1. Introduction
Worldwide there is an effort being made to spray dry fruit juices. It is now well understood that the combination
of fruit sugars (fructose, maltose, sucrose and glucose) and fruit acids (citric acid and malic acid) leads to a powder
which cannot be dried in a spray drier due to the low glass transition temperature (Tg) of the amorphous mixture that
results from the rapid drying of the juices. To overcome this problem, the common solution is to add a drying aid,
such as maltodextrin, to raise the glass transition temperature to a point where the mixture can be spray dried. This is
usually tackled by trial and error methods although Bhandari et al (1997) (1) used a semi-empirical approach. These
powders are also known to be extremely hydroscopic and cake easily.
A lemon juice powder was produced by first predicting the glass transition temperature of the mixture (Tgmix) by
using a simple weighted additive method of the individual Tg values using the Gordon Taylor approach to estimate
individual Tg values from moisture contents determined by their adsorption isotherms. Where experimental
isotherms for individual components could not be found, as a first approximation, the adsorption isotherm for lactose
was used. The predicted Tgmix for the surface of the powder particles was calculated from the outlet air temperature
and relative humidity conditions which were estimated by a mass and energy balance around the spray drier which
was calibrated for heat losses by running the drier with pure water. This initial work was followed up by drying a
second batch of lemon juice obtained much later in the season. For reasons of availability of the spray drier, it was
necessary to conduct the spray drying runs before it was possible to check the composition of the new batch, so it
was assumed that the composition was the same as the first batch.
This paper characterises the powder obtained from the spray drier in terms of the actual Tgmix as measured on a
DSC compared to the predicted Tgmix values using both the method outlined above and using the Fox equation, and
its tendency to cake when the glass transition temperature is exceeded. It also quantifies the caking behaviour of the
lemon juice powder with time using both the blow test method (2) and a penetration test method (3).