Multichannel Detection and Differentiation ofExplosives with a Quantum Dot ArrayWilliam J PevelerdaggerDagger Alberto RoldanDaggerpara Nathan HollingsworthDagger Michael J PorterDagger

and Ivan P ParkinDagger

daggerDepartment of Security and Crime Science University College London 35 Tavistock Sq London WC1H 9EZ United KingdomDaggerDepartment of Chemistry University College London 20 Gordon St London WC1H 0AJ United KingdomparaSchool of Chemistry Cardiff University Main Building Park Place Cardiff CF10 3AT United Kingdom

S Supporting Information

ABSTRACT The sensing and differentiation of explosivemolecules is key for both security and environmentalmonitoring Single fluorophores are a widely used tool forexplosives detection but a fluorescent array is a morepowerful tool for detecting and differentiating suchmolecules By combining array elements into a singlemultichannel platform faster results can be obtained fromsmaller amounts of sample Here five explosives aredetected and differentiated using quantum dots asluminescent probes in a multichannel platform 24-dinitrotoluene (DNT) 246-trinitrotoluene (TNT) tetryl(246-trinitrophenylmethylnitramine) cyclotrimethylene-trinitramine (RDX) and pentaerythritol tetranitrate (PETN) The sharp variable emissions of the quantum dots froma single excitation wavelength make them ideal for such a system Each color quantum dot is functionalized with a differentsurface receptor via a facile ligation process These receptors undergo nonspecific interactions with the explosives inducingvariable fluorescence quenching of the quantum dots Pattern analysis of the fluorescence quenching data allows forexplosive detection and identification with limits-of-detection in the ppb range

KEYWORDS quantum dot TNT sensor luminescence explosive multichannel

Detection of small amounts of explosive material is akey challenge in both the securing of public spacesand vulnerable targets and the environmental

monitoring of drinking and waste waters The application ofcross-reactive arrays to chemical sensing is a major advance forthe detection of a range of these analytes without the need forhighly specific antibody or similar tests1minus3 Array-based sensingcan differentiate between multiple components within complexmixtures as well as allowing detection of previously unknownor unexpected analytes without the need for a new sensorPrevious examples have focused on using electrochemicalsensors and colorimetric dyes and advances in computingpower allow for application of machine learning techniques torapidly classify sensing results with a variety of multivariatestatistical techniques4minus8 The use of nanoparticle-based sensorsby Rotello et al for biosensing has introduced another powerfultool to array-based chemical sensing and the new direction ofthis work focuses on multichannel sensing combining elementsof the cross-reactive array in a single test with multiple outputsfor example multiple color channels to reduce the samplevolume required and increase sample throughput9minus11

Explosive detection is a problem that lends itself well to suchan array-based technique Monitoring of types and levels ofexplosive in the environment is an active challenge in thesecurity and environmental safety domains1213 with the USEnvironmental Protection Agency (EPA) setting limits on 24-dinitrotoluene (DNT) 246-trinitrotoluene (TNT) and cyclo-trimethylenetrinitramine (RDX) in drinking water (lt01 mgL) Fluorescent nanoparticle-based sensing systems lendthemselves to multichannel sensing due to their wide varietyof colors and sharp emission peaks14 It has been shown thatquantum dots (QDs) bind a range of explosives in differentways bringing the explosive into close proximity with thefluorescent nanoparticles causing electron-transfer-mediatedfluorescence quenching15minus20 The differential binding betweendifferent QD surfaces gives rise to variable fluorescencequenching allowing a sensing array to be created Althoughseveral simple QD systems have been described for nitro-aromatic explosives such as TNT and picric acid particularly

Received October 13 2015Accepted November 18 2015Published November 18 2015

Artic

lewwwacsnanoorg

copy 2015 American Chemical Society 1139 DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

This is an open access article published under a Creative Commons Attribution (CC-BY)License which permits unrestricted use distribution and reproduction in any mediumprovided the author and source are cited

focusing on the formation of Meisenheimer complexes betweenthe electron-poor aromatics and electron-rich amines21 fewhave tried to discriminate between multiple types ofexplosive22minus24

