Development of A Cheap and Rapid Method To Determine Calcium in Milk Fractions in An Industrial Environment
Development of A Cheap and Rapid Method To Determine Calcium in Milk Fractions in An Industrial Environment
Development of A Cheap and Rapid Method To Determine Calcium in Milk Fractions in An Industrial Environment
Daljit Kaur
March 2007
ii
Acknowledgements
I would like to acknowledge the financial support of the Auckland University of Technology,
Auckland, and of New Zealand Dairy Foods Limited, Takanini, Auckland. I also
acknowledge the financial support provided by the Foundation for Research, Science and
Technology (Technology New Zealand, Technology for Industry Fellowships) throughout the
research year. I would like to acknowledge my friend and classmate Iris Ying Yu Li for her
continuous help in computer-assisted methods. I would like to thank my husband and my
sons for being patient and supportive throughout my studies. I would like to acknowledge my
primary supervisor Assoc Prof Owen Young for being an inspiration and for providing me
with enough courage and support to complete this thesis. Similarly I acknowledge my
secondary supervisor Mark Duxbury for his help and support.
iii
Abstract
Milk contains high concentrations of calcium. It occurs in two forms, a free ionic form, and
calcium associated with milk proteins (caseins). The latter association is called colloidal
calcium phosphate. New Zealand Dairy Foods of Takanini is marketing a range of
commercial milks in supermarkets. The company uses ultrafiltration to concentrate milk
proteins and calcium in different milk products. During ultrafiltration, the fraction that is
retained by the membrane is rich in calcium bound to proteins and the portion that passes
through the membrane is richer in the free ionic form. The company wanted to develop a
quick and an economical method that can be applied in industrial settings to determine
calcium in both these fractions and in other milk products.
This research aimed to develop a quick, wet chemistry method to measure calcium in milk
fractions and to trial it in an industrial environment.
Two methods, the so-called EDTA method and the atomic absorption spectrophotometric
method (AA) were trialled as potential reference methods against which to compare results
obtained by the method to be developed. The AA method was chosen due to its ease,
accuracy and precision. (This could not be selected as the industrial method for a number of
reasons.)
A colorimetric method was favoured over other contenders. Two colorimetric dyes, Arsenazo
I and o-cresolphthalein-complexone (CPC) were chosen to work with. Arsenazo I forms a
purple complex with calcium in a suitable buffer at a range of pHs. o-Cresolphthalein-
complexone also forms purple-coloured complexes at alkaline pHs.
During method development with Arsenazo I, different buffers were trialled and a NaOH/
KCl buffer was selected for further development at pH 12. The method worked well during
the development phase but with some inconsistent results at times. o-Cresolphthalein-
complexone formed clear purple complexes with Clark and Lubs and 2-amino-2-
methylpropanol (AMP) buffers. The key advantage of the CPC dye with AMP buffer was
that when 8-hydroxyquinoline was included in the reaction mixture, it successfully masked
coloured complex formation due to CPC with magnesium, which is present in milk at about
1/3 the calcium concentration. This effect did not work with Arsenazo I. However, the
results obtained with the CPC method were lower than claimed values of most milks trialled
during development. Both methods were compared for their precision and it was found that
CPC method has better precision and was chosen for further development.
iv
To improve the accuracy and precision, various denaturing reagents were used to
(hypothetically) release calcium from the caseins. Trichloroacetic acid at 25 % was more
effective than the several other denaturing treatments tested. The finalised CPC method,
using trichloroacetic acid, AMP and 8-hydroxyquinoline, was then used to monitor calcium
concentration over four months in three milk products, skim, Xtra (retentate) and permeate.
For all milks, the CPC values were lower than the AA reference values, and the values
reported by a commercial analytical laboratory. The reasons for this are discussed, as are
other changes in calcium concentration in the three milks throughout the trial.
The correlation between the CPC and AA values was poor for Xtra, better for skim, and best
for permeate. A chemical model to explain this is discussed.
The method developed is cheap and quick, and sample and reagent preparation is simple. The
method could be applied in an industrial environment, but a proportionality factor would have
to be applied to account for the difference in mean values between the CPC and AA methods.
v
Table of contents
Acknowledgements ............................................................................................................ II
Abstract ............................................................................................................................ III
Table of contents ................................................................................................................ V
List of figures .................................................................................................................. VIII
List of tables ...................................................................................................................... X
Statement of originality ..................................................................................................... XI
Intellectual property rights/confidential material .............................................................. XII
1 Introduction .................................................................................................................. 1
1.1 Statement of intent .................................................................................................. 1
1.2 Milk synthesis and its biological role ...................................................................... 2
1.3 Milk proteins .......................................................................................................... 3
1.4 Salts in milk ........................................................................................................... 5
1.5 Bioavailability of bound and unbound calcium ....................................................... 8
1.6 Calcium in the human body and health issues ......................................................... 9
1.7 Sources of calcium in the human diet .................................................................... 10
1.8 Production, processing and consumption of milk in New Zealand ......................... 11
1.9 Ultrafiltration and calcium concentration in milks................................................. 12
1.10 The problem ......................................................................................................... 13
6 Conclusion .................................................................................................................. 61
List of figures
Figure 1: Serine residues esterified to phosphate in caseins .................................................. 4
Figure 2: Submicelle model of casein micelle ...................................................................... 5
Figure 3: Three inorganic phosphate entities in single phosphate containing cluster ............. 7
Figure 4: The calibration curve of calcium standard with AA Elmer Perker ....................... 19
Figure 5: Structure of Arsenazo I ....................................................................................... 24
Figure 6: Structure of o-cresolphthalein-complexone ......................................................... 24
Figure 7: Absorbances of calcium with Arsenazo I dye at 562 nm with Pipes buffer at pH
5.9, 6.9 and pH 7.9.............................................................................................. 28
Figure 8: Absorbances of calcium complexed to Arsenazo I dye at 562 nm with Tris Maleate
buffer at pH 6.0, pH 7.0 and pH 8.0. ................................................................... 29
Figure 9: Calcium and magnesium complexed with Arsenazo I dye at a range of pHs from
highly acidic to strongly alkaline conditions ....................................................... 30
Figure 10: Absorbances of calcium due to complexation with Arsenazo I at various pHs
ranging from highly acidic to highly alkaline from 400 to 800 nm ...................... 31
Figure 11: Absorbances of magnesium due to complexation with Arsenazo I at various pHs
ranging from highly acidic to highly alkaline from 400 to 800 nm ...................... 31
Figure 12: Absorbance spectra of lithium, potassium, magnesium and calcium cations at pH
12 with NaOH/KCl buffer................................................................................... 32
Figure 13: Absorbance spectra of lithium, potassium, magnesium and calcium cations at pH
10 with Clark and Lubs buffer. ........................................................................... 32
Figure 14: Calibration curve for magnesium and calcium when complexed with Arsenazo I
dye in NaOH/KCl buffer at pH 12 at wavelength 562 nm. .................................. 34
Figure 15: Calibration curves for calcium and magnesium with CPC at pH 10.2 with Clark
and Lubs buffer at wavelentgth 575 nm .............................................................. 39
Figure 16: Absorbances for calcium and magnesiumin in the presence of 8-hydroxyquinoline
with CPC at pH 10.2 with Clark and Lubs buffer ................................................ 40
Figure 17: a. Absorbances due to calcium and magnesium with CPC in AMP buffer at pH
10.5 at lower CPC concentration. b. Spectral scans at maximum cation
concentration, 6.25 μM. ...................................................................................... 40
Figure 18: a. Absorbances due to calcium and magnesium with CPC in AMP buffer at pH
10.5 at lower concentraions of CPC and 8-hydroxyquinoline. b. Spectral scans at
maximum cation concentraion 6.25 μM ............................................................. .41
Figure 19: a. Absorbance for calcium and magnesium with CPC in AMP buffer at pH 10.5.
The CPC concentration was 37.5 μM. b. Spectral scans at maximum cation
concentration, 18.7 μM. ...................................................................................... 41
Figure 20: a. Absorbances for calcium and magnesium with CPC in AMP buffer at pH 10.5 in
the presence of 8-hydroxyquinoline. The CPC concentration was 37.5 μM. b.
Spectral scans at maximum cations concentration, 18.7 μM. ............................... 42
ix
Figure 21: Monitoring of calcium in three milk products during the 2005/06 season by the
CPC and AA methods, with periodic comparison to a commercial laboratory
method................................................................................................................ 54
Figure 22: Correlation between AA and the CPC methods for calcium determination in Xtra
milk .................................................................................................................... 56
Figure 23: Correlation between AA and the CPC methods for calcium determination in skim
milk .................................................................................................................... 56
Figure 24: Correlation between AA and the CPC methods for calcium determination in
permeate ............................................................................................................. 57
Figure 25: Correlation between the AA and CPC methods for calcium determination in all
three products. .................................................................................................... 57
Figure 26: Correlation between the Gribbles and AA method for calcium determination in all
three products ..................................................................................................... 58
x
List of tables
Statement of originality
I hereby declare that this submission is my own work and that, to the best of my knowledge
and belief, it contains no material previously published or written by another person, nor
material which to a substantial extent has been accepted for the qualification of any other
degree or diploma of a university or other institution of higher learning, except where due
acknowledgement is made.
Signed…………………………….
Date……………………………….
xii
The parties’ rights to intellectual property owned or licensed to each at the date the student
commenced research for the Company remain untouched. The student’s thesis will be
examined under confidentiality. Any publications from the thesis are subject to an embargo
of 18 months after the date of thesis submission. This will allow the New Zealand Dairy
Foods the time and opportunity to seek patent protection should it so choose. Any other
disclosure of information related to this project in which the Company has a proprietary
interest must be approved by the Company in writing.
1
1 Introduction
This research aimed to develop a quick and economical method in industrial environment to
measure unbound and bound calcium in milk for New Zealand Dairy Foods Limited (NZDF)
Takanini, Auckland, New Zealand.
Calcium is an important mineral in human bones. In Western countries such as New Zealand,
60 % of dietary calcium is obtained from dairy sources, so it is important to dairy processors
to be able to quantify the amount of calcium in dairy products, especially since consumers
recognise the need for calcium to maintain bone health. The industrial environment also
demands an economical, industrially viable and quick method, so that on-site decisions can be
made.
Calcium exists in two main forms in milk. Roughly 66 % is present as colloidal calcium
phosphate (CCP), where the calcium ion is ionically bound to phosphate that is in turn
covalently linked to caseins at a number of serine residues. The remainder is present as free
calcium ion, ionically balanced with phosphate, chloride, carbonate, bicarbonate, sulphate and
citrate (Fox and McSweeney, 1998, p. 249-257)
However, the proportion of calcium as CCP and free ions varies from the 2:1 ratio in response
to changes in overall concentration of calcium in milk, temperature, ionic strength, pH and the
origins of the milk (diet, breed, time of year) (Fox and McSweeney, 1998, p. 243-246).
Different milk processing methods such as ultrafiltration (UF) of milk, yoghurt and cheese
manufacturing, can affect the amount of calcium in the final product. In particular, UF is used
as a means to increase the calcium in specialty milk by concentrating the proteins, so in turn
the bound calcium. As is discussed later there are two views on the importance of calcium in
CCP versus free ions. Some research supports the view that calcium bound to proteins is
more bioavailable than the ionic calcium. Other research contradicts this. Whatever the truth,
NZDF wants a rapid method to determine calcium in its dairy products. There are two
reasons for this. First, it suspects that cheese properties are affected by the free calcium and
CCP concentrations. Second, it may wish to make promotional claims on the ratio of calcium
in the two forms should the bioavailability differ.
• To research and review the current methods for determining bound and unbound
calcium in milk.
• To develop a rapid and cheap method for bound and unbound calcium
determination and to compare it against reference methods for total calcium.
Milk is a fluid secreted by the mammary glands of female mammals to meet the complete
nutritional and several physiological requirements of the neonate. It contains a sugar, fat,
anions and cations, and proteins and peptides, which include immunoglobulins, enzymes and
enzyme inhibitors, binding or carrier proteins and peptides, growth factors and antibacterial
agents (Fox and McSweeney, 1998, p. 1).
In a sense, milk can be viewed as modified secreted blood serum. Only single cell layer
separates the capillaries of the mother from the alveolus lumen, the primary collection cavity
of the mammary gland. Whereas this single cell layer is metabolically very active, some milk
components are transmitted unchanged from serum to milk, and because of the close cellular
connection between blood and milk, the two fluids have identical osmolarity (Swaisgood,
1996, p. 842-846). Significantly, however, milk contains no erythrocytes. The function of
milk is nutrition, not oxygen transport.
Milk is a white or yellow-white, opaque liquid. The colour and opacity arise from scattering
and absorption of light by milk fat globules as a discontinuous emulsion phase and protein
micelles in colloidal suspension within the continuous aqueous phase. The physical
properties of milk are similar to water but are modified due to the presence of various solutes
in the continuous phase and by the degree of dispersion of the emulsified and colloidal
components (Fox and McSweeney, 1998, p. 437).
Milk is principally comprised of about 3.2 % protein, 4.6 % carbohydrate (as lactose), 3.9 %
fat, 0.2 % ash and rest water. It also contains organic acids such as citrate, and a range of
vitamins that do not fit conveniently into the four major categories. Typical composition is
summarised in Table 1 (Swaisgood, 1996, p. 846).
