sensors

Article The Identification of Seven Mimics Using a Colorimetric Array

Michael J. Kangas * , Adreanna Ernest, Rachel Lukowicz, Andres V. Mora , Anais Quossi, Marco Perez, Nathan Kyes and Andrea E. Holmes * Department of Chemistry, Doane University, Crete, NE 68333, USA; [email protected] (A.E.); [email protected] (R.L.); [email protected] (A.V.M.); [email protected] (A.Q.); [email protected] (M.P.); [email protected] (N.K.) * Correspondence: [email protected] (M.J.K.); [email protected] (A.E.H.)

 Received: 8 November 2018; Accepted: 3 December 2018; Published: 6 December 2018 

Abstract: Chemical warfare agents pose significant threats in the 21st century, especially for armed forces. A colorimetric detection array was developed to identify warfare mimics, including mustard and nerve agents. In total, 188 sensors were screened to determine the best sensor performance, in order to identify warfare mimics 2-chloro ethyl ethylsulfide, 2-20-thiodiethanol, trifluoroacetic , methylphosphonic acid, dimethylphosphite, diethylcyanophosphonate, and diethyl (methylthiomethyl)phosphonate. The highest loadings in the principle component analysis (PCA) plots were used to identify the sensors that were most effective in analyzing the RGB data to classify the warfare mimics. The dataset was reduced to only twelve sensors, and PCA results gave comparable results as the large data did, demonstrating that only twelve sensors are needed to classify the warfare mimics.

Keywords: warfare mimics; ; colorimetric array; principle component analysis; RGB data

1. Introduction Improving the non-presumptive detection of chemical warfare agents for soldiers in the field is a pressing need [1]. Warfare agents include chemical and explosive compounds, such as nerve agents like , , and , which are clear, colorless liquids without strong , and can cause loss of consciousness, seizures, and eventual death [2]. Other chemicals of war include vesicants or , such as mustard gas, , , , cyanide, , , and made of capsaicin [2]. Additionally, the detection of the precursors and degradation products of these materials can be equally important [3]. Existing technology has shown that these devices can effectively detect such analytes, but requires high-tech, expensive, and large instrumentation, which makes them less than ideal for field applications [4–7]. Colorimetric and fluorescent sensors have the potential to be effective detection systems for these analytes, but require minimal instrumentation. New sensors for these analytes have recently been reviewed [8]. One drawback of sensors is that they can be limited in the versatility or specificity of the analytes that can be identified. This can be overcome by combining multiple sensors into a sensor array, and the pattern of color changes can be used to identify and quantify a wide range of analytes [9–12]. A few sensor arrays chromofluorogenic supramolecular complexes for the chromogenic or fluorogenic sensing of nerve agents have been reported [13], as well as color changes of organophosphorus warfare mimics in the gas phase [14,15]. Exposure guidelines by the Center Disease Control suggest that mustard gas of 3.9 mg/m3 cause life-threatening effects or death. Therefore, there is a pressing need to identify sensors that can quickly detect this life-threatening agent. Herein, we examined the screening

Sensors 2018, 18, 4291; doi:10.3390/s18124291 www.mdpi.com/journal/sensors Sensors 2018, 18, x FOR PEER REVIEW 2 of 9 Sensors 2018, 18, 4291 2 of 8 sensors that can quickly detect this life-threatening agent. Herein, we examined the screening of 188 colorimetric sensors, in solution, to simulate an array for the detection of 144 samples of mustard gas mimics,of 188 colorimetric 2-chloro-ethyl-ethylsulfide sensors, in solution, (0.1 M), to simulate2-2′-thiodiethanol an array (1 for M), the and detection nerve ofagent 144 samplesmimics, 0 trifluoroaceticof mustard gas acid mimics, (1 2-chloro-ethyl-ethylsulfideM), methylphosphonic (0.1acid M),(1 2-2M),-thiodiethanol dimethylphosphite (1 M), and (1 nerveM), diethylcyanophosphonateagent mimics, trifluoroacetic (1 M), acid and (1 diethyl M), methylphosphonic (methylthiomethyl)phosphonate acid (1 M), dimethylphosphite (0.1 M) (Figure 1) (1 [14– M), 16].diethylcyanophosphonate The identity and CAS (1 numbers M), and diethylof all 188 (methylthiomethyl)phosphonate sensors are located in the Supplemental (0.1 M) (Figure Information1)[ 14–16]. The(Table identity S1). and CAS numbers of all 188 sensors are located in the Supplemental Information (Table S1).

