Journal of Food Engineering 114 (2013) 1421

Contents lists available at SciVerse ScienceDirect

Journal of Food Engineering

journal homepage: www.elsevier.com/locate/jfoodeng

Comparative study of deteriorative changes in the ageing of milk powder

Nima Yazdanpanah , Tim A.G. Langrish
Drying and Process Technology Group, School of Chemical & Biomolecular Engineering, The University of Sydney, NSW 2006, Australia

a r t i c l e

i n f o

Article history:
Received 17 May 2012
Received in revised form 17 July 2012
Accepted 27 July 2012
Available online 4 August 2012
Keywords:
Milk powder
Crystallisation
Shelf-life
Protein modication
Egg-shell particles
Physical properties

a b s t r a c t
Skim milk powders produced by spray drying (mostly amorphous) are commonly used in downstream
food industries and have to be stored for long times after production before nal use. However, their
physicochemical qualities and properties decline during long-term storage due to the hygroscopicity of
amorphous lactose. Surface modication of particles by crystallisation of the outer layer, giving so called
egg-shell particles, has shown signicant improvements in the physical properties of the powder in
their fresh state. In this study, the raw powder and processed powder (with modied surface) were stored
at around 33% relative humidity and 2530 C for 30 weeks to investigate the effect of ageing on these
two types of powders. Agglomeration, large lactose crystal formation on the surface, surface composition
changes and protein modications were studied. The changes between the raw powder and process powder were compared after ageing. The non-hygroscopic crystalline surface layer showed signicant benets in maintaining the physicochemical qualities of the powders over long storage times. The aged raw
powder showed a 24% change in the crystallinity, a 3% change in the lactose/protein ratio on the surface
and 6% protein denaturation compared with the aged processed powder with a 4% change in the crystallinity, a 1.5% change in the lactose/protein ratio on the surface and 2% protein denaturation, respectively,
after 30 weeks storage.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
Milk powders are widely used in food processing industries or
for re-constituting and often must be stored for months. The shelf
life for a food product is marked by an increase in undesirable
physicochemical qualities or microbial levels. Spray-dried milk
powders, in amorphous states, are very unstable. They have a tendency to sorb moisture and form caked powders that are not free
owing. The increase in moisture content, due to the elevated
water activity, causes bridge formation between particles, protein
denaturation and modication (Haque et al., 2010; Liu and Chaudhary, 2011), the progress of Maillard reactions and decolouration
(Bell, 2008), bacteria growth (Tapia et al., 2008) and quality loss
(Sablani et al., 2007). These deteriorative changes start by sorbing
moisture form environment, leading to a decrease in the glasstransition temperature of amorphous components to below ambient conditions, which initiates crystallisation and an increase in
the stickiness of the powders. The expelled water from crystallisation further increases the water activity and enhances the changes.
In addition, crystallisation of lactose from the matrix rearranges
the proteinlactose bonds and de-stabilizes the proteins, meaning
that less amorphous lactose is available around the proteins to be
attached to the H-bond active sites of the proteins (Buera et al.,

Corresponding author. Tel.: +61 02 93515661; fax: +61 02 93512854.

E-mail address: nyaz3239@uni.sydney.edu.au (N. Yazdanpanah).
0260-8774/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jfoodeng.2012.07.026

2005; Lpez-Dez and Bone, 2000). Amorphous component of powders readily sorb moisture from the environment, which leads to
high moisture-content gradients near the surface, initiating moisture migration into the particles.
There are different methods that are commonly used to prevent
the powders from sorbing moisture, such as keeping them in cool
and dry conditions and using sealed impermeable packaging.
Crystallisation of lactose in milk powders in pre-crystallisation or
post-crystallisation facilities has been suggested (Hynd, 1980;
Yazdanpanah and Langrish, 2011b) to improve the powder
stability against moisture sorption and enhancing the physical
properties. Yazdanpanah and Langrish (2011b) showed the
improvement (decrease) in moisture sorption for the crystallised
lactose and milk powders that were crystallised in a uidizedbed dryer/crystallizer. The mostly crystalline powders that were
produced by that technique have a very much reduced tendency
to sorb moisture from the environment (even under very humid
conditions for a long time).
Amorphous materials have greater solubility, porosity (Trivedi
and Axe, 2001) and bioavailability (Yang et al., 2010), compared
with crystalline materials. The lower solubility and higher stability
of the milk powders with crystallised lactose (mostly crystalline
particles) could be disadvantages when the powders are needed
to be reconstituted or dispersed in dairy processing industries.
Sandiness and rough texture in ice-cream and chocolate that is
made from this kind of crystallised milk powder are not desirable
in sensory analysis (McSweeney and Fox, 2009).

