Recent advances in the infrared analysis of milk 1 - 1 - Holroyd... · 2015-04-02 · Recent advances in the infrared analysis of milk ... • Screening methods for detection of five

Recent advances in the

infrared analysis of milk

Steve Holroyd Directorate of Science & Technology

Confidential to Fonterra Co-operative Group

April 2015

Page 2 Confidential to Fonterra Co-operative Group

Overview

• Infrared analysis of milk

• Mid infrared spectroscopy

– Measurement of minor components

–Targeted and non-targeted analyses

– Instrument developments

• Near infrared - current applications

• A vision of future applications


Principle of IR-Spectroscopy

d

I 0 I

Light-Source

Detector

Transmission T = —

1

T

I 0

I

Sample

Absorption = A = log — = · c · d

I = intensity of reference beam 0

I = intensity of sample beam

d = sample thickness

c = concentration

= molar absorption coefficient


What’s in liquid milk?

• Evolved as a complete food

• Complex mixture of fat, protein

and carbohydrates

• Extensive microstructure 1-100µm

• Fat globules and casein micelles


Milkfat globules – scatter light

0

2

4

6

8

10

12

14

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10

Volume (%)

Size (µm)


MIR and NIR

• Mid IR

- Transmission through thin sample

- Discrete peaks

- More sensitive

• Near IR

- Good for reflectance

- Broad overlapping peaks

- More potential light scattering

Page 7 Confidential to Fonterra Co-operative Group 7

Near vs. mid infrared for milk

• Mid IR

– Transmission through 37-50um cells using Fourier transform infrared (FTIR) wavelength range 1000-4000cm-1

– Widely used for liquid dairy products

– Both in- and at-line

– Traditionally measures fat, protein and lactose

• NIR

– FTNIR, diode array and dispersive systems, wavelength range 700-2500nm

– Reflectance or transmission at- or in-line

– Widely used for solid and semi-solid products, also for liquids

– Traditionally measures moisture, fat and protein


Where infrared spectroscopy fits in….

Final product

grading

In process

Before processing

For payment


Mid IR spectroscopy

1. Measurement of “minor components” in

milk

2. The use of targeted and untargeted

calibration models to assess milk quality

3. The development of lower cost

instrument platforms


Mid IR spectra of liquid milk

0.0

0.1

0.2

0.3

0.4

1000 1100 1200 1300 1400 1500 1600

wavenumber

ab

so

rba

nce


1. Measurement of “minor 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

• 2010: Individual fatty acids and proteins

• Adulterants at concentrations below 1000ppm

–What is the correlation model (calibration) based on

–Beware of covariance!


MIR new applications 2014-5 published

• Estimation of genetic parameters by FTIR – energy balance (Journal of Dairy Science

98, McParland et al, 2014)

• Detection of whey in milk/whey quality (Food chemistry, 174/176, de Carvalho et al and

Kucheryavkiy et al 2014)

• Within milking variation of milk composition and fatty acid profile (Journal of Dairy

Science, 97/7, Rico et al 2014)

• Estimation of genetic and cross breeding parameters of fatty acid concentrations in milk

fat (Journal of Dairy Science, 81/3, Holroyd et al, 2014)

• Screening methods for detection of five adulterants by FTIR (Food chemistry, 181, 31,

Botelho et al 2015)


2. The use of targeted and untargeted

calibration models

• Melamine crisis (2008) with contaminated milk powder

• Liquid milk deliberately adulterated for economic gain

• Collaborative project between Fonterra, Arla and Foss

• Resulted in development of targeted and untargeted models for detecting milk

adulteration at economic levels

• In 2014 Foss has commercial release of FTIR models using abnormal

spectrum module (ASM) as well as targeted models for specific adulterants


1. Quantitative (targeted analysis)

C:\valhalla\MSCFT2\PCA\DK\New Folder\pre0

0 100 200 300 400 500 600 700 800 900 1000

0

100

200

300

400

500

600

700

800

900

1000

RMSEP=28.111 SEP=28.158 SEPCorr=28.237 SDrep=12.757 Mean=338.09

Predicted (Melamine F/C=9/22)

