research article a smart high-throughput experiment...

Research ArticleA Smart High-Throughput Experiment Platform forMaterials Corrosion Study

Peng Shi Bin Li Jindong Huo and Lei Wen

National Center for Materials Service Safety University of Science and Technology Beijing Beijing China

Correspondence should be addressed to Peng Shi shipengustbsinacom

Received 6 August 2016 Accepted 4 October 2016

Academic Editor Wenbing Zhao

Copyright copy 2016 Peng Shi et alThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Materials corrosion study is based on plenty of contrast experiments Traditional corrosion experiments are time-consuming andrequire manual corrosion grade evaluating during the experiment To improve the efficiency of experiment a high-throughputexperiment platform is designed and accomplished The platform mainly consists of high-throughput corrosion reaction facilitydata acquisition system and data processing systemThe corrosion reaction facility supports high-throughput materials corrosionreactions under various conditions The data acquisition system is mainly responsible for capturing the images of samplesrsquo surfacecollecting electrochemical signals and storing them into the computer in real time The data processing system treats the acquireddata and evaluates the degree of materials corrosion in real time by program automatically The platform not only reduces theoccupation of the equipment but also improves the efficiency of sample preparation and experiment occurrenceThe experimentaldata shows that the platform can accomplish high-throughput corrosion contrast experiment easily and reduce the time costobviously

1 Introduction

Corrosion is one of themost popularmaterialsrsquo failure causesIt will significantly reduce materialsrsquo strength plastic andtoughness properties shorten the service life and causecatastrophic accidents [1] To investigate the factors andmechanisms ofmaterials corrosion a lot of experimentsmustbe conducted A small change of the experimental parametersoften means a series of new similar experiments whichwill undoubtedly consume a large amount of resources Toimprove the efficiency of this kind of experiments high-throughput method is adopted

The idea of high-throughput originates from combinato-rial chemistry [2] aiming to shorten the product develop-ment period and reduce costs by large-scale chemical syn-thesis technology Its essence is to accomplish large amountof repetitive experiments in parallel and obtain experimentalresults through the novel design of experiment scheme andequipment High-throughput characterization technologyhas been widely used in gene sequencing drug screeningand so on In recent years new materials development

begins to adopt high-throughput technology to improvedevelopment efficiency [3] However there are few reportson high-throughput materials experiment method In thispaper we try to improve the efficiency of materials corrosionexperiment by high-throughput technology

A high-throughput corrosion experiment platform hasbeen designed and accomplished The platform mainlycomprises high-throughput corrosion reaction facility dataacquisition system and data processing system The cor-rosion reaction facility supports high-throughput materi-als corrosion reactions under various conditions The dataacquisition system is mainly responsible for capturing theimages of samplesrsquo surface collecting electrochemical signalsand storing them into the computer in real time The dataprocessing system treats the acquired data and evaluatesthe degree of materials corrosion in real time by programautomatically

The platform has accomplished some high-throughputmaterials corrosion experiments which not only reduces theoccupation of the equipment but also improves the efficiencyof sample preparation and experiment conduction Different

Hindawi Publishing CorporationScientific ProgrammingVolume 2016 Article ID 6876241 9 pageshttpdxdoiorg10115520166876241

2 Scientific Programming

from traditional artificial corrosion evaluation the platformcan treat the collected data and automatically generate corro-sion evaluation resultsThe contribution of ourmethod is thatit provides the possibility of large-scale materials corrosionexperiment and version-based data analysis

The rest of the paper is arranged as follows Section 2introduces the related works about high-throughput experi-ment and corrosion image processing Section 3 describes thestructure and implementation of the platform The key algo-rithm and programing method are presented in Section 4The experimental procedure and result analysis are shown inSection 5 Section 6 analyzes the performance of our methodAt last Section 7 summarizes our contributions and pointsout the future work

2 Related Works

21 High-Throughput Technology High-throughput is a tech-nology to execute parallel task with multicharacterizationmethods It has been applied in several fields includinghigh-throughput sequencing high-throughput screeningand combination technology High-throughput sequencing[4] brings great promotion on the development of biologicalgene sequencing It can measure millions of DNAmoleculesrsquosequence at the same time This makes it possible to analyzethe transcriptome and the genome of a species in detailHigh-throughput screening [5] is a method for scientificexperimentation especially used in drug discovery and rel-evant to the fields of biology and chemistry It allows aresearcher to quickly conduct millions of chemical geneticor pharmacological tests with robotics data processing andcontrol software liquid handling devices and sensitive detec-tors [6] Combination technology with multichannel parallelsynthesis and high-throughput rapid characterization cansynthetize samples of different compositions fleetly by finitesteps investigate its structure and properties efficiently anddevelop new materials with required properties finally [7]