Herein we report a multichannel nanoparticle array for thedetection of explosives in a rapid single fluorometric test It isbased on a system of cross-reactive surface functionalitiescomprised of two macrocyclic (calixarene and cyclodextrin) andtwo simple (minusOH and minusOMe) surfaces on multicoloredfluorescent QDs The system is designed to respond to a rangeof explosives through supramolecular interactions such ashostminusguest binding electrostatics and πminusπ stacking causingfluorescence quenching of the QDs to create an analyticalfingerprint (Figure 1) It is tested against five explosives DNT

TNT tetryl RDX and pentaerythritol tetranitrate (PETN)using single channel analysis demonstrating 100 specificityThe system is created in methanol for compatibility with widelyused methanolic solid-phase extraction techniques Multi-channel explosive sensing is then demonstrated using three ofthe QD sensors with high sensitivity and specificity

RESULTS AND DISCUSSIONSystem Design and Synthesis Hostminusguest interactions

mediated by macrocyclic hosts are a powerful tool for cross-reactive binding studies25 Here two macrocycles were selectedto form cross-reactive QD surfaces calix[4]arene (CX) and β-cyclodextrin (CD) Calixarenes have small but open and flexiblearomatic cavities (sim3 Aring)26 allowing for πminusπ interactions witharomatic guests27 Cyclodextrins have a larger more rigid cavity(sim57 Aring)28 and are entirely aliphatic cyclic glucopyranosideoligomers giving very different receptor characteristics25 Tosupplement these OH and OMe surface functionalities werecreated to give differential reactivity (Figure 2)Red green and blue QDs were synthesized by decom-

position of CdO and trioctylphosphine-Se at high temperatureby varying reaction times then shelled with ZnS via

decomposition of Zn-dithiocarbamate onto the surface (FigureS7)29minus31

The surface bound receptors in this study were designed witha lipoic acid (LA) headgroup a short triethylene glycol (TEG)spacer for solubility and the surface functionality itself (CXCD OH or OMe) A short ethylene glycol was used to ensureclose proximity of the bound analyte to the nanoparticlesurface facilitating better electron transfer An azido-clickapproach was taken to the ligand synthesis Azide terminatedLA-TEG was produced and the macrocycles were mono-functionalized with alkynes using literature procedures3233 Aclick-coupling was performed using copper immobilized oncarbon in the presence of base34 This strategy allows for futurevariation in the library of surface functionalities potentiallyenabling attachment of any alkyne functionalized moietyThe QD ligand exchange was achieved with 365 nm UV-light

on the LA moiety in a biphasic hexaneMeOH system3536

This QD functionalization does not require protected thiols orborohydride salts and gives good control over ratios of differentreceptors on the surface of the nanoparticle This proved usefulfor tailoring solubility of the particles CX capped QDsproduced with this method were not fully soluble in MeOHso a 32 mix of CX to OMe was used to give full solubility Allother ligand surfaces were fully composed of the named ligandIt has been shown that this procedure can introduce up to 200ligands per QD however in this case it is likely to be fewer dueto the larger size of the terminal functionality and shorter chainlength37

Initially each surface ligand (CD CX OH and OMe) wasassembled onto green QDs (em 544 nm) to create CD544CX544 OH544 and OMe544 In addition red QDs (em 608 nm)CX608 OMe608 and blue QD (em 516 nm) OH516 wereproduced and tested The QDs produced were air stable andstored in the refrigerator without significant loss of fluorescencefor 3 months

Sensing Methanolic solutions of each QD were titratedwith one of five explosives of interest DNT TNT tetryl RDXand PETN in a single channel fashion - one type of QD pertest These explosives were selected on the basis of currentthreats and to test a range of different analyte functionalities(structures given in Figure 2(ii)) On mixing the ligand shouldbind the explosive and an electron-transfer mechanismbetween the QD and the electron-deficient explosive causesQD fluorescence quenching The final concentration ofexplosive in the solution was varied between 15 and 85 μMwith the aim of obtaining a linear quenching regime(approximately 1minus30 ppm dependent on explosive mass)Figure 3a illustrates the quenching results of each singlechannel QD for each analyte The rate of quenching wasmeasured and found to be lt10 s in all cases An exemplarkinetic curve is given in Figure S12Of the green channel QDs CX544 showed good quenching

for TNT and tetryl and to a lesser extent DNT For exampleTNT achieved maximal Quenching [Q = 100 times (1 minus II0)]of 73Q and had a limit of detection (LOD) of lt01 μgmLTetryl had a maximal 47Q and a LOD = 12 μgmL (TableS2) The response of the CX surface is justified through π-system interactions between the calixarene and the electron-deficient arene systems of TNT tetryl and DNT causingstrong binding between analyte and QD and the relativeefficiencies of the ET process between the QD and the analyteAlthough PETN was hard to detect often causing the leastquenching of all the analytes CD544 achieved a greater response

