new sensors for food industry - université · pdf filenew sensors for food industry...

NEW SENSORS FOR FOOD INDUSTRY Corrado Di Natale1, Ed Jones2, Gustavo Alcuri3 1 University of Rome Tor Vergata, Italy 2 CTech; Chester, UK 3 Alctra; Montreuil, France

www.cafe-project.org


Outline

• Extending sensing capabilities in CAFÉ project • Approach to new sensors

• Ultrasounds • Electric impedance • Gas sensors

• Electronic nose

• Applications to case studies • Conclusions

2


Sensors available in case study plants

Case study 1: wine

Temperature sensors, flowmeter for injected O2, flowmeter for produced CO2, on-line GC

Case study 2: beer

Viscosity, pH, conductivity, VIS-NIR, particle size distribution, turbidity

Case study 3: freeze-drying

Temperature (product and refrigerating liquid), chamber pressure

Case study 4: Ice cream

Temperature, pressure, power consumption, dasher speed, Focused beam reflection, on-line viscometer

3


CAFÉ approach to extend sensing capabilities

• During the processes a number of physical and chemical properties are altered.

• Considering the complexity of food composition and the changes occurring during the process the use of non-specific broadly selective sensors was considered

• Physical properties • Ultrasound probes • Electric impedance

• Chemical properties • Gas sensor arrays

4


Ultrasonic probes • Ultrasonic sensors investigate the propagation of acoustic waves in a media • Waves are reflected, transmitted and absorbed. • The acoustic properties of materials depend generally on the density (amount of

material moved by the wave) and the mechanical resistances (resistance to movement).

• There are several examples of applications of ultrasounds to the characterization of foodstuff.

Sound velocity vs. concentrations of simple and complex food ingredients Data from Saggin and Coupland, J. Food Sci. 2001

5


Emission-receiver and electronic unit

piezocomposite transducer Emission – reception – treatment unit

delay time

level E R

d E/R

d

Measurement modes: transmission pulse/echo Measurement features

Delay: sound velocity Attenuation: impedance loss

6


Freeze-drying process

Ultrasonic unit A in emission / reception mode.

Ultrasonic unit B in echoes mode.

Emission transducer

Reception transducer

Emission / reception transducer

Ultrasound frequency 1.25 MHz

7


Ultrsonic monitoring of Freeze-drying

-30

-20

-10

0

10

20

30

0

5

10

15

20

25

30

35

40

10:3

110

:37

10:4

810

:53

11:0

411

:13

11:2

211

:30

11:3

411

:36:

…11

:38

11:4

511

:49

11:5

311

:57

12:0

412

:09

12:1

712

:21

12:2

512

:38

13:0

113

:35

13:4

114

:11

14:4

715

:06

15:2

415

:37

tem

pera

ture

(C°)

prop

agat

ion

time

(µs)

mode E/R mode echo temperature

Transition zone at T=-10.6°C Abrupt change in E/R mode. Slow transition of echo mode signal due a progressive stratification down to T=-14.8°C.

8


Electric impedance

• The dielectric properties of foods are studied because of their implications with microwave cooking and processing.

• However, the impedance even at lower frequency can provide interesting information about the content and the structure of the food.

• Dielectric properties depend on a wide range of physical and chemical properties of the material, including: • conductivity • viscosity • density • composition • temperature …

9


Impedance probe

• Simple probe design to allow impedance measurements in the presence of dielectrics under test

• Impedance measurements can be calibrated to give real and complex permittivities: ε�r(f) and ε�r(f).

• Two dielectric standards are required for calibration (usually air and water)

• For CAFÉ monitoring actual dielectric values not required, just signals that change with target processing properties

sensor

analyser

sample fringing field

1.234

conditioning unit

10

Optimized probe

Fringe field diffusion


Fingerprint method with the reflection coefficient

• 1) The frequency dependence of the reflection coefficient of the initial states is measured • The reflection coefficient is related to the

change of dielectric constant with respect to the external medium.

• 2) The frequency dependence of the reflection coefficient of the ongoing state is measured

• 3) The frequency dependence of the difference between the two reflection coefficients is produced

11


Beers mixtures

Flat Hoegarden beer added to flat Leffe beer Bubbly Hoegarden added to bubbly Leffe

Flat Hoegarden added to fizzy Leffe Fizzy Hoegarden added to fizzy Leffe

12


Physical and chemical changes heating stirring

Ethanol addition Water addition

13


Test in real conditions Pilot plant in Wageningen

14

Real permittivity at 100 MHz

• A difference between filtered and unfiltered sample is observed

• Sensitivity to temperature disturbance • Necessity of a better screening from

environmental noise.