3
Because physiological and nutritional requirements of each species are different, the
composition of milk shows marked interspecies differences. For example, the fat content of
cold-adapted mammals (grey seal, 53 % and polar bear, 33 %) is much higher than that of
others (cow 3.8 %, horse 1.9 %) presumably because of the greater energy-generating
capacity of fat compared with lactose (milk sugar) and proteins (Fox and McSweeney, 1998,
p. 2).
With the domestication of animals, it became possible to include milk of other species in the
diets of adult humans as well as neonates. In much of the world, particularly in the Western
world, bovine milk – commonly called cow milk – accounts for nearly all the milk processed
for human consumption (Swaisgood, 1996, p. 842). In this thesis all investigation and
discussion will refer to bovine milk.
Milk contains 30 to 36 g.L-1 of protein and it rates highly in nutritive quality (Swaisgood,
1996, p. 846). The milk proteins have very high proline content, and certain milk proteins are
rich in lysine that is deficient in many plant proteins. The proline content of proteins makes
them susceptible to proteolysis without heat denaturation, possibly an important characteristic
for neonatal nutrition. However, lack of sulphur amino acids in most of the milk proteins
(unlike egg albumin) limits their biological value (Fox and McSweeney, 1998, p. 166-167).
Milk proteins are usually classed as either caseins or whey proteins. When milk is acidified to
pH 4.6 or exposed to the mammalian stomach enzyme chymosin (also known as rennin),
about 80 % of total bovine milk protein precipitates out of suspension. This fraction is
collectively termed casein, whereas the protein which remains in solution or suspension is
called whey or serum protein, sometimes called non-casein nitrogen (Fox and McSweeney,
1998, p. 149-153). There is a direct analogy between casein precipitation by chymosin and
4
blood clotting. Proteinaceous matter (that includes red cells in the case of blood) forms clots
under the influence of enzyme activity, yielding a soluble fraction termed serum. The name
serum is sometimes used as the equivalent of whey.
Bovine caseins contain four main gene products, designated as αs1, αs2, β and, κ which
represent approximately 37, 10, 35 and 12 % of the casein. All four kinds of caseins are
phosphorylated to variable degrees but characteristic to each form. αs1 has eight or
occasionally nine phosphorylated residues, αs2 may have 10,11, 12 or 13, β caseins five or
occasionally four, and κ caseins can have one, two or perhaps three phosphorylated residues.
Phosphorus is mainly covalently bound to the caseins through ester bonds on serine residues.
Those phosphates in turn bind calcium (Figure 1). The amino acid profiles of the caseins are
unremarkable but the amino acid sequence of each is critically important to the functional
roles of these proteins. The number and position of mainly serine residues in the primary
structure is particularly important in respect of phosphorylation of caseins.
Phosphorylation occurs in Golgi membranes of the mammary cell, catalysed by two serine
specific casein kinases. Only certain caseins are phosphorylated, requiring Ser-X-Y as
recognition sites where Y is a glutamyl and occasionally an aspartyl residue. X may be any
amino acid (Fox and McSweeney, 1998, p. 173).
In milk, proteins do not exist as individual molecules. Caseins, together with phosphate,
calcium and traces of citrate form structures called casein micelles. These are roughly
spherical particles with diameters up to 600 nm. A typical micelle contains 104-105 casein
molecules (Coultate, 2002, p. 142). The biological importance of micelle structure relates to
the comparative colligative properties of individual casein molecules on one hand and
micelles on the other. If casein molecules were present as monomers, the viscosity of milk
would prohibit secretion (Swaisgood, 1996, p. 856), and the osmolarity of milk would
obviously exceed that of its progenitor, blood.
5
The structure of casein micelles has been subject of investigation for years. Various models
of the casein micelles have been proposed and refined over the past 40 years. The widely
accepted model is based on the concept; core coat, internal structure and subunit or sub
micelles (Figure 2).
Minerals in milk are represented by the ash fraction in Table 1. The ash content in Table 1
(0.7 %) does not truly represent all milk salts (Table 2), because organic salts of organic acids
are also present in milk, principally citrate. Being carbohydrates these are destroyed on
ashing. The quantities of milk salts are shown in Table 2 (Belitz and Grosch, 1999, p. 486).
Milk salts exist in many forms ranging from free ions and ion complexes to colloidal forms.
The principle salts are chlorides, phosphates, citrates and bicarbonates of sodium, calcium and
magnesium (Swaisgood, 1996, p. 853). Whereas the sodium and potassium are sufficiently
soluble to be present almost entirely as free ions, other minerals, in particular calcium and
phosphate are present in higher concentrations than that can be maintained in true solution at
the normal pH of milk. Consequently, these exist partly in truly soluble form and partly in a
colloidal form associated with caseins, the major milk proteins.
The division between soluble and colloidal form of milk salts is somewhat arbitrary but a
fairly sharp separation between two phases is not difficult to achieve by physical processes
6
such as ultrafiltration, because insoluble salts are mainly associated with the colloidal casein
micelles (Fox and McSweeney, 1998 , p. 249).
Among all the salts in milk, calcium has been the main focus of interest due to health issues
(rickets, osteoporosis etc.). It is the second major salt present in milk after potassium. In the
soluble phase, calcium exists mainly as a citrate complex bearing a single negative charge (Ca
Citr-1). Other soluble complex forms include CaPO4-1, and CaHCO3+1 (Fox, 2001).
About 67 % of calcium and 25 % of phosphate are in the colloidal phase (Table 3) and in
this phase they are referred to as colloidal calcium phosphate (CCP). However, some
potassium, sodium, magnesium and citrate are also present in the colloidal phase. CCP is
closely associated with casein micelles and the nature of its association with the casein
micelles has been intensely studied. To date there has been no agreement as the exact
structure(s).
The most recent theory supports the concept of three different inorganic entities in bovine
casein micelles (Kolar et al., 2000). These authors used isotopic exchange method (32P) to
demonstrate a ratio of 2.1:1.0:1.0. Because there are not three kinds of caseins, phosphoserine
residues or residue clusters where this ratio occurs, the authors proposed these three inorganic
phosphate entities are present within every single calcium phosphate-containing ion cluster
(Figure 3).
Figure 3: Three inorganic phosphate entities in single phosphate containing cluster (Kolar et
al., 2002)
Whatever the exact structure, the close association of CCP with casein protects it from
precipitation in the milk matrix. According to McGann et al. (1983), the storage of calcium
and inorganic phosphate in mitochondria and casein micelles as a solid is a way of packing
large amounts of these ions in a form which can be rapidly mobilized but which provide much
higher ion concentrations than found in biological solutions. This is particularly relevant to
the matter of bioavailability.
8
Bioavailability can be broadly defined as the absorption and utilisation of nutrient, both of
which may be affected by such host factors as gender, physiological state and coexisting
pathologic conditions. It is a degree to which the amount of an ingested nutrient is absorbed
and is available to the body. There are numerous factors that effect bioavailability of a
nutrient, such as chemical form, the food or supplement matrix in which the nutrient is
consumed and other foods in diet (Krebs, 2001).
Calcium absorption mainly occurs in the upper part of the small intestine (duodenum),
because calcium needs a pH below 6 to stay in solution in an ionic state (Ca2+). By the time
acidic stomach contents reach the duodenum, they have been partially neutralised by the
bicarbonates released by the pancreas, but are still acidic. This provides a suitable
environment for calcium absorption. There is another factor, the presence of vitamin D
hormone calciferol that also helps greatly in absorption of calcium. As food passes through
small intestine it becomes progressively more alkaline so absorption of calcium also decreases
as it moves to the lower part of small intestine (Wardlaw, 1999, p. 484).
As discussed earlier (Section 1.4) milk calcium exists as free divalent cation and in its
colloidal form that is bound directly and indirectly to phosphate, citrate and caseins. The free
form here will be referred to as unbound calcium (typically 1/3 of total) while the calcium in
CCP – associated with caseins – will be referred to as bound calcium (2/3).
Some research supports the view that calcium bound to milk proteins is more bio available
than the free calcium. As explained earlier (Section 1.3) the αs1, αs2, and β caseins have
phosphorylated serine residues that bind calcium and phosphate. In the small intestine
gastrointestinal proteinases hydrolyse caseins at certain residues (vulnerable lysine and
arginine residues) to yield casein phosphopeptide fragments (Sato et al., 1983; Naito et al.,
1972) that are soluble by virtue of their component amino acid environment. The argument is
that this action maintains the calcium in a soluble, bioavailable form by inhibiting its
precipitation as calcium phosphate (Mykkanen and Wasserman, 1980; Naito et al., 1972). In
this respect, Delisle et al. (1995) found that the form of dietary milk and milk products (skim
evaporated, yoghurt, cheese) affected calcium uptake, but in different ways in young and old
rats.
In contrast, other researchers claim that the form of calcium is unimportant for bioavailability.
(Ho et al., 2003) found that the uptake of calcium, supplied to rats as 45CaCl2, had no effect on
9
calcium incorporation into bone. Likewise Weaver and Boushey (2003) reported that the
availability of calcium from milk is not remarkably different than its availability from other
sources like tofu and milks fortified with different calcium salts. They used a crossover
design in Caucasian and Asian women to compare the absorption of calcium from milk and
tofu. Within each ethnic group the absorption of calcium was the same for both foods.
However, Caucasian women absorbed more than Asian women.
In summary, there is no established evidence that the calcium bound to caseins is more
bioavailable than the ionic calcium.
Calcium is the most abundant mineral in human body. It comprises about 40 % of the total
mineral mass in body, translating to about 1.5 % of body mass. Ninety nine percent of it is
located in bones and teeth where it has mainly a structural function. The remaining 1 % is
present in intracellular and extra cellular fluids where it plays more functional roles like the
release of neurotransmitters, contraction of muscles, regulation of heart beat, and the clotting
of blood etc. (Wildman and Medeiros, 2000, p. 226). Calcium is lost due to normal
metabolism, through urine, skin and faeces, and therefore the body needs to replenish calcium
continuously (Nordin, 1997).
Mother’s milk is unquestionably the best source of food for the neonate but its utility in older
humans is less clear. Arguments that support life-long consumption of (ruminant) milk are
generally accepted by nutritionists and population at large, but nonetheless must be seen in
light of a huge global industry seeking profits.
Various health claims are made about, for example, calcium and bone health, and calcium
intake and osteoporosis. Many are substantiated by research. These claims also extend to
other milk nutrients, including bioactive peptides, which affect osteoblasts and osteoclasts
directly. A study on bone health in children who were milk avoiders showed that they tend to
be short and overweight and they fractured their bones more often than frequent and regular
milk drinkers (Goulding, 2003). Thus, calcium-rich foods, particularly milk, have long been
promoted in diets of children to enhance peak bone mass (bone density achieved at maturity)
to its maximum within the genetic potential. If the diet is calcium deficient during skeletal
growth it will decrease peak bone density. This has no immediate harmful consequences but
is associated with increased chance of fractures in later life (Nordin and Heaney, 1990).
10
The USA’s National Health and Nutrition Examination Survey (NHANES) between 1999
and 2002 was analysed by Wiley (2005) for correlation between milk consumption and height
in American children. The analysis focused on the frequency of milk consumption in
childhood and its relation to height in adulthood. The other analysis by the same team on
children aged 5 to 18 yrs was related to the reported frequency of milk consumption and
height. The results indicated that the adult height was positively correlated with milk
consumption at ages 5 to 12 and 13 to 17 after adjusting data for gender, education and
ethnicity.
Bone mineral status in children with cow milk allergies – and who avoided milk – was
investigated by Jensen et al. (2004). They all had reduced bone mineral density and lower
bone mineral content than equivalent children. They were all shorter than their siblings.
Osteoporosis is important health issue and is the single most common cause of fractures in the
old and middle-aged costing health services millions of dollars every year. Loss of bone starts
in women at the time of menopause and in men at about 55 years and increases the rate of
fractures in both genders (Nordin, 1997). Murphy et al. (1995) studied the effect of historical
milk consumption on current bone mineral density and found that the frequent milk
consumption before age of 25 had a favourable influence on hip bone mass in middle-aged
and elderly women. There is little evidence available about the relationship between
osteoporosis and calcium intakes in elderly and middle-aged men. One such study by Owusu
et al. (1997) showed that there is no consistent relationship between calcium intake and
incidence of forearm or hip fractures in men.
In the New Zealand population almost half the dietary calcium comes from milk (37 %) and
cheese (11 %) (Russell et al., 1999). Other than dairy products, good sources of calcium
include tofu, molasses, almonds, calcium fortified foods, and leafy vegetables, particularly
brassica family plants like broccoli and turnip greens (Wildman and Medeiros, 2000, p. 226).
An analysis was recently reported by Cook and Friday (2003) on calcium intake by 18,000
persons in the U.S.A. Data were from food intake survey databases developed in the 1990s
by the U.S. Department of Agriculture. The survey reported that milk, cheese and yoghurt –
as separate food items in the dairy category – contributed 42 % of calcium in diet. An
additional 21 % of dietary calcium came from dairy ingredients in mixed foods such as
macaroni and cheese, pizza, sandwiches, and desserts. The remaining dietary calcium sources
11
were single grains (16 %), vegetables (7 %), meat, poultry and fish (5 %), fruit (3 %), and
miscellaneous foods (7 %).
Due to inherent low costs of production, an increasing world demand, the volume of milk
produced in New Zealand has been steadily increasing from at least 1990 to the present day.