Figure 1 Scheme of sensing mechanism Differential analytebinding across the multicoloured QD array gives a fluorescentfingerprint for the analyte that can be analyzed computationally

ACS Nano Article

DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

1140

to PETN than for DNT The PETN response from CD544 issmall (16Q at maximum) implying poor electron transferbetween analyte and QD but the increased response ascompared to DNT or RDX gives a good discriminant for thiscompound It is possible that the PETN could bind one arminside the hydrophobic CD cavity due to its greater flexibilitycompared to the rigid aromatic explosives3839

QDs OH544 and OH516 gave a similar response to the fiveanalytes likewise CX544 and CX608 implying that for the mostpart the different color QDs are interchangeable with thedifferent functional surfaces However OMe608 gave far higherquenching responses than its green counterpart in particular toTNT and tetryl achieving maxima of 85Q and 77Qrespectively This behavior makes OMe608 a useful stand-alonesensor for nitro-aromatics with a limit of detection of lt02 μgmL for DNT TNT and tetryl giving rise to the potential fornaked-eye detection (Figure 4) Test strips impregnated withOMe608 could successfully detect TNT and tetryl clearlyExplosives DNT RDX and PETN gave some degree ofquenching but particularly with the latter two there was notenough for naked-eye detection on the stripThis deviation from the expected response could be due to

the larger and less spherical (Figure S7c) red QDs having alower ligand surface coverage allowing the more rigid planararene analytes to effectively slip down the gaps between theOMe ligands and interact more closely with the QD surfacethus facilitating a stronger quenching It has been recently

stated that dithiol headgroups such as dihydrolipoic acidprovide less surface protection than monothiols at lowadsorbate concentrations but are less likely to be displacedthus ensuring longer term colloidal stability40 These larger Qvalues are not observed for CX608 due to the bulky ligandheadgroup providing more effective surface sheilding Toexplore this we modeled how the analytes might interact withthe QD surface

Ab Initio Modeling Ab initio methods (described in full inthe Supporting Information (SI) Section 4) were applied toprobe whether binding of the analytes to the particle surfacewas feasible and to measure the resulting distortion of theelectronic structure Despite the small size of the CdSeZnSQD they are large enough to be modeled as a periodic slab ofZnS Thus a model of wurzite ZnS (110) was developed withdensity functional methods as described in the SI41 Reducedfunctional approximations of each of the five analytes were thentested on the surface to measure the extent of the adsorptionon the pristine ZnS (110) surface (Figures 5 S8 and S9) Theresults summarized in Table 1 show that thiols as expectedform strong surface bonds and although the analytes can bindto the surface it is unlikely the thiols are displacedNitroaromatics showed little interaction through the aromaticring but did bind to the surface through the nitro-groupsIncrease in the number of nitro groups did not significantlyincrease the binding potential Finally nitroaliphatics were

Figure 2 Synthesis and molecules used for surface creation and testing (i a) Mesyl chloride triethylamine (TEA) THF RT then NaN3H2O 100 degC then PPh3 1 M HCl EtOAc RT 65 (b) LA DCC DMAP DCM RT 97 (c) CuC catalyst monopropargyl cavitandTEA dioxane 70 degC 3 days (d) as (b) 01 equiv LA 60 and (e) methoxyPEG 350 as (b) 82 (ii) Analytes tested

