review article microfluidic sample preparation methods for

Review ArticleMicrofluidic Sample Preparation Methods forthe Analysis of Milk Contaminants

Andrea Adami Alessia Mortari Elisa Morganti and Leandro Lorenzelli

Centre for Materials and Microsystems Fondazione Bruno Kessler 38123 Trento Italy

Correspondence should be addressed to Andrea Adami andadamifbkeu

Received 14 August 2015 Accepted 5 November 2015

Academic Editor Kourosh Kalantar-Zadeh

Copyright copy 2016 Andrea Adami et alThis is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

In systems for food analysis one of themajor challenges is related to the quantification of specific species into the complex chemicaland physical composition of foods that is the effect of ldquomatrixrdquo the sample preparation is often the key to a successful application ofbiosensors to real measurements but little attention is traditionally paid to such aspects in sensor research In this critical review wediscuss several microfluidic concepts that can play a significant role in sample preparation highlighting the importance of samplepreparation for efficient detection of food contamination As a case study we focus on the challenges related to the detection ofaflatoxin119872

1in milk and we evaluate possible approaches based on inertial microfluidics electrophoresis and acoustic separation

compared with traditional laboratory and industrial methods for phase separation as a baseline of thrust and well-establishedtechniques

1 Introduction

In systems for food analysis one of the major challengesis related to the quantification of specific species into thecomplex chemical and physical composition of foods thatis the effect of ldquomatrixrdquo in addition target contaminantsmust be often detected at very low concentrations typicallyranging from ngkg to 120583gkg The scenario is then quitedifferent from other routine analysis for example fat andprotein content in milk for quality evaluation of productsduring production which is often based on high throughputIR spectroscopy methods or cytometry for somatic cells orbacteria quantification

In the cases where the detection is based on surfaceinteraction like in biosensors food analysis is particularlysensitive to matrix interaction ranging from unwanted sur-face coating or interaction by matrix components to segrega-tion of target analytes into specific fractions of the samplethus reducing the efficiency of capture by biorecognitionmolecules on the sensor Therefore it is common for sensorsto have low sensitivity specificity reproducibility and veryshort lifetime in food matrixes and it is often required toprocess the sample before quantification of the chemicalspecies under examination in order to increase the analytical

accuracy of a method In the field of biosensors the samplepreparation often relies on time-consuming labour intensiveand sometimes complex laboratory procedures Especiallyin the perspective of applied research where sensors areused in real settings the key performance indicators includethe robustness and easiness of application of the detectionmethod It is then required to provide a sample appropriatefor the sensor which often requires a laboratory preparationnot compatible with most of industrial applications

In the case ofmilk the presence of fat proteins andmanyother components in a complex phase equilibrium posesparticular challenges for analysis The two major elementsin milk microstructure are casein micelles and fat globulesCasein micelles are a mixture of caseins (1205721- 1205722- 120573- and k-casein) which is the most abundant protein family in milk(78) with associated calcium phosphate The structure ofthe micelle can be modelled as a cluster of submicelles linkedtogether by calcium bridges or alternatively flexible array ofcasein molecules interlinking calcium phosphate nanoclus-ters In both models the micelle is stabilized by a surfacecovering of k-casein providing both steric and electrostaticrepulsion Casein micelles have colloidal dimensions (50ndash250 nm) Caseins are known as blocking agent for surfacesand therefore may interfere with sensor surfaces

Hindawi Publishing CorporationJournal of SensorsVolume 2016 Article ID 2385267 9 pageshttpdxdoiorg10115520162385267

2 Journal of Sensors

Fat globules are composed of triglycerides with minorcontributions from diacylglycerols monoacylglycerols free(unesterified) fatty acids phospholipids and sterols andare bounded by a membrane Trace amounts of fat-solublevitamins carotene and fat-soluble flavouring compounds arealso present The native fat globule membrane is complexconsisting of an inner phospholipid unit membrane 01 nmthick with associated lipoproteins glycoproteins enzymesand carotene Although protein accounts for 2 of the weightof the fat globule the total interfacial area of the membraneis 80m2L or 2m2g fat and can have a significant influenceduring milk processing The fat globule size is between 01and 10 120583m in diameter (the range and mean vary with thespecies breed and health of the animal stage of lactationetc) where 90 of globules are between 1 and 8 120583m Theremaining serum phase contains dispersed whey proteinsand dissolved lactose as major components [1ndash3] Fat canaccumulate on surfaces or even clog narrow channels inparticular conditions

The water-based fraction includes several componentswhere sugars and water-soluble proteins (or whey proteins)are the primary part Lactose is the principal carbohydrate inthe milk of most species (0ndash10) and it is unique to milk [4]Serum or whey proteins include 120572-lactalbumin 120573-lactoglob-ulin blood serum albumin (BSA) and immunoglobulins (Ig)and account for 17 of proteins Milk proteins also includemilk fat globule membrane proteins and a large variety ofenzymes (about 60) and hormones [5]

In addition to its complexity the phase structure andcomposition can be influenced by processing For instancethe temperature influences the status of immunoglobulinswhich are linked to the behaviour of fat and casein mem-branes Physical processing is also used to modify the milkmicrostructure for example impact of a high pressure milkstream on a surface results in the homogenisation of thefat globule size during industrial processing This processprevents cream separation in milk usually available in themarket

In this paper we analyse the state of the art of samplepreparation for milk by comparing traditional industrialprocedure with potential for miniaturisation and emergingmicrofluidic methods with a particular emphasis on con-taminant detection and especially for aflatoxin 119872

1as case

studyMost of techniques discussed in the paper however arerelated to phase and chemical separation and can be appliedto other applications

The aflatoxin 1198721is of particular interest because it is a

milk contaminant and potent carcinogen classified in group1 of the International Agency for the Research on Cancer[6] It is characterised by carcinogenic genotoxic teratogenichepatotoxic immunosuppressive and antinutritional effects[7 8] Aflatoxin 119872