Nearly all production is linked to the global cooperative Fonterra, headquartered in Hamilton.
Most milk in New Zealand is produced for export, but New Zealand is minor producer when
compared with E.U. and the U.S.A. (Johnson, 2005). Nonetheless, Fonterra is the world’s
largest trader of milk products.
Although New Zealand is a large producer on a population basis, it ranks only 5th in the
world for consumption (Table 4). Countries like Ireland, Sweden and Austria are still the top
consumers in the dairy products (Table 4).
Country Litre
Ireland 190
Sweden 155
Austria 142
Finland 135
New Zealand 128
U.K. 128
Australia 104
U.S.A. 100
France 75
Japan 41
Source: Scrimgeour (1998)
Milk consumption in New Zealand declined in last twenty years (1982 to 1997) (Wham and
Worsley, 2002). The decreased consumption can be due to many factors like discontinuation
of the subsidised school schemes in 1967, removal of government subsidy available on milk
in 1985, and the impact of dairy export returns on milk pricing. The other influences may be
the increased consumption of juices and soft drinks, which are cheaper to produce and are
promoted and advertised on a large scale worldwide.
Faced with competition from the plethora of alternative drinks on the market, the New
Zealand dairy industry has responded by its own promotional activity and a new range of fluid
12
milk products, such as Mega, Xtra, Super Trim etc. (Table 5). This has been accomplished
with conventional cream separation to control fat content (appealing to the diet conscious),
and the relatively new technology, ultrafiltration with membranes. Table 5 below displays
different fat contents and calcium claimed in different milk products available in supermarkets
in New Zealand marketed by NZDF, Takanini under the Anchor brand name.
Membrane technology is applied to fractionate and concentrate milk proteins. The technology
involves ultrafiltration (UF) process which was developed for milk in the 1970s (Scrimgeour,
1998). The process involves forcing skim milk across and through porous membrane that
allows only some of the water, lactose and minerals to pass through. During ultrafiltration,
the fraction that is retained by the membranes is called retentate and the product that passes
through the membrane is called permeate. Permeate is an aqueous solution of lactose,
minerals, non-protein nitrogen, vitamins and other low molecular weight compounds (Rattray
and Jelen, 1996).
At this time, the question as to the relative bioavailability of calcium as a free ion and CCP is
unimportant in the consumer mind. The claims made are simply to calcium concentration and
this is a minimum legal requirement (Food Standards Australia New Zealand). No claims are
13
made as to the relative proportions of CCP and free calcium ion. As discussed earlier
(Section 1.5) the relative bioavailability of calcium as a free ion and CCP is an unresolved
question. But if it is eventually shown that CCP is a better source, then the high calcium
milks could be promoted on that basis.
NZDF also produce a wide variety of cheeses, yoghurts and other milk products like sour
creams, salad dips etc. When milk is processed into cheese, pH and temperature changes alter
the balance between CCP and free calcium in curd and whey fractions (Singh, 2004). In the
case of cheeses, NZDF suspects that cheese properties may be affected by free calcium and
CCP concentrations.
In light of these calcium issues NZDF wants to develop a quick and easy method to measure
calcium quantity in retentate and permeate the raw materials for the milks (Table 5) and other
milk products. The industry reference method – described in the next chapter – is too
laborious to be applied in a near-production line situation.
NZDF wants to develop a quick wet chemistry method to measure calcium either in its bound
or unbound form in milk or calcium retained in the retentate and lost in the permeate during
use of membranes in concentrating the milk proteins and also calcium concentrations. This
method may be applied to fulfil regulatory demand, and fulfil consumer health demands,
whether real or perceived. It may also be applied to industrial monitoring on the basis that
calcium concentration may affect the attributes of various milk products.
14
This method can be used to measure metal ions in any milk product (Chen and Jiang, 2002).
The advantage of the method is that sample preparation is simple. The sample is digested
with HNO3 and then passed through a concentric nebuliser with cyclonic spray chamber to the
ICP-MS. However, ICP equipment is expensive and unavailable in milk processing factories,
so this method was not examined.
Calcium is precipitated as the insoluble oxalate (Kirk and Sawyer, 1991, p. 32-33). The
precipitate is solubilised with sulphuric acid and the oxalate is titrated with potassium
permanganate. Although a cheap method, it involved ashing and various steps thus it was
unsuitable as a rapid method.
Methods are well documented for this technique but NZDF has no AA equipment.
The literature describes an ion selective field effect transistor (ISFET)-based calcium sensor
(Bratov et al., 2000). This method is being used in medical and biological laboratories and
also has a food industry application to control the calcium ion in cheese production. Research
publications claim high accuracy and precision, but in industrial situation with relatively
unskilled labour, data can often be unreliable (O.A Young, personnel communication) due to
poor calibration, slow response time and electrode contamination.
2.1.5 Protein bound fraction of calcium by inductively coupled plasma optical emission
spectrometry (ICP)
The method used 10 % trichloroacetic acid (TCA) and 5 % pepsin solution to precipitate the
proteins out of bovine milk samples prior to determination by this advanced technique (Silva
et al., 2001). The contents of calcium, zinc and magnesium were found to agree well with
15
values claimed in literature. However, the equipment involved is very expensive and not
affordable industrially.
The method involves microwave-assisted sample preparation and can be used to introduce the
sample for analysis as a suspension or as slurry (Oliveira et al., 2000). The method uses
liquid chromatography equipment and involves about 20 steps. Because of the equipment
requirement and the complexity of the procedure, the method is not industrially applicable.
Ekinci et al. (2004) have applied this method to determine calcium concentrations in human
milk. The method is not suitable for industry due to the high capital cost involved.
EDTA is a dairy industry standard method for measuring calcium in liquid and powdered
milks and was available from NZTM 3 Chemical Methods Manual. EDTA is widely used as
a chelator and forms strong 1:1 complexes with most of the metal ions (Harris, 2003, p. 259).
Method was trialled as a reference method.
The Sigma catalogue (2002-2003) offers two diagnostic kits both yielding purple complexes
that can be measured in a colorimeter (cheap option) or spectrophotometer. The instrument
operates by passing a beam of light of a selected wavelength through a sample and measuring
the amount of light absorbed. At first sight these rapid methods appear to be entirely suitable,
but their applicability to potentially opaque samples remains unknown. Interestingly one test
16
requires an alkaline environment and the other an acid environment suggesting scope for
clarifying opaque samples. Both mediums were tested for milk samples.
A suitable reference method is necessary to compare the results of the rapid method(s) to be
developed. Two methods were tested for suitability, the EDTA method and atomic absorption
spectrophotometry (AA), both noted above. In the EDTA method, which is the current
industry standard method available from NZTM 3: Chemical Methods Manual (2000),
calcium is determined by complexometric titration with the chelating chemical
ethylenediaminetetraaceticacid (EDTA). This is applied to liquid milks, milk powders and
protein products. However, it is a laborious method and is therefore unsuited to routine
industrial applications.
The AA method used here is derived from Methods for Chemical Analysis of Water and
Wastes from manual of U.S. Environmental Protection Agency (1983) and from general
chemical methods in Pearson’s Composition and Analysis of Foods (Kirk and Sawyer, 1991,
p. 33-34).
2.2.1.1 Principle
The method is based on the titrimetric method discussed in Vogel’s Textbook of Quantitative
Analysis (Bassett et al., 1978, p. 223-324). The principle of the method is as follows:
Hydrochloric acid is used to liberate metal ions from milk. Subsequently an excess of EDTA
solution is added which will serve to complex calcium along with other divalent cations like
magnesium. Magnesium sulphate solution is then added. The pH is then adjusted to about 10
by addition of concentrated sodium hydroxide. Under these conditions all the magnesium is
present as Mg(OH)2, with none as an EDTA complex because calcium forms a more stable
complex with EDTA than does magnesium. All the excess EDTA is free to complex with the
back titrant, CaCl2 solution. The excess EDTA does not react with the magnesium (present as
Mg(OH)2) until all the free calcium and the calcium-indicator complex have been complexed
with EDTA. At this point the indicator changes colour from blue to pink. This is the end
point. The calculation involves four values: the volume of the standard EDTA solution
added; the titre to achieve the end point colour change; the weight of original sample; and a
conversion factor relating calcium equivalents to EDTA. Details of the method are given in
17
Appendix 1. All chemicals used were analytical grade from recognised suppliers like
BDH, Merck, and Aldrich Sigma, and are not detailed in this thesis except where the chemical
is not in common use.
The EDTA method was used to determine calcium in seven different milk products made by
NZDF, most with a claimed calcium concentration.
The milks were tested in quadruplicate. The quantity of EDTA added to the different milks
required a molar excess as noted in the previous section. Thus between 3 and 7.5 ml of
EDTA solution were used on the basis of the claims. The official method requires that the
milks be sampled by weight as was done in these tests.
2.2.2.1 Principle
Because of the time required to perform the EDTA method, another reference method was
developed. AA was found to be rapid, easy and reliable. In AA a small quantity of sample is
introduced to very hot flame via a nebuliser. The solvent evaporates in the flame and metal
atoms are produced. The vast majority of these atoms are in the ‘ground state’ that is, the
electrons are in orbitals that give the lowest possible energy to the atom (Fifield and Kealey,
1995, p.320 -333).
The more metal atoms there are in the flame, the more light from the lamp is absorbed. A
light detector measures the decrease in light caused by the absorbance of the metal atoms.
The absorbance is proportional to the concentration of the metal in solution (Mcguire, 2004).
As with the EDTA method above, all chemicals were analytical grade from recognised
suppliers.
18
CaCO3 was dried to constant weight at 180°C. A stock solution of 1000 mg.L-1 Ca2+ was
prepared by dissolving 2.497 g of CaCO3 in a minimum amount of 1 M HCl then made to 1 L
with deionised water. Trichloroacetic acid solution (TCA) (25 % w/v) was prepared as was
lanthanum chloride solution. The latter at 5 % (w/v) was prepared from LaCl3.7H2O. The
concentrations of trichloroacetic acid solution and lanthanum solution were not critical. TCA
in excess serves to precipitate the milk proteins.
In AA, chemical interference can originate during dissociation of the analyte in the flame,
where it is volatilised at a different rate than the standard. One method of alleviating this
interference is through the addition of a releasing agent, a metal or salt, which forms a more
stable compound with the interferent than the analyte (Van Loon, 1980, p. 41-42). In calcium
analysis the addition of lanthanum, which acts as a releasing agent, reduces interference with
the calcium peak from Al, Si, PO43-, and SO42- (Trudeau and Freier, 1967). When in excess it
serves to bind phosphates that would otherwise complex with calcium (Van Loon, 1980, p.
41).
A reagent blank for standard solutions was made by adding 20 mL of 25 % TCA solution and
20 ml of 5 % lanthanum chloride solution to 1L with deionised water. A series of standard
calcium solutions containing 2, 4, 6, 8, 10, 12, 14 and 16 ppm Ca2+ in volumetric flasks were
prepared from the stock solution of calcium using the reagent blank as diluent.
2.2.2.3 Instrumentation
The AA instruments used were a Perkin Elmer 3110 (Wellesley, Massachusetts, U.S.A.) for
initial work, and later a GBC Avanta (GBC Scientific, Dandenong, Victoria, Australia). The
inlet on the Perkin Elmer 3110 proved to be narrow to the point that samples could block it, so
leading to erroneous results. This was not a problem with the GBC Avanta.
2.2.2.4 AA conditions
The lamp for calcium was used (422.7 nm) to pass light through an air-acetylene flame. The
lamp current was set to 7 mA (Perkin Elmer) or to 5 mA (GBC Avanta). Readings were taken
in continuous absorption mode. The dynamic range was linear between 0 and 16 ppm
calcium, linearly translating to absorptions between 0 and 0.8.
Initially the procedure followed involved sampling a small volume of milk, 0.3 mL, to which
TCA solution was added followed by lanthanum solution. This was then made up to 50 mL.
19
This approach was deleted for two reasons. The small volume of the relatively thick milk
suspension was difficult to pipette accurately and the fixed volume led to off-scale results for
calcium-rich milks as the calcium varied from 1150 to 2000 mg.L-1 in different milks to be
tested.
The following protocol was adopted and varied slightly to account for different calcium
concentrations. By way of example, 5 mL of The Milk, which contained a claimed 1150
mg.L-1 of calcium, was diluted to 50 mL in volumetric flask. A 3 mL aliquot of this
suspension was treated with 1 mL of TCA solution, allowed to stand for 10 min, shaken,
mixed with 1 mL of lanthanum solution, and then finally diluted to 50 mL. The resulted
treated mixture contained approximately 6 to 7 mg.L-1 calcium, which was in the middle of
the dynamic range. About 10 mL of this mixture was centrifuged at 1500 gravities (3000
rpm) for a period of 10 min in a Heraeus 400e centrifuge (Hanau, Germany) to let the
precipitates settle at the bottom of centrifuge tube and obtain the clear supernatant for tests.
The supernatant solution was immediately used for calcium determination.
A calibration curve was prepared from the standard calcium solutions described in Section
2.2.2.2 (Figure 4). Quadruplicate samples of milks to be tested were prepared as described
above and analysed straight after. The results were adjusted to account for dilutions.
0.6
0.5
mn 7.224 ta ecnabrosbA
0.4
0.3
0.2
0.1
0.0
0 2 4 6 8 10 12 14 16 18
Ideally a comparison of the EDTA and AA methods should be done with the same bottle of
milk for a number of milk products. This was not possible due to time constraints and the
perishability of milk. Thus the data for EDTA and AA were obtained on separate batches on
different days. Two features of the results were of primary interest. Are the results plausible
(a rough measure of accuracy) and is precision acceptable?