N. Yazdanpanah, T.A.G. Langrish / Journal of Food Engineering 114 (2013) 1421

The destabilisation effect of lactose crystallisation during storage and the elevated water activity also cause protein denaturation,
releasing encapsulated ingredients, such as fat, and decreasing the
nutritional value of the stored powder (Augustin et al., 2007; Fyfe
et al., 2011). Proteins are stable in their low-energy, native (folded)
state. Denaturation or unfolding, which are changes in the secondary structures of proteins, increase the hydrophobicity of the proteins (Haque et al., 2010) and promote aggregation and
destabilisation (Allison et al., 1999). Most dairy powders have
shelf-lives of around 36 months in cool and dry conditions when
packed in sealed moisture-proof containers. The moisture sorption
and diffusion rate, and consequential lactose crystallisation and
protein modication, increase the rate of quality loss in the
powders.
Most of the physical properties in spray-dried milk powders,
such as owablity, stickiness, agglomeration and caking, are surface-dominated properties (Fldt and Bergensthl, 1994;
Yazdanpanah and Langrish, 2010). Surface amorphicity and composition of the powder particles may then be expected to play an
important role in their ow behaviour, because owability involves overcoming the surface attractions between the particles.
Amorphous and hygroscopic materials on the surfaces have very
signicant effects on the agglomeration, stickiness and hygroscopicity of the particles. Moisture sorption by crystalline lactose is
signicantly lower than by amorphous lactose (Bronlund and Paterson, 2004), and this lower amount of moisture on the surface
causes fewer moisture-induced changes inside particles. Also, particles with crystalline surfaces have weak tendencies to form
bridges with other particles due to the high thermodynamic stability of the crystalline state. Yazdanpanah and Langrish (2010) have
demonstrated that a thin layer of modied molecular structure on
the surfaces of particles can greatly improve the physical properties of the powders. They created a new milk powder structure that
was called an egg-shell structure, with a crystalline surface and an
amorphous core, which has the good owability and stability characteristics of crystalline powders, while preserving the desirable
characteristics of the amorphous core. The previous report shows
signicant improvements in stability, owability and physical
properties of the freshly-processed powders with egg-shell structures compared with raw commercial powders, while maintaining
the same functionality. There are few reports in the literature
about long-term studies of processed powders, and this study addresses this issue.
The aim of this study was to investigate the water-induced
changes in the skim milk powders during storage, comparing the
raw commercial powder with the processed powders, with an
egg-shell structure, during long storage. The effect of the crystalline outer layer on the physical properties of the fresh powder
has been studied before (Yazdanpanah and Langrish, 2010); in this
research, the effects of the crystalline surface layer on the moisture
sorption rate and the deteriorative changes in milk powder have
been investigated.

2. Materials and methods

2.1. Materials
2.1.1. Raw skim milk powder
Skim milk powder (medium heat) has been supplied by Murray
Goulburn Cooperative Co. Ltd. (Brunswick, VIC, Australia). The
manufacturers specication data sheet has showed that the composition on a dry weight basis was lactose (59% w.w 1), protein
(39% w.w 1) and fat (0.9% w.w 1). The moisture content of the
raw milk powder has been measured to be 3.8% w.w 1 from oven
drying tests. This powder will be called raw powder hereafter.

15

2.1.2. Processed powder

Processed powder with egg-shell structure particles has been
made by a uidized-bed crystallisation technique described before
(Yazdanpanah and Langrish, 2010). The processing condition was
60 C, 40% RH and 20 min processing time. After processing, the
powder was transferred to a vacuum dryer and left under vacuum
at room temperature for 4 h to slow down further crystallisation.
The processed powder has been analysed by previously-described
methods to conrm that it has an egg-shell type structure, and the
amorphicity of the core of the particles has been checked.
2.2. Methods
2.2.1. Experimental setup
The raw powder and processed powder have been stored under
controlled humidity and temperature conditions at 2530 C and
3235% RH for 30 weeks. Both raw and processed powders have
been kept separately in the same sorption box at these constant
temperature and humidity conditions for 30 weeks to age the powders. The powders have been left under vacuum for 4 h to remove
the initial moisture content prior to placing them in the sorption
box containing saturated MgCl2 solutions giving 33% relative
humidity. A Tiny Tag Extra TGX3580 datalogger device from Gemini Data Loggers has been used for recording the temperature and
the relative humidity during the storage period. The equilibrium
moisture content of skim milk powder at 33% relative humidity
is 7.3% (w.w 1), and the glass-transition temperature was reported
as 22 C (Schuck et al., 2005). The samples have been examined
when fresh and after 15 and 30 weeks of storage to assess any
changes in crystallinity and protein modications. The tests after
15 weeks storage time have showed that this time period was
not adequate to develop the changes to an extent that was readily
detectable with the sensitivity of the equipment used here, such as
XRD. Therefore the 30 weeks aged powders were compared with
the fresh state of the powders.
2.2.2. Scanning electron microscopy
A scanning electron microscope was used to observe the surface
morphology of the powders. The samples were prepared by placing
a small amount of each sample on a carbon tape that was placed on
an aluminium sample disc. The sample was coated by a standard
30 nm gold layer to produce the conductive surface (Emitech,
K550X, Quorum Technologies, UK). The electron micrographs were
produced using a Zeiss ULTRA plus (Carl Zeiss SMT AG, Germany)
scanning electron microscope (SEM) in the In-Lens mode with an
operating voltage of 2 keV. A range of 50030,000 times magnication was used in the images.
2.2.3. X-ray diffraction (XRD)
X-ray diffraction (XRD) was used to investigate the bulk crystallinity of raw and processed powders by using a Siemens D5000 diffractometer. The scanning range was set to 530, the step size was
0.02 with a scanning rate of 1 step/s, and the operating conditions
were 40 kV and 30 mA. The EVA evaluation program (DIFFRAC
Plus, Bruker analytical X-ray system, GmbH) was used for peak
searching and calculating under-peak areas as part of the quantitative crystallinity analysis. Each scan has been repeated three times,
and the average value and the standard deviation have been
reported.
2.2.4. Fourier transform infrared spectroscopy (FTIR)
Attenuated total reectance (ATR) spectra were acquired using
a single bounce diamond ATR (Universal ATR) in a nitrogen purged
Nicolet 6700 FTIR spectrometer (Thermo Fisher Scientic Inc) controlled by OMNIC 8.2.387. The FTIR spectra were collected at a resolution of 4 cm 1 with 32 scans over a range of wavelengths from