1(Figure 1) is a metabolite of aflatoxin 119861

1

Aflatoxin1198611is produced by the ubiquitous fungusAspergillus

flavus In favourable climatic condition the fungus producesaflatoxin 119861

1and contaminates animal feed Following the

ingestion of spoiled feed aflatoxin1198611is hydroxylated creating

in the liver aflatoxin 1198721 Aflatoxin 119872

1is secreted into milk

with an elapsed time of about 12 hours and a peak time of

O

OO

O

O

O

OH

CH3

Figure 1 Chemical structure of aflatoxin 1198721

about 24 hours It is thermostable and thus is not deactivatedby pasteurisation or UHT treatment [9] and it is found indairy products and concentrated in cheese [10]

In order to protect public health and safety in theEuropean Commission (EC) regulation number 18812006the maximum level of aflatoxin 119872

1contamination in milk

is set to 50 ppt and to 25 ppt for infant formulae as avulnerable group of the population Similar regulatory limitsare implemented in other countries in Africa Asia andLatin America where aflatoxins incidence is higher than inthe EU countries due to favourable warm and wet climaticconditions In other countries such as the United States andsome countries inAsia andLatinAmerica the suggested limitis only 500 ppt [11 12]

Aflatoxin is slightly soluble in water (10ndash30 120583gmL) butyet much higher than maximum allowed limit in milk itis insoluble in nonpolar solvents which results in a verysmall fractionmdashabout 5 to 20 depending on the source andconditionsmdashpartitioned in the cream part of milk [10] Inmilk it is mainly weakly absorbed onto milk proteins [13]mainly casein The small molecule size of aflatoxin (MW3283 gmol) is an additional challenge if detection is basedon label-free methods monitoring the change of surfaceproperties like measurement of refractive index (photonicsensors) or impedance (electrochemical capacitive sensors orimpedance spectroscopy) at the surface

With this background it is clear that phase and chemicalseparation are needed for an efficient detection of aflatoxinMany microfluidic methods are available for this purpose[14] but they are usually developed for small samples (oftenblood) Since contaminants like aflatoxin are found at verylow concentration it is often required to work with samplevolumes in the range 1ndash100mL to have a sufficient recoveryfor the analysis and to get a representative and uniform sam-ple from the batch Scaling up processing rate of microfluidicmethods to deal with 1ndash100mL samples in minutes is notpractical in most cases and therefore a careful selection of thetechnology must be performed by including performances ofseparation and processing time

The paper is organised as follows in Section 2 wedescribe the techniques used for phase separation and samplepreparation in the specific case of milk and aflatoxin Wecompare the techniques used in laboratories to prepare thesamples for analysis with state-of-the-art methods like ELISAand HPLC and lateral flow to industrial methods for phaseseparation and microfluidic techniques A discussion of

Journal of Sensors 3

advantages and applications of different methods is reportedin Section 3 where we discuss the role and perspective ofmicrofluidic methods

2 Separation and SamplePreparation Techniques

21 Laboratory Techniques Laboratory analysis of samplecan be performed with different aims and with a range oftechniques Analytical techniques used are HPLC which isthe standard confirmatory method in regulations fluorime-try and ELISA or lateral flow kits for screening In certifiedlaboratories analyses are usually performed with referencemethods to provide the best accuracy and recovery achievableand with the goal to confirm the results of screening onfield Due to the complex phase structure of milk andpurpose of such procedures extraction methods are usedto reduce matrix interference and to provide the best toxinrecovery from the sample Time and complexity are usuallynot an issue for such methods since the number of samplesprocessed is lower than in screening tests The fat fraction isusually discarded beforemeasurement of aflatoxin content toavoid performance reduction due to clogging of structuresinterference or inactivation of immunoaffinity receptors andso forth Other milk components may interact with aflatoxinby reducing the recovery of toxin from sample and interferewith detection sample preparation is often a key procedurefor good analytical accuracy in the detection of AFM1 Inthis paragraph we describe a summary of the most commonpreparation step

211 Centrifugation Ultracentrifugation and UltrafiltrationPhase Separation Centrifugation can be used to fractionatemilk into individual phases for further analysis or processingDefatting milk by centrifugation is a common procedure(typical procedure 2000timesg 101015840 or 4000timesg 151015840) but it isalso possible to use an ultracentrifuge to remove casein frommilk due to the nanometric particle structure where a typicalprocedure is 80000timesg at 4∘C for 45min In addition the useof a cut-off membrane (eg 10000Da) can further removemilk components above a threshold weight which is typicallyused to remove proteins Procedures are batch proceduresand usually time-consuming [15]

212 Clean-Up Columns Solid phase extraction coatedliquidliquid phase separation and immunoaffinity are alter-native techniques for sample preparation and suspensionThey are traditionally used for extraction of trace compo-nents from large volumes in combination with analyticaltechniques Preparation columns for aflatoxin are availableon the market using immunoaffinity (aptamers molecularlyimprinted polymers or antibodies) Similarly solid phaseextraction uses fibres or adsorbent phasemade of appropriatematerials to trap specific molecules with a similar approachto chromatography capillary columns with appropriate sta-tionary phase ranging from all-purpose C18 coating to spe-cific coatings Laboratory instrumentations able to performsuch preparation are commonly available on the market [1617]

One key point for clean-up columns especially whenused in combination with biosensors is the definition ofrelease buffer which should not be an organic solvent as con-ventionally used in clean-up columns for HPLC and shouldnot interfere with the immunoaffinity sensor Release fromimmunoaffinity columns can be performed by pH shock andreequilibration or temperature in a specific suspension buffer

213 Solvent Extraction Solvent extraction is the traditionaltechnique for aflatoxin separation and concentration inHPLC analysis procedure [18] Extraction is traditionallyperformed in chloroform or other organic solvents which arepurified with columns evaporated and resuspended in thesolvent of choice for the analysis (typically watermethanolfor HPLC)

As reported in the literature it is possible to extractthe aflatoxin without the use of organic solvents in milkpreparation by adding sodium citrate to milk and applying atemperature treatment [19]With such an approach chelationof calcium ions by citrate or EDTA [20] results in an increaseof hydration of casein micelles and proteins are resuspendedinto water with an increase of aflatoxin availability fordetection Organic solvents should be avoided in extractionprocedure to reduce the compatibility issues of polymericstructures and safety issues in the management of consum-ables and compatibility with aptamers on the sensor whichare denatured in nonwater solutions Extraction procedurestypically provide a factor 2 improvement in binding to ELISAplates as reported in [19]