In Table 6 the units for both results are not same because in EDTA method the milks analysed
were weighed in during the procedure whereas in AA method the milks were measured by
volume.
Although results in Table 6 were calculated in different units, they were still a good measure
of precision and accuracy against the claimed calcium by NZDF in the milk products tested.
The last three products are not being marketed by NZDF as such; therefore no claims made
were available for them. However retentate ultimately is marketed as Xtra.
In the EDTA method most of the milk solutions turned clear during the procedure other than
Lite milk. The clarity of the milk sample solutions was important in obtaining the clear
endpoints. Acid digestions and prolonged heatings were recommended to overcome this
problem. Lite milk was allowed to stand for longer periods than the others but clarity of the
sample mixture could not be achieved. As method was time consuming and laborious acid
digestions were not tried.
21
2.3 Conclusion
Although the AA method is more expensive as a reference method it was still selected as the
method of choice looking at the ease of use, precision and accuracy of results as compared to
EDTA.
22
3.1 Introduction
For this research a colorimetric method was chosen as the best approach for routine calcium
determination in milk processing factories. The basis for this decision was cost and
simplicity.
Two different dyes were selected for this work. This chapter will discuss the principle of the
colorimetric determinations, and describe in detail the two dyes and method development with
them.
When light of a suitable wavelength passes through a solution containing an analyte, some of
the photons may be absorbed by the analyte. An analyte can be a molecule or element
whether in its native form or complexed in some way. The difference between the light
intensity before and after passage through the sample is a measure of the absorbance due to
the analyte. For light of a single or very narrow wavelength, the light absorbed by the analyte
varies with the concentration of the analyte. This technique is called spectrophotometry and
its behaviour can be described mathematically according to the Lambert-Beer’s Law (Skoog
et al., 1994, p.403-406):
Where
I0 is the intensity of the incident light
It is the intensity of the transmitted light
A is absorbance, a dimensionless quantity
ε is the molar absorptivity constant (L.mole-1.cm-1) that is constant for a given analyte
c is the concentration of the analyte in the solution (mole. L-1) and
l is the length of the light path through the solution (typically cm).
23
Since spectrophotometer determinations are carried out in cuvettes for which the path
length is constant (usually 1 cm), and the molar absorptivity is also a constant for a given
analyte under set conditions, then according to the above equation the absorbance due to
analyte is linearly proportional to its concentration. Linearity usually, but not always, occurs
when absorbance values are between 0 and 2 (Higson, 2003, p. 116). Determinations of
unknown concentrations of an analyte are made after a standard curve is prepared from a
concentration series of the analyte. It is important to note that spectrophotometry is valid only
for clear solutions, and as will be seen this has important consequences in this research.
3.3.1 Arsenazo I
According to the Sigma catalogue (2002-2003) Arsenazo dyes form purple complexes with
calcium ions in an acidic environment. Two forms are available, Arsenazo I and Arsenazo III.
Arsenazo III forms three kinds of complexes with calcium (Palade and Vergara, 1983; Dorogi
and Neumann, 1981; Clemen et al., 1988; Gratzer and Beaven, 1977) whereas Arsenazo I
forms a simple 1:1 complex (Dorogi et al., 1982). An extensive literature reports that
Arsenazo III forms coloured complex with calcium in a pH range from neutral to highly
alkaline (Blasco et al., 1999; Gratzer and Beaven, 1977; Malcik et al., 2005; Dorogi and
Neumann, 1981). Contrary to the claim in the Sigma catalogue that coloured complexes form
in acidic solutions, most research reports that a pH around 7 is the most suitable for calcium
complexation with Arsenazo III (Clemen et al., 1988; Dorogi and Neumann, 1981; Ohnishi,
1979). Similarly Dorogi et al. (1982) reported the complexation of Arsenazo I with calcium
at neutral pH.
Arsenazo I was chosen over Arsenazo III because of the reported simplicity of its
complexation with calcium (Dorogi et al., 1982). No clear representation of the 1:1 complex
was presented in that paper, or in others listed above, but it seems very likely that the single
arsenate group is involved on the basis that Arsenazo III has two arsenate groups and forms
more than one type of complex (three reported).
Arsenazo I (Figure 5) has the chemical formula C16H11As N2 Na2 O11S2 with a molecular
weight of 592 g.mole-1. It was bought from Sigma Aldrich, Steinheim under the name
Neothorin. The claimed purity was 99.9 %.
24
The literature was searched to get a starting point for suitable method development. Among
the many methods reported for calcium and Arsenazo I and III that by Dorogi et al. (1982)
was the clearest and so was chosen.
o-Cresolphthtlein complexone (CPC) (Figure 6) also forms a purple colour complex with
calcium. It is generally recognised that the complex forms only in alkaline solution.
It is one of the most sensitive metallochromic ligands available for the determinations of
alkaline earth metal ions. Toei (1988) used CPC as a component of an ion-exchange eluant
and applied the system to the separation and detection of alkaline earth metals. For this work
stable metal complexes were determined by absorption at 572 nm at pH 10.2.
CPC forms the complexes: Ca(CPC)4-, CaH(CPC)3-, and Ca2(CPC)2- with Ca2+. A weakly
absorbing complex H2(CPC)4- is also formed. According to Staden and Rensburg (1990) the
colour formed by the CPC complex is due to the lactone ring in the phthalein molecule, but in
25
exactly what way is not clear from their paper or other research publications. The reaction
between CPC and calcium between pH 10 and 11 can be written as:
CPC has chemical formula C32H32N2O12 and molecular weight of 637 g.mole-1. It was bought
from Sigma Aldrich, Steinheim. The dye was kept at room temperature in the dark.
Given that an industrial method was being developed, simplicity and ease of use were
important. Several decisions were made at the outset. Because the absorbance is in the
visible range for aqueous solutions, disposable plastic cuvettes for a conventional 1 cm-path
length spectrophotometer were deemed to be adequate. The spectrophotometer was defined to
be a simple single beam instrument with only visible range capability. In every case the final
volume was to be 3 mL, of which 2 mL would be buffer and the other 1 mL of reagents.
Generally the final diluted and treated with reagents, sample volume to be tested was 0.5 mL,
a convenient volume for the sorts of pipette guns common in quality control laboratories. It
was decided that the concentration of the dye in the cuvette would never be less than double
than the concentration of calcium whether in a test or calibration.
A stock solution of 15 mM Arsenazo I was prepared by dissolving the dye in deionised water
at the rate of 0.886 g.100 mL-1. It was held in the dark at room temperature. Analytical-grade
calcium carbonate from BDH, Poole, dried to a constant weight was used to make 7.5 mM
calcium standard (half the concentration of Arsenazo I). To prepare this, 0.1875 g of CaCO3
was dissolved in minimum amount of 1 M HCl then made to 250 mL in a volumetric flask
with deionised water.
Two sources of magnesium ion were used, analytical grades of MgSO4.7H2O and
4MgCO3.Mg(OH)2 5H2O both from BDH, Poole. The standard solution for the former was
prepared by dissolving 0.1845 g in deionised water by making the final volume to 100 mL.
26
For the latter 0.0724 g was dissolved in deionised water adding several drops of 1 M HCl
until fully dissolved and made upto final volume of 100 mL.
The lithium standard was prepared by dissolving 0.032 g of analytical grade LiCl (BDH,
Poole) in deionised water and then making the final volume to 100 mL. The potassium
standard was similarly prepared from analytical grade KCl (BDH, Poole) by dissolving 0.060
g in deionised water to final volume of 100 mL. All these standards were made as stock
solutions of 7.5 mM then dilutions were made as required. The standards were used to test
the interference of other monovalent and divalent cations during the complexation of calcium
with Arsenazo I dye.
Trichloroacetic acid 25 % was prepared from analytical grade CCl3.COOH (Ajax Chemicals
N.S.W.). The acid was dissolved in deionised water and made to 100 mL.
Pipes buffer, piperazine-N N’-bis-2-ethane sulphonic acid (BDH, Poole) was prepared at 0.03
M by titration with sodium hydroxide to pHs 5.9, 6.9 and 8.0. Tris-maleate buffer (0.05 M)
was prepared by dissolving 24.2 g of tris-(hydroxymethyl)-aminomethane (Serva Fine,
Heidelberg) and 23.2 g maleic acid (May and Baker, Dagenham) in 1 litre. To 25 mL of this
were added varying volumes of 0.2 M NaOH to yield pHs 6.0, 7.0 and 8.0. Other buffers
specified in Table 7 were also prepared according to the methods described by Dawson et al.
(1986, p. 417-449). Five buffers were chosen to cover the pH range of 3.2 to 12.0 (Table 7).
Cuvettes were prepared as follows. Two millilitres of a prepared buffer solution was added to
cuvettes, followed by 0.5 mL of Arsenazo I dye and 0.5 mL of the salt standards. The final
concentrations of buffers in cuvettes are given in Table 7. The final cation concentrations and
Arsenazo I were changed at times according to claims made in literature and to get
absorbances between zero and unity.
27
An Ultraspec 2100 Pro UV/visible spectrophotometer from Biochrom (Cambridge) was used
for spectrophotometric titrations. This is a single beam spectrophotometer.
In all cases the spectrophotometer was zeroed with water. Other cuvettes were filled in the
sequence: buffer, dye then standard or test solutions, to 3 mL. Absorbances were read directly
after mixing. The mean absorbances of the duplicate cuvettes containing no calcium were
subtracted from all values. Thus the curves pass through zero by definition. Absorbances
were measured at single wavelengths in the visible range and scans were made as required.
Different commercial milks like The Milk, Bluetop, Lite, Super Trim and Trim were tested
with the Arsenazo I method. All the milks claim different amounts of calcium in them (Table
5). The milks were routinely diluted with water according to the calcium content claimed on
the bottles to bring their calcium concentrations within the dynamic range of the calibration
curve.
Four replicates of each milk were used and from each replicate three cuvettes were prepared.
Thus 12 measurements contributed to the mean. Results were manipulated on a spreadsheet
to calculate the estimated calcium concentration in the original milk.
Because about 2/3 of calcium in milk is bound to protein, it was thought that liberation of
calcium from this matrix by protein denaturation might increase the measured values.
Two ways were tried to denature proteins in milk with TCA. In first method, to 0.5 mL of
The milk and Trim, 0.5 mL of 25 % TCA was added. The mixture was shaken and allowed to
stand for 10 minutes and final volume was made to100 mL. In second treatment, Trim milk
was diluted in different ways (Table 11) to avoid the clumping of precipitates after treating
with TCA. In each trial, for every 0.5 mL of Trim milk, 0.5 mL of 25 % TCA was used for
protein denaturation and final volumes were made to 100 mL. From the diluted TCA - treated
sample mixtures 0.5 mL was taken for absorbance readings.
28
Contrary to claims made in the literature (Clemen et al., 1988; Dorogi and Neumann, 1981;
Ohnishi, 1979) the absorbance due to complexation of calcium with Arsenazo I was poor
around neutrality.
0.16
0.14
0.12
mn 265 ta ecnabrosbA
0.10
0.08
0.06
0.04
0.02
0.00
-0.02
0 2 4 6 8 10 12 14
Concentration of calcium in µM
Figure 7: Absorbances of calcium with Arsenazo I dye at 562 nm in 0.03 M Pipes buffer at
pH 5.9 ■, 6.9 ▲, and pH 7.9 ●
Another buffer (Tris maleate) was tested in the same pH range (Grater and Beaven, 1977),
with similar results (Figure 8), except that the absorbance at pH 8 was comparatively lower
than the approximately equal value in Pipes buffer (Figure 7).
29
0.16
0.14
0.10
0.08
0.06
0.04
0.02
0.00
0 2 4 6 8 10 12 14
Concentration of calcium in µM
At this point it was thought possible that the dye might be heavily contaminated with calcium
or other alkaline earth metal, so causing it not to respond to added calcium although (the
claimed purity was 99.9 %). The concentration of calcium in the Arsenazo I was tested by
atomic absorption spectrophotometry. It was found that the mole ratio of calcium to Arsenzo
I was a negligible 1 to 1920. This was clearly not the cause of the problem. Attention was
turned to pH.
In the previous section, increasing concentrations of calcium in both the buffers used showed
increasing absorbance at 562 nm to around 0.05 to 0.14 (Figure 7 and 8) at slightly alkaline
pHs with final calcium concentration in the cuvette 12.5 μM.. Preliminary trials showed that
this absorbance could be greatly increased at pH 9.0 with Pipes buffer (results not shown).
The absorbances were about 10 fold higher.
However, Pipes does not buffer well at pH 9, so it was decided to explore the colour changes
of Arsenazo I dye over a wide range of pH values for cations significant in milk and for
lithium.
30
Figure 9 is a panoramic view of the three of the five treatments (color change for lithium and
potassium ion not shown) in cuvettes taken with hand-held digital camera. A potentially
useful change in colour occurred at higher pHs for both calcium and magnesium.
CaC2+
a
Mg2+
Figure 9: Calcium and magnesium complexed with Arsenazo I dye at a range of pHs from
highly acidic to strongly alkaline conditions showing significant colour change at
highly alkaline pHs where cation concentration were 37.5 µM each and Arsenazo
I 75 µM
In the case of calcium, Figure 9, pHs 9, 10 and 12 gave useful differences. pH 11 appeared to
break a trend from pH 9 to 12, and this may be due to a specific buffer effect (Table 7).