16

N. Yazdanpanah, T.A.G. Langrish / Journal of Food Engineering 114 (2013) 1421

2000500 cm 1. The collected data were analysed using the software Grams/AI 8.0 (Thermo Electron Corporation) for further data
processing, such as least-squares curve tting and analysis (Gaussian + Lorenziane) and the calculation of the under-peak areas for
quantifying the secondary structure of the proteins in the Amide
I and Amide III regions. Second-derivative spectra were calculated
from normal spectra by means of the SavitskyGolay algorithm
with a nine data point smoothing factor. The powders were kept
in good optical contact by the diamond surface with a consistent
application of low pressure. The powders have been kept under
vacuum for four hours prior to scanning to remove the free moisture content and to minimize the effects of the H-bonds from water
vapour. The analysis has been done for the shortest possible time
to avoid the particles being exposed to humid air. Each experiment
has been repeated three times, and the average value and the standard deviation have been reported.
2.2.5. X-ray photoelectron spectroscopy (XPS)
An XPS system of SPECS-XPS (SPECS, Germany) was used to
analyse the element concentrations and binding energies of the
particle surfaces. The XPS system was equipped with a PHOIBOS
150-9 MCD energy analyser, and an Al K-alpha X-ray monochromatic source (1486.74 eV) was used. Since we are using skim milk
powder, it is possible to assume the surface components are
mainly proteins and lactose with small amounts of fat. Therefore
the equations from the literature (Fldt and Bergensthl, 1994)
and the average of the standard component values for lactose, protein and fat (Fldt and Bergensthl, 1994; Gaiani et al., 2006; Kim
et al., 2005; Vignolles et al., 2009) were used to calculate the surface compositions of the powders.
2.2.6. Raman spectroscopy
Raman spectroscopy was used to evaluate the crystallinity of
lactose on the surfaces of the particles. The Raman spectra were
collected using a Raman Station 400F (PerkinElmer, CA, USA). The
samples were analysed using a power of 100% with a 785 nm laser
and a ve second exposure time, with ve exposures over a range
of wavelengths from 200 to 3200 cm 1. The spectra were analysed
using the software Spectrum v6.3.4.0164. The intensities of characteristic peaks of lactose at 355, 445, 865 and 1100 cm 1 were used
as references (Kirk et al., 2007; McGoverin et al., 2010; Murphy
et al., 2005; Susi and Ard, 1974). Each experiment has been repeated two times.

3. Results and discussion

3.1. Surface morphology and changes on the surface of the particles
The surface of the freshly processed powder has a crystalline
texture covered by ne lactose crystals, which were expected for
egg-shell structure particles on the basis of previous work
(Yazdanpanah and Langrish, 2010) (Fig. 1 FP). The particles of fresh
raw powder have smooth surfaces with no lactose crystals (Fig. 1
FR). SEM micrographs show some morphological changes for the
powders during the storage time. Large lactose crystals have appeared on the surfaces of the raw powder particles. The particles
are agglomerated together substantially and the large lactose crystals have made bridges between particles. The amount and size of
the lactose crystals formed during the storage time on the raw
powders (Fig. 1 AR) were larger than those on the aged processed
powders (Fig. 1 AP). The growth of lactose crystals on the surfaces
of aged processed particles is evident, but the size and amount are
very small; there are no bridge formed between the particles and
the powders are still free-owing.

The signicant lactose crystal formation on the surface of the

raw powder is due to the high availability of amorphous lactose
there, the hygroscopicity of amorphous lactose on the surface
and the higher permeability of amorphous lactose. The hygroscopic
amorphous lactose of the surface on the raw powder sorbs moisture and increases the water activity (Bronlund and Paterson,
2004), make the molecules more mobile and lowers the glass-transition temperature. This higher water activity and signicant
amount of expelled water from crystallisation on the surface
causes other changes subsequently, such as component migration
or protein deformation, which will be discussed in detail in the
coming sections. The small lactose crystals on the surfaces of the
aged processed powders might have been formed from joining
small lactose crystals that were formed in the processing stage.
The ne lactose crystals might have acted as nuclei to increase
the crystal growth rate, but the low molecular mobility, because
of the low amount of sorbed moisture and the high viscosity at
the higher glass-transition temperature, might have prevented
the formation of large crystals. The internal microstructure of the
particles appeared to remain similar during storage.
3.2. Crystallinity of the powders
The sorbed moisture on the surfaces of the particles and the
moisture diffusion through the amorphous surface layers into the
particle core decreases the glass-transition temperature and viscosity and increases molecular mobility, which induces lactose
crystallisation. As SEM micrographs showed large lactose crystals
on the surface of the aged powder particles (more on the surfaces
of the aged raw powders), the surface crystallinity of the powders
have changed during the storage time. Surface crystallinity can be
measured by Raman and FTIR spectroscopy; XRD is a technique for
determining mainly the bulk crystallinity (Yazdanpanah and
Langrish, 2010).
Kirk et al. (2007), Murphy et al. (2005), Seyer et al. (2000) and
Susi and Ard (1974) found characteristic peaks for a- and b-lactose
in their study, which are due to the characteristic stretching and
bending vibrations of the COC bridge grouping. However, the characteristic and clear peaks are difcult to recognise in these spectra.
The Raman spectra, shown in Figs. 2 and 3, show the changes in
crystallinity of the powders after 30 weeks ageing compared with
the fresh powders. The surfaces of the fresh processed particles,
which were covered by crystalline lactose, Fig. 2 (bottom curve),
show sharp peaks at characteristic Raman shift points for lactose
at around 355, 445, 865, 1100 cm 1 (Murphy et al., 2005; Seyer
et al., 2000).
The increase in the peak intensity with more surface crystallinity is expected, and while there may be a slight shift in the peak position and changes in the sharpness/width of the peaks due to
differences in geometry, size and orientation of the newly-formed
(or grown) crystals. It is very difcult to use Raman spectra in a
quantitative way since there is no reference to which peak is more
dominant for different amounts of crystallinity. Murphy et al.
(2005) used the Raman depolarisation ratio to determine the symmetry of a molecular vibration in the measured sample form and
tried to numerically analyse the degree of crystallinity. Katainen
et al. (2005) used the peaks at 440 and 470 cm 1 for evaluating
amorphous lactose in pure lactose (Katainen et al., 2005). The measurement of crystallinity based on their approach was not successful here, possibly due to the presence of other complex materials
(proteins and minerals), which masked the crystals and suppressed/depressed the bond vibrations (intensity) causing a lack
of repeatability for those regions. The same peaks at the same
places in the processed powders were observed (355, 445,
865,1100 cm 1), but these peaks are weaker and have lower intensities, possibly due to smaller lactose crystals on the surfaces.