More recently in several new commercial rapid kitsthe extraction procedure is often avoided to speed up thedetection since the time-for-result is a key performanceparameter for users However the kits are prone to matrixinterference and must be validated for a specific matrix stillproviding accuracy issues and a general overestimation inthe quantification of the analyte [21 22] although this isacceptable for commercial screening tools

Lateral flow is both a laboratory method used for screen-ing of samples and amicrofluidic technique coupling intrinsicsample preparation and detection The approach providesinexpensive and easy-to-use analysis and it is used by manycommercial devices like Charm ROSA [23] for integration ofa competitive ELISA test The use of a membrane typicallyin nitrocellulose allows for intrinsic filtering while the useof surfactants in the coupling pad to support the flow intothe nitrocellulose membrane and avoid clustering of nanodotreporters is also supposed to perform an intrinsic aflatoxinrelease from complex with casein (Figure 2)

The technique is becoming a diffused approach for low-accuracy semiquantitative testing including sample prepara-tion in a user-friendly format The simplified sample prepa-ration is compatible with a semiquantitative evaluation ofsample contaminations above the required limit by regulationfor screening purposes where degradation of performancesor repeatability of the test is a minor issue as long as samplesabove the threshold are identified The typical approach forscreening methods is to minimise the occurrence of falsenegatives which results in an overestimation of results anda potentially large number of false positives The volume of


1

2 3 4 5

Figure 2 Lateral flow format 1 membrane and support layer 2sample pad 3 coupling pad 4 reporting areas and 5 adsorptionpadmdashelaborated from [23]

Figure 3 Centrifugal separation by rotative particle separatorSource [24]

sample that is possible to process with such method is verylimited which may result in nonrepresentative samples insome food applications

22 Industrial Standard Techniques for Phase Separation

221 Rotative Particle Separation Rotative particle separa-tion is a well-established separation technique used to skimmilk in process lines with very large throughput It is aseparation technique based on densityinertia and improvesthe hydrocyclone separation technique by the use of a rotorto improve the efficiency and speed of separationMilk entersin the separation bowl from the bottom and is distributed inthe rotating separation plates through distribution holesThefat fraction is concentrated at the centre of the centrifuge andcollected at top of the inverted bowl (Figure 3)

The remaining fat content in the skimmedmilk is usuallybetween 004 and 007 corresponding to 1-2 of the full-milk fat content (about 35 fat) [24]

222 Cross-Flow Filtering and Ultrafiltration Cross-flowfiltering and ultrafiltration are a well-established procedurefor separation of milk fractions It is often adopted in dairiesfor mild pasteurisation by removing bacteria by filtering with

appropriate filter size (Figure 4) [25] A wide range of sep-aration techniques ranging from microfiltration to reverseosmosis allows the separation of each fraction of milk Inparticular microfiltration removes fat and bacteria ultrafil-tration removes casein micelles ultrafiltrationnanofiltrationcan further remove proteins from whey and reverse osmosiscan separate salts and inorganic compounds leaving onlywater in the final permeate

Transverse-flow microfilters are designed to work withproduction volumes and are typically in the form of tubularceramic polymeric or stainless steel filters A longitudinalflow is maintained in the multiple bores or spiral woundedmembrane of the filter in order to provide a cross-flow withrespect to separation membrane and therefore remove anydeposit at filter surface [26]

A further improvement of performances is related tothe possibility of using the high-speed cross-flow to ldquopinchrdquothe flow through lateral pores similar to what is used incontinuous flow separation [14] The concept is that the largeparticles cannot enter the thin boundary layer at channelwall that is separated through the lateral pores becauseof size exclusion The lateral flow is then free of particlesabove a threshold size which can be considerably lower thanactual pore size thus increasing the filter efficiency Furthereffect of cross-flow is the exploitation of microfluidic forcesas described in Section 3 for inertial separation where thelarger particles tend to move to the centre of flow becauseof interaction with channel walls thus reducing the filterclogging by larger particles [27]

223 Air or Gas Floatation Air or gas floatation is a methodwidely used in waste processing especially for the separationof phases of emulsions typically oil in water or suspensionsSince many food samples are emulsions the method can beused as a solution for sample clean-up at large volumes Thestandard process includes the following steps

(1) Coagulation or flocculation for solid suspension ispromoted by chemical agents pH variations and soforth in order to speed up the procedure in a laterstage

(2) The flow is pressurised with air or N2flow in order to

obtain a supersaturated solution of gas in the liquidphase Gas content follows Henryrsquos law 119875 = 119862119883where 119875 is partial pressure of the gas 119862 is solubilitycoefficient and119883 is dissolved gas in the liquid phase

(3) The liquid phase is transferred to inlet into a vessel atroompressure thus producing a fine dispersion of gasbubbles in the bulk Bubbles nucleate at the interfaceof heterogeneities of the emulsion and float to the sur-face carrying the oil droplets or particles with them

(4) Then there is mechanical or fluidic separation of toplayer with specific structures engineered to increasethe throughput


Reverseosmosis

Nanofiltration

Ultrafiltration

Microfiltration

Particlefiltration

Gelatin

Bacteria

Redbloodcells

Virus

Oilemulsion

Sugars

Aminoacids

Aqueoussalts

Antibiotics

Mol wt (Da) 100

Pressurebar 27ndash69 bars 7ndash41 bars 5ndash14 bars 1ndash86 bars

001 101 100001Size (120583m)

1k 100 k 1M

Figure 4 Separation ranges by different filtration techniquesmdashelaborated from the following source [26]

Separation follows the sedimentation law where the sedi-mentation speed is

V119904=

1198892(120588119901minus 120588119897)