22 Corrosion Image Processing and Evaluation Image pro-cessing is a technology to analyze images and obtain somedesired features by computer It generally includes imagecompression enhancement restoration matching descrip-tion and recognition Image description matching andrecognition are usually adopted to accomplish some smartapplications

During the study of materials corrosion the appearanceof surface is important to evaluate the corrosion degree Theidea of image processing can be used in the grade evaluationof materials corrosion In 1981 Itzhak et al scanned thesurface of AISI 304 material by digital scanner which hasbeen soaked for 20 minutes in 50∘C 10 FeCl

3solution [8]

The corrosive pitting is counted and the corrosion rate iscalculated by computer program according to the scannedimages Codaroa et al analyzed the appearance of pittingof Al-Ti alloy and gave a quantitative description methodfor pitting The method can be utilized to represent theevaluation procedure of corrosion [9] Wang et al collectedthe morphology of sea water corrosion of carbon steel

(1)

(1)

(2)(3)

(4)

(5)

Figure 1 The structure of high-throughput corrosion experimentplatform ((1) corrosion reaction facility (2) specimen (3) electro-chemical signal acquisition system (4) image acquisition systemand (5) computer)

[10] They adopted grey correlation method to establish therelationship between the corrosion degree and the apparentcolor and edge Xu et al investigated the relationship betweenthe apparent grey value of image and the depth of corrosivepitting in the specimen by fractal dimension method [11]They found out that the relationship was nearly linear In thispaper corrosion images are automatically processed in realtime to evaluate the corrosion grade of specimens

3 Platform Structure

With the idea of high-throughput a smart materialsrsquo cor-rosion experiment platform is designed It consists of high-throughput corrosion reaction facility data acquisition sys-tem and data processing system shown in Figure 1 All thesystems are connected to transfer signal and data

31 High-Throughput Corrosion Reaction Facility High-throughput corrosion reaction facility indicates the containerwhere the corrosion reaction occurs We design two kindsof reaction units for high-throughput corrosion reactionsingle-solution reaction unit and multiple-solution reactionunit

311 Single-Solution ReactionUnit Theappearance of single-solution reaction unit is shown in Figure 2 It uses corrosionresistant materials such as epoxy resin to package severalspecimens into one component Single-solution reaction unitcan only investigate the corrosion behaviors of differentspecimens in the same solution The specimens in onecomponent can be distinguished from materials roughnessand other properties Since the specimens are packagedthey can be treated together to improve the sample makingefficiency such as polishing and surface processing

312 Multiple-Solution Reaction Unit The appearance ofmultiple-solution reaction unit is shown in Figure 3 It is areaction vessel consisting of a set of parallel tactic grooveliquid pool arrays Each liquid pool can contain different

Scientific Programming 3

Figure 2 Single-solution reaction unit

Figure 3 Multiple-solution reaction unit

solutions during one experiment Every specimen can bearranged into one liquid pool After one specimen is putin the bottom of the pool should be sealed up to avoid theleaking of solution During the experiment procedure eachliquid pool can be plugged with an electrode Thus the elec-trochemical signals of every specimen can be collected inde-pendently The advantage of multiple-solution reaction unitis that different specimens can be tested in different solutionsat the same time One experiment may produce independentimage and electrochemical data of every specimen It canobviously improve the corrosion reaction efficiency

32 Data Acquisition System

321 Electrochemical Signal Acquisition System Electro-chemical signal is an important characterization parameterduring materials corrosion reaction It is usually detectedby electrochemical workstation representing the current andpotential of the specimen and solution The electrochemicalsignal acquisition system is composed of an electrochem-ical workstation a reference electrode and a multiplexerThe electrochemical workstation and reference electrode are

working together to collect the signal of current andpotentialIn our high-throughput materials corrosion experimentsthe two-electrode system was used to form a closed loopwith a working electrode and a reference electrode We useldquoInterface 5000rdquo electrochemical workstation from GamryCompany The working electrode is a bolt type materialsrsquoelectrode and the reference electrode is commonly AgAgClelectrode used in the study of electrochemical corrosionAgAgCl electrode can be utilized as a probe part of referenceelectrode in the solution

Since high-throughput experiment needs to get theelectrochemical signals of all the specimens a multiplexeris designed to accomplish the function of parallel signalcollection The multiplexer is developed based on the circuitboard It consists of several input ports and one output portHere we show a multiplexer with 16 input ports All thespecimens in the corrosion reaction unit can be connectedwith the input ports The output port is connected withan electrochemical workstation to analyze the signal fromthe input ports After circuit programing we can realizethe connecting between one input port and the output portat one moment Thus the multiplexers can take turns toquery 16 input ports in accordance with the way specified by


(a) (b)

Figure 4 Multiplexer and its connection mode (a) Multiplexer (b) Connecting with specimens