ACS Nano Article

DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

1141

modeled and nitrate esters and nitroamines were shown to alsobind however nitroamines bound more favorablyIn the case of OMe608 if each analyte was reaching the

surface then we would expect nitroaromatics to affectfluorescence substantially due to reasonable binding andelectronic structure modification However we predict a lowresponse from nitrate esters due to low binding despite their

good electron withdrawal and strong binding of nitroaminesbut no charge transfer This model is simplified because it doesnot account for additional electron-withdrawing groups on thedifferent analytes for example TNT vs DNT nor the stericsbetween molecules and capping ethylene glycol chains or otherlateral interactions Accounting for the extra NO2 groups onTNT and tetryl and the steric bulk of PETN and RDX theexpected series is tetryl ≃ TNT gt DNT gt RDX ≃ PETNwhich is as observed This adds weight to the theory of analytepenetration of the ligand surface of OMe608

Chemical NoseTongue for Explosives Sensing Thedifferential quenching of each QD by each analyte was used tobuild up fingerprints for the analytes in the range ofconcentrations (full details given in Methods) The fluores-cence quenching was expressed as either SternminusVolmerquenching (SV = I0I) or percentage quenching (Q) Thesefingerprints were then analyzed with machine learningtechniques to investigate the classification accuracy achievedThe models were blind to the concentration data Lineardiscriminant analysis (LDA) was employed and as acomparator a support vector machine (SVM) utilizing 10-fold cross validation was tested4 but returned poorerclassification results (Tables S5 and S6)We examined first whether SV or Q data provided better

classification power LDA analysis of the full data set attained100 classification accuracy in each case but the minus2logLikeli-hood score for the SV data of 0389 indicated that it performedslightly worse than analysis of Q data with minus2logLikelihood= 0001 (see SI for discussion of negative loglikelihoods but

Figure 3 Results of single channel sensing with seven red green and blue QDs (a) Quenching for each QD for a range of analyteconcentrations Lines indicate best linear fit by least-squares and shading indicates confidence of fit (b) Canonical plot (LDA) for Q dataset Ellipses indicate 95 CI of the mean Data are classified with 100 accuracy (c) Jackknifed analysis of same data set showingclassification for several subsets of the array indicated by ticks to illucidate the discriminatory power of each QD indicated by the ofcorrect classification for that subset

Figure 4 Detection by eye (a) Five test strips prepared withOMe608 on filter paper exposed to drops of water MeOH tetryland TNT (detailed in SI) Quenching is observed for tetryl andTNT although TNT still has some residual brightness in thecenter As the camera did not record the brightness of the QDs wellunder blacklight postprocessing was performed to improvecontrast against background to simulate what is seen by eye (b)

ACS Nano Article

DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

1142

simply put the smaller the number the better) This differenceis further illustrated in the canonical plots for each analysis(also explained further in the SI) The Q plot ellipses (95confidence limit of the data mean) are all well separated(Figure 3b) whereas for the SV data (Figure S10) there is closeproximity between the TNT and DNT ellipsesTo investigate which QDs had the most discriminatory

power in the array a jackknifed analysis was conducted usingthe Q data set running the LDA multiple times includingonly a subset of the particles in turn (Figure 3c) As expectedremoving quenching information decreased classificationaccuracy and indicated that the most powerful discriminatingelements were OMe608 due to its large quenching response toaromatics and CD544 due to its differential response to PETN

Figure 5 Modeling of surface interactions (a) First approximations of analytes and surface bound thiol Example of nitrate ester bound toslab in (b) side on and (c) top down profile

Table 1 Summary of Adsorption Parameters BindingEnergy Per Site (EB) Charge Transfer (Δq) Band Gap (Eg)and Work Function (ϕ) Compared with Pristine ZnS(110)Surface

approximation EB (eV) Δq (eV) Eg (eV) ϕ (eV)

single thiol minus072 minus01 206 35dihydrolipoic acid minus050 minus01 205 35benzene minus015 00 206 37nitrophenol minus019 00 132 40dinitrophenol minus020 02 063 47nitrate ester minus017 08 182 36nitroamine minus048 minus02 203 34

Figure 6 Results of multichannel sensing (a) Canonical plot for LDA analysis of single channel data approximating the multichannel examplewith OH516 CX544 and OMe608 Ellipses show 95 CI on the mean Classification accuracy is 80 (b) Calculated fit of the three QDs to givecomposite curve relative to obtained fluorescence spectrum (c) Canonical plot for multichannel array with Q data for 516 520 533 544and 608 nm Classification accuracy is 100 (d) Sample titration curves for tetryl and RDX against the multichannel system with increasingconcentration decreasing the fluorescence