18120583 (1)

where 120588 is the density of particle and liquid 119889 is the particlediameter and 120583 is the viscosity When air floatation is used afixed ratio of dissolved gas versus particle typically 00065ndash008mg airmg particles is used to separate particles effi-ciently The floatation rate is calculated from sedimentationlaw where density of the particle is corrected for air bubblesattached to the particle thus typically reducing fat densityfrom 08 g cmminus3 to equivalent density of 0015 g cmminus3 (assum-ing uniform distribution of gas on particles) With suchmodel a floating velocity of about 90 120583ms was estimated for10 120583m fat droplets In order to achieve a real separation suchseparation speed must be larger than average speed of Brow-nianmote V

119861= (119896119879119898)

05This is typically found for particleslarger than 10 120583m which does not make this approach conve-nient for the purpose of milk separation without coagulationof particles before processing A similar technique is the foamfractionation which is reported for protein enrichment [28]based on surface affinity of bubbles toward the component

to be separated where bubbles are produced by flowing gasthrough a glass frit in the sample vessel

3 Discussion of the Role ofMicrofluidic Methods

In principle the peculiarity of microfluidic devices can beexploited to miniaturise standard separationmethods or newconcepts In food applications not all traditional advantagesof microfluidics are necessarily an advantage For instancesample volume is often irrelevant or even worse the samplemust be large enough to be representative of a large nonuni-form batch to be analysed One of the major advantages ofmicrofluidics is very precise control of the interfaces betweenlaminar flows or phases [29] and high surface-to-volumeratioThis can be used tominiaturise the solvent extraction byputting in contact different parallel laminar flows transferringaflatoxin from a sample to a solvent stream thus providinghigher diffusion rates leading to faster extraction procedure

The tight control of temperature flow rates and volumesinmicrofluidic devices can provide the integration ofmultiplefunctionalities into the device like fast thermal processing orcycling due to small thermalmass faster diffusion to surfaces


for molecule trapping or detection due to high surface-to-volume ratio Clean-up columns can be also included asappropriate fluidic chambers with active coating similarly tominiaturise integrated chromatography chips [30] Thermalrelease from clean-up column can be largely sped up by thesmall volumes used which is used for instance for fast ther-mal cycling of PCR procedures on chip Working with largeactive areas over volume ratios and fine control of flow rateand temperature allows achieving concentration of analyteswhich are not possible with standard clean-up concentrationcolumns due to large volumes of standard devices

The new concepts of continuous flow inertial separationand electrophoretic separation are possibilities which aremade possible by a scale reduction of flow thus increasingthe speed and effectiveness of separation with approacheswhich are not applicable at macroscopic scale On theother hand microscale fluidic forces and microfluidic con-cepts are already exploited to improve the performances ofmacroscopic systems like the role of inertial forces in theoptimisation of performances of cross-flow filters workingon the scale of litresminute in industrial plants

In addition to peculiar functionalities microfluidic sys-tems can automatize bioassays and immunoaffinity proce-dures for instance by dispersing beadswith appropriate coat-ing and functionalities (eg magnetic beads) into a sampleand recovering them by magnetic forces Full-assay inte-gration by mixing beads agitation of sample bead captureand target analyte release and detection was demonstrated[31 32] The direct integration of detection is possible thusexploiting the small channel volumes reduced dead volumesand fast diffusion to surface [33] Further down to real appli-cations the possibility of having high parallelisation of pro-cedures into multiplexed devices can lead to high throughputby parallel processing of several samples at the same time

31 Microfluidic Approaches to Milk Analysis Differentmicrofluidic approaches can be used to separate mixturecomponents andparticles especially based onparticle size byexploiting forces generated by the interaction of particles withflow conditions and microfluidic structures In the specificcase of fat separation from milk the required performancesare

(1) to be able to separate fat particles that is in the sizerange 1ndash10 120583m

(2) to have an output throughput compatible with userrequirements that is processing 1ndash10mL in minutes

(3) to be operated with simple equipment to be self-cleaning or compatible with long term use or to beeasily replaced and cheap if disposable

(4) low susceptibility to clogging and performance degra-dation

Not all microfluidic approaches are suitable to fit with theseconstraints here we discuss a selection of best candidateoptions available and their foreseen performances Whenfurther separation is considered for instance chemical orcasein separation a combination of different techniques

can be integrated in a single device to reach the requiredperformances In this section we discuss the most promisingapproaches for sample preparation in milk

311 Inertial Microfluidics Inertial microfluidics is a branchof microfluidics dealing with flow condition in which alaminar flow in a microchannel is coupled with intermediate(1ndash100) Reynoldrsquos numbers thus providing a limited butstill effective inertial force on flow stream thus exploitingpeculiar effects and opportunities for separation of emulsionsor suspensions At that scale there is a complex interactionof particles with flow and channel walls resulting in a varietyof behaviours depending on channel geometry (size aspectratio and eventual bending) particle properties (size particlesize-to-channel ratio and particle concentration) and flowconditions (Reynoldrsquos number shear rate and Pecletrsquos num-ber) The position and conditions for focusing are related totwo major forces that is shear lift and wall repulsion [34]

Shear lift can be expressed as

119865119871=

41205881198621198711198802

1198911198864

31205871205831198632

ℎ

(2)

where 120588 and 120583 are density and viscosity of fluid a particledimension 119863

ℎis hydraulic diameter of channel 119880

119891is flow

velocity and 119862119871is lift coefficient Since 119862

119871is proportional to

1198822(1198862Re05) with 119882 channel dimension the overall shear

force moving particles toward walls is inversely proportionalto 1198862 The wall repulsion depends on a similar equation but

119862119871is proportional to 119867

2(119886Re05) where 119867 is the minimum

channel dimension and it is proportional to 1198863120575 where 120575

is the distance from channel wall Larger particles are thenconcentrated to a position toward the centre of flow streamwith respect to smaller particles

Length required for focusing is

119871 =3120587120583119863

2

ℎ119871119898

21205881198621198711198801198911198863

(3)

where 119871119898is the travelling distance Optimal condition for

particle focusing that is minimisation of focusing lengthis usually achieved for Re in the range 20 to 100 dependingon particle size (from experimental results Figure 3 in [34])Typical focusing length for particles with diameter around10 120583m is in the order of millimetres for shear focusing towardchannel walls and about 1-2 centimetres for shear focusingtoward channel centre Particles then focus to equilibriumpositions at the centre of channel walls especially at thecentre of wider channel sides for high aspect ratio sections