Table 1 Daheng camera parameters

Name ParameterModel MER-500-7UCInterface Mini USB 20Resolution 2592 (H) times 1944 (V)Frame rate 7 fpsSensor 12510158401015840 CMOSPixel size 22 120583m times 22 120583mSpectrum Black-and-whitecolor

the program The appearance of the multiplexer is shown inFigure 4(a) Figure 4(b) shows the connecting mode with theexperimental sample With the help of the multiplexer onesingle channel electrochemical workstation can accomplishthe function of multichannel electrochemical workstationAlthough there may have time delay on signal collection ithardly influence the result because of slow corrosion action

322 Image Acquisition System Since the surface appearanceof specimen is one important characterization for corrosionstudy specimen surface image must be collected continu-ouslyThe image acquisition system is composed of a cameraa ring light and a light shield

We choose ldquoMER-500-7UCrdquo camera from DahengGraphics CompanyThe specific parameters of the camera areshown in Table 1The camera adopts USB 20 port to connectwith computer and provides programing interface to controlits action Its size is only 29mm lowast 29mm lowast 29mm and it iseasy to be used in various environments

For enough light on the specimens we take LED ring lightas a light source It can supply an even light for the measuredobject To avoid the influence from the environmental lighta light shielding cover was used The stable light conditionmakes the direct comparison of images at different stagespossible Light shield selection was relatively simple as longas it can cover the whole system and block the environmentallight in

33 Data Processing System Data processing of our plat-form is automatically accomplished by programing whichis running on a computer The data produced from theplatform mainly includes the electrochemical signals andspecimen surface images The program can separate theelectrochemical data of each specimen from the electrochem-ical workstation file which contains the mixed data of allthe specimens The program can also process the surfaceimage of specimens The corrosion grade of each specimenis evaluated by image processing based program The detailsof the method and algorithm will be introduced in Section 4

4 Algorithm and Programing

41 Electrochemical Signal Extracting In our high-through-put experiment the electrochemical data of all specimens arecollected together The electrochemical workstation collectssignals sequentially and stores them into oneDAT format fileTo analyze the electrochemical signal of every specimen thesignal data of every specimen should be extracted from theDAT file Based on the time stamp in the file and multiplexereach specimenrsquos electrochemical data can be attained by aprogram

To avoid the error of data collecting we set the datacollecting with a higher frequency than the realistic demandIn the electrochemical data split program the data with bigerror is picked out An average data value of the remainingdata with small error is adopted as the electrochemical signalvalue of the specimen by the program After the programtreatment the data in DAT file is separated into severalelectrochemical signals Every group of independent signalsis corresponding to one specimen

42 Image-Based Smart Corrosion Evaluation The surfacemorphology and corrosion grade evaluation are importantfor the study of corrosion mechanism During traditionalcorrosion experiment the specimen corrosion grade is evalu-ated manually With our smart high-throughput experimentplatform the specimen corrosion grade can be evaluated bythe program automatically


In the past decades image processing technology hasbeen successfully applied in corrosion description and eval-uation Most of the related works focused on the analysis ofgrey image Here we give a quick corrosion degree evaluationmethod for color corrosion image For the RGB image eachpixel point has three color channels red green and blueEach color channel can take any value from 0 to 255 (with 8bitsrsquo format) so the whole color space is a total of 2563 (about16 million) kinds of color combination The calculation ofcomparing the distribution of these 16 million combinationsis too large so we design a quickmethod to evaluate the colorimage by characteristic valueThe RGB value from 0 to 255 isdivided into four ranges 0 to 63 is the first range 64 to 127 isthe second range 128 to 191 is the third range and 193 to 255is the fourth range Each color channel can be divided intofour districts so 43 (64) kinds of color combinations can bereformedThe 64 kinds of combinations can be regarded as aspace of 64 dimensions

Here we define the concept of characteristic vector 119862 ofan image 119862 is a 64-dimension vectorThe element of 119862 is thetotal number of pixels with the color of certain dimensiondenoted as 119888

119894 Thus a color image can be represented by a 64-

dimensional vector For example the characteristic vector ofthe left top brass specimen in Figure 8(a) can be denoted as119862 = (7414 230 0 0 8 109 0 0 3415 53929) (1)

The details of RGB district and characteristic vector areshown in Table 2 The elements in column ldquo119888rdquo compose avector that is the characteristic vector It can be regardedas a ldquofingerprintrdquo of an image The similarity between twoimages can be determined by comparing their characteristicvectors [12] In this way we can improve the efficiency ofimage similarity determination while ensuring the relativeuniqueness

To quickly evaluate the corrosion grade of a specimen weestablish a standard corrosion image database The databaseconsists of a series of standard corrosion images with manualcorrosion grade evaluation The characteristic vectors of theimages are also stored in the databaseThemain idea of quickcorrosion grade evaluation is to specify one standard imagersquosgrade to an image whose characteristic vector is the nearestto the standard image The procedure of corrosion gradeevaluation based on characteristic vector similarity mainlyconsists of the following five steps