Lithium and potassium behaved rather like the control except that there was some colour
change in the acidic range (not shown) but not at higher pH values, as will be examined later.
Difference spectra (figures shown on next two pages) showed that potentially useful colour
changes could be obtained at the higher alkaline pH values for calcium (Figure 10). Similar
results were obtained for magnesium (Figure 11).
However for magnesium, absorbances were higher than calcium at the wavelength of interest,
562 nm. Moreover at pHs 5.9, 6.9, and 8.0, the colour changed little but the absorbance was
high (Figure 11). As with calcium, the more acidic pHs showed no colour change (Figure
10).
.
31
0.6
0.4
pH 10
pH 12
ecnabrosbA
0.2
pH 9
pH 11
0.0
Acidic pHs
-0.2
Wavelength (nm)
Figure 10: Absorbances of calcium due to complexation with Arsenazo I at various pHs
ranging from highly acidic to highly alkaline from 400 to 800 nm. The calcium
concentration was 37.5 µM with Arsenazo I at 75 µM
0.6
pH 6.9
pH 5.9
pH 10
0.4
pH 8
pH 12
ecnabrosbA
0.2 pH 9
pH 11
0.0
-0.2
Wavelength (nm)
Figure 11: Absorbances of magnesium due to complexation with Arsenazo I at various pHs
ranging from highly acidic to highly alkaline from 400 to 800 nm. The
magnesium concentration was 37.5 µM with Arsenazo I at 75 µM
32
0.6
0.4
Mg2+
Li+
Ca2+
ecnabrosbA
0.2
0.0
-0.2 K+
Wavelength (nm)
Figure 12: Absorbance spectra of lithium, potassium, magnesium and calcium cations at pH
12 with NaOH/KCl buffer. The concentration of Arsenazo I was 75 μM and the
cation concentrations were each 37.5 μM.
0.6
Mg2+
0.4
Ca2+
ecnabrosbA
0.2
Li+
0.0
K+
-0.2
Wavelength (nm)
Figure 13: Absorbance spectra of lithium, potassium, magnesium and calcium cations at pH
10 with Clark and Lubs buffer. The concentration of Arsenazo I was 75 μM and
the cation concentrations were each 37.5 μM
33
The most useful absorbance differences were obtained at pH 12 and 10 (Figure 12 and 13)
so further work was focused on these pHs. Figure 12 (pH 12) shows that the magnesium
complex has the highest absorbance. Lithium was unresponsive, whereas potassium showed a
loss of colour over a range of wavelengths at pH 12. At pH 10 (Figure 13) potassium showed
negative absorbance and lithium showed some absorbance when compared to pH 12 (Figure
12), where both were non-responsive.
Both pH 10 and 12 looked equally good for calcium absorption but both these pHs showed
similar colour response for magnesium. At pH 12, both lithium and potassium were almost
non-responsive so pH 12 was selected for further trials. At wavelength 562 nm, which has
been used by Dorogi et al. (1982), both showed maximum absorption for calcium and
magnesium.
As noted in Section 3.4.1.2, two magnesium salts were used in method development. As
work progressed it was realized that the magnesium salt initially used was not completely
homogeneous suggesting variable waters of crystallization.
Therefore in subsequent work, the dry salt (4Mg CO3 Mg (OH)2 5H2O) was used.
Absorbances obtained in standard curves do not exactly equal absorbances obtained in various
figures presented earlier.
Figure 14 shows the calibration curves obtained for calcium and magnesium at 562 nm at pH
12 with new magnesium salt used. On occasions, the calibration curves showed a slight
tendency to saturate, but generally a linear fit was best.
34
0.7
0.6
0.5
mn 265 ta ecnabrosbA
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40
Concnentration of calcium or magnesium in µM
Figure 14: Calibration curve for magnesium ■ and calcium ● (maximum molarities for both
in cuvettes 37.5 μM) when complexed with 75 μM Arsenazo I dye in NaOH/KCl
buffer at pH 12 at wavelength 562 nm.
The standard curves above show that the magnesium-Arsenazo I complex has a very similar
absorbance to that due to calcium. According to Malcik et al. (2005), interference from
competing divalent cations during calcium Arsenazo III complexation at pH 9 is not
significant until the competing cation concentration is very high compared with calcium.
The molar concentration of magnesium in milk is about 1/3 that of calcium, and has the
potential to compete with calcium during complexation with Arsenazo I. This raised the
question of how magnesium would interfere with calcium complexation with Arsenazo I. To
test this, a series of cuvettes were assembled where the calcium concentration was maintained
at constant 9.37 μM, and the concentration of magnesium was varied from zero to 3.13 μM
(about 1/3 the calcium concentration) (Table 8). In both experiments the Arsenazo I
concentration was in excess at 37.5 μM. Table 8 shows that the effect of magnesium was not
as great as would be expected if each ion acted independently, and in this respect Arsenazo I
appears to behave similarly to Arsenazo III.
35
The method was then tested with commercial milk samples diluted in water with no
denaturation treatment. The samples were also measured by AA according to the reference
method discussed earlier in Section 2.2.2.2 to 2.2.2.5. The results paralleled the claimed
calcium concentrations but were higher, especially as measured by the Arsenazo I method
(Table 9).
As discussed earlier, about 2/3 of calcium in milk is bound to proteins. The AA method
adopted here includes a denaturing step with concentrated trichloroacetic acid (TCA), the idea
36
being to liberate calcium from the protein (Rodriguez et al., 2001). It was thought that
TCA might similarly be useful in the Arsenazo I procedure. This idea was explored with The
Milk and Trim.
When 25 % TCA was mixed with an equal volume of undiluted milk, severe clumping was
observed. The results obtained after treating these milk samples with TCA were low (Table
10).
This issue was further explored with the same sample of Trim milk but used TCA applied in
different ways (Table 11). This time the results for TCA treatment were good and again
higher than the claimed calcium with all the methods tried. There appears to be an inherent
inconsistency.
Table 11: Calcium concentration found in Trim milk after treating with TCA in different
ways using Arsenazo I Method
The literature was searched for a colorimetric method available for the dye, but only one was
found first described by Stern and Lewis (1957). In contrast to the experiments with
Arsenazo I where the choice of buffer was made empirically, the buffers used here were as
defined for CPC and calcium by earlier researchers. These were Clark and Lubs buffer (Stern
and Lewis, 1957; Paull et al., 1997) and 2-amino-2-methyl-1-propanol (AMP) buffer
(Moorehead and Biggs, 1974; Mesquita and Ranngel, 2004; Stedan and Rensburg, 1990).
Moorehead and Biggs (1974) reported that interferences from magnesium could be removed
by using 8-hydroxyquinoline in presence of AMP buffer.
3.5.1.2 Reagents
The stock solution of 15 mM CPC was prepared by dissolving 0.9549 g in minimum amount
of 5 M HCl and then making the final volume to 100 mL with deionised water. Dilutions
were made as required.
8-Hydroxyquinoline (analytical grade) was obtained from Beijing Chemical Works, Beijing.
To prepare CPC solution with 8-hydroxyquinoline, the latter was mixed with CPC prior to the
dissolution procedure with HCl and water.
38
As for Arsenazo I work, the reference was water. Cuvettes were prepared as follows. Two
millilitres of a prepared buffer solution was added to cuvettes, followed by 0.5 mL of CPC
dye or CPC plus 8-hydroxyquinoline of required molarity and 0.5 mL of the salt standards or
test solutions. Absorbances were read at 575 nm (Paul et al., 1997) directly after mixing.
CPC and 8-hydroxyquinoline solutions made according to Herraro et al. (1992) were used to
see the effect on magnesium masking with AMP buffer.
Several milks were diluted in water to ensure a calcium concentration in the dynamic range of
the assay. The details were as for the Arsenazo 1 method Section 3.4.1.5.
When Clark and Lubs buffer and CPC were mixed with calcium or magnesium ion, the clear
colourless solutions became clear purple resulting in linear calibration curves. The typical
calibration curves were generated with 25 μM maximum calcium standard and 50 μM of CPC
dye (Figure 15).
39
0.7
0.6
0.5
mn 575 ta ecnabrosbA
0.4
0.3
0.2
0.1
0.0
0 5 10 15 20 25 30
Figure 15: Calibration curves for calcium (●) and magnesium (■), when maximum
concentrations for each were 25 μM and CPC 50 μM at pH 10.2 with Clark and
Lubs buffer
The literature claims that magnesium can be masked with 8-hydroxyquinoline in the presence
of AMP buffer (Section 3.5.1.1). This was tested with Clark and Lubs buffer (concentration
of CPC and 8-hydroxyquinoline used according to Herraro et al. 1992). Over the cation
concentration range 0 to 6.25 μM, 1.15 mM 8-hydroxyquinoline quenched the usual colour
change for both calcium and magnesium with Clark and Lubs buffer (Figure 16).
The absorbance for calcium at 6.25 μM without 8-hydroxyquinoline was about 0.14 (Figure
15), but fell to a low 0.04 with 8-hydroxyquinoline (Figure 16) so rendering the assay useless.
Attention was therefore switched to AMP buffer.
40
0.20
0.15
mn 575 ta ecnabrosbA
0.10
0.05
0.00
-0.05
0 1 2 3 4 5 6 7
Figure 16: Absorbances for calcium and magnesium (6.25 μM maximum concentration for
both) in the presence of 8-hydroxyquinoline with CPC (12.5 μM concentration in
the cuvettes) at pH 10.2 with Clark and Lubs buffer
Figure 17 shows the relationships between calcium and magnesium concentration and
absorbances at 575 nm in the absence of 8-hydroxyquinoline with AMP buffer.
0.20 0.20
(a) (b)
0.15 0.15
mn 575 ta ecnabrosbA
Ca2+
ecnabrosbA
0.10 0.10
Mg2+
0.05 0.05
0.00
0 1 2 3 4 5 6 7
0.00
400 500 600 700 800
Concentratrion of calcium or magnesium in µM Wavelength (nm)
Figure 17: a. Absorbances due to calcium (●) and magnesium (■) with CPC in AMP buffer
at pH 10.5. The CPC concentration was 12.5 μM. b. Spectral scans at maximum
cation concentration, 6.25 μM.
41
0.20 0.20
(a)
(a) (b)
0.15
0.15 Ca2+
mn 575 ta ecnabrosbA
0.10
ecnabrosbA
0.10
0.05
0.05
0.00
Mg2+
0.00 -0.05
0 1 2 3 4 5 6 7 400 500 600 700 800
Figure 18: a. Absorbances due to calcium (●) and magnesium (■) with CPC in AMP buffer
at pH 10.5 in the presence of 1.15 mM 8-hydroxyquinoline. The CPC
concentration was 12.5 μM. b. Spectral scans at maximum cation concentration,
6.25 μM.
0.6
0.6
0.4 Ca2+
mn 575 ta ecnabrosbA
0.4
ecnabrosbA
0.3
0.3
0.2
0.2 Mg2+
0.1
0.1 0.0
-0.1
0.0
400 500 600 700 800
0 5 10 15 20
Wavelength (nm)
Concentration of calcium or magnesium in µM
Figure 19: a. Absorbance for calcium (●) and magnesium (■) with CPC in AMP buffer at pH
10.5. The CPC concentration was 37.5 μM. b. Spectral scans at maximum cation
concentration, 18.7 μM.
42
0.6 0.6
(a) (b)
0.5 0.5
Ca2+
0.4
mn 575 ta ecnabrosbA
0.4
ecnabrosbA
0.3
0.3
0.2 Mg2+
0.2
0.1
0.1
0.0
0.0 -0.1
0 5 10 15 20 400 500 600 700 800
Figure 20: a. Absorbances for calcium (●) and magnesium (■) with CPC in AMP buffer at
pH 10.5 in the presence of 3.45 mM 8-hydroxyquinoline. The CPC concentration
was 37.5 μM. b. Spectral scans at maximum cations concentration, 18.7 μM.
Figure 19 and 20 compares the absorption for both cations at higher concentrations of CPC
and 8-hydroxyquinoline as the cation concentrations were increased. The figures showed the
similar trend as before in previous two figures (Figure 17 and 18). No negative absorption
observed in the latter scans.
Table 12 compares the claimed concentration in milks by NZDF with the results of the CPC
method. For all the milks the value determined by the CPC method was lower than the
claimed value, in three cases markedly so. Standard deviations were also high considering the
large number of replicates.
3.6 Conclusion
The origin of the claimed values in the milk brands tested in both methods is unknown, but is
likely to stem from determinations made at the time the product and its process was defined.
At this point of the research the fact that the CPC value was markedly lower than the claimed
was disturbing but interesting. The Arsenazo I method suffered from inconsistent calibration.
Neither assay was suitable for long term monitoring of calcium in milk. It was decided to
compare both methods for precision, even if accuracy was still poor.
44
4.1 Introduction
The previous chapter described the development of two colorimetric assays for calcium in
milk. Both suffered interference from magnesium, but with one assay, o-cresolphthalein
complexone (CPC), this could be minimised by the addition of 8-hydroxyquinoline to the
assay mixture. At the outset it appeared that CPC assay would be more useful than the
Arsenzo I assay, but that had to be tested.