The flow regimewhere separation can be achieved followsthe assumptions [35]

120582 =119886

119863ℎ

gt 007 (4)

Then maximum hydraulic diameter is fixed for a givenparticle size to be separated as a reference for particles of1 120583m 119863

ℎmust be less than 14 120583m and for 4 120583m 119863

ℎlt 56 120583m

The other condition for particle Reynoldrsquos number Re119901is

Re119901= Re 1205822 gt 1 (5)


thus setting Re gt 204 unless 120582 is large (ie small hydraulicdiameter versus particle size whereminimumRe can be egas low as 20)

When channels are bended effect of inertial forces in flowregimes at intermediate Re can result in a lateral asymmetricfocusing due to inertial effect [36] namely Dean forcesrelated to Dean number De = Re 12057512 where the curvatureratio 120575 = 119863

ℎ2119903 compares the hydraulic diameter and

curvature radius 119903 According to experimental results in [35]optimal condition for lateral migration is 120582120575

12= 2 that is

120575 gt 0001225 or 2119903 less than 11ndash45 cm with119863ℎof 14ndash60120583m

respectivelyA combination of inertial and hydrodynamic forces can

be therefore used for the separation of fat particles at highflow rate since they are in the 1 to 10 120583m size range while thereis no expected separation of casein which is much smallerand it is not effectively separated with this technique

As an additional opportunity acoustic separation incontinuous flow devices can provide a further separationmethod by exploiting the fat acoustic contrast factor versusproteins and matrix Transversal separation as in [37] maybe tailored to perform high efficiency separation of fatand proteins from milk Separation of particles with suchtechniques was demonstrated for beads and proteins Theseparation force is

119865119877

= minus(

1198750119881119901120573119898120587

2120582)120601 (120573 120588) sin(

4120587119909

120582) (6)

where the contrast factor Oslash depends on density 120588 andcompressibility 120573 119875

0is pressure amplitude 119881

119901is particle

volume and 120582 is wavelength Although casein enrichmentwas demonstrated in the cited paper the flow rate was quitelow (not reported in the paper for best performing case butinferred from the text as 10ndash30 120583Lmin) By applying thismodel it was not possible to demonstrate the scalability of thedesign to achieve separation of particles down to 1120583m size atflow rates in themLmin range as required by the application

312 Electrophoresis Electrophoresis is a common techniqueused for the analysis of mixtures where the different mobilityof molecules under an electric field is used to separate thecomponents and analyse them with an appropriate detectorIn such configuration the volume of sample is quite smalland separation rate is low Electrophoresis can be also used topurify or separate species for production or sample prepara-tion with a different configuration of the device in the case ofsample preparation the use of electrophoresis in continuousflow devices can select chargedmolecules and particles in thesample and move them into a specific stream of flow thatcan be separated downstream Electrophoretic separationin milk can provide a casein-enriched or depleted phaseby exploiting the charged nature of proteins at milk pHElectrophoresis can be used to separate effectively moleculesand it is expected to scale better than body forces like inertialseparation for small particles While inertial force scaleswith 119871

3 surface charge scales with 1198712 and drag resistance

with 119871 therefore electrophoresis scales better than bulkforces like acoustic or inertial separation for instance and

Table 1 Example of mobility of ions sugars and proteins in milk

Mobility UnitCasein 850 10minus11 m2sLactose 490 10minus10 m2sClminus 203 10minus9 m2s

the efficiency of separation for small particles performsbetter Transversal separation as in [38] may be tailored toperform the separation of protein frommilkThe system wasalready demonstrated for protein separation in the literaturealthough challenges related to milk composition and toxinpartition into milk phases need to be investigated and aspecific protocol must be set up

The electrophoretic mobility as function of ionic strengthcan bemodelled by theOnsager and Fuossmodel [39] and inparticular mobility is reduced with (ionic concentration)05and at the ionic strength of milk 119868 = (12)sum 119888

11199112

119894= 008M

[40] the expected reduction of mobility of species can be upto about 50 as reported in [39 Figure 1]

Drag velocity is calculated as V = 120583119864 where themobility120583

of a charged chemical species in a solution is approximated as

120583119894=

10038161003816100381610038161199111198941003816100381610038161003816 119890

(6120587120578119903119894) (7)

where 119911 is the ion charge 119890 is the electronic charge 120578 is the sol-vent viscosity and 119903 is the ion radius [41] By interacting withthe solution an electrical double layer is formed thus alteringthe charge distribution in the molecule or particle surround-ings For proteins in particular it is common practice toevaluate the electrical mobility as a consequence of doublelayer formation around the molecule thus resulting in a dragby electric field proportional to Zeta potential of the particle

120583119894=

120577120576

120578 (8)

The Zeta potential of the chemical compound dependson pH of the solution where the separation is performedFor proteins in particular the isoelectric point that is thepH where the molecule has neutral charge in solution isfound as tabulated data Typical mobility of different milkcomponents is reported in Table 1

The use of transverse separation allows efficient separa-tion with electrophoretic displacements in the order of 10to 100 microns which should allow for high performanceseparation with sufficient throughput

4 Conclusions

In conclusion microfluidics can play a significant role in ana-lytical system by providing methods to increase sensitivityaccuracy and reproducibility of detection in real samplesThesmall size of microfluidic structures and peculiar phenomenaat microscale can be exploited to improve the speed of pro-cessing with respect to conventional laboratory techniqueswhile the increased system integration can reduce the work


burden to perform laboratory test on food matrices Thisapproach follows the research trend of increasing blendingof biochemistry physics engineering and food science toproduce new micronanobiosystems able to take over real-lifechallenges related to food safety health and food production

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgment

This work has received funding from the European UnionrsquosSeventh Framework Programme FP7-ICT-2013-10 underGrant Agreement no 610580 STREPProject ldquoSYMPHONYrdquo

References

[1] M Auty ldquoMicroscopy (microstructure of milk constituents andproducts)rdquo in Encyclopaedia of Dairy Sciences JW Fuquay EdAcademic Press 2nd edition 2011