(1) Load the corrosion images to be evaluated(2) Calculate the characteristic vector of the image to be

evaluated(3) Compare the calculated characteristic vector with

that of the standard image one by one(4) Find out the standard image with the nearest charac-

teristic vector(5) Output the corrosion grade of the standard image as

that of the input imageBased on the procedure above we have developed a

program to calculate corrosion grade of collected imagesConsequently the high-throughput experiment platform canoutput the corrosion grade of every specimen in real time

Table 2 RGB and characteristic vector of image

R G B 119888

0 0 0 74140 0 1 2300 0 2 00 0 3 00 1 0 80 1 1 3720 1 2 880 1 3 00 2 0 00 2 1 00 2 2 100 2 3 10 3 0 01 0 0 8911 0 1 131 0 2 01 0 3 01 1 0 5921 1 1 34621 1 2 3551 1 3 01 2 0 01 2 1 1011 2 2 8821 2 3 531101 3 0 110532 0 0 11462 0 1 02 0 2 02 0 3 02 1 0 25522 1 1 90402 1 2 472 1 3 02 2 0 02 2 1 88082 2 2 82 2 3 02 3 0 163 0 0 113 0 1 03 0 2 03 0 3 03 1 0 8563 1 1 13763 1 2 03 1 3 03 2 0 03 2 1 36503 2 2 62603 2 3 1093 3 0 0

5 Experiments and Data Analysis

With the smart high-throughput experiment platform wehave conducted several experiments ofmaterials corrosion inNaCl solution


(a) (b)

Figure 5 The solution container on single-solution corrosion unit (a) Sticking method (b) Solution container with an electron

Table 3 The polishing sandpaper types and surface roughness

Sandpaper type 200 400 800 1500Surface roughness120583m 750 350 218 126

51 Experiment Preparation Four kinds of materials includ-ing aluminum brass copper and steel are selected as theobjects to be investigated Every kind of material is made intofour specimens as parallel reference separatelyThey aremadeinto bar sample with 5mm diameter and 20mm lengthThe intersecting surface is polished as the corrosion surfaceby 200 400 800 and 1500 sandpaper respectivelyThus the effect of surface roughness on materials corrosioncan be studied The polishing sandpaper types and surfaceroughness are shown in Table 3

The corrosion experiment is conducted in 35 NaClsolution Here we use the single-solution corrosion reactionunit as reaction container After the specimens are polisheda pipe made of PTFE (Polytetrafluoroethylene) is stuck tothe package surface by 704 silicone rubber Then the NaClsolution can be contained in the pipe space The appearanceof stuck single-solution unit is shown in Figure 5(a) Areference electrode is plugged into the container to collect theelectrochemical signals shown in Figure 5(b)

52 Experimental Data

521 Electrochemical Signal The experiment monitors theopen circuit potential signal of the specimens by two-electrode system AgAgCl electrode is adopted as the ref-erence electrode of the corrosion system The referenceelectrode is connected with the electrochemical workstationThe specimens can be regarded as working electrode All theworking electrodes are connected with the input ports ofthe multiplexer one by one with extension wire The outputport of the multiplexer is directly connected with the Gamryelectrochemical workstation Thus the specimens electro-chemical workstation multiplexer and reference electrodecompose a connected loop

A B500040003000

minus1000 V

minus5000 mV

0000 V

T (s)

Vf

(V v

ersu

s Ref

)

Figure 6 Open circuit potential of a full sampling period

In order to obtain stable data output we maintain 5-second period of switching on with each specimen bymultiplexer control program At the same time the Gamrypotential signal scanning period is set to 05 seconds Con-sequently the signal scanning may complete about 10 timesin each specimenrsquos switching on interval After removing thedata with big offset the average of the remaining data iscalculatedThe average value is specified as the potential valueof the specimen in the switching on period The platformcan accomplish a complete scan circle for all the specimensin 80 seconds The data of one complete circle is shown inFigure 6 Every point in the diagram is original data from theelectrochemical workstation

522 Surface Image During the experiment the imageacquisition system records the specimensrsquo surface imageautomatically Since the corrosion action is not fast thesystem captures the surface image once per hour by theprogramThe corrosionmorphology images of the specimenswithin 0sim7 hours are shown in Figure 7 From the imagesresearchers can investigate the corrosion morphology Thecorrosionmorphology comparison on differentmaterials anddifferent corrosion time can be executed easily


(a) (b) (c) (d)

(e) (f) (g) (h)

Figure 7 High-throughput corrosion images of one-hour interval (a) 119905 = 0 h (b) 119905 = 1 h (c) 119905 = 2 h (d) 119905 = 3 h (e) 119905 = 4 h (f) 119905 = 5 h (g)119905 = 6 h and (h) 119905 = 7 h