N. Yazdanpanah, T.A.G. Langrish / Journal of Food Engineering 114 (2013) 1421

17

Fig. 1. (FR) Surfaces of fresh raw powder (fully amorphous). (AR) Aged raw powder after 30 weeks in storage. (FP) Surfaces of fresh processed powder, textured structure of
crystalline outer layer with ne lactose crystals. (AP) Aged processed powder after 30 weeks in storage.

Fig. 2. Raman spectra of the processed milk powder in the fresh condition and after
ageing, showing the similarity in the surface crystallinity.

Fig. 2 shows the processed powder for fresh and aged conditions. The spectra include the sharp peaks due to crystalline lactose
in both samples with slightly higher intensities for the aged processed powders. In contrast, as Fig. 3 shows, the differences in
intensity (and crystallinity) are much higher for the raw powder
under fresh and aged conditions. The difference is clearly signicant (qualitatively) between the bottom line for the fresh raw powder and top line for the aged raw powder in the Fig. 3, when the
representative peaks have higher intensity in the aged powders
with greater degree of crystallinity.
The surfaces of the raw powders after storage for 30 weeks have
changed signicantly compared with the aged fresh powders. Sim-

Fig. 3. Raman spectra of the raw milk powder in the fresh condition and after
ageing, showing the signicant difference in the surface crystallinity.

ilar behaviours have been observed in the FTIR analysis of the powders at 8501200 cm 1 (results not shown here) where the aged
raw powder has more noisy spectra with sharper peaks in the characteristic lactose region (Hogan and OCallaghan, 2010). The sharp
peaks at 1260, 900 and 875 cm 1 of the FTIR spectra have been reported to differentiate crystalline from amorphous lactose (Listiohadi et al., 2009).
The XRD technique is a well-established method for quantifying
the degree of bulk crystallinity and the crystal types in terms of the
a/b lactose ratios in the powders. The most noted representative
peaks were located at 12.5, 16.4, 20.1 for a-lactose monohy-

18

N. Yazdanpanah, T.A.G. Langrish / Journal of Food Engineering 114 (2013) 1421

drate; 10.5, 21 for anhydrous b-lactose; 18.2, 19.1, 21.1 for the
mixture of anhydrous a/b with a molar ratio of 5:3, and 19.5,21.2
for mixture of anhydrous a/b with a molar ratio of 4:1 (Barham
et al., 2006; Gombas et al., 2002). The full spectra have not been
shown here, and relative crystallinity has been estimated according to the method of Nara and Komiya (1983), from the ratio of
the peak areas, which is also called the two-phase method
(Kim et al., 1997; LeCorre et al., 2011; Nara and Komiya, 1983),
for different powders relative to the mostly-crystalline skim milk
powder as produced by techniques that have been described before (Yazdanpanah and Langrish, 2011a). This summation does
not take into account the recrystallisation between the different
polymorphs. In this case, individual peaks should be considered.
Table 1 shows the measured bulk crystallinity for the fresh and
aged powders by the XRD technique. The fresh raw powder was
completely amorphous, and the spectrum was very noisy with no
characteristic peak (not shown here), therefore the associated peak
area was assumed to be zero for the fresh raw amorphous powder
in the X-ray spectra. The freshly processed powder has 8.9% crystallinity, which was mainly located on the surface of the particles,
as Raman spectra shows and as described before (Yazdanpanah
and Langrish, 2010). The aged raw powder showed 23.7 0.1%
crystallinity and the aged processed powder showed 12.7 0.1%
crystallinity. The crystallinity of the raw powder changed from
0% to 23.7% and the crystallinity of the processed powder changed
from 8.9% to 12.7%. It is difcult to determine the location of crystals and determine if all the crystallinity refers to the surface or
some crystals developed in the core of the particles. Since the
moisture diffusion induces the crystallisation within the particle
matrix, the crystal formation could have occurred inside the particle. The change in crystallinity of the aged powders relative to the
raw powder was 23.7 0.1% compared with the 3.8 0.1% change
for the aged processed powder. This change is signicantly higher
than the processed powder and shows the effect of the non-hygroscopic outer layer in decreasing moisture sorption and reducing
the diffusivity of moisture into the particle. T-tests have shown
that there are signicant differences between the fresh and aged
powders at a 95% condence level.
3.3. Surface composition
XPS (also known as ESCA) was used to determine the surface
composition of the powders. Milk powder is a multi-component
mixture, with lactose, protein and fat being the main components.
The main purpose of using this technique was to measure the
changes in the lactose and protein concentrations on the surfaces.
The powders were analysed before and after storage. The changes
in the percentage compositions of different components on the
surface were calculated by a matrix formula method described
by Fldt and Bergensthl (1994) and Kim et al. (2009). Table 2
shows the surface composition analysis for the different powders
in this study. In all the powders, the bulk fat was released to the
surface. The freshly-processed powder and aged powders have
more fat on their surface. This observation agrees well with a
recent study for releasing fat during high-humidity and hightemperature processing for producing crystalline powders in a