[2] M W Taylor and A K H MacGibbon ldquoMilk lipidsmdashgeneralcharacteristicsrdquo inEncyclopaedia of Dairy Sciences JW FuquayEd Elsevier 2nd edition 2011

[3] P F Fox ldquoFat globules in milkrdquo in Encyclopaedia of DairySciences J W Fuquay Ed 2nd edition 2011

[4] P F Fox ldquoMilkmdashintroductionrdquo in Encyclopaedia of DairySciences J W Fuquay Ed Academic Press 2nd edition 2011

[5] D Dupont R Grappin S Pochet and D Lefier ldquoMilkproteinsmdashanalytical methodsrdquo in Encyclopaedia of Dairy Sci-ences J W Fuquay Ed Academic Press 2nd edition 2011

[6] International Agency for Research on Cancer (IARC) Mono-graphs on the Evaluation of the Carcinogenic Risk of Chemicals toHumans International Agency for Research on Cancer (IARC)Lyon France 1993

[7] J H Williams T D Phillips P E Jolly J K Stiles C MJolly and D Aggarwal ldquoHuman aflatoxicoses in developingcountries a review of toxicology exposure potential heathconsequences and interventionsrdquo The American Journal ofClinical Nutrition vol 80 no 5 pp 1106ndash1122 2004

[8] P B Wangikar P Dwivedi N Sinha A K Sharma and A GTelang ldquoEffects of aflatoxin B

1on embryo fetal development in

rabbitsrdquo Food and Chemical Toxicology vol 43 no 4 pp 607ndash615 2005

[9] M Carvajal A Bolanos F Rojo and I Mendez ldquoAflatoxinM1in pasteurized and ultrapasteurized milk with different fat

content in Mexicordquo Journal of Food Protection vol 66 no 10pp 1885ndash1892 2003

[10] L Anfossi C Baggiani C Giovannoli and G Giraudi ldquoOccur-rence of aflatoxinM1 in dairy productsrdquo inAflatoxins I Torres-Pacheco Ed InTech Rijeka Croatia 2011

[11] FAO ldquoWorldwide regulations for mycotoxins in food andfeed in 2003rdquo Corporate Document Repository Food andAgriculture Organisation 2011

[12] US-FDA Whole Milk Low fat Milk Skim Milk Aflatoxin M1CPG section 527400 2005

[13] IARCTheMonographs vol 100F IARC 2012[14] N Pamme ldquoContinuous flow separations in microfluidic

devicesrdquo Lab on a Chip vol 7 no 12 pp 1644ndash1659 2007

[15] A Barbiroli F Bonomi S Benedetti et al ldquoBinding of aflatoxinM1to different protein fractions in ovine and caprine milkrdquo

Journal of Dairy Science vol 90 no 2 pp 532ndash540 2007[16] ldquoWebsite ofThermoScientificrdquo 2015 httpswwwthermoscien-

tificcomenproductsautomated-solid-phase-extractionhtml[17] Agilent ldquoSimultaneous determination of 26 mycotoxins in

sesame butter using modified QuEChERS coupled withUHPLC-ESItriple quadrupole mass spectrometryrdquo Applica-tion Note 5991-5347EN Agilent 2015

[18] Macrae HPLC in Food Analysis Academic Press 1988[19] L Anfossi M Calderara C Baggiani C Giovannoli E Arletti

and G Giraudi ldquoDevelopment and application of solvent-freeextraction for the detection of aflatoxin M

1in dairy products

by enzyme immunoassayrdquo Journal of Agricultural and FoodChemistry vol 56 no 6 pp 1852ndash1857 2008

[20] M C A Griffin R L J Lyster and J C Price ldquoThe disaggrega-tion of calcium-depleted casein micellesrdquo European Journal ofBiochemistry vol 174 no 2 pp 339ndash343 1988

[21] W Li S Powers and S YDai ldquoUsing commercial immunoassaykits for mycotoxins lsquoJoys and sorrowsrsquordquo World MycotoxinJournal vol 7 no 4 pp 417ndash430 2014

[22] A J Alldrick ldquoLooking for best compromise in rapid foodmycotoxin tests speed sensitivity precision and accuracyrdquoWorld Mycotoxin Journal vol 7 no 4 pp 407ndash415 2014

[23] S J Saul M E Tess and R J Markovsky ldquoLateral flow testkit and method for detecting an analyterdquo US Patent no US7785899 B2 Charm Sciences Lawrence Mass USA 2010

[24] Tetra Pak ldquoCentrifugal separators andmilk fat standardisationrdquoin Dairy Processing Handbook chapter 62 p 98 Tetra Pak2003

[25] G T Vladisavljevic M Shimizub and T Nakashima ldquoPrepa-ration of monodisperse multiple emulsions at high productionrates by multi-stage premix membrane emulsificationrdquo Journalof Membrane Science vol 244 no 1-2 pp 97ndash106 2004

[26] GEAfiltrationAugust 2015 httpwwwgeafiltrationcomtech-nologycross flow filtrationasp

[27] A M C van Dinther C G P H Schroen A Imhof H MVollebregt and R M Boom ldquoFlow-induced particle migra-tion in microchannels for improved microfiltration processesrdquoMicrofluidics and Nanofluidics vol 15 no 4 pp 451ndash465 2013

[28] J Merz H Zorn B Burghoff and G Schembecker ldquoPurifi-cation of a fungal cutinase by adsorptive bubble separation astatistical approachrdquo Colloids and Surfaces A Physicochemicaland Engineering Aspects vol 382 no 1ndash3 pp 81ndash87 2011

[29] J Atencia and D J Beebe ldquoControlled microfluidic interfacesrdquoNature vol 437 no 7059 pp 648ndash655 2005

[30] L Sainiemi T Nissila R Kostiainen S Franssila and R AKetola ldquoA microfabricated micropillar liquid chromatographicchip monolithically integrated with an electrospray ionizationtiprdquo Lab on a Chip vol 12 no 2 pp 325ndash332 2012

[31] LA Sasso IH JohnstonMZheng RKGupte A Undar andJ D Zahn ldquoAutomated microfluidic processing platform formultiplexed magnetic bead immunoassaysrdquo Microfluidics andNanofluidics vol 13 no 4 pp 603ndash612 2012