(a) (b)

Figure 8 High-throughput image segmentation (a) Image before segmentation (b) Single specimen image

53 Data Processing and Analysis The data produced byhigh-throughput experiment is automatically processed inreal time Electrochemical signal processing is to separate theoriginal data into 16 independent signals of every specimenImage processing is mainly to evaluate the corrosion grade ofevery specimen

531 Electrochemical Signal Processing The electrochemicaldata of all the specimens collected by the electrochemicalworkstation are stored in a DAT format file According tothe program setting the workstation collects data twice inone second shown in Figure 6 The potential data of all the

specimens is in one file The data of each specimen is sep-arated into several electrochemical signals by the programTheprogramconsiders two factorsOne is the collecting timeby which the certain specimen can be located The other isthe data value Because of the error of detecting there maybe some data with big offset The program removes the offsetdata and calculates the average of the remaining normal dataThe average value is specified as the potential value of thespecimen in the switching on period Thus every specimencontains a series of independent potential data

The open circuit potential data after processing is shownin Figure 9 It presents the electrochemical signals of high-throughput experiment of carbon steel brass copper and


Carbon steel independent Carbon steel Brass

Copper Stainless steel

minus09

minus08

minus07

minus06

minus05

minus04

minus03

minus02

minus01

Ope

n ci

rcui

t pot

entia

l (m

V)

minus2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 340Time (min)

Figure 9 Electrochemical signals after processing

stainless steel To verify the correctness of high-throughputexperiment we conduct a carbon steel corrosion experimentwith traditional independent experiment The data of theindependent experiment is also shown by black squares inFigure 9 From the comparison result the independent exper-imental potential data is consistent with the data obtainedby high-throughput experiment of the same materials inthe same condition Therefore the phenomenon shows thecorrectness of high-throughput experiment in one respect

532 Image Processing and Corrosion Grade Since high-throughput experiment produces images withmultiple speci-mens the image of every single specimen should be extractedBased on image edge detection method the image of singlespecimen can be easily attained The high-throughput imagesegmentation is shown in Figure 8 After the treatment by theprogram the image of every specimen is extracted shown inFigure 8(b)

Based on the standard GBT 6461-2002 China standardof corrosion evaluation the corrosion grade can be evaluatedThe standard proposes two basic parameters corrosion arearatio and corrosion grade to evaluate the corrosion gradeThe corresponding relationship between corrosion grade andcorrosion area ratio is shown in Table 4 Corrosion area ratiois the percentage ratio between corrosion surface area andtotal surface area The corrosion grade is from 1 to 10 Thehigher the grade is the more serious the corrosion status is

To verify the correctness of corrosion evaluation thecorrosion images are also submitted to an expert of materialscorrosion The evaluation result of 7 h experiment by ourprogram and the expert is shown in Table 5 From thetable it can be seen that there are some differences of theevaluated grade It is because the expert uses hisher eyes toevaluate the corrosion grade by hisher experiences After theconfirmation of the expert the program produced result isbetter than manual result

Table 4 The corresponding relationship of grade value and corro-sion area ratio

Corrosion area ratio Corrosion grade valueNo corrosion 100 lt 119860 le 01 901 lt 119860 le 025 8025 lt 119860 le 05 705 lt 119860 le 10 610 lt 119860 le 25 525 lt 119860 le 50 450 lt 119860 le 25 325 lt 119860 le 50 250 lt 119860 1

Table 5 Corrosion evaluation result of 7 h experiment

Specimen position inthe sample

Grade value byprogram

Grade value byexpert

11 7 712 7 713 7 714 6 721 6 622 6 623 6 624 5 631 8 832 7 833 8 834 8 841 2 242 3 343 2 244 4 4

Table 6 Time cost comparison of one-by-one parallel and high-throughput experiment mode

Steps One-by-oneminute Parallelminute High-

throughputminutePolish 960 960 240Encapsulation 0 0 30Experiment 6720 420 420Handle 140 140 140Total 7820 1520 830

6 Performance Analysis and Discussion

Compared with traditional corrosion experiments the high-throughput experiment platform shows high efficiency espe-cially in time cost Table 6 shows the time and equipment costcomparison of a typical high-throughput corrosion experi-ment with parallel experiment and one-by-one experiment


Here the high-throughput experiment with 16 specimens istaken as an example From the data in Table 6 it is clearthat the high-throughput experiment costs much less timethan traditional experiment wayThe parallel experiment andhigh-throughput experiment cost less time than one-by-oneexperiment However more testing equipment is occupiedby parallel experiment than one-by-one experiment andhigh-throughput experiment Although the high-throughputmethod takes some time for sample packaging it reduces thesample polishing time The parallel experiment costs moretime for polishing because the high-throughput sample canbe polished together From the total time cost data in Table 6it is obvious that high-throughput experiment has the highestefficiency From the analysis above it can be concluded thatthe high-throughput experiment method designed in thispaper has a great advantage in the equipment occupation andthe overall time consumption