Thus, the first part of this chapter compares the accuracy and precision of the three methods –
atomic absorption (AA), Arsenazo I, and CPC – on several milks replicated over five days.
This experiment served to identify the colorimetric assay that would be used to monitor the
calcium content of ultrafiltered milks over a period of months.
Once identified (and it was the CPC method) it was developed further. Those results are
reported here, in the second part of this chapter.
4.2.1.1 Milks
The Milk, Trim, Mega, Xtra, raw milk and permeate were obtained from NZDF, Takanini for
five consecutive days, each from a different production run. Budget whole milk powder was
bought from Foodtown supermarket, Mt Wellington. A sample was sent to Gribbles
Analytical Laboratories, Auckland for calcium determinations. They used NZTM3: 9.21 the
standard dairy industry method to determine calcium in whole milk powder. The milk was
intended to use as a reference milk at times as the liquid milks were perishable and were
unable to be preserved for longer periods.
4.2.1.2 Assays
The AA assay protocol was as described in the Section 2.2.2.2 to 2.2.2.5. The Arsenazo I and
CPC assay protocols followed the order of addition described in Section 3.4.1.4, with final
cuvette concentrations as follows: the final concentration of Arsenazo I and CPC were both
45
37.5 μM and the maximum final calcium concentration was 18.75 μM. The final 8-
hydroxyquinoline concentration in the CPC assay was 3.45 mM. The final buffer
concentrations were 0.03 M NaOH/KCl and 0.80 M AMP for Arsenazo I and CPC,
respectively.
The milk powder was reconstituted according to the recipe on the pack, and then diluted to get
the calcium concentration in the dynamic range of the calibration curve.
Liquid milk dilutions were made according to the claimed or known calcium concentrations,
again to match milks to the dynamic range of the calibration curve. For each liquid milk, four
replicate dilutions were made for Arsenazo I and CPC assays, and three replicate
determinations were made on each, making 12 values for every milk. For AA milk four
replicates were used for each milk tested.
The analysis of whole milk powder (Table 13) showed reasonable agreement between the
several methods. AA showed the lowest mean value, but the lowest percent coefficient of
variation (CV %) of three methods. The highest CV % was 7 % for the CPC method, but the
mean was close to the Gribbles mean, which is the industry standard.
In principle, these results from reconstituted milk powder showed that any of the methods
might be expected to work well with whole fresh milks. This was tested in the next section.
46
Analysis of six liquid milks over five days of sampling explored means and within-day and
between-day variation for three analytical methods (Table 14). The within-day percent
47
Table 14: Calcium determined by AA, the Arsenazo I and the CPC method in different commercial milks to compare precision
of both methods
2
AA 1134 2 1362 2 1923 2 1487 2 1045 2 291 1
AA 1262 0 1297 1 2344 2 1564 2 1200 2 336 0
AA 1027 2 1232 2 1828 3 1290 3 1158 0 298 2
AA 1092 1 1267 3 1878 0 1388 2 1114 3 285 2
AA 1114 2 1286 1 1945 2 1378 1 1134 1 309 3
Day-to-day CV % 8 5 4 7 10 7
3
Arsenazo I 2160 5 1648 6 2556 6 2672 2 2193 3 611 8
Arsenazo I 267 7 190 8 309 25 270 12 265 14 68 15
Arsenazo I 1794 7 1592 8 2382 6 2017 9 1711 4 484 10
Arsenazo I 1773 4 1544 8 2301 6 2027 4 1859 6 483 15
Arsenazo I 713 6 605 7 931 6 779 9 301 7 221 5
Day-to-day CV % 60 72 60 59 60 64
CPC 590 6 857 4 1379 3 1042 7 868 5 227 5
CPC 889 6 922 7 1764 9 1317 4 1043 4 255 7
CPC 868 5 771 15 1254 4 1047 7 948 5 192 19
CPC 900 5 860 8 1335 5 1043 4 964 9 182 8
CPC 886 4 855 6 1307 4 1006 7 987 9 196 10
Day-to-day CV % 16 7 6 14 15 12
1
CV % is coefficient of variation percentage; 2AA Atomic absorption means are averages of four values; 3Arsenazo I mean
values for Arsenazo I and CPC method are average of 12 values. Claimed calcium for The Milk, Trim, Xtra and Mega were
1150, 1400, 2000 and 1600 mg.L-1 respectively. No calcium claims were made for raw milk and permeate.
48
coefficients of variation for the AA method were low and consistent across all milks. By
contrast the equivalent values for the Arsenazo I and CPC methods were high and variable.
The mean values for CPC determinations were always lower than the same-day AA
determinations for all milks. Summed over five days, the mean percent difference was 27 %,
with a range of 15 to 34 %. The between-day coefficients of variation for CPC were up to
twice as high as those for AA. Thus by any measure the AA method was better than the CPC
method. However, the CPC method performed far better than the Arsenazo I method.
Although the Arsenazo I method returned high and variable within-day coefficients of
variation, these were less than the extremely high between-day variability. Percent
coefficients of variation ranged between 59 and 72. This variability was totally unacceptable.
What is surprising was that although the method worked reasonably well during development,
it failed in this comprehensive trial. The CPC method was chosen for subsequent work,
principally a long term monitoring experiment in a commercial environment.
Prior to starting the monitoring trial, some further refinements were made to the CPC method
in an effort to reduce variability.
It was thought that the release of calcium from caseins might be the cause of variation, so the
following experiments were done to identify suitable denaturing reagents, if any, that might
aid the release of calcium from caseins.
Glucono-δ-lactone has been used in many milk research papers as a progressive acidification
agent for the release of calcium from caseins (Roesch et al., 2004; Dalgleish et al., 2004 and
Dalgleish et al., 2005). Herraro et al. (2001) used acetate buffer to precipitate caseins from
milk samples. Sodium tungstate and sulphuric acid was used by Sarkar and Chauhan (1967)
for protein removal in the milk samples.
Highly alkaline 8 M NaOH used in EDTA method almost dissolved the milk proteins during
EDTA procedure in this investigation, so it was tested as well. Trichloroacetic acid was
already used in Arsenazo I method where trichloroacetate interacts with positive protein
groups R-NH3+ producing a white precipitate to denture the proteins (Rodriguez et al.,
2001;Silva et al., 2001). Milk was heated to see any effect on calcium measurement.
Concentrated nitric acid and perchloric acids were used for digestion of milk proteins.
49
Aqueous sodium tungstate solution (10 %, w/v) was prepared from Na2WO4.2H2O (Scharlau,
Chemie, Barcelona). H2SO4 (0.3 M) was prepared from concentrated acid. Perchloric and
nitric acid were used as supplied, 11.7 M and 15.9 M, respectively. Aqueous glucono-δ-
lactone (1.5 % w/v) was prepared from solid glucono-δ-lactone (Sigma Aldrich, Steinheim).
0.2 M sodium acetate trihydrate (Sharlau Chemie, Barcelona) solution was made by
dissolving 2.72 g of it in deionised water and final volume made to 100 mL. To 7 mL of 0.2
M NaOAc solution 3 mL of glacial acetic acid (BDH, Poole) was added to make pH 5 sodium
acetate-acetic acid buffer.
A single sample of three milks was selected for this work, skim, Xtra and permeate.
(Pasteurised skim milk is ultrafiltered into a casein- and calcium-enriched fraction, Xtra, and a
depleted permeate.)
Except for two treatments (heat, wet digestion) the three milk products were first diluted
before any denaturing treatment. Three millilitres of Xtra and 5 mL of skim and permeate
were each diluted to 50 mL in volumetric flasks. All denaturing treatments were done in
quadruplicate followed by three determinations from each denaturation, making 12
determinations for each milk product.
The treatments were as follows. For the control, the milks were diluted by a further 1 in 50 by
volume. For the heat treatment, undiluted milks were heated on hot plates and magnetically
stirred continuously to boiling. The milks were then accurately diluted to match the dynamic
range before determination.
For one TCA treatment, 1 mL of diluted milks was mixed with 1 mL of 25 % TCA in
volumetric flasks. The mixture was shaken thoroughly, allowed to stand for 10 minutes and
was made to 50 mL. An aliquot was centrifuged at 1500 gravities and the determinations
were made on the supernatant. In a variation of this treatment, the TCA-treated samples were
held overnight before centrifugation.
50
For wet digestion, 3 mL of Xtra and 5 mL of skim and permeate were mixed with 25 mL of
the concentrated nitric and 5 mL of the concentrated perchloric acid. The mixtures were
heated to boiling on a hot plate until fumes appeared. Final volumes were made to 50 mL
with water after adjusting the pH to neutrality with 2 M NaOH. Further dilutions were made
before calcium determinations. There was no need to centrifuge because the solutions cleared
on digestion.
The alkali treatment employed 1:1 dilutions of initially diluted milks with 8 M NaOH. After
10 minutes at room temperature, the mixture was clear and further dilutions were made. The
mixtures were neither neutralised nor centrifuged.
Glucono-δ-lactone (GDL) treatment employed 1:1 dilutions of initially diluted milks with 1.5
% GDL. The mixture was shaken and held for 10 minutes before further dilution and
centrifugation prior to determinations. For the sodium tungstate treatment, 1 mL of initially
diluted milks were mixed with 2 mL of 10 % sodium tungstate and 2 mL of 0.3 M sulphuric
acid, shaken and held for one minute. After dilution to 50 mL, the mixture was centrifuged
before determinations on the supernatant.
For the combined acetic acid and sodium acetate treatments, 1 mL of each of the initially
diluted milks was mixed with 1 mL of sodium acetate and acetic acid buffer, shaken and held
for 10 minutes before the dilution. After centrifugation determinations were performed.
The means for the CPC method across the several denaturation treatments were usually
different from the equivalent means determined by the AA method, the reference in this
experiment (Table 15). At this time however, accuracy was not as important as precision, on
the basis that the means for two methods may prove to be well correlated.
In Table 15, the AA results had low percent coefficients of variation, about 1 %, and markedly
better than those for the control (no denaturation treatment) with CPC, ranging between 4 and
19 %. Those for the wet digestion treatment were also poor (6 to 16 %) and completely
unacceptable for an analytical method.
Looking at the results (Table 15) it was seen there were two to three possible options that
could be used to collect final data in the monitoring trial. A short treatment with TCA
appeared to be the best across the three milks, with a coefficient of variation range between 3
and 5 %. Highly alkaline conditions also worked well.
51
Table 15: Effects of various treatments on calcium concentration in different milks by the
CPC method and compared to AA method
The decision was made to choose the quick TCA treatment over NaOH denaturation because
the coefficients of variation were lower and TCA has previously been used to denature protein
in calcium determimation (Rodriguez et al., 2001; Silva et al., 2001).
4.4 Conclusion
The CPC method proved to be more precise than the Arsenazo I method, where very high
day-to-day variation was seen. The Arsenazo I method worked well with the reference whole
milk powder, but with liquid milks its variability was totally unacceptable for routine testing.
Both the methods were quick and easy, but possibly acceptable precision was only achievable
by the CPC method. The denaturation of milk proteins improved the results in Xtra milk over
those obtained in a previous trial (Table 14). The CPC method was chosen over the Arsenazo
I method for long term monitoring of calcium in milk.
52
5.1 Introduction
The previous chapter defined a rapid colorimetric method to measure calcium in milk. The
method is here applied to mid-season milk supply in NZDF’s Takanini plant to test the
method’s utility in an industrial setting. At this plant, pasteurised milk is first separated into
cream and skim milk. The skim milk is then passed over ultrafiltration membranes to
concentrate the milk proteins at the expense of the permeate. The concentrate fraction of the
skim milk is NZDF’s Xtra brand. The permeate is low in caseins and other proteins but high
in free calcium and other salts. The three fractions tested were thus skim milk, Xtra, and
permeate.
The aim was to compare the results of the AA and CPC methods from mid of September to
mid January, 2005/06, with periodic monitoring by the Gribbles reference method up to mid
November, 2005.
To cope with the expected wider range of calcium concentrations in this monitoring trial, the
final cuvette concentration of CPC and 8-hydroxyquinoline were doubled for this work.
There was no loss of linearity in calibration curves.
Reagents and standards were prepared for AA as detailed in Section 2.2.2.2 and the AA
procedure were followed in as discussed in subsequent Sections 2.2.2.3 to 2.2.2.5 in Chapter
2.
Milk samples of skim, Xtra and permeate were collected daily from NZDF Takanini, on
typically Monday through Thursday when the dairy prepared Xtra. The milks were
refrigerated in 1 L bottles, the content of which were analysed twice weekly.
53
Milk as sampled off the production line was sent to Gribbles Analytical Laboratories,
Auckland, weekly to mid-November 2005. The laboratory determined calcium by AOAC
984.27, the official food industry method for determining calcium in milk.
Two sequential dilutions were made for each milk. In the primary dilution, 3 mL of Xtra, 5
mL of skim and of permeate, were made to 50 mL. This applied to both the CPC and AA
methods.
For CPC, quadruplicate 2 mL aliquots of each primary dilution were treated with 1 mL of 25
% TCA, shaken and held for 10 minutes before final volumes were made to 50 mL.
For the AA method, quadruplicate 5 mL aliquots of each primary dilution were mixed with 1
ml of 25 % TCA, shaken and held for 5 minutes before the addition of 1 mL of 5 %
lanthanum chloride solution and further standing for 5 minutes. Final volumes were made to
50 mL.