Headspace composition

• In principle gas sensors can be used to monitor the evolution of headspace during the processing of foods.

• But: • Headspace build-up may requires a large time with respect to the typical

time scale of the process. • Physical conditions of food during the process (e.g. temperature and

pressure) do not allow for the formation of a readable headspace. • Nonetheless, gas sensors can be used in selected processes and

they are valuable to determine the quality of the final product. • In CAFÉ the properties of sensors based on metalloporphyrins were

considered.

15


Richness and versatility of porphyrins chemistry

16

Hydrogen Bond Acceptor

Hydrogen bond acceptor Hydrogen bond donor

Hydrophobe

Aromatic Hydrophobe

Coordination

Porphyrins building blocks


Porphyrins coated Quartz microbalances

17

• Mass to electric frequency converters

• TSMR sensors are simple to be prepared and the signal read-out can be done with great accuracy.

• Since any interaction results in a variation of the mass, the mass transducer is in principle non-specific.

• The resolution is of the order of 1 ng • About 1012- 1014 molecules have to be

absorbed to have 1 Hz of frequency shift.

Brunink et al. Analytica Chimica Acta 1996


Porphyrins coated QMB based Electronic Nose

18


Some applications

19

Food Quality and Control Eifler et al. PLoS ONE, 2011 Santonico et al. Food Chem, 2010 Santonico et al. Sens Act. B 2008 Olafdsottir et al. Trends Food Sci Tech. 2004 Saevels et al. Posth. Sc. Techn. 2003 ••••••

Medical Diagnosis Santonico et al. Lung cancer, 2012 Montuschi et al. Chest, 2010 D’Amico et al. Skin Res. Techn. 2007 Di Natale et al. Bios. Bioelec. 2003 ••••••

Spacecraft air quality control Martinelli et al. Microgravity Sci Techn. 2008

Fortezza et al. Acta Astronautica, 2006 •••••• ISS 2005

ISS 2011


Application to ice-cream volatile compounds vs. temperature

SPME

-10 -5 0 5 10 150

0.5

1

1.5

2

2.5

3

3.5 x 106

Temperature [°C]

Ab

un

dan

ce [

a.u

.]

Heneicosane, 3-methyl- 1,3,8-p-Menthatriene D-Limonene Crithmene Nerol acetate

20


Monitoring of ice-cream making

• Lemon sorbet • 5 experimental conditions changing process

parameters : • Flow ice-cream machine (35 Kg/h) • Tevap refrigerant liquid (from -9°C to -16°C)

• aim: to predict from gas sensors signals the ice crystals size and the apparent viscosity • ice crystals size measured with Focused Beam

Reflection • viscosity was model estimated • Ice fraction

Crystal size 6.53 μm

Crystal size 6.63 μm

21


Gas sensors measurements

• A measurement chamber is fixed at the output of the ice-cream machine.

• The chamber holds a fixed amount of ice-cream into a closed volume whose headspace is delivered into the sensors cell.

22


Crystal Size Prediction

RMSETrain : 0.1133 µm RMSETEST : 0.1980 µm about 8% of the range of variability of crystals size.

• Feed Forward Neural Network. • Single hidden layer (with 3

neurons) • 2/3 of the dataset been used

for training and 1/3 for test.

23


Estimation of viscosity and Ice fraction

24

RMSETrain : 0.3920 RMSETEST : 0.9926

RMSETrain : 0.0211 RMSETEST :0.0326


Application to freeze-dryed bacteria quality

25

• Concentrates of lactic acid bacteria are used as starters of fermentation processes for the production of cheeses, yogurts,……

• Freeze-drying is the method of choice to improve the shelf-life of concentrates.

active ingredient (Bacteria)

excipients (polymers, sugars,…)

Freeze-drying

lyophilized product storage

Quality criteria • water content (<3%)

• activity • cells viability


Experimental

26

• six batches of lactobacillus delbrueckii sbsp. bulgaricus • samples were prepared with different salts to

favour the drying • different freeze-drying conditions were applied to

change the quality of the products. • Nucleation LyoGamma Telstar

• half of each batch was used to measure water activity, acidification activity, and cells viability

• half for headspace analysis: electronic nose and GC/MS


Measurement setup

27

• Tor Vergata Electronic nose equipped with eight porphyrins coated quartz microbalances

• room temperature measurements


Measured quality parameters

28

• PCA of parameters indicates two groups of samples separated by the acidification activity.

• As expected, acidification activity is anticorrelated to the cells viability

cells viability

water activity

Tmax


Electronic nose data

29

• The PCA of electronic nose data is largely distorted with respect to the quality parameters.