7 Conclusion and Future Work

In this paper we propose a smart high-throughput experi-ment platform for the research of materials corrosion Intel-ligent analysis of the corrosion of materials was successfullyaccomplished by the experimental platform The corrosiongrade can be calculated by the program automatically Thecalculated results are matched closely with the manual ana-lyzed results Unfortunately for the experiment with littlesurface changing the image-based corrosion evaluation maynot work well

In the future we will try to collect other types of dataduring materials corrosion such as microstructure solutionchanges and other characterization data By analyzing thenew types of data it is expected to accomplish more com-prehensive understanding of the mechanisms of materialscorrosion At the same time through adding the new type ofdata to the standard library the corrosion evaluation resultscan be more accurate

Competing Interests

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

Acknowledgments

This work is supported by ldquoThe Fundamental ResearchFunds for the Central Universities (FRF-TP-14-060A2)rdquo ofChinamdashldquoThe Research on Parallel Data Processing Methodfor Materials High-Throughput Experimentrdquo

References

[1] N Canter ldquoMaterials corrosion preventives protect materialsand specific applicationsrdquo Tribology amp Lubrication Technologyvol 718 pp 5ndash6 2015

[2] P Dydio P-A R Breuil and J N H Reek ldquoDynamic com-binatorial chemistry in chemical catalysisrdquo Israel Journal ofChemistry vol 53 no 1-2 pp 61ndash74 2013

[3] P de Lima-Neto A N Correia R A C Santana et alldquoMorphological structuralmicrohardness and electrochemicalcharacterisations of electrodeposited Cr and Ni-W coatingsrdquoElectrochimica Acta vol 55 no 6 pp 2078ndash2086 2010

[4] L Chen L-YWang S-J Liu et al ldquoProfiling of microbial com-munity during in situ remediation of volatile sulfide compoundsin river sediment with nitrate by high-throughput sequencingrdquoInternational Biodeterioration amp Biodegradation vol 85 pp429ndash437 2013

[5] L M Junker and J Clardy ldquoHigh-throughput screens forsmall-molecule inhibitors of Pseudomonas aeruginosa biofilmdevelopmentrdquo Antimicrobial Agents and Chemotherapy vol 51no 10 pp 3582ndash3590 2007

[6] G R Eldridge H C Vervoort C M Lee et al ldquoHigh-throughputmethod for the production and analysis of large nat-ural product libraries for drug discoveryrdquo Analytical Chemistryvol 74 no 16 pp 3963ndash3971 2002

[7] PAWhite A EHughes S A Furman et al ldquoHigh-throughputchannel arrays for inhibitor testing proof of concept forAA2024-T3rdquo Corrosion Science vol 51 no 10 pp 2279ndash22902009

[8] D Itzhak I Dinstein and T Zilberberg ldquoPitting corrosionevaluation by computer image processingrdquo Corrosion Sciencevol 21 no 1 pp 17ndash22 1981

[9] E N Codaroa R Z Nakazatoa A L Horovistizb L MF Ribeirob R B Ribeirob and L R O Heinb ldquoAn imageprocessingmethod formorphology characterization andpittingcorrosion evaluationrdquoMaterials Science and Engineering A vol334 no 1-2 pp 298ndash306 2002

[10] SWang D Kong and S Song ldquoDiagnosing corrosionmodalitysystem of metallic material in seawater based on fuzzy patternrecognitionrdquoActaMetallrugica Sinica vol 37 no 5 pp 517ndash5212001

[11] S Xu Y Weng and X Li ldquoCharacterization for corrosion pitdistribution by using fractal dimension of imagerdquo Journal of theChinese Society of Corrosion and Protection vol 27 no 2 pp109ndash113 2007

[12] K Belaid O Echi and R Gargouri ldquoTwo classes of locallycompact sober spacesrdquo International Journal of Mathematicsand Mathematical Sciences vol 2005 no 15 pp 2421ndash24272005

Submit your manuscripts athttpwwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



from traditional artificial corrosion evaluation the platformcan treat the collected data and automatically generate corro-sion evaluation resultsThe contribution of ourmethod is thatit provides the possibility of large-scale materials corrosionexperiment and version-based data analysis

The rest of the paper is arranged as follows Section 2introduces the related works about high-throughput experi-ment and corrosion image processing Section 3 describes thestructure and implementation of the platform The key algo-rithm and programing method are presented in Section 4The experimental procedure and result analysis are shown inSection 5 Section 6 analyzes the performance of our methodAt last Section 7 summarizes our contributions and pointsout the future work