Act

ual

57 samples in 3 replicates

%CV: RMSEP=8.31 SEP=8.33 SEPCorr=8.35 SDrep=3.77

Slope: 1.0013

Intcpt: -1.8636

r: 0.9969

Bias : -1.4116

FTIR predicted melamine (ppm)

Refe

ren

ce

te

st m

ela

min

e (

pp

m)


Qualification (untargeted analysis)

• Defines a single “good” group

• Anything outside is by definition

“abnormal”

• Advantage – you can find what you

are not looking for


Untargeted 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

pri

ncip

al c

ompo

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

pri

ncip

al c

ompo

nent

2

Score on principal component 1

New sample

Normal

Short distance

New sample

Abnormal

Long distance


Results – untargeted analysis of liquid milk

by FTIR

Threshold


Fingerprint FTIR analysis of NZ milk

0

1

2

3

4

5

6

7

8

9

10

145

89

133

177

221

265

309

353

397

441

485

529

573

617

661

705

749

793

837

881

925

969

101

3105

7110

1114

5118

9123

3127

7132

1136

5140

9145

3149

7154

1158

5162

9167

3171

7176

1180

5184

9189

3193

7198

1202

5206

9211

3215

7220

1224

5228

9233

3237

7

AS

M s

co

re

Sample number

FT2 - ASM Score 07.01.13


Challenge with untargeted analyses

0

200

400

600

800

1000

1200

0 1 2 3 4 5

Limit o

f Det

ectio

n (p

pm)

% Acceptable false positives

Melamine

Ammonium sulphate

Urea


Combining Untargeted and Targeted models

Does the untargeted (FTIR)

model indicate a deviating

sample?

Is there a targeted (FTIR) model

indicating the nature of the

deviation?

The sample is most likely

adulterated Confirmatory chemistry

Is there a targeted (FTIR) model

indicating adulteration with a

specific adulterant?

No Yes

Yes Yes

The deviating sample

must be investigated

further with confirmatory

chemistry

The sample is normal

No No


3. 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

• In widespread use


Lower cost mid IR/FTIR

• Smaller, less automated = substantially cheaper

• Measure gross composition, but less accurate

• Use proprietary technology to reduce interference

from lack of homogenisation

• IndiFoss MilkoScreen

• MIRIS - The Dairy Milk Analyser (filter instrument)

• Foss MilkoScan Mars

Images from Foss and MIRIS


Milkoscreen in use at Fonterra Sri Lanka


Prevention of economically motivated adulteration

• Use FTIR to manage risks

associated with accessing

larger part of global milk pool

• Business support systems for

effective decision making


NIR for liquid milks

• More commonly used for solids and semi-solids

• Liquid milk applications can be challenging

• NIR used a number of applications

– On farm analysis of composition

– Differentiation of milk from cows with different diets

– Organic/non-organic milk

– Different calibration protocols for dealing with light

scattering effects due to fat globule size

– Scattering shown to contribute to 50% total absorbance

in a liquid milk sample

– Milk coagulation

– Somatic cell detection

Source - The use of NIR on milk and milk products (Holroyd,

Journal of NIR, 21/5 2013)


Liquid Sampling Module (LSM)

26

Attachment to existing NIR system

LSM features

• Homogenizer for raw milk

• Peristaltic pump for high viscousity dairy

products, e.g. concentrates and premixes

• Flow cell with 1mm (1000mm) pathlength

• Automated sampling and repeated

measurements

• Basic and advanced cleaning (0.5 and 2%

detergent)

• Similar accuracy to FTIR systems

Info courtesy of Bruker Optics


Vision of the future

• Mid IR - wide range of FTIR systems with accuracy

dependent upon value chain in specific location

– Cost vs accuracy

– Calibration models used for range of quality parameters

– Targeted and non targeted models common

• Near IR systems fill niche roles

– Single instrument for multiple uses, liquids and solids

– On farm applications

• Used with guidance from international standard setting

groups

Documents

Recent advances in the infrared analysis of milk 1 - 1 - Holroyd... · 2015-04-02 · Recent advances in the infrared analysis of milk ... • Screening methods for detection of five