For CPC and AA, 10 mL from each final dilution (i.e. 3 milks x 4 fully-diluted replicates)
were centrifuged at 1500 gravities for 10 minutes and the supernatants recovered.
On each analysis day, calcium was determined without delay after centrifugation. For each
milk, the four replicate dilutions were sampled in triplicate for the CPC method while single
determinations were made by the AA method. Thus, each data point for CPC represents the
mean and standard deviation of 12 values and each for AA represents four values.
To 1 cm-path length cuvettes, reagents and the sample were added in this order: 2 mL of
AMP buffer (1.2 M) 0.5 mL of CPC reagent (450 μM) (includes 8-hydroxyquinoline, 40 mM)
and 0.5 mL of the milk sample from the second dilution. For calibration curves, 0, 100, 200,
300, 400 and 500 μL of calcium standard (225 μM) was added in lieu of milk, and the final
volume made to 3 mL with deionised water. Absorbances were measured at 575 nm.
For AA, light from a calcium lamp (422.7 nm) was directed through an air-acetylene flame.
The lamp current was set to 5 mA. Readings were taken in continuous absorption mode.
54
5.3 Results
Figure 21 shows the results of the monitoring trial for calcium in three milk products, skim
Xtra, and permeate between mid-September 2005 and mid-January 2006. The ultrafiltation
process is clearly effective because in the monitoring period, there was consistent and very
large differences between the calcium content of Xtra and of permeate (Figure 21, Table 16).
2500
2000
)1-L.gm( .ncnoc muiclaC
1500
1000
500
0
Oct Nov Dec Jan
Figure 21: Monitoring of calcium in three milk products during the 2005/06 season by the
CPC and AA methods, with periodic comparison to a commercial laboratory
method. For AA, ○ is Xtra, □ is skim and ∆ is permeate. For CPC, ● is Xtra, ■
skim and ▲ permeate. Gribbles results are in colour. Standard deviations are
vertical lines.
55
Inspection of the Figure 21 shows that AA values agree well with the results from the
Gribbles Analytical Laboratory for all three products. Values for the CPC method were
generally lower than for AA and Gribbles.
Close inspection of Figure 21 suggests that the differences between AA and CPC values in
Xtra and skim were less in the period late September to mid-October, but there was scant
evidence for this in permeate. In January, there was possibly a similar narrowing of the gap
between AA and CPC values for Xtra and skim.
There was some day-to-day variation in calcium concentrations, the cause of which is not
known. Because volume data for the three products were not obtained, it has not been
possible to compare mass balances of calcium as determined by the three methods on a daily
basis.
Inspection of Figure 21 also suggests that there is a gradual long-term decline in calcium
concentration in Xtra as the season progressed. This decline was not evident in skim or
permeate.
5.3.2 Mean values for calcium measured in three products by both methods
The mean values for calcium measured by both methods in all three products were
significantly different (Table 16). The mean value obtained by the AA method was close to
the claimed value, but the mean CPC value was significantly lower. At first sight it appears
that the CPC values in permeate are much closer to the AA values than for skim and Xtra. On
an absolute value basis this is true, but is not true when data are expressed on a ratio basis.
Thus, the CPC:AA ratio as extracted from Table 16 were 0.88 for Xtra, 0.92 for skim and 0.79
for permeate.
Table 16: Mean values of calcium concentration measured by AA and CPC method in three milk
products summed over the monitoring period
The within-day, between-method correlation graphs plotted for Xtra, skim and permeate,
Figures 22, 23 and 24, show that there was no significant correlation for Xtra (r2 = 0.05 and P
< 0.11), an improving correlation for skim (r2 = 0.08 and P < 0.048), and significant
correlation for permeate (r2 = 0.33 and P < 0.0001).
2200
1-L.gm
ni AA yb artX ni muiclaC
2000
1800
1600
Figure 22: Correlation between AA and the CPC methods for calcium determination in Xtra
milk
1400
1-L.gm
1300
ni AA yb miks ni muiclaC
1200
1100
1000
1000 1100 1200 1300 1400
-1
Calcium in skim by CPC in mg.L
Figure 23: Correlation between AA and the CPC methods for calcium determination in
skim milk
57
450
1-L.gm
ni AA yb etaemrep ni muiclaC
400
350
300
250
250 300 350 400 450
Figure 24: Correlation between AA and the CPC methods for calcium determination in
permeate
2500
1-L.gm
Xtra
2000
ni AA yb stcudorp eerht lla ni muiclaC
1500
Skim
1000
500
Permeate
0
0 500 1000 1500 2000 2500
-1
Calcium in all three products by CPC in mg.L
Figure 25: Correlation between the AA and CPC methods for calcium determination in all
three products
58
2500
1-L.gm
Xtra
ni selbbirG yb stcudorp eerht ni muiclaC
2000
1500
Skim
1000
500
Permeate
0
0 500 1000 1500 2000 2500
Figure 26: Correlation between the Gribbles and AA method for calcium determination in all
three products
A rather different picture emerged when all AA data were plotted against all CPC data (Figure
25), where there was a strong correlation between the two methods (r2 = 0.98, P < 0.0001).
This plot also reveals the better correlation in permeate than in skim than in Xtra.
An equivalent correlation plot between the AA method and the Gribbles results showed a
similar pattern in that the correlation with Xtra was the poorest and with permeate the best
(Figure 26).
5.4 Discussion
There are two main issues that emerge from the data in Figure 21. The first is an explanation
as to why the values obtained by the CPC method are lower than those obtained by AA. The
second is why the within-day, between-method variability as revealed by the correlation plots
is greatest for Xtra and least for permeate. The two issues may also be related.
Consider the results for Xtra. According to Fox and McSweeney (1998, p. 257) about 33 %
of the calcium in milk is present as free ions and the remaining 67 % is bound to caseins in the
form of colloidal calcium phosphate. The fraction of calcium partitioning in the permeate
fraction is about 18 % (calculated from Table 16), indicating that about half the free calcium
in the original skim is retained in Xtra on a concentration basis. But after ultrafiltration,
59
nominally all the milk proteins – with caseins dominating – will be present in the Xtra
fraction. All the colloidal calcium phosphate will therefore be in Xtra. In the AA method,
there is a denaturation step with TCA followed by exposure to transient very high
temperatures in the acetylene flame. These events may be sufficient to fully liberate the
colloidal calcium. In the CPC method there was a similar denaturation step with TCA, but no
exposure to extreme temperature. Arguably, not all the colloidal calcium phosphate may be
liberated by the CPC method, so accounting for the low CPC:AA ratio of 0.88. Progressing
now to skim, where there is relatively more free calcium, the ratio improved to 0.92, as would
be expected from this model. However, the model is not tenable for permeate, where
nominally all the calcium will be free. The equivalent ratio for permeate was worse at 0.79.
The correlation plots show that the within-day, between-method variability is greatest for Xtra
and least for permeate. This explanation for this may be the proportion of colloidal calcium
phosphate. When it is relatively low, as in permeate, the results will more reflect the
concentration of free calcium, for which there are no complexities due to protein.
Thus the proportion of colloidal calcium appears useful to explain within-day, between-
method variability, but not the ratio of results from the two methods for the three milks. The
reason remains unknown.
The AA method is not the standard reference method, but was used in this research as an
alternative to the laborious EDTA method. The AA data closely agreed with Gribbles data,
obtained with their official food industry method AOAC 984.27. The fact that the same
pattern in variability occurred between AA and Gribbles (compare Figure 25 and 26) is not
inconsistent with colloidal calcium model discussed above. However, it is worth noting that
Gribbles results appeared to be rounded to nearest 100 for Xtra and skim and for permeate to
the nearest 10 mg.L-1. Correlations would be closer for unrounded data.
The concentration of calcium in skim milk was essentially constant throughout the monitoring
period. According to Fox and McSweeney (1998 p .244) the composition of milk salts is
influenced by a number of factors including breed, individuality of the cow, stage of lactation,
feed, mastitic infection and season of the year. However, calcium concentrations are usually
constant if early and late lactation periods are excluded as is the case in the present monitoring
trial. Again according to Fox and McSweeney (1998, p. 244), diet has little effect on minerals
in milk because the skeleton acts as a reservoir. However, Yves (1995) reported a marked
increase in the concentration of colloidal and free calcium during winter months in bovine
milk in Quebec, Canada, when temperatures drop markedly. Rodriguez et al. (2001) reported
60
almost constant levels of calcium during the 12 months of sampling milks where the
temperature ranged from 18 to 26°C. Therefore the data obtained from milk from grazing
cows, mid-season in a temperate zone, are consistent with the literature.
With no trend in calcium concentration in skim as received in the period under study, the
apparent decline in Xtra’s calcium concentration has to be explained in another way. Factors
linked to ultrafiltration performance may be involved, or the association of calcium between
free and various bound forms may vary subtlely at different times of the year. In this respect
the narrowing of differences between AA and CPC values in Xtra and skim between the
period late September to mid-October is interesting. However, an ultrafiltration effect cannot
be excluded.
61
6 Conclusion
The requirement from New Zealand Dairy Foods was to develop a cheap, simple and rapid
method to measure calcium in a variety of product forms, principally skim milk – the basis of
all processed products – and the ultra filtration products Xtra and permeate. Xtra is enriched
in caseins and thus bound calcium, while permeate is correspondingly low in bound but
higher in free calcium. In this research four methods were used to determine calcium.
Initially the atomic absorption (AA) and EDTA methods were trialled for precision and
accuracy, so that one of them could be adopted as a reference method for this research. For
both methods the results agreed well with the claimed calcium concentration in commercial
milks. The EDTA method was, however, laborious and time consuming whereas AA was
quick and easy. The AA method was chosen over EDTA method.
Colorimetric methods were chosen for development over more costly and complex methods
involving expensive equipment. Two colorimetric dyes Arsenazo I and o-cresolphthalein
complexone (CPC) were selected. These formed coloured complexes with calcium that could
be used to determined calcium concentrations in a spectrophotometer.
For the Arsenazo I method, different buffers were first tested to find a suitable buffer and pH.
Of the various buffers trialled, NaOH/ KCl buffer at pH 12 was found to be suitable. In
colorimetric methods it is important to work with clear samples. Although the milk samples
used in the tests were very dilute, they were not fully transparent due to presence of milk
proteins in colloidal suspension. Trichloroacetic acid was thus used to denature the proteins
(and was routinely used in the AA method as well). The Arsenazo I method worked well
during the development phase but when it was applied in routine to test its precision,
inconsistent results were obtained. Day-to-day coefficients of variation were unacceptably
high.
The CPC dye was complexed with calcium at a pH of 10.5 with AMP buffer. The method
developed here gave consistent calibration curves. Magnesium is present in milk at about 1/3
the concentration of calcium, and also produces a similarly coloured complex with CPC,
although the molar extinction about 1/2 that of calcium. The reaction with magnesium could
be also completely masked with 8-hydroxyquinoline included in the reaction mixture. The
results during the early trials with CPC method were consistently lower than claimed values
on several liquid milk products.
62
Both methods were then trialled on reconstituted milk powder, and the results compared
with those of the AA method and the method used by Gribbles. Agreement was good among
all methods. However, when the colorimetric methods were compared with AA in six liquid
milks over five days, the Arsenazo method proved highly variable. CPC results had
acceptable precision, but the means were variably lower for each liquid milk. Because these
milks contained varying concentrations of calcium-binding caseins and fat, it was reasoned
that agreement with the AA method might be improved by denaturation of caseins, and
removal by centrifugation before colour development. Of the seven methods tried in three
milks, the best results were obtained with brief treatment with 25% TCA. The means were
close to the AA means, and the percent coefficients was low.
The long term monitoring trial on three milk products, skim, Xtra and permeate, showed clear
differences in the amount of calcium in all three products as expected, but there were no clear
long term changes in calcium concentration in any of the fractions. CPC results were
consistently lower than AA results, but this was probably not due to the concentration of
casein and thus bound calcium. Nonetheless, the best correlations between AA and CPC
results were obtained for permeate, suggesting that the state of the calcium was important in
some way to the success of the CPC method.
The correlations obtained were statistically significant only for permeate, but this does not
negate the value of the method because there was no long term change in calcium
concentration. Thus within-fraction plots cover a narrow range of calcium concentration. A
clearer example of the method’s utility is seen when data are pooled. The resulting plot with
three clusters of data (skim, Xtra, permeate), shows very strong correlation with AA. In short
the method works, but with a consistent accuracy bias.
The CPC method here can be applied in an industrial situation. The method is cheap and
involves simple steps and minimal sample preparation. It uses a calcium standard, a TCA
solution, CPC plus 8-hydroxyquinoline, and AMP buffer, all being shelf-stable. The
equipment required is a single beam ,visible-range spectrophotometer or a colorimeter with a
suitable filter, disposable plastic cuvettes, and possibly a bench top clinical centrifuge.
63
Reference list
Bassett J., Denney R. C., Jefferey G. H. & Mendham J. (1978). Vogel's Textbook of
Quantitative Analysis (4th edition). New York: Longman Group Limited.
Belitz H. D. & Grosch W. (1999). Food Chemistry (2nd edition). Berlin: Springer.
Blasco G. F., Bosch R. F. & Campins F. P. (1999). A chemometric study of the simultaneous
determination of calcium and magnesium in natural waters. Talanta, 49(1), 155-163.
Bratov A., Abramova N., Dominguez C. & Baldi A. (2000). Ion-selective field effect
transistor (ISFET)-based calcium ion sensor with photocured polyurethane membrane
suitable for ionised calcium determination in milk. Analytica Chimica Acta, 408(1-2),
57-64.