2 Related Works

21 High-Throughput Technology High-throughput is a tech-nology to execute parallel task with multicharacterizationmethods It has been applied in several fields includinghigh-throughput sequencing high-throughput screeningand combination technology High-throughput sequencing[4] brings great promotion on the development of biologicalgene sequencing It can measure millions of DNAmoleculesrsquosequence at the same time This makes it possible to analyzethe transcriptome and the genome of a species in detailHigh-throughput screening [5] is a method for scientificexperimentation especially used in drug discovery and rel-evant to the fields of biology and chemistry It allows aresearcher to quickly conduct millions of chemical geneticor pharmacological tests with robotics data processing andcontrol software liquid handling devices and sensitive detec-tors [6] Combination technology with multichannel parallelsynthesis and high-throughput rapid characterization cansynthetize samples of different compositions fleetly by finitesteps investigate its structure and properties efficiently anddevelop new materials with required properties finally [7]

22 Corrosion Image Processing and Evaluation Image pro-cessing is a technology to analyze images and obtain somedesired features by computer It generally includes imagecompression enhancement restoration matching descrip-tion and recognition Image description matching andrecognition are usually adopted to accomplish some smartapplications

During the study of materials corrosion the appearanceof surface is important to evaluate the corrosion degree Theidea of image processing can be used in the grade evaluationof materials corrosion In 1981 Itzhak et al scanned thesurface of AISI 304 material by digital scanner which hasbeen soaked for 20 minutes in 50∘C 10 FeCl

3solution [8]

The corrosive pitting is counted and the corrosion rate iscalculated by computer program according to the scannedimages Codaroa et al analyzed the appearance of pittingof Al-Ti alloy and gave a quantitative description methodfor pitting The method can be utilized to represent theevaluation procedure of corrosion [9] Wang et al collectedthe morphology of sea water corrosion of carbon steel

(1)

(1)

(2)(3)

(4)

(5)

Figure 1 The structure of high-throughput corrosion experimentplatform ((1) corrosion reaction facility (2) specimen (3) electro-chemical signal acquisition system (4) image acquisition systemand (5) computer)

[10] They adopted grey correlation method to establish therelationship between the corrosion degree and the apparentcolor and edge Xu et al investigated the relationship betweenthe apparent grey value of image and the depth of corrosivepitting in the specimen by fractal dimension method [11]They found out that the relationship was nearly linear In thispaper corrosion images are automatically processed in realtime to evaluate the corrosion grade of specimens

3 Platform Structure

With the idea of high-throughput a smart materialsrsquo cor-rosion experiment platform is designed It consists of high-throughput corrosion reaction facility data acquisition sys-tem and data processing system shown in Figure 1 All thesystems are connected to transfer signal and data

31 High-Throughput Corrosion Reaction Facility High-throughput corrosion reaction facility indicates the containerwhere the corrosion reaction occurs We design two kindsof reaction units for high-throughput corrosion reactionsingle-solution reaction unit and multiple-solution reactionunit

311 Single-Solution ReactionUnit Theappearance of single-solution reaction unit is shown in Figure 2 It uses corrosionresistant materials such as epoxy resin to package severalspecimens into one component Single-solution reaction unitcan only investigate the corrosion behaviors of differentspecimens in the same solution The specimens in onecomponent can be distinguished from materials roughnessand other properties Since the specimens are packagedthey can be treated together to improve the sample makingefficiency such as polishing and surface processing

312 Multiple-Solution Reaction Unit The appearance ofmultiple-solution reaction unit is shown in Figure 3 It is areaction vessel consisting of a set of parallel tactic grooveliquid pool arrays Each liquid pool can contain different










(a) (b)



























R G B 119888

0 0 0 74140 0 1 2300 0 2 00 0 3 00 1 0 80 1 1 3720 1 2 880 1 3 00 2 0 00 2 1 00 2 2 100 2 3 10 3 0 01 0 0 8911 0 1 131 0 2 01 0 3 01 1 0 5921 1 1 34621 1 2 3551 1 3 01 2 0 01 2 1 1011 2 2 8821 2 3 531101 3 0 110532 0 0 11462 0 1 02 0 2 02 0 3 02 1 0 25522 1 1 90402 1 2 472 1 3 02 2 0 02 2 1 88082 2 2 82 2 3 02 3 0 163 0 0 113 0 1 03 0 2 03 0 3 03 1 0 8563 1 1 13763 1 2 03 1 3 03 2 0 03 2 1 36503 2 2 62603 2 3 1093 3 0 0




(a) (b)








A B500040003000

minus1000 V

minus5000 mV

0000 V

T (s)

Vf

(V v

ersu

s Ref

)





(a) (b) (c) (d)

(e) (f) (g) (h)


(a) (b)









minus09

minus08

minus07

minus06

minus05

minus04

minus03

minus02

minus01

Ope

n ci

rcui

t pot

entia

l (m

V)

minus2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 340Time (min)












