rose,lemongrass) from apricot brandy and the same probably happens in Scotch whisky filtration. Studies on Glenlivet malt whisky indicate that significant amount of gallic acid was removed in filtering, but the same study found no difference on nosing aroma.
Plate and frame filter in a bottling plant
A typical filtering method used in Scotch whisky industry is a plate and frame filtration with cellulose sheets, medium pressure and cool temperature. Depending on the source the temperature of whisky is lowered usually to +5 - +10C, although some apparently use even colder temperatures (down to -10C?). According to Glenmorangie the cold stabilisation time is 3 hours and apparently up to 24h in some other bottling plants. In the 1990s all of the filters used in Scotch whisky industry were of the plate and frame type. Some distillers have at least experimented with easier and cheaper membrane filters, but there is little information about whether they are in use at a larger scale. The pressure use in the plate and frame filters is usually between 20-60 psi (138-414 KPa), a higher pressure means faster but less complete filtration. The particle retention size is 5-7µm, at least in one bottling plant. Compared to the modern nanofilters at 10nm (0,001µm), that is quite crude.

Plate and frame filter (whisky.com)
Because the filtration processes are not uniform in the Scotch industry, it is not clear how much the process affects the flavour. Certainly you can change the flavour profile and the mouthfeel of an estery and oaky whisky with slow cold filtration. A too strict filtration would likely dimish waxy/oily mouthfeel and floral, fruity and oaky aromas. Whether the current filtering practices do that, is uncertain.

References and further reading.
Braus H et al. Isolation and identification of a sterol glucoside from whiskey. Agr Food Chem 1957;5(6);458-9
Da Porto C. Effects of chill filtration on the composition of grape spirit. Wein-Wissenschaft 2000;55(1);7-12
Duarte FC et al. Physicochemical and sensory changes in aged sugarcane spirit submitted to filtering with activated carbon filter. Ciênc Tecnol Aliment.,2012;32(3);471-477
Glaub, R et al. Effects of various filter systems on sensory quality of fruit brandies. Kleinbrennerei 1998;50 (1);6–12
Himmelstein L. The king of vodka. Harper 2010
Hsieh CW et al. Develop a novel method for removing fusel alcohols from rice spirits using nanofiltration. Food Sci 2010;75(2); 25-29
Ko W et al. Removal of higher fatty acid esters from Taiwanese rice-spirits by nanofiltration.  p353. In Distilled spirits, ed Walker GM et al NOttingham Univ Press 2012.
Lachenmeier D et al. Defining maximum levels of higher alcohols in alcoholic beverages and surrogate alcohol products. Reg Tox Pharm 2008;50;313-321
Miljic UD et al. The application of sheet filters in treatment of fruit brandy after cold stabilisation. Acta per tech 2002;44;87-93
Taylor AJ, Mottam DS. Flavour science: Recent developments, RCS 1996
Persson KM. Med kol och kolonn. Spiritus 2005;7
Piggott JR et ak. The science and technology of whiskies. Longman 1989
Pirie G et al. Membrane filtration of whisky. Food & Drink 2000; p9-13. IChemE 2000.
Pokhlebkin W. A history of vodka. Verso, 1992
Puskas V et al. Influence of cold stabilisation and chill membrane filtration on volatile compounds of apricot brandy. Food Bioprod Proc 2013;91;348-351
Schidrowitz P, Kaye F. The determination of higher alcohols in spirits. Analyst 1906;31;181-194
Singer DD. The analysis and composition of potable spirits. Analyst 1966;91;127-134
Wisniewski I. Filtering out. Whisky Magazine 2011;97;27
Malt Maniacs E-pistle: http://www.maltmaniacs.net/E-pistles/Malt-Maniacs-2011-06-Chill-filtration-and-cloud-formation-in-whisky.pdf
Whisky.com study: http://www.whisky.com/information/knowledge/science/study-on-the-chill-filtration-of-scotch-single-malt-whiskies.html


  1. A very interesting, detailed and scientific explanation - this is certainly the type of science many whisky enthusiasts would like to read about.

  2. A great article. I'd be interested to know if you had any reference to the micron size, i.e. maximum particle size allowed through, in a typical non chill barrier filter. This is a step nearly all manufacturers will be taking and there will be a balance between polish and the risk of removing any terpenes, fatty acids, etc,