Chen K.L. & Jiang S.J. (2002). Determination of calcium, iron and zinc in milk powder by
reaction cell inductively coupled plasma mass spectrometry. Analytica Chimica Acta,
470(2), 223-228.
Clemen S., Rabl C.R. & Neumann E. (1988). Kinetic calibration of the calcium indicator
Arsenazo III. I. Stopped-flow spectroscopy. Biochimica et Biophysica Acta (BBA) -
Biomembranes, 938(3), 456-460.
Cook A. J. & Friday J. E. (2003). Food mixture or ingredient sources for dietary calcium:
Shifts in food group contributions using four grouping protocols. Journal of the
American Dietetic Association, 103(11), 1513-1519.
Coultate T. P. (2002). Food: The Chemistry of its Components (4th edition). Cambridge:
Royal Society of Chemistry.
Dalgleish D., Alexander M. & Corredig M. (2004). Studies of the acid gelation of milk using
ultrasonic spectroscopy and diffusing wave spectroscopy. Food Hydrocolloids, 18(5),
747-755.
Dalgleish D. G., Verespej E., Alexander M. & Corredig M. (2005). The ultrasonic properties
of skim milk related to the release of calcium from casein micelles during acidification.
International Dairy Journal, 15(11), 1105-1112.
Dawson R. M. C., Elliot D. C., Elliot W. H. & Jones K. M. (1986). Data for Biochemical
Research (3rd edition). Oxford: Clarendon Press
Delisle J., Amiot J. & Dore F. (1995). Biological availability of calcium and magnesium from
dairy products. International Dairy Journal, 5(1), 87-96.
Dorogi P. L. & Neumann E. (1981). Spectrophotometric determination of reaction
stoichiometry and equilibrium constants of metallochromic indicators :II the Ca2+-
Arsenazo III complexes. Biophysical Chemistry, 13(2), 125-131.
Dorogi P. L., Santarius U. & Neumann E. (1982). Arsenazo I and tetramethylmurexide as
optical calcium indicators. Analytical Biochemistry, 124(1), 27-36.
Ekinci N., Ekinci R., Polat R. & Budak G. (2004). The determination of calcium
concentrations in human milk with energy dispersive X-ray fluorescence. Journal of
Quantitative Spectroscopy and Radiative Transfer, In Press, Uncorrected Proof.
Fifield F. W. & Kealey D. (1995). Principles and Practice of Analytical Chemistry. London:
Blackie Hall.
64
Fox P. F. (2001). Milk proteins as food ingredients. International Journal of Dairy Tech,
54(2), 41-55.
Fox P. F. & McSweeney P. L. H. (1998). Dairy Chemistry and Biochemistry (1st edition).
London: Blackie Academic and Professional.
Goulding A. (2003). Milk components and bone health. Australian Journal of Dairy
Technology, 58(2), 73.
Gratzer W. B. & Beaven G. H. (1977). Use of the metal-ion indicator, Arsenazo III, in the
measurement of calcium binding. Analytical Biochemistry, 81(1), 118-129.
Harris D. C. (2003). Quantitative Chemical Analysis. New York: W.H. Freeman and
Company.
Higson S. A. (2003). Analytical Chemistry. Oxford: New York Oxford University Press.
Ho T.Y., Quigg A., Finkel Z. V., Milligan A. J., Wyman K., Falkowski P. G et al. (2003).
The elemental composition of some marine phytoplankton. J Phycol, 39(6), 1145-1159.
Jensen V. B., Jorgensen I. M., Rasmussen K. B., Molgaard C. & Prahl P. (2004). Bone
mineral status in children with cow milk allergy. Pediatric Allergy and Immunology,
15(6), 562-565.
Johnson, J. (2005). Export profile U.S. Dairy Export Council, Arlington, U. S. A. Retreived
on 13 January 2006 from: http://www.usdec.org/
Kirk R. S. & Sawer R. (1991). Pearson's Composition and analysis of Foods (9th edition).
Harlow, Essex: Longman Scientific and Technical.
Kolar Z. I., Verburg T. G. & Van Dijk H. J. M. (2002). Three kinetically different inorganic
phosphate entities in bovine casein micelles revealed by isotopic exchange method and
compartmental analysis. Journal of Inorganic Biochemistry, 90(1-2), 61-66.
Krebs N. F. (2001). Bioavailability of dietary supplements and impact of physiologic state:
Infants, children and adolescents. The Journal of Nutrition, 131(4S), S1351.
Malcik N., Ferrance J. P., Landers J. P. & Caglar P. (2005). The performance of a microchip-
based fiber optic detection technique for the determination of Ca2+ ions in urine.
Sensors and Actuators B: Chemical, 107(1), 24-31.
McGuire N. (2004). Chemistry 101: atomic absorption spectroscopy. Retrieved on 22nd
March 2006 from:
http://www.chemistry.org/portal/a/c/s/1/feature_tea.html?id=c373e9fc5e4e9a828f6a4fd
8fe800100
Mesquita R. B. R. & Ranngel A. O. S. S. (2004). A sequential injection system for the
spectrophotometric determination of calcium, magnesium and alkalinity in water
samples. Analytical Sciences, 20(8), 1205-1210.
Moorehead W. R. & Biggs H. G. (1974). 2-Amnio-2-methyl-1-propanol as the alkanising
agent in an improved continous flow cresolphthalein copmlexone procedure for calcium
in serum. Clinical Chemistry, 20(11), 1458-1460
Murphy S., Khaw K. T., May H. & Compston J. E. (1995). 94115616 Milk consumption and
bone mineral density in middle aged and elderly women. Maturitas, 21(1), 79.
Mykkanen H. M. & Wasserman R. H. (1980). Enhanced absorption of calcium by casein
phosphopeptides in rachitic and normal chicks. Journal of Nutrition, 110, 2141-2148.
65
Naito H., Kawakami A. & Imamura T. (1972). In-vivo formation of phosphopeptide with
Ca binding property in small intestine tract of rat fed on casein. Agric. Biol. Chem., 36,
409-415.
New Zealand Dairy Industry. (2000). Calcium by complexometric titrations for liquid milks,
milk powders and protein products. NZTM 3: Chemical Methods Manual Section 9:
Major minerals (7), p.9.12.2.
Nordin B. E. C. (1997). Calcium and osteoporosis. Nutrition, 13(7-8), 664-686.
Nordin B. E. C. & Heaney R. P. (1990). Calcium supplementation of the diet justified by
present evidence. British Medical Journal, 300(6731), 1056-1060.
Nyman, J. & Ivaska A. (1995). Spectrophotometric determination of calcium in paper
machine white water by sequential injection analysis. Analytica Chimica Acta, 308(1-
3), 286-292.
Ohnishi T. S. (1979). A method of estimating the amount of calcium bound to the
metallochromic indicator Arsenazo III. Biochimica et Biophysica Acta (BBA) - General
Subjects, 586(2), 217-230.
Oliveira C. C., Sartini R. P. & Zagatto E. A. G. (2000). Microwave-assisted sample
preparation in sequential injection: spectrophotometric determination of magnesium,
calcium and iron in food. Analytica Chimica Acta, 413(1-2), 41-48.
Owusu W., Willett W. C., Feskanich D. & Ascherio A et al. (1997). Calcium intake and the
incidence of forearm and hip fractures among men. The Journal of Nutrition, 127(9),
1782.
Palade J. & Vergara J. (1983). Stoichiometries of Arsenazo III-Ca complexes. Biophysical
Journal, 43, 355-369.
Paull B., Macka M. & Haddad P. R. (1997). Determination of calcium and magnesium in
water samples by high-performance liquid chromatography on a graphitic stationary
phase with a mobile phase containing o-cresolphthalein complexone. Journal of
Chromatography A, 789(1-2), 329-337.
Rattray W. & Jelen P. (1996). Thermal stability of skim milk with protein content
standardized by the addition of ultrafiltration permeates. International Dairy Journal,
6(2), 157-170.
Rodriguez R. E. M., Sanz Alaejos M. & Diaz Romero C. (2001). Mineral concentrations in
cow's milk from the canary island. Journal of Food Composition and Analysis, 14(4),
419-430
Roesch R., Juneja M., Monagle C. & Corredig M. (2004). Aggregation of soy/milk mixes
during acidification. Food Research International, 37(3), 209-215.
Russell D., Parnell W. & Wilson N. (1999). NZ food: NZ people key results of the 1997
national nutrition survey: Ministry of Health. Retreived on March 15th 2006 from
www.moh.govt.nz/moh.nsf/49ba80c00757b8804c256673001d47d0/8f1dbeb1e0e1c70c4
c2567d80009b770?OpenDocument
Sarkar B. C. R. & Chauhan U. P. S. (1967). A new method for determining micro quantities
of calcium in biological materials. Analytical Biochemistry, 20(1), 155-166.
66
Sato R., Noguchi T. & Naito H. (1983). The necessity for the phosphate portion of casein
molecules to enhance Ca absorption from the small intestine. Agricultural and
Biological Chemistry, 47, 2415-2417.
Scrimgeour F. (1998). Reform of the domestic market for dairy produce: an analysis of
progress and issues: New Zealand Business Roundtable. Retrieved on April 10th 2006
from: www.nzbr.org.nz/topic.asp?Topic=IndustryRegulatory
Sigma (2002-2003). Biochemical and reagents for life science research. Electronic source
http://www.aldrichsigma.com.
Skoog D. A., West D. M. & Holler J. F. (1994). Analytical Chemistry: an introduction (6th
edition). Philadelphia: Saunders College Publishing
Silva F. V., Lopes G. S., Nobrega J. A., Souza G. B. & Nogueira A. R. A. (2001). Study of
the protein-bound fraction of calcium, iron, magnesium and zinc in bovine milk.
Spectrochimica Acta Part B: Atomic Spectroscopy, 56(10), 1909-1916.
Singh H. (2004). Heat stability of milk. Int J Dairy Tech, 57(2-3), 111-119.
Stedan J F. & Rensburg A. V. (1990). Simultaneous determination of total and free calcium
in milk by flow injection. Analyst, 115, 605-608.
Stern J. & Lewis W. H. P. (1957). The colorimetric estimation of calcium in serum with o-
cresolphthalein complexone. Clinica Chimica Acta, 2(6), 576-580.
Swaisgood H. E. (1996). Characteristics of milk. In O. R. Fennema’s Food Chemistry (3rd
edition). New York: Marcel Dekker.
Toei J. (1988). High-performance liquid-chromatography using a colour-forming agent as a
component of the mobile phase - separation and determination of magnesium and
calcium in clinical-samples with ortho-cresolphthalein complexone. Analyst, 113(2),
247-250.
Trudeau D. L. & Freier E. F. (1967). Determination of calcium in urine and serum by atomic
absorption spectrophotometry (AAS). Clinical Chemistry 13(2), 101-114.
U.S. Environmental Protection Agency Manual (1983). Methods for chemical analysis of
water and wastes. Calcium: method 215.1 (Atomic absorption, direct aspirations) p.
215.1-215.2
Van Loon J. C. (1980). Analytical Atomic Absorption Spectroscopy: Selected methods. New
York: Academic Press.
Wardlaw G. M. (1999). Perspectives in nutrition (4th edition). Boston: WCB/McGraw-Hill.
Weaver C. M. & Boushey C. J. (2003). Milk good for bones, good for reducing childhood
obesity? Journal of the American Dietetic Association, 103(12), 1598-1599.
Wham C. A. & Worsley A. (2002). New Zealanders' attitude to milk: Implications for public
health. Public Health Nutrition, 6(1), 73-78
Wildman R. E. C. & Medeiros D. M. (2000). Advanced Human Nutrition. Boca Raton.: CRC
Press LLC.
Wiley S. A. (2005). Does milk make children grow? Relationship between milk consumption
and height in NHANES 1999-2002. American Journal of Human Biology 17, 424-441.
67
Appendix 1
Liquid milk (any milk sample from different brands being marketed by NZDF) was weighed 4
grams in 150 mL erlenmeyer flask to 2 decimal places. 25 mL of reagent grade water (25
mL) for liquid milks for every 4 grams was added and the mixture swirled well to disperse the
sample.
For each milk 10 mL 0.1 M hydrochloric acid (HCL) was added, heated to 60o C and the flask
was swirled until the entire sample was dissolved. The mixture was then cooled to room
temperature.
10 ml of EDTA was required for every 1-% m/m of calcium in the sample. The volume of
EDTA (V) was calculated for different milks according to the Ca content claimed in the brand
of the milk
A volume (V) of 0.010 M EDTA, which was 5-10 mL in excess of that, required to complex
all the calcium present in the sample was added by means of a pipette.
1 mL of 0.05 M magnesium sulphate solutions was added. The mixture was swirled to
dissolve the product as completely as possible, then 2 mL of 8.0 M sodium hydroxide added.
The mixture swirled thoroughly and allowed to stand for about half an hour shaking
occasionally until the mixture cleared.
68
Sufficient Patten and Reeder’s indicator was added to produce a distinct purple blue colour.
The mixture was titrated immediately with 0.010 M calcium chloride solution (T1) until a
pink colouration persisted on standing for at least 15 seconds.
Calculations:
The EDTA method was repeated using same amount of 0.1 M trichloro acetic acid in place of
0.1 M HCl to get the clear solution but the results were not found to be reliable.