11 7 712 7 713 7 714 6 721 6 622 6 623 6 624 5 631 8 832 7 833 8 834 8 841 2 242 3 343 2 244 4 4











Competing Interests


Acknowledgments


References




















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in












(a) (b)



























R G B 119888

0 0 0 74140 0 1 2300 0 2 00 0 3 00 1 0 80 1 1 3720 1 2 880 1 3 00 2 0 00 2 1 00 2 2 100 2 3 10 3 0 01 0 0 8911 0 1 131 0 2 01 0 3 01 1 0 5921 1 1 34621 1 2 3551 1 3 01 2 0 01 2 1 1011 2 2 8821 2 3 531101 3 0 110532 0 0 11462 0 1 02 0 2 02 0 3 02 1 0 25522 1 1 90402 1 2 472 1 3 02 2 0 02 2 1 88082 2 2 82 2 3 02 3 0 163 0 0 113 0 1 03 0 2 03 0 3 03 1 0 8563 1 1 13763 1 2 03 1 3 03 2 0 03 2 1 36503 2 2 62603 2 3 1093 3 0 0




(a) (b)








A B500040003000

minus1000 V

minus5000 mV

0000 V

T (s)

Vf

(V v

ersu

s Ref

)





(a) (b) (c) (d)

(e) (f) (g) (h)


(a) (b)









minus09

minus08

minus07

minus06

minus05

minus04

minus03

minus02

minus01

Ope

n ci

rcui

t pot

entia

l (m

V)

minus2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 340Time (min)












11 7 712 7 713 7 714 6 721 6 622 6 623 6 624 5 631 8 832 7 833 8 834 8 841 2 242 3 343 2 244 4 4











Competing Interests


Acknowledgments


References




















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in

















R G B 119888

0 0 0 74140 0 1 2300 0 2 00 0 3 00 1 0 80 1 1 3720 1 2 880 1 3 00 2 0 00 2 1 00 2 2 100 2 3 10 3 0 01 0 0 8911 0 1 131 0 2 01 0 3 01 1 0 5921 1 1 34621 1 2 3551 1 3 01 2 0 01 2 1 1011 2 2 8821 2 3 531101 3 0 110532 0 0 11462 0 1 02 0 2 02 0 3 02 1 0 25522 1 1 90402 1 2 472 1 3 02 2 0 02 2 1 88082 2 2 82 2 3 02 3 0 163 0 0 113 0 1 03 0 2 03 0 3 03 1 0 8563 1 1 13763 1 2 03 1 3 03 2 0 03 2 1 36503 2 2 62603 2 3 1093 3 0 0




(a) (b)








A B500040003000

minus1000 V

minus5000 mV

0000 V

T (s)

Vf

(V v

ersu

s Ref

)





(a) (b) (c) (d)

(e) (f) (g) (h)


(a) (b)









minus09

minus08

minus07

minus06

minus05

minus04

minus03

minus02

minus01

Ope

n ci

rcui

t pot

entia

l (m

V)

minus2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 340Time (min)












11 7 712 7 713 7 714 6 721 6 622 6 623 6 624 5 631 8 832 7 833 8 834 8 841 2 242 3 343 2 244 4 4











Competing Interests


Acknowledgments


References




















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




(a) (b)








A B500040003000

minus1000 V

minus5000 mV

0000 V

T (s)

Vf

(V v

ersu

s Ref

)





(a) (b) (c) (d)

(e) (f) (g) (h)


(a) (b)









minus09

minus08

minus07

minus06

minus05

minus04

minus03

minus02

minus01

Ope

n ci

rcui

t pot

entia

l (m

V)

minus2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 340Time (min)












11 7 712 7 713 7 714 6 721 6 622 6 623 6 624 5 631 8 832 7 833 8 834 8 841 2 242 3 343 2 244 4 4











Competing Interests


Acknowledgments


References




















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




(a) (b) (c) (d)

(e) (f) (g) (h)


(a) (b)









minus09

minus08

minus07

minus06

minus05

minus04

minus03

minus02

minus01

Ope

n ci

rcui

t pot

entia

l (m

V)

minus2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 340Time (min)












11 7 712 7 713 7 714 6 721 6 622 6 623 6 624 5 631 8 832 7 833 8 834 8 841 2 242 3 343 2 244 4 4











Competing Interests


Acknowledgments


References




















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






minus09

minus08

minus07

minus06

minus05

minus04

minus03

minus02

minus01

Ope

n ci

rcui

t pot

entia

l (m

V)

minus2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 340Time (min)












11 7 712 7 713 7 714 6 721 6 622 6 623 6 624 5 631 8 832 7 833 8 834 8 841 2 242 3 343 2 244 4 4











Competing Interests


Acknowledgments


References




















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in








Competing Interests


Acknowledgments


References




















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



research article a smart high-throughput experiment...

Documents