[32] C T Lim and Y Zhang ldquoBead-based microfluidic immunoas-says the next generationrdquo Biosensors and Bioelectronics vol 22no 7 pp 1197ndash1204 2007

[33] A Mortari and L Lorenzelli ldquoRecent sensing technologiesfor pathogen detection in milk a reviewrdquo Biosensors andBioelectronics vol 60 pp 8ndash21 2014


[34] J Zhou and I Papautsky ldquoFundamentals of inertial focusing inmicrochannelsrdquo Lab on a Chip vol 13 no 6 pp 1121ndash1132 2013

[35] J M Martel and M Toner ldquoParticle focusing in curvedmicrofluidic channelsrdquo Nature Scientific Reports vol 3 article3340 2013

[36] D Di Carlo ldquoInertial microfluidicsrdquo Lab on a Chip vol 9 no21 pp 3038ndash3046 2009

[37] C Grenvall P Augustsson J R Folkenberg and T LaurellldquoHarmonic microchip acoustophoresis a route to online rawmilk sample precondition in protein and lipid content qualitycontrolrdquo Analytical Chemistry vol 81 no 15 pp 6195ndash62002009

[38] N Narayanan A Saldanha and B K Gale ldquoA microfabricatedelectrical SPLITT systemrdquo Lab on a Chip vol 6 no 1 pp 105ndash114 2006

[39] S S Bahga M Bercovici and J G Santiago ldquoIonic strengtheffects on electrophoretic focusing and separationsrdquo Elec-trophoresis vol 31 no 5 pp 910ndash919 2010

[40] F Gaucheron ldquoMilk salts distribution and analysisrdquo in Encyclo-pedia of Dairy Sciences J W Fuquay Ed p 908 Elsevier 2ndedition 2011

[41] C G Zoski Handbook of Electrochemistry Elsevier 2007

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics


Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks



Fat globules are composed of triglycerides with minorcontributions from diacylglycerols monoacylglycerols free(unesterified) fatty acids phospholipids and sterols andare bounded by a membrane Trace amounts of fat-solublevitamins carotene and fat-soluble flavouring compounds arealso present The native fat globule membrane is complexconsisting of an inner phospholipid unit membrane 01 nmthick with associated lipoproteins glycoproteins enzymesand carotene Although protein accounts for 2 of the weightof the fat globule the total interfacial area of the membraneis 80m2L or 2m2g fat and can have a significant influenceduring milk processing The fat globule size is between 01and 10 120583m in diameter (the range and mean vary with thespecies breed and health of the animal stage of lactationetc) where 90 of globules are between 1 and 8 120583m Theremaining serum phase contains dispersed whey proteinsand dissolved lactose as major components [1ndash3] Fat canaccumulate on surfaces or even clog narrow channels inparticular conditions

The water-based fraction includes several componentswhere sugars and water-soluble proteins (or whey proteins)are the primary part Lactose is the principal carbohydrate inthe milk of most species (0ndash10) and it is unique to milk [4]Serum or whey proteins include 120572-lactalbumin 120573-lactoglob-ulin blood serum albumin (BSA) and immunoglobulins (Ig)and account for 17 of proteins Milk proteins also includemilk fat globule membrane proteins and a large variety ofenzymes (about 60) and hormones [5]

In addition to its complexity the phase structure andcomposition can be influenced by processing For instancethe temperature influences the status of immunoglobulinswhich are linked to the behaviour of fat and casein mem-branes Physical processing is also used to modify the milkmicrostructure for example impact of a high pressure milkstream on a surface results in the homogenisation of thefat globule size during industrial processing This processprevents cream separation in milk usually available in themarket

In this paper we analyse the state of the art of samplepreparation for milk by comparing traditional industrialprocedure with potential for miniaturisation and emergingmicrofluidic methods with a particular emphasis on con-taminant detection and especially for aflatoxin 119872

1as case

studyMost of techniques discussed in the paper however arerelated to phase and chemical separation and can be appliedto other applications

The aflatoxin 1198721is of particular interest because it is a

milk contaminant and potent carcinogen classified in group1 of the International Agency for the Research on Cancer[6] It is characterised by carcinogenic genotoxic teratogenichepatotoxic immunosuppressive and antinutritional effects[7 8] Aflatoxin 119872

1(Figure 1) is a metabolite of aflatoxin 119861

1

Aflatoxin1198611is produced by the ubiquitous fungusAspergillus

flavus In favourable climatic condition the fungus producesaflatoxin 119861

1and contaminates animal feed Following the

ingestion of spoiled feed aflatoxin1198611is hydroxylated creating

in the liver aflatoxin 1198721 Aflatoxin 119872

1is secreted into milk

with an elapsed time of about 12 hours and a peak time of

O

OO

O

O

O

OH

CH3

Figure 1 Chemical structure of aflatoxin 1198721

about 24 hours It is thermostable and thus is not deactivatedby pasteurisation or UHT treatment [9] and it is found indairy products and concentrated in cheese [10]

In order to protect public health and safety in theEuropean Commission (EC) regulation number 18812006the maximum level of aflatoxin 119872

1contamination in milk

is set to 50 ppt and to 25 ppt for infant formulae as avulnerable group of the population Similar regulatory limitsare implemented in other countries in Africa Asia andLatin America where aflatoxins incidence is higher than inthe EU countries due to favourable warm and wet climaticconditions In other countries such as the United States andsome countries inAsia andLatinAmerica the suggested limitis only 500 ppt [11 12]

Aflatoxin is slightly soluble in water (10ndash30 120583gmL) butyet much higher than maximum allowed limit in milk itis insoluble in nonpolar solvents which results in a verysmall fractionmdashabout 5 to 20 depending on the source andconditionsmdashpartitioned in the cream part of milk [10] Inmilk it is mainly weakly absorbed onto milk proteins [13]mainly casein The small molecule size of aflatoxin (MW3283 gmol) is an additional challenge if detection is basedon label-free methods monitoring the change of surfaceproperties like measurement of refractive index (photonicsensors) or impedance (electrochemical capacitive sensors orimpedance spectroscopy) at the surface

