Terahertz Science and Technology, ISSN 1941-7411 Vol.3, No.4, December 2010
between 0.1 – 0.5 THz. Hence, measurements employing terahertz radiation are, in the absence of
significant scattering, dominated by absorption due to moisture content and are relatively
immune to small fluctuations in the concentration of other constituents of the food sample.
Another advantage of this method is the minimal power used for THz-TDS measurements
(typically below 10 μW), which does not result in any significant heating of the sample.
THz-TDS can thus be considered both as a non-invasive and non-intrusive method.
In this paper we present a method for determining the moisture content of intact food wafer
samples by using terahertz transmission measurements. The method is simple, robust and
requires minimal calibration as terahertz absorption is dominated by water content. We also
briefly discuss the possible extrapolation of this method onto mobile, cheaper hardware and its
potential for rapid on-line measurement.
2. Materials and Method
2.1 Wafer Samples
The samples used in this study are rectangular pieces of intact wafer provided by Nestlé, a
photograph and cross-sectional schematic of which are shown in Fig. 1. Wafers are essentially
starch foams produced by depositing wheat flour batter and heating it between two plates [2]. The
two metal plates have an engraved pattern to ease the removal of the wafer sheet once it is baked.
This produces the reeding pattern on the surface of the wafer, as seen in Fig. 1. In addition, these
structures also contribute to the overall texture of the finished product. The flour is predominately
composed of starch with a small amount of gluten and water. Yeast and sodium bicarbonate are
used as aerating agents. Different types of starch can be used for different products depending on
the final desired properties. The wafers are highly porous structures (typically 80 – 90 %
porosity). There is a large distribution of pore sizes across the cross-section of the wafer, with the
larger pores being of the order of hundreds of micrometers, and which are located mostly in the
centre of the sample, while the smallest pores, of the order of a few micrometers, are located
predominately near the edges. High resolution Scanning Electron Micrograph (SEM) and
Confocal Laser Scanning Microscopy (CLSM) images of such wafers have been published
elsewhere [2] and illustrate this typical pore size distribution in the cross-section of the samples.
The wafer samples were initially dried in a dry nitrogen environment over a period of several
weeks to ensure evaporation of the free water; at that point only water that is bound in the starch
matrix was present in the samples. The samples were then exposed to moist air for variable
time periods followed by a THz-TDS transmission measurement. After the terahertz
measurements, the samples were placed in a Sartorius MA-45 moisture analyser (Sartorius AG,
Göttingen, Germany) where the exact moisture content was measured and the initial bound
moisture content was determined. This moisture analyser is equipped with a motorised ceramic
heater generating infrared (IR) radiation and uses gravimetric analysis to determine moisture loss.
The resolution of the balance is 0.1 mg. The accuracy of the moisture analysis with regards to
small samples masses (< 200 mg) has been reported to be ±0.2 % [28]. Two different sets of
samples were used in this study covering the lower and higher end of porosity found in a typical
wafer production line. Type I is a denser type of wafer with an overall porosity of 77 vol %, while
Type II is the lighter wafer sample with porosity of 88 vol %. Type I and II wafers are produced
174