ACS Nano Article

DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

1143

It was also shown that by choosing a subset of the sevenQDs good classification accuracy could still be achieved Asingle channel array using the four 544 nm QDs could classifythe data with 87 accuracy Thus a multichannel (mixedelement in a single test) sensor was designed to utilize threedifferent colored QDs each bearing a different surfacefunctionality Other considerations were selecting the mostdiscriminating elements and choosing those that gave thelargest dynamic response to provide clear peaks in the morecomplex spectral environment OH516 CX544 and OMe608 wereselected for the reasons above and when modeled as a subset ofthe single channel data gave a classification accuracy of 80suggesting that a multichannel sensor would be successful(Figure 6a)To create a multichannel sensor for explosives the three

QDs (OH516 CX544 and OMe608) were mixed The finalspectrum was confirmed by recombination of individual QDspectra to ensure there were no interaction effects between theindividual elements (Figure 6b) On excitation at 365 nm theanalytes were titrated at the same concentrations as before andthe principal QD emission wavelengths were monitored as wellas two intermediate wavelengths in the overlap regions betweenpeaks (516 520 533 544 and 608 nm) LOD analysis on thedata was hard to perform for RDX and PETN due to nonlinearresponses but for nitroaromatics tetryl and TNT LODs of lt1μgmL and 02 μgmL were obtained respectively across thewhole array (Table S3)The change in fluorescence at the five points at different

concentrations was then used to train and test LDA and SVMmodels with SV and Q data sets as before LDA analysis onthe Q data set performed better than a simple combination ofthe three individual channels from the single channel data setscoring 100 classification accuracy (minus2logLikelihood = 151)(Figure 6c) The availability of intermediate wavelengths in themultichannel setting allows for this improved classificationaccuracy by probing the cross-reactivity between the threesensor elements directly in a single test The SV data set againperformed worse only scoring 833 accuracy (Figure S11)Jackknifed analysis was performed with the most successful

(Q LDA) model to investigate the classification power ofeach wavelength (Table S4) It was found that no onewavelength contributed particularly overall to the predictivepower of the array with most individual QD or intermediateemissions scoring giving 40 classification accuracy bythemselves Removing the intermediate wavelengths (520 and533 nm) leaving the three principal emission peaks in the dataset did score 100 classification accuracy but with minus2logLikeli-hood = 798 indicating a greater risk of misclassificationHowever the fact that perfect accuracy is achievable with thesethree wavelengths in comparison to the 80 achieved by thecombination of the three single channel results highlights theadvantages of a multichannel arrayFinally the effect of contaminants on the array was

examined The methanolic array was dosed with eitheruntreated London tap water (containing trace Clminus NO3

minusNa+ etc) or a solution of nitrobenzene In each case thecontaminant addition did not substantially alter the arrayfluorescence and on addition of TNT quenching was stillobserved of the same magnitude as in ideal conditions (FigureS13) However it was observed that red OMe608 channel wasthe least stable element to contaminants with more variationon dilution than with the other channels

CONCLUSIONSWe have created and demonstrated the first multichannelfluorescent nanoparticle array for explosives detection Thearray was constructed from multicolour QDs featuring severalnovel surface ligands with supramolecular functionality Thesewere synthesized with a facile methodology easily extendableto other functionalities Five nitroaromatics and nitroaliphaticscould be reliably detected and discriminated by the array atpart-per-million or lower concentrations The use of array-based sensing with a multichannel platform has created a singlesample test for these materials with a limit-of-detection of lt02μgmL The multichannel nature of the test ensures that lesssample is required and more accurate and rapid results areobtained This work is the first step toward devising amonitoring system for explosive residues in waste waters withenvironmental and security applications It could for examplebe coupled with solid-phase extraction of waters and soils intomethanol for a rapid diagnostic test This work furtherhighlights the power of differential multichannel arrays forsensing a wide range of chemicals

METHODSCdSeZnS QDs LA-PEG-OMe LA-TEG-OH and other startingmaterials prepared by literature methods as described in SupportingInformation LA-TEG-CX and -CD were created by the clickconjugation of monopropargyl per-methyl-β-cyclodextrin or mono-propargyl calix[4]arene to LA-TEG-N3 with Cu on an activated carbonsupport34