• Nonetheless, the two groups of samples found in the PCA of quality parameters and GC abundances are visible also in electronic nose data.


PLS prediction of quality parameters

30

AW viability Tmax

• PLS was used to estimate the three quality parameters • leave-one-out cross-validation was used for latent variables choice • 4 latent variables minimize the RMSECV


GC PCA

31

high Tmax

lowTmax


compounds identification

32

1 2 3 4 5 6 7 8 9 10 11 12 13 1,5-Hexadiene, 3,3,4,4-tetrafluoro- 53535 19979 45573 0 0 0 42421 0 0 0 0 0 0 2-Butanone, 3-hydroxy- 205615 982019 0 0 0 0 0 0 0 0 0 0 2,3-Butanediol, [S-(R@,R@)]- 665136 280857 1161775 0 0 0 0 0 0 0 0 0 0 Cyclopenta[e]tetrahydropyran-2-one, 5-(5-hydroxyindan-1-yl)-4a-methyl- 95293 114202 0 0 0 0 0 0 0 0 0 0 0 2-Propanol, 2-methyl- 698935 782189 227943 676121 556372 83147 0 146754 155908 196459 127855 0 222490 Oxime-, methoxy-phenyl-_ 197155 0 38095 0 0 0 0 0 0 0 0 158988 0 di-tert-Butyl dicarbonate 116648 119728 28742 132645 93394 12427 0 30913 33652 16980 13465 0 18794 1,1-Hexylenedioxybutane 295974 333641 132045 167240 125220 55304 0 40502 27932 81308 80091 20121 81936 Undecane, 2,2-dimethyl- 0 0 0 0 0 0 0 0 0 0 0 0 0 3-Hexen-2-one 128945 91092 60622 39834 8773 0 0 0 0 39564 25263 0 16199 1-Hexanol, 2-ethyl- 129539 0 0 0 0 0 0 0 0 0 0 39113 0 2,6-Heptadienal, 2,4-dimethyl- 7091 8144 0 8208 6528 0 0 0 0 0 0 0 0 3-Methyl-4-oximido-2-pyrazolin-5-one 0 3219 0 0 0 0 0 0 0 0 0 0 0 2-Propanol, 2-methyl- 0 5133 0 0 0 0 0 0 0 0 0 0 0 Di(1,2,5-oxadiazolo)[3,4-b;3,4-E]pyrazine, 4,8-diacetyl- 5707 0 0 0 0 0 0 0 0 0 0 0 0 2-Butanone, 3-methoxy-3-methyl- 11075 0 0 0 0 0 0 0 0 0 0 0 0 Fenproporex 13720 22489 12173 15452 13682 0 0 0 0 19893 19334 11620 14978 2,4,4-Trimethyl-1-pentyl methylphosphonofluoridate 81356 100433 63158 76281 74496 0 0 14381 8990 89896 103058 64750 86772 3-Butene-1,2-diol, 1-(2-furanyl)- 23937 29765 26056 18556 21781 10271 7458 7682 5516 14569 16054 14813 12565 3,5-Dimethyl-4-octanone $$ 42970 0 0 0 0 0 0 0 0 0 0 11694 0 Dibutyl phthalate 220206 0 28636 0 0 36198 35805 0 39562 0 167130 0

3-hexen-2-one butane-diol 3-hexen-2-one • The chemical functional descriptors

of VOCs more correlated with the acidification activity suggest a good detectability with porphyrins coated QMB.


Detection of filtration stages in beer production

• three formulations from a local brewery • Rajah: indian pale ale with large hop

content • Kerala: indian pale ale spiced and

falvored with orange, apple and cinnamon

• Spiga: weiss beer (31% filtered in the final product)

• three samples for each formulation

• unfiltered (as is from brewery) • paper filtered • siringe filtered

1mm sample of unfiltered

33


Principal component analysis of ENose data

non filtered

paper filtered

syringe filtered

• Filtration affects the ENose signal

• The removal of particulate increases the differences between samples

• The discrimination among beers largely increases after filtration.

• Results shows that the QMB

electronic nose is sensitive to filtration

• On line application in filtration unit is hindered by low-temperatures and high pressure.

34


Conclusions

• Broadly-selective sensors for physical and chemical quantities can provide useful information to monitor the evolution of processes in food industry.

• Electric and acoustic impedance can be used to estimate composition, density and texture properties

• The headspace composition can reveal changes in activity in the liquid material (e.g. due to crystalization) and to signal the presence of compounds affecting the quality.

• Eventually, these sensors can complement the sensors aimed at controlling the process parameters with feedback information about the quality of the processed product.

35

new sensors for food industry - université · pdf filenew sensors for food industry...

Documents