uidized-bed dryer (Yazdanpanah and Langrish, 2012), increasing

surface fat content during storage of milk powder (Fldt and
Bergensthl, 1996) and releasing lipids in the storage of proteins
(Gaiani et al., 2009). Fyfe et al. (2011) reported signicant changes
in the surface lactose content of milk protein concentrate (MPC85)
after more than 90 days storage (Fyfe et al., 2011). The lactose concentrations in the fresh raw powder and freshly processed powder
were similar, 37% and 36%, respectively, which shows restricted
migration of the small molecules (lactose in this case) from the
matrix to the surface; but the fat concentration is higher on the
processed powder surfaces due to the melting of fats during
high-temperature processing in the uidized-bed dryer. This
contrast could be due to the short processing time (30 min) when
lactose molecules have insufcient time to move to the surface,
compared with the fat that diffuses in the liquid state with a higher
diffusion rate than solid lactose.
On this basis, the fresh raw powder and aged raw powders have
37% lactose on the surface, but the lactose to protein ratios were
74% and 77%, respectively. Statistical analysis show, that this three
percent change, which develops just 1.5 percent in the lactose concentration, is signicant compared with the aged processed powder. The processed powder was more stable during storage, in
terms of lower component migration. This stability again shows
the role of the non-hygroscopic crystalline layer in retaining the
original state of the matrix by reducing moisture diffusion. The results have been supported by SEM micrographs that show fewer
lactose crystals on the surfaces of the processed particles. This
composition change could occur because of molecular migration
towards the surface or inwards to the centre due to differences
in the solubilities of protein and lactose. The more soluble component (lactose in this case) could form a highly-saturated solution
by the sorption of moisture, diffuse to the surface and create more
lactose crystals. Less soluble materials are likely to behave the
other way (Meerdink and vant Riet, 1995).
The increase in non-polar bonds, crystallinity and cross-linked
proteins on the surface can affect the rehydration rate of the particles (Fyfe et al., 2011; Gaiani et al., 2006, 2009; Haque et al., 2010).
The changes are associated with an increase in the hydrophobicity
of the protein on the surface and the decrease in solubility of the
large lactose crystals.
3.4. Protein conformation modication
Fourier transform infrared spectroscopy (FTIR) have been used
to study lactose crystallisation in milk powder (Lei et al., 2010;
Listiohadi et al., 2009) and changes in protein secondary structures
(Cai and Singh Bal, 1999; Chittur, 1998; Dong et al., 1990; Farrell
et al., 2001; Goormaghtigh et al., 2009; Haque et al., 2010; Huson
et al., 2011). Three broad peaks, with ranges around 16001700,
15001600 and 12001320 cm 1 stand for the vibrations of amide
I, II and III groups in proteins. The area between 800 and 1200 cm 1
presents the characteristic peaks of various CO vibrations in
carbohydrates.
Types of secondary structure of polypeptide chains, such as ahelices, b-sheets, b-turns, and random coils in protein, are most often extracted from the IR spectra in the amide I region (Fig. 4),

Table 1
Calculation of crystallinity by summation of peak areas under the characteristic peaks of XRD spectra for different powders.
Powders

Peak area

Amount of crystallinity (% of total)

Change in crystallinity (aged-fresh/fully crystalline, %)

Fresh raw
Aged raw
Fresh processed
Aged processed
Fully crystalline

0
32.1 0.1
12.0 0.05
17.1 0.1
135.0 0.7

0
23.7 0.1
8.9 0.05
12.7 0.1
100

23.7 0.1

3.8 0.1

19

N. Yazdanpanah, T.A.G. Langrish / Journal of Food Engineering 114 (2013) 1421

Table 2
Different surface compositions of powders and atomic percentages of components detected by XPS analysis.
Powders

Fresh raw
Aged raw
Fresh processed
Aged processed
a

Relative atomic concentration (%)

Surface composition (%)

Fat

Protein

Lactose

L/Pa

65.29
66.56
66.73
66.92

26.84
26.83
26.54
26.48

6.25
6.00
6.12
6.00

12
15
15
16

50
48
49
48

37
37
36
36

74.0
77.0
73.4
75.0

Lactose to protein ratio.

which represents primarily the C@O stretching vibrations of amide

groups, and the amide III region, which has been described as arising predominantly from CN stretching vibrations coupled with N
H bending vibrations, with weak contributions from CC stretching
and C@O bending (Bandekar, 1992; Cai and Singh Bal, 1999; Haris
and Severcan, 1999). There is no H2O interference in the amide III
region; therefore, even though the signal from the amide III bands
are much weaker than from the amide I bands, the amide III region
is very useful for estimating protein secondary structure contents
(Fig. 5). In this research, the populations of the secondary structure
elements were studied in both the amide I and amide III regions.
Fig. 6 shows the least-squares analysis for the consistency of the
data, which shows the secondary structure recognised from the
amide I region supported by the results from the amide III region
(R2 = 0.971).
Since a protein usually contains different secondary structural
elements, such as a-helices, b-sheets, b-turns, and random coils,
the amide I and III bands are composite bands consisting of overlapping signals. Fourier self-deconvolution or second derivatives
can be used to identify the number and the positions of the bands
underlying these amide bands. Generally, the second derivative IR
spectra can enhance the resolution and amplify even small differences in ordinary IR spectra. Least-squares curve tting techniques
are particularly useful for precisely following changes in protein
conformations as a function of adsorption time. The technique
has been used for quantitative estimation of different secondary
structures in proteins as a linear combination of individual component bands, by iterative adjustment of the heights, widths, and
positions of these bands. The fractional areas of the individual
bands give the fraction of the secondary structure elements. This
procedure has provided a very good estimate of the protein secondary structure (Chittur, 1998; Haris and Severcan, 1999). Therefore the percentages of b-sheets were calculated by adding the

Fig. 5. FTIR spectrum showing the amide III band of fresh-processed powder. The
outer envelope is the original spectrum, and the individual component peaks
underneath are the results of regression analysis. The peaks are associated with
different secondary structures.

Fig. 6. PLS regression of amide I vs amide III percentage for the population of
different structural components in the secondary structure of proteins.

