Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Fine milk compositional analysis by FTIR
Steve Holroyd New Technology Development
Confidential to Fonterra Co-operative Group
February 2018
Page 2Confidential to Fonterra Co-operative Group
Bovine Milk Composition
Water 86.5%
Lactose 4.5%Minerals 0.8%
Protein 3.5%
Fat 4.7%
FatFat
ProteinProtein
TriaglycerideFat-soluble vitamins
Diglyceride
Monoglyceride
Phospholipid
Free Fatty Acid
Sterol
α
- Casein
κ
- Casein
β
- Lactoglobulin
β
- Casein
Serum Albumin
α
- Lactalbumin
IIgg’’s, LPO, GFss, LPO, GFsOtherOther, eg, eg FBP,FBP, LfLf
WaterWater--soluble vitaminssoluble vitamins
FatFat--soluble vitaminssoluble vitamins
Page 3Confidential to Fonterra Co-operative Group
FTIR: Fourier transform infrared spectroscopy
FTIR is widely used for rapid compositional analysis of liquid dairy products.
•Centralised milk testing common globally.
•Allows rapid quantification for gross composition.
•Uses absorption of infrared frequency radiation.
•Fourier transform – mathematical function used for collecting data.
Page 4Confidential to Fonterra Co-operative Group
Principle of infrared spectroscopy
d
I0 I
Light-SourceDetector
Sample
Beer’s Law: Absorbance is proportional to concentration
Page 5Confidential to Fonterra Co-operative Group
FTIR instrument platforms
• Traditional FTIR based instruments – Foss FT1, FT2, FT+
– Bentley DairySpec
– Delta Lactoscope
• Highly accurate (0.02% for fat and protein)• Automated flow/cleaning system with
temperature control and precise homogenisation for consistent sample presentation and high thru put.
• Lower costs systems coming to market.
Page 6Confidential to Fonterra Co-operative Group
Spectroscopy – FTIR spectra of liquid milk
0.0
0.1
0.2
0.3
0.4
1000 1100 1200 1300 1400 1500 1600
wavenumber
abso
rban
ce
Page 7Confidential to Fonterra Co-operative Group
Analysis
1
2
3
4Absorbance
Concentration
1
3
2
4Absorbance
Wavelength
X X
Traditional calibration - quantitative analysis; regression
Page 8Confidential to Fonterra Co-operative Group
Historic measurement of components in milk
• 1980-90s: Traditional quantitative calibration are for gross composition.– Fat, protein, lactose, total solids, solids not fat.
• 2000: Major fractions of fat and protein.– Casein, saturated and unsaturated fatty acids.
• 2008 onwards – adulterants.
• 2012 onwards: Individual fatty acids and proteins and adulterants at concentrations 100-1000ppm.
0.0
0.1
0.2
0.3
0.4
1000 1100 1200 1300 1400 1500 1600
wavenumber
abso
rban
ce
Page 9Confidential to Fonterra Co-operative Group
2015-2017 publications FTIR/milk composition
• Estimation of genetic parameters by FTIR – energy balance.
• Detection of whey in milk/whey quality.
• Within milking variation of milk composition and fatty acid profile.
• Estimation of genetic and cross breeding parameters of fatty acid concentrations in milk fat.
• Screening methods for detection of five adulterants by FTIR.
• Impact of feed on milk composition.
• Estimation of oestrous cycles from milk compositional changes.
Page 10Confidential to Fonterra Co-operative Group
Centralised milk testing by FTIR: Opportunities
• Collection of thousands milk FTIR spectra/day.
• Networking software allows rapid collation of information.
• FTIR instruments can give outputs from multiple complex calibrations simultaneously.
Page 11Confidential to Fonterra Co-operative Group
Specific parameter in raw milk assessed by FTIR 3.6 million Fonterra samples 2014/5
Page 12Confidential to Fonterra Co-operative Group
Details on milk composition at farm level daily
Page 13Confidential to Fonterra Co-operative Group
Rapid fine milk composition analysis by FTIR
• Not as accurate as “traditional” calibrations.
• Useful for trending but must understand fitness for purpose.
• Still require careful calibration and validation.
• Quantitative vs. qualitative approaches.
Page 14Confidential to Fonterra Co-operative Group
What are non-targeted methods?
Page 15Confidential to Fonterra Co-operative Group
Non-targeted evaluation of spectra
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-5 -4 -3 -2 -1 0 1 2 3
Scor
e on p
rincip
al co
mpo
nent
2
Score on principal component 1
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-5 -4 -3 -2 -1 0 1 2 3
Scor
e on p
rincip
al co
mpo
nent
2
Score on principal component 1
New sampleNormalShort distance
New sampleAbnormalLong distance
Page 16Confidential to Fonterra Co-operative Group
Milk “quality” by FTIR
• Expand the range of compositional information available.
• Extend milk fingerprinting through the use of complimentary data sets.
Page 17Confidential to Fonterra Co-operative Group
Non-targeted methods - process
• Gather database – “fingerprints”
• Do statistics.
• Measure new sample and make a decision.
0.0
0.1
0.2
0.3
0.4
1000 1100 1200 1300 1400 1500 1600
wavenumber
abso
rban
ce
Page 18Confidential to Fonterra Co-operative Group
The use of targeted and non-targeted calibration models
• Melamine crisis (2008) liquid milk deliberately adulterated for economic gain.
• Resulted in development of targeted and untargeted FTIR models for detecting milk adulteration at economic levels.
• Use of much more FTIR spectral information.
Page 19Confidential to Fonterra Co-operative Group
Results –non-targeted analysis of liquid milk by FTIR
Threshold
Page 20Confidential to Fonterra Co-operative Group
Fingerprint FTIR analysis of NZ milk
Page 21Confidential to Fonterra Co-operative Group
Challenges of FTIR for detailed milk composition• How accurate can the measurement be?
• What is actually being measured?
• Will the calibration remain reliable over time and how to check it?
• Can different FTIR instruments give the same performance over time?
0.0
0.1
0.2
0.3
0.4
1000 1100 1200 1300 1400 1500 1600
wavenumber
abso
rban
ce
Page 22Confidential to Fonterra Co-operative Group
Vision of the future
• Wide range of FTIR systems with accuracy dependent upon value chain in specific location.
– Targeted and non-targeted models using full FTIR spectral region.
– Systems networked and data integrated and accessible.
– Cost vs. accuracy understood.
– Calibration models used for range of quality parameters.
– Validation appropriate for application.
• Used with guidance from international standard development organisations.
0.0
0.1
0.2
0.3
0.4
1000 1100 1200 1300 1400 1500 1600
wavenumber
abso
rban
ce
Page 23Confidential to Fonterra Co-operative Group
Acknowledgements
• Gavin Scott, Scott Middleton, Callie Smith, Jonathan Crook.
• Michel Nieuwoudt, Cushla McGoverin, Cather Simpson, Chris Trigg.
• Fonterra on-farm R&D. Bridget Maclean, Alana McCool, Paul Jamieson.