To create the QD-ligand conjugates a protocol based on Palui etal35 was used Thirty μL of an as-synthesized QD solution was addedto 710 μL n-hexane in a glass vial Methanolic tetramethylammoniumhydroxide solution (20 mM 100 μL) and 250 μL of 100 mMmethanolic ligand solution were added to form a biphasic mixture Astirrer bar was added and the atmosphere changed to nitrogen beforethe vial was capped and stirred under 5 times 18 W 365 nm UV-bulbs(Philips TL-D) The progress of the reaction was monitored byobserving the fluorescence in the methanolic layer and once thetransfer was complete the QDs were mixed with 1 mL 101 ethanolchloroform and 9 mL hexane to precipitate before separation andwashing via centrifugation three times (4500 rpm 5 min) Theprecipitated QDs were resuspended in 1 mL MeOH and stored at 4degC for further use

Dilute solutions of each QD were prepared in MeOH (15 μLmL)and titrated with fixed amounts of 1 mM DNT TNT tetryl RDX andPETN to create solutions at 152 298 441 580 714 and 845 μMThe fluorescence of each solution was measured multiple times toattain a stable average reading The intensity of the fluorescence peak(I) was used with the initial sample fluorescence (I0) to measure theSternminusVolmer quenching (SV = I0I) and Q was calculated as [Q= 100 times (1 minus II0)]

LDA and SVM analysis were performed on a data set consisting of 1attribute per QD (eg Q-CX544 or SV-516 nm) and the class Theconcentration data were removed to blind the computer model LODfitting and LDA analysis were performed in the JMP 11 softwarepackage SVMs were performed using WLSVM in WEKA For theSVM an RBF kernel was applied and a grid search was used tooptimize the cost and gamma kernel values for each data set followedby 10-fold cross validation with the optimized values4243 Otheranalysis and modeling was performed in OriginPro

ASSOCIATED CONTENTS Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI 101021acsnano5b06433

Full experimental procedures and characterization datafor all new materials synthesis of known materials

ACS Nano Article

DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

1144

Nanoparticle TEM and EDS Materials modeling detailsLOD fitting and additional machine learning discussion(PDF)

AUTHOR INFORMATIONCorresponding AuthorsE-mail williampeveler11uclacukE-mail ipparkinuclacuk

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTSThe authors thank Dr Steve Firth for useful discussion and DrKersti Karu for assistance with mass spectrometry Donation ofthe fluorescence spectrometer was gratefully received from MrRichard H Unthank WJP is grateful for an EPSRC DoctoralPrize Fellowship (EPM5064481) Via our membership of theUKrsquos HPC Materials Chemistry Consortium which is fundedby the EPSRC (EPL000202) this work made use of theARCHER facility part of the UKrsquos national high-performancecomputing services which are funded by the Office of Scienceand Technology through EPSRCrsquos High End ComputingProgramme

REFERENCES(1) You C-C Miranda O R Gider B Ghosh P S Kim I-BErdogan B Krovi S A Bunz U H F Rotello V M Detection andIdentification of Proteins using NanoparticleminusFluorescent PolymerrsquoChemical Nosersquo Sensors Nat Nanotechnol 2007 2 318minus323(2) De M Rana S Akpinar H Miranda O R Arvizo R RBunz U H F Rotello V M Sensing of Proteins in Human Serumusing Conjugates of Nanoparticles and Green Fluorescent ProteinNat Chem 2009 1 461minus465(3) Diehl K L Anslyn E V Array Sensing using Optical Methodsfor Detection of Chemical and Biological Hazards Chem Soc Rev2013 42 8596minus8611(4) Peveler W J Binions R Hailes S M V Parkin I P Detectionof Explosive Markers using Zeolite Modified Gas Sensors J MaterChem A 2013 1 2613minus2620(5) Evans G P Buckley D J Skipper N T Parkin I P Single-walled Carbon Nanotube Composite Inks for Printed Gas SensorsEnhanced Detection of NO2 NH3 EtOH and Acetone RSC Adv2014 4 51395minus51403(6) Askim J R Mahmoudi M Suslick K S Optical Sensor Arraysfor Chemical Sensing The Optoelectronic Nose Chem Soc Rev 201342 8649minus8682(7) Ivy M A Gallagher L T Ellington A D Anslyn E VExploration of Plasticizer and Plastic Explosive Detection andDifferentiation with Serum Albumin Cross-Reactive Arrays ChemSci 2012 3 1773minus1779(8) Ponnu A Edwards N Y Anslyn E V Pattern RecognitionBased Identification of Nitrated Explosives New J Chem 2008 32848minus855(9) Bajaj A Miranda O R Kim I B Phillips R L Jerry D JBunz U H F Rotello V M Detection and Differentiation ofNormal Cancerous and Metastatic Cells Using Nanoparticle-PolymerSensor Arrays Proc Natl Acad Sci U S A 2009 106 10912minus10916(10) Elci S G Moyano D F Rana S Tonga G Y Phillips R LBunz U H F Rotello V M Recognition of GlycosaminoglycanChemical Patterns using an Unbiased Sensor Array Chem Sci 2013 42076minus2080(11) Rana S B L D Mout R Saha K Tonga G Y S B EMiranda O R Rotello C M Rotello V M A MultichannelNanosensor for Instantaneous Readout of Cancer Drug MechanismsNat Nanotechnol 2015 10 65minus69

