Near infra-red spectroscopy for identifying malting quality


Cereals & Oilseeds
Project code:
01 May 1987 - 30 April 1990
AHDB Cereals & Oilseeds.
AHDB sector cost:
£136,664 From HGCA Project No (0029/1/87)
Project leader:
E D Baxter, and W McGill Brewing Research Foundation



About this project



The objective of this project was to investigate the use of Near Infra-Red (NIR) spectroscopy of whole grain barley to predict the concentrations of components which will be present after malting and thus to assess the malting quality of that barley.

Areas of investigation

The measurement of components by NIR relies on correlations between the sample spectra and the reference analysis. These correlations will hold for future samples only if all possible sources of variation are present in the calibration set. To achieve this a large number of samples, including different varieties, geographical sites and harvest years, were used to develop calibrations.

The diffuse reflectance spectrum of each sample was recorded using a Perstorp Analytical 6250 scanning spectrophotometer. All samples were then malted, using a standard malting regime, and analysed by recommended Institute of Brewing methods.

Calibrations have been developed for several varietal subsets, from a single variety to a range of malting and feed varieties. Winter and spring barleys were analysed separately.

Within each subset the samples were divided into those used for calibration and those used for testing the equations generated.

The spectra were derivatised, to aid peak resolution and reduce interferences, such as those caused by particle size effects. Multiple linear regression analysis was then carried out on the calibration samples. This technique identifies groups of wavelengths which correlate well with the reference data and incorporates them into predictive equations. The analytical performance of the equations was then determined by testing them on the prediction sets.

Key findings

1. Hot Water Extract (HWE), Total Nitrogen (TN), and Total Soluble Nitrogen (TSN) of the malt can be predicted from scans of whole grain barley. Once calibrations have been developed, approximately 10-12 samples/hour can be analysed for all 3 parameters.

2. Calibrations developed using several malting varieties had similar levels of error to those developed using a single variety. It is not, therefore, necessary to develop equations for each individual variety.

3. Calibrations developed using malting and feed barleys are accurate enough to provide a useful screening tool, for example, to help selection during barley breeding programmes. Spring barleys from the 1987 harvest, which were judged acceptable for malting on the basis of reference HWE values, were correctly identified by NIR in 89% of cases. Similarly, 81% of those unacceptable for malting were correctly identified.

4. In most cases equations can be used, with similar levels of accuracy, for samples from new seasons. Combining data from more than one harvest sometimes resulted in more accurate equations.

5. Calibrations for combined malting and feed winter barleys are less stable to seasonal variations than are spring barley calibrations and combining data from more than one harvest was unsuccessful.

By using near infra-red (NIR) analytical methods the malting potential of small samples of barley can be analysed in minutes instead of days.

The research evaluated the technique for a wide range of anlaytical data from hundreds of barley samples drawn from numerous varieties over three successive harvests grown on sites from Scotland to the south of England. The aim was to compare the results of NIR analysis of whole grains with the traditional evaluation of micro-malting.

In the past small barley samples have been tested by first malting them and then mashing the resultant malt and analysing the wort. The technique investigated in this project takes a handful of whole grains. This is analysed by NIR spectroscopy which uses a light source to scan the sample and provide an analysis of several of the complex components of the grain such as starch and protein. The measurements are compared, by computer, against standard data to predict the sample's malting quality in terms of the Hot Water Extract, Total Nitrogen and Total Soluble Nitrogen. The new technique takes minutes compared to the eight or nine days required for micro-malting.

The technique is sufficiently accurate to be used as a scanning tool in barley breeding programmes and should allow selections for malting quality to be made at least one year earlier than is currently possible.