Fig. 4. FTIR spectrum showing amide I bands of fresh processed powder. The outer
envelope is the original spectrum, and the individual component peaks underneath
are the results of regression analysis. The peaks are associated with different
secondary structures.

areas of all bands assigned to the b-sheets and expressing the ratio
of this sum as a fraction of the total band area, as the population of
b-sheets. The same procedure has been carried on for b-turns and
a-helices in the amide I and III regions. The peak assignment of
deconvolved amide I bands (17001600 cm 1) and amide III bands
(13201200 cm 1) was done using the procedures outlined in previous reports (Bandekar, 1992; Cai and Singh Bal, 1999; Chittur,
1998; Dong et al., 1990; Farrell et al., 2001; Haris and Severcan,
1999; Singh Bal, 1999). The results for the fresh processed powder
have been shown in the Figs. 4 and 5, where Fig. 4 shows the amide
I bands of the freshly processed powder and the associated bands
for the different secondary structural elements; and Fig. 5 shows
the amide III bands of freshly-processed powder with the assigned

20

N. Yazdanpanah, T.A.G. Langrish / Journal of Food Engineering 114 (2013) 1421

Table 3
Population (relative percentages) of different structural components in the secondary structure of the proteins in amide I and amide III regions for different powders.
Powders

Fresh raw
Aged raw
Differences in raw powder
Fresh processed
Aged processed
Differences in processed powder

Amide I

Amide III

a-Helices

b-Sheets

b-Turns

a-Helices

b-Sheets

b-Turns

28.0 0.05
22.1 0.1
6
26.3 0.05
24.0 0.1
2.3

52.9 0.1
57.0 0.05
4
53.8 0.1
54.7 0.1
0.9

19.1 0.1
21.0 0.2
1.9
20.1 0.1
21.4 0.1
1.3

26.1 0.05
20 0.1
6
24.8 0.1
22.4 0.3
2.4

53.8 0.1
56.2 0.1
2.5
54.5 0.3
55.3 0.1
0.8

20.5 0.05
24.1 0.2
3.6
20.9 0.1
22.5 0.3
1.6

bands. The peak at1648 cm 1 was considered to represent a loop

or helix and the peak at 1658 cm 1, a large loop and assigned as
an a-helix and the peak of 1650 cm 1 as an a-helix in the tting
routine for the amide I region; and in the amide III region, the peak
at 1305 cm 1 was assigned as an a-helix. The full list of the peak
positions for all the secondary structures of polypeptide chains,
such as a-helices, b-sheets, b-turns, and random coils in protein,
are available in the before-mentioned references.
The FTIR analyses of the amide I and III regions show some
changes in the secondary structure of the proteins, as shown in
Table 3. The transformations (unfolding) of helical and loop components (folded structure) to sheet and turn components indicate
the changes in the secondary structure of the milk proteins in
the powders after storage. The decrease in the helical components
and a corresponding increase in the b-components is signicant in
terms of the ageing of the powders due to the protein conformational modications and the destruction of the secondary structure
(native structure) of the proteins. After long storage, the proteins
are not completely denatured since they are in the solid form,
but storage in high water activity conditions caused some changes,
in terms of denaturation, modication or cross-linking. The presence of the shoulders, the broadened bands, the appearance of further bands and misplaced bands may be attributed to these
aggregated structures, but the exact explanations are unclear.
The abnormal bands were mostly detected in the aged powders.
Mauerer and Lee (2006) reported some unusual peaks associated
with intermolecular H-bonding in extended polypeptide chains,
non-H-bonded amide groups and high-frequency inter- or intramolecular H-bonding for spray-dried Poly-L-lysine powder (Mauerer and Lee, 2006). T-tests have shown that there are signicant
differences between the fresh and aged powders at a 95% condence level for each amide I and amide III regions separately. In
the raw powder, the population of a-helices signicantly decreased in the aged raw powder (6%) compared with the fresh
raw powder. The amide I shows the decrease from 28.0 0.05%
in the fresh raw powder to 22.1 0.1% in the aged raw powder,
and amide III shows the same 6% decrease in a-helices but from
26.1 0.05% to 20 0.1% in the aged powder. The population of
b-sheets and b-turns increased signicantly in the aged powder
(Table 3). The amide I region showed a 4 % increase in b-sheets
and a 1.9% increase in b-turns , while the amide III region showed
2.5% increase in b-sheets and a 3.6% increase for b-turns. In the
aged processed powder, the decrease in the amount of a-helices
was around 2.3%,the amount of b-sheets increased by about 0.9%
and the b-turns increased by the average of 1.5% (the amide I
showed a 1.3% increase and the amide III region showed a 1.6%).
The reason for the differences in the amount of b-sheets and
b-turns between the amide I and amide III regions was not clear,
but it could be due to multiple bands existing in these regions.
The results for a-helix have been very repeatable for all the powders in the amide I and III, which were used as reference points
for assessing the changes in the secondary structure of the proteins
(unfolding). The reduction in the a-helices content of the processed powder has been signicantly lower than the population
of the a-helices in the raw powder, which suggests more loss of

the native structure for the proteins in the raw powder during storage compare with the processed powders. There was a slight
change in the secondary structure of the protein during processing
at high temperatures and humidities, which was represented by
the decrease in the a-helix population from the raw to the processed powder from 28% to 26.3%. However, the results of the aged
powders showed that the processed powder maintains the native
structure of the proteins better, since the relative fraction of
a-helices in the aged processed powder is higher than in the aged
raw powder, and consequently the relative fractions of b-sheets
and b-turns are lower. The a-structure to b-structure ratio has
been used as a reference for the native structure of the proteins,
and a decrease in helical components has been reported as indicating the loss of the native secondary structure (Chittur, 1998; Goormaghtigh et al., 2009; Haris and Severcan, 1999; Huson et al.,
2011). The role of the crystalline surface layer in maintaining the
secondary structure of the proteins by keeping low internal water
activity appears to have been signicant during storage.

4. Conclusions
This research has assessed the effect of storage on raw skim
milk powders and processed powders, particularly regarding the
characterisation of the powder surface. The paper reports the role
of the crystalline surface layer of the processed powder in preventing (or slowing) deteriorative changes in the particles. The hygroscopicity of amorphous lactose on the surface of the raw powder
and the higher permeability of amorphous lactose cause moisture
sorption and increase the water activity, which provides higher
mobility for the molecules and lowers the glass-transition temperature. This higher water activity and large amount of expelled
water from crystallisation on the surface cause other changes subsequently, such as component migration and/or protein deformation. The processed powder with a higher amount of crystalline
lactose on the surface, has been found to sorb less moisture reducing these subsequent material changes. Surface analysis reports,
crystallinity, composition and protein modication have all indicated that there were much smaller changes in the ageing of the
processed powder in contrast to signicant changes in the aged
raw powder.
Acknowledgements
The authors would like to thank Dr. Elizabeth Carter from the
School of Chemistry, University of Sydney for her helps with the
FTIR analysis. The authors would also like to thank Murray Goulburn Cooperative Co. Ltd. for providing the dairy powders for this
research.
References
Allison, S.D., Chang, B., Randolph, T.W., Carpenter, J.F., 1999. Hydrogen bonding
between sugar and protein is responsible for inhibition of dehydration-induced
protein unfolding. Archives of Biochemistry and Biophysics 365 (2), 289298.