(12) Abdul-Karim N Blackman C S Gill P P Wingstedt E MM Reif B A P Post-Blast Explosive Residue minus A Review ofFormation and Dispersion Theories and Experimental Research RSCAdv 2014 4 54354minus54371(13) Jurcic M Peveler W J Savory C N Scanlon D O KenyonA J Parkin I P The Vapour Phase Detection of Explosive Markersand Derivatives using Two Fluorescent MetalminusOrganic Frameworks JMater Chem A 2015 3 6351minus6359(14) Michalet X Pinaud F F Bentolila L A Tsay J M DooseS Li J J Sundaresan G Wu A M Gambhir S S Weiss SQuantum Dots for Live Cells in vivo Imaging and Diagnostics Science2005 307 538minus544(15) Freeman R Finder T Bahshi L Willner I Beta-Cyclodextrin-Modified CdSeZnS Quantum Dots for Sensing andChiroselective Analysis Nano Lett 2009 9 2073minus2076(16) Niu Q Gao K Lin Z Wu W Amine-Capped Carbon Dotsas a Nanosensor for Sensitive and Selective Detection of Picric Acid inAqueous Solution via Electrostatic Interaction Anal Methods 2013 56228minus6233(17) Pazhanivel T Nataraj D Devarajan V P Mageshwari VSenthil K Soundararajan D Improved Sensing Performance fromMethionine Capped CdTe And CdTeZnS Quantum Dots for theDetection of Trace Amounts of Explosive Chemicals in Liquid MediaAnal Methods 2013 5 910minus916(18) Carrillo-Carrion C Simonet B M Valcarcel M Determi-nation of TNT Explosive Based on its Selectively Interaction withCreatinine-Capped CdSeZnS Quantum Dots Anal Chim Acta 2013792 93minus100(19) Tian X Chen L Qing X Yu K Wang X Wang X HybridCadmium Tellurium Quantum Dots for Rapid Visualization of Trace-Level Nitroaromatic Explosives Anal Lett 2014 47 2035minus2047(20) Sun X Wang Y Lei Y Fluorescence Based ExplosiveDetection From Mechanisms to Sensory Materials Chem Soc Rev2015 44 8019(21) Zhang K Zhou H Mei Q Wang S Guan G Liu RZhang J Zhang Z Instant Visual Detection of TrinitrotolueneParticulates on Various Surfaces by Ratiometric Fluorescence of Dual-Emission Quantum Dots Hybrid J Am Chem Soc 2011 133 8424minus8427(22) Freeman R Finder T Bahshi L Gill R Willner IFunctionalized CdSeZnS QDs for the Detection of Nitroaromatic orRDX Explosives Adv Mater 2012 24 6416minus6421(23) Freeman R Willner I Optical Molecular Sensing withSemiconductor Quantum Dots (Qds) Chem Soc Rev 2012 414067minus4085(24) Enkin N Sharon E Golub E Willner I Ag NanoclusterDNA Hybrids Functional Modules for the Detection of Nitroaromaticand RDX Explosives Nano Lett 2014 14 4918minus4922(25) Ponnu A Anslyn E V A Fluorescence-Based CyclodextrinSensor to Detect Nitroaromatic Explosives Supramol Chem 2010 2265minus71(26) Gutsche C D In Calixarenes Stoddart J F Ed Royal Societyof Chemistry Cambridge UK 1989(27) Montmeat P Veignal F Methivier C Pradier C MHairault L Study of Calixarenes Thin Films as Chemical Sensors forthe Detection of Explosives Appl Surf Sci 2014 292 137minus141(28) Del Valle E M M Cyclodextrins and Their Uses A ReviewProcess Biochem 2004 39 1033minus1046(29) Dethlefsen J R Doslashssing A Preparation of a ZnS Shell onCdSe Quantum Dots Using a Single-Molecular ZnS Precursor NanoLett 2011 11 1964minus1969(30) Bear J C Hollingsworth N McNaughter P D Mayes A GWard M B Nann T Hogarth G Parkin I P Copper-DopedCdSeZnS Quantum Dots Controllable Photoactivated Copper(I)Cation Storage and Release Vectors for Catalysis Angew Chem IntEd 2014 53 1598minus1601(31) Bear J C Hollingsworth N Roffey A McNaughter P DMayes A G Macdonald T J Nann T Ng W H Kenyon A JHogarth G et al Doping Group IIB Metal Ions into Quantum Dot