With this background it is clear that phase and chemicalseparation are needed for an efficient detection of aflatoxinMany microfluidic methods are available for this purpose[14] but they are usually developed for small samples (oftenblood) Since contaminants like aflatoxin are found at verylow concentration it is often required to work with samplevolumes in the range 1ndash100mL to have a sufficient recoveryfor the analysis and to get a representative and uniform sam-ple from the batch Scaling up processing rate of microfluidicmethods to deal with 1ndash100mL samples in minutes is notpractical in most cases and therefore a careful selection of thetechnology must be performed by including performances ofseparation and processing time

The paper is organised as follows in Section 2 wedescribe the techniques used for phase separation and samplepreparation in the specific case of milk and aflatoxin Wecompare the techniques used in laboratories to prepare thesamples for analysis with state-of-the-art methods like ELISAand HPLC and lateral flow to industrial methods for phaseseparation and microfluidic techniques A discussion of














1

2 3 4 5


















Reverseosmosis

Nanofiltration

Ultrafiltration

Microfiltration

Particlefiltration

Gelatin

Bacteria

Redbloodcells

Virus

Oilemulsion

Sugars

Aminoacids

Aqueoussalts

Antibiotics

Mol wt (Da) 100


001 101 100001Size (120583m)

1k 100 k 1M



V119904=

1198892(120588119901minus 120588119897)

18120583 (1)


119861= (119896119879119898)



















119865119871=

41205881198621198711198802

1198911198864

31205871205831198632

ℎ

(2)



119891is flow










119871 =3120587120583119863

2

ℎ119871119898

21205881198621198711198801198911198863

(3)




120582 =119886

119863ℎ

gt 007 (4)



ℎlt 56 120583m


Re119901= Re 1205822 gt 1 (5)






12= 2 that is





119865119877

= minus(

1198750119881119901120573119898120587

2120582)120601 (120573 120588) sin(

4120587119909

120582) (6)



119901is particle









11199112

119894= 008M




120583119894=

10038161003816100381610038161199111198941003816100381610038161003816 119890

(6120587120578119903119894) (7)


120583119894=

120577120576

120578 (8)



4 Conclusions






Acknowledgment


References
























1in dairy products



























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation






















1

2 3 4 5


















Reverseosmosis

Nanofiltration

Ultrafiltration

Microfiltration

Particlefiltration

Gelatin

Bacteria

Redbloodcells

Virus

Oilemulsion

Sugars

Aminoacids

Aqueoussalts

Antibiotics

Mol wt (Da) 100


001 101 100001Size (120583m)

1k 100 k 1M



V119904=

1198892(120588119901minus 120588119897)

18120583 (1)


119861= (119896119879119898)



















119865119871=

41205881198621198711198802

1198911198864

31205871205831198632

ℎ

(2)



119891is flow










119871 =3120587120583119863

2

ℎ119871119898

21205881198621198711198801198911198863

(3)




120582 =119886

119863ℎ

gt 007 (4)



ℎlt 56 120583m


Re119901= Re 1205822 gt 1 (5)






12= 2 that is





119865119877

= minus(

1198750119881119901120573119898120587

2120582)120601 (120573 120588) sin(

4120587119909

120582) (6)



119901is particle









11199112

119894= 008M




120583119894=

10038161003816100381610038161199111198941003816100381610038161003816 119890

(6120587120578119903119894) (7)


120583119894=

120577120576

120578 (8)



4 Conclusions






Acknowledgment


References
























1in dairy products



























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










Reverseosmosis

Nanofiltration

Ultrafiltration

Microfiltration

Particlefiltration

Gelatin

Bacteria

Redbloodcells

Virus

Oilemulsion

Sugars

Aminoacids

Aqueoussalts

Antibiotics

Mol wt (Da) 100


001 101 100001Size (120583m)

1k 100 k 1M



V119904=

1198892(120588119901minus 120588119897)

18120583 (1)


119861= (119896119879119898)



















119865119871=

41205881198621198711198802

1198911198864

31205871205831198632

ℎ

(2)



119891is flow










119871 =3120587120583119863

2

ℎ119871119898

21205881198621198711198801198911198863

(3)




120582 =119886

119863ℎ

gt 007 (4)



ℎlt 56 120583m


Re119901= Re 1205822 gt 1 (5)






12= 2 that is





119865119877

= minus(

1198750119881119901120573119898120587

2120582)120601 (120573 120588) sin(

4120587119909

120582) (6)



119901is particle









11199112

119894= 008M




120583119894=

10038161003816100381610038161199111198941003816100381610038161003816 119890

(6120587120578119903119894) (7)


120583119894=

120577120576

120578 (8)



4 Conclusions






Acknowledgment


References
























1in dairy products



























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation






















119865119871=

41205881198621198711198802

1198911198864

31205871205831198632

ℎ

(2)



119891is flow










119871 =3120587120583119863

2

ℎ119871119898

21205881198621198711198801198911198863

(3)




120582 =119886

119863ℎ

gt 007 (4)



ℎlt 56 120583m


Re119901= Re 1205822 gt 1 (5)






12= 2 that is





119865119877

= minus(

1198750119881119901120573119898120587

2120582)120601 (120573 120588) sin(

4120587119909

120582) (6)



119901is particle









11199112

119894= 008M




120583119894=

10038161003816100381610038161199111198941003816100381610038161003816 119890

(6120587120578119903119894) (7)


120583119894=

120577120576

120578 (8)



4 Conclusions






Acknowledgment


References
























1in dairy products



























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation














12= 2 that is





119865119877

= minus(

1198750119881119901120573119898120587

2120582)120601 (120573 120588) sin(

4120587119909

120582) (6)



119901is particle









11199112

119894= 008M




120583119894=

10038161003816100381610038161199111198941003816100381610038161003816 119890

(6120587120578119903119894) (7)


120583119894=

120577120576

120578 (8)



4 Conclusions






Acknowledgment


References
























1in dairy products



























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation













Acknowledgment


References
























1in dairy products



























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation




















RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









review article microfluidic sample preparation methods for

Documents