N. Yazdanpanah, T.A.G. Langrish / Journal of Food Engineering 114 (2013) 1421

Augustin, M.A., Udabage, P., Steve, L.T., 2007. Inuence of processing on
functionality of milk and dairy proteins. Advances in Food and Nutrition
Research, vol. 53. Academic Press, pp. 138.
Bandekar, J., 1992. Amide modes and protein conformation. Biochimica et
Biophysica Acta (BBA) Protein Structure and Molecular Enzymology 1120
(2), 123143.
Barham, A.S., Haque, M.K., Roos, Y.H., Hodnett, B.K., 2006. Crystallization of spraydried lactose/protein mixtures in humid air. Journal of Crystal Growth 295 (2),
231240.
Bell, L.N., 2008. Moisture Effects on Foods Chemical Stability. In: Barbosa-Cnovas,
G.V. (Ed.), Water Activity in Foods. Blackwell Publishing Ltd., pp. 173198.
Bronlund, J., Paterson, T., 2004. Moisture sorption isotherms for crystalline,
amorphous and predominantly crystalline lactose powders. International
Dairy Journal 14 (3), 247254.
Buera, M.D.P., Schebor, C., Elizalde, B., 2005. Effects of carbohydrate crystallization
on stability of dehydrated foods and ingredient formulations. Journal of Food
Engineering 67 (12), 157165.
Cai, S., Singh Bal, R., 1999. Determination of the secondary structure of proteins
from amide I and amide III infrared bands using partial least-square method.
Infrared Analysis of Peptides and Proteins, vol. 750. American Chemical Society,
pp. 117129.
Chittur, K.K., 1998. FTIR/ATR for protein adsorption to biomaterial surfaces.
Biomaterials 19 (45), 357369.
Dong, A., Huang, P., Caughey, W.S., 1990. Protein secondary structures in water from
second-derivative amide I infrared spectra. Biochemistry 29 (13), 33033308.
Fldt, P., Bergensthl, B., 1994. The surface composition of spray-dried protein
lactose powders. Colloids and Surfaces A: Physicochemical and Engineering
Aspects 90 (23), 183190.
Fldt, P., Bergensthl, B., 1996. Changes in surface composition of spray-dried food
powders due to lactose crystallization. Lebensmittel-Wissenschaft undTechnologie 29 (56), 438446.
Farrell Jr, H.M., Wickham, E.D., Unruh, J.J., Qi, P.X., Hoagland, P.D., 2001. Secondary
structural studies of bovine caseins: temperature dependence of b-casein
structure as analyzed by circular dichroism and FTIR spectroscopy and
correlation with micellization. Food Hydrocolloids 15 (46), 341354.
Fyfe, K.N., Kravchuk, O., Le, T., Deeth, H.C., Nguyen, A.V., Bhandari, B., 2011. Storage
induced changes to high protein powders: inuence on surface properties and
solubility. Journal of the Science of Food and Agriculture 91 (14), 25662575.
Gaiani, C., Ehrhardt, J.J., Scher, J., Hardy, J., Desobry, S., Banon, S., 2006. Surface
composition of dairy powders observed by X-ray photoelectron spectroscopy
and effects on their rehydration properties. Colloids and Surfaces B:
Biointerfaces 49 (1), 7178.
Gaiani, C., Schuck, P., Scher, J., Ehrhardt, J.J., Arab-Tehrany, E., Jacquot, M., et al.,
2009. Native phosphocaseinate powder during storage: lipids released onto the
surface. Journal of Food Engineering 94 (2), 130134.
}s, I., 2002. Quantitative
Gombas, ., Szab-Rvsz, P., Kata, M., Regdon, G., Ero
determination of crystallinity of a-lactose monohydrate by DSC. Journal of
Thermal Analysis and Calorimetry 68 (2), 503510.
Goormaghtigh, E., Gasper, R., Bnard, A., Goldsztein, A., Raussens, V., 2009. Protein
secondary structure content in solution, lms and tissues: redundancy and
complementarity of the information content in circular dichroism, transmission
and ATR FTIR spectra. Biochimica et Biophysica Acta (BBA) Proteins &
Proteomics 1794 (9), 13321343.
Haque, E., Bhandari, B.R., Gidley, M.J., Deeth, H.C., Mller, S.M., Whittaker, A.K.,
2010. Protein conformational modications and kinetics of water protein
interactions in milk protein concentrate powder upon aging: effect on
solubility. Journal of Agricultural and Food Chemistry 58 (13), 77487755.
Haris, P.I., Severcan, F., 1999. FTIR spectroscopic characterization of protein
structure in aqueous and non-aqueous media. Journal of Molecular Catalysis
B: Enzymatic 7 (14), 207221.
Hogan, S.A., OCallaghan, D.J., 2010. Inuence of milk proteins on the development
of lactose-induced stickiness in dairy powders. International Dairy Journal 20
(3), 212221.
Huson, M.G., Strounina, E.V., Kealley, C.S., Rout, M.K., Church, J.S., Appelqvist, I.A.M.,
et al., 2011. Effects of thermal denaturation on the solid-state structure and
molecular mobility of glycinin. Biomacromolecules 12 (6), 20922102.
Hynd, J., 1980. Drying of whey. International Journal of Dairy Technology 33 (2), 52
54.
Katainen, E., Niemel, P., Harjunen, P., Suhonen, J., Jrvinen, K., 2005. Evaluation of
the amorphous content of lactose by solution calorimetry and Raman
spectroscopy. Talanta 68 (1), 15.
Kim, E.H.J., Chen, X.D., Pearce, D., 2005. Effect of surface composition on the
owability of industrial spray-dried dairy powders. Colloids and Surfaces B:
Biointerfaces 46 (3), 182187.