ACS Nano Article

DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

1145

Shells via the One-Pot Decomposition of Metal-DithiocarbamatesAdv Opt Mater 2015 3 704minus712(32) Chetcuti M J Devoille A M J Othman A B Souane RThuery P Vicens J Synthesis of Mono- Di- and Tetra-AlkyneFunctionalized Calix[4]arenes Reactions of these Multipodal Ligandswith Dicobalt Octacarbonyl to give Complexes which Contain up toEight Cobalt Atoms Dalton Trans 2009 2999minus3008(33) Faiz J A Spencer N Pikramenou Z Acetylenic Cyclodextrinsfor Multireceptor Architectures Cups with Sticky Ends for theFormation of Extension Wires and Junctions Org Biomol Chem 20053 4239minus4245(34) Lipshutz B H Taft B R Heterogeneous Copper-in-Charcoal-Catalyzed Click Chemistry Angew Chem Int Ed 2006 45 8235minus8238(35) Palui G Avellini T Zhan N Pan F Gray D Alabugin IMattoussi H Photoinduced Phase Transfer of Luminescent QuantumDots to Polar and Aqueous Media J Am Chem Soc 2012 13416370minus16378(36) Zhan N Palui G Grise H Tang H Alabugin I MattoussiH Combining Ligand Design with Photoligation to Provide CompactColloidally Stable and Easy to Conjugate Quantum Dots ACS ApplMater Interfaces 2013 5 2861minus2869(37) Zhan N Palui G Merkl J-P Mattoussi H Quantifying theDensity of Surface Capping Ligands on Semiconductor QuantumDots Proc SPIE 2015 93381Aminus93381A-11(38) Cahill S Bulusu S Molecular Complexes of Explosives withCyclodextrins I Characterization of Complexes with the NitraminesRDX HMX and TNAZ in Solution by 1H NMR Spin-LatticeRelaxation Time Measurements Magn Reson Chem 1993 31 731minus735(39) Zhang M Shi Z Bai Y Gao Y Hu R Zhao F UsingMolecular Recognition of β-Cyclodextrin to Determine MolecularWeights of Low-Molecular-Weight Explosives by MALDI-TOF MassSpectrometry J Am Soc Mass Spectrom 2006 17 189minus193(40) Takeuchi H Omogo B Heyes C D Are Bidentate LigandsReally Better than Monodentate Ligands For Nanoparticles NanoLett 2013 13 4746minus4752(41) Afzaal M Malik M A OrsquoBrien P Preparation of ZincContaining Materials New J Chem 2007 31 2029minus2040(42) Chang C-C Lin C-J LIBSVM A Library for Support VectorMachines ACM Trans Intell Syst Technol 2011 2 1minus27(43) El-Manzalawy Y Honavar V WLSVM Integrating LIBSVMinto WEKA Environment 2005 Software available at httpwww csiastate eduyasserwlsvm

ACS Nano Article

DOI 101021acsnano5b06433ACS Nano 2016 10 1139minus1146

1146