21

Kim, E.H.J., Chen, X.D., Pearce, D., 2009. Surface composition of industrial spraydried milk powders. 1. Development of surface composition during
manufacture. Journal of Food Engineering 94 (2), 163168.
Kim, J.O., Kim, W.S., Shin, M.S., 1997. A comparative study on retrogradation of rice
starch gels by DSC, X-ray and a-amylase methods. Starch Strke 49 (2), 7175.
Kirk, J.H., Dann, S.E., Blatchford, C.G., 2007. Lactose: a denitive guide to polymorph
determination. International Journal of Pharmaceutics 334 (12), 103114.
LeCorre, D., Vahanian, E., Dufresne, A., Bras, J., 2011. Enzymatic pretreatment for
preparing starch nanocrystals. Biomacromolecules 13 (1), 132137.
Lei, Y., Zhou, Q., Zhang, Y.-L., Chen, J.-B., Noda, S.-Q., 2010. Analysis of crystallized
lactose in milk powder by Fourier-transform infrared spectroscopy combined
with two-dimensional correlation infrared spectroscopy. Journal of Molecular
Structure 974 (13), 8893.
Listiohadi, Y., Hourigan, J.A., Sleigh, R.W., Steele, R.J., 2009. Thermal analysis of
amorphous lactose and a-lactose monohydrate. Dairy Science and Technology
89 (1), 4367.
Liu, H., Chaudhary, D., 2011. The moisture migration behavior of plasticized starch
biopolymer. Drying Technology 29 (3), 278285.
Lpez-Dez, E.C., Bone, S., 2000. An investigation of the water-binding properties of
protein + sugar systems. Physics in Medicine and Biology 45 (12), 3577.
Mauerer, A., Lee, G., 2006. Changes in the amide I FT-IR bands of poly-L-lysine on
spray-drying from [alpha]-helix, [beta]-sheet or random coil conformations.
European Journal of Pharmaceutics and Biopharmaceutics 62 (2), 131142.
McGoverin, C.M., Clark, A.S.S., Holroyd, S.E., Gordon, K.C., 2010. Raman
spectroscopic quantication of milk powder constituents. Analytica Chimica
Acta 673 (1), 2632.
McSweeney, P.L.H., Fox, P.F., 2009. Signicance of lactose in dairy products. In:
McSweeney, P., Fox, P.F. (Eds.), Advanced Dairy Chemistry. Springer, New York,
pp. 35104.
Meerdink, G., vant Riet, K., 1995. Modeling segregation of solute material during
drying of liquid foods. AIChE Journal 41 (3), 732736.
Murphy, B.M., Prescott, S.W., Larson, I., 2005. Measurement of lactose crystallinity
using Raman spectroscopy. Journal of Pharmaceutical and Biomedical Analysis
38 (1), 186190.
Nara, S., Komiya, T., 1983. Studies on the relationship between water-satured state
and crystallinity by the diffraction method for moistened potato starch. Starch
Strke 35 (12), 407410.
Sablani, S.S., Al-Belushi, K., Al-Marhubi, I., Al-Belushi, R., 2007. Evaluating stability
of vitamin C in fortied formula using water activity and glass transition.
International Journal of Food Properties 10 (1), 6171.
Schuck, P., Blanchard, E., Dolivet, A., Mjean, S., Onillon, E., Jeantet, R., 2005. Water
activity and glass transition in dairy ingredients. Lait 85 (45), 295304.
Seyer, J.J., Luner, P.E., Kemper, M.S., 2000. Application of diffuse reectance nearinfrared spectroscopy for determination of crystallinity. Journal of
Pharmaceutical Sciences 89 (10), 13051316.
Singh Bal, R., 1999. Basic aspects of the technique and applications of infrared
spectroscopy of peptides and proteins. Infrared Analysis of Peptides and
Proteins, vol. 750. American Chemical Society, pp. 237.
Susi, H., Ard, J.S., 1974. Laser-raman spectra of lactose. Carbohydrate Research 37
(2), 351354.
Tapia, M.S., Alzamora, S.M., Chirife, J., 2008. Effects of water activity (aw) on
microbial stability: as a hurdle in food preservation. In: Barbosa-Cnovas, G.V.
(Ed.), Water Activity in Foods: Fundamentals and Applications. Blackwell
Publishing Ltd., pp. 239271.
Trivedi, P., Axe, L., 2001. Ni and Zn sorption to amorphous versus crystalline iron
oxides: macroscopic studies. Journal of Colloid and Interface Science 244 (2),
221229.
Vignolles, M.-L., Lopez, C., Ehrhardt, J.-J., Lambert, J., Mjean, S., Jeantet, R., et al.,
2009. Methods combination to investigate the suprastructure, composition and
properties of fat in fat-lled dairy powders. Journal of Food Engineering 94 (2),
154162.
Yang, J., Grey, K., Doney, J., 2010. An improved kinetics approach to describe the
physical stability of amorphous solid dispersions. International Journal of
Pharmaceutics 384 (12), 2431.
Yazdanpanah, N., Langrish, T.A.G., 2010. Egg-shell like structure in dried milk
powders. Food Research International 44 (1), 3945.
Yazdanpanah, N., Langrish, T.A.G., 2011a. Crystallization and drying of milk powder
in a multiple-stage uidized bed dryer. Drying Technology 29 (9), 10461057.
Yazdanpanah, N., Langrish, T.A.G., 2011b. Fast crystallization of lactose and milk
powder in uidized bed dryer/crystallizer. Dairy Science and Technology 91 (3),
323340.
Yazdanpanah, N., Langrish, T.A.G., 2012. Releasing fat in whole milk powder during
uidized bed drying. Drying Technology 30 (10), 10811087.