S Cl HO OH S 2-2'-thiodiethanol 2-chloro-ethyl-ethylsulfide

N H O P HO O O PO O O O F F F diethylcyanophosphonate dimethylphosphite trifluoroacetic acid

O S O P O OH O P OH

diethyl-(methylthiomethyl)-phosphonate methylphosphonic acid Figure 1. ChemicalChemical structures structures of mustard gas mimics (red) and nerve gas mimics (blue).

2. Materials and Methods All reagentsreagents werewere purchased purchased from from various various chemical chemical supply supply companies companies at technical at technical grade grade or better, or better,and were and used were as received.used as Thereceived. universal The indicator universal was indicator composed was of Vancomposed Urk’s and of Yamada’sVan Urk’s recipe, and Yamada’sand contained recipe, methyl and red,contained methyl methyl orange, red, phenolphthalein, methyl orange, andphenolphthalein, bromothymol and blue bromothymol powders in a weightblue powders ratio of approximatelyin a weight ratio 1:3:7:8 of [17 approximately]. One percent weight-by-weight1:3:7:8 [17]. One solutionspercent ofweight-by-weight all sensors were preparedsolutions byof dissolvingall sensors each were into prepared aliquots of by a solvent dissolving mixture each consisting into aliquots of acetate of buffera solvent (0.1 M,mixture pH 5), consistingethylene glycol, of acetate triethylene buffer glycol(0.1 M, monobutyl pH 5), ethyle ether,ne andglycol, glycerol, triethylene in a ratio glycol of 14:1.6:1:3.2. monobutyl The ether, sensors and ◦ wereglycerol, dissolved in a ratio by of sonication 14:1.6:1:3.2. in The a bath sensors sonicator were (30dissolvedC) for by 1 h,sonication followed in by a bath 5 min sonicator mixing (30 with °C) a probefor 1 h, sonicator, followed and by then5 min vacuum mixing filteredwith a onceprobe through sonicator, Whatman and then #1 vacuum filter paper, filtered and once through Whatman nylon#1 filter filter paper, membranes and once (0.2 throughµm). Whatman nylon filter membranes (0.2 µm). A solution of of methyl methyl phosphonic phosphonic acid acid (1 (1 M) M) wa wass prepared prepared dissolving dissolving an an appropriate appropriate amount amount of theof theanalytes analytes in milli-Q in milli-Q water (18 (18MΩ M-cm).Ω-cm). Solutions Solutions of 2 ofchloroethyl 2 chloroethyl ethyl ethyl sulfide (0.1 (0.1M), 2,2 M),′ 0 thiodiethanol2,2 thiodiethanol (1 M), (1 M),diethylcyanophosphonate diethylcyanophosphonate (0.1 (0.1 M), M), dimethylphosphite dimethylphosphite (1 (1 M), M), and and diethyl (methylthiomethyl)phosphonate (0.1 (0.1 M) M) were were prepar prepareded by diluting the reagents with milli-Q water (18 M MΩΩ-cm).-cm). The The concentrations concentrations were were selected selected to be to near be near the the solubility limit limitor 1 M. or Solutions 1 M. Solutions of 2,2′ 0 thiodiethanolof 2,2 thiodiethanol and methylphosphonic and methylphosphonic acid were acid stored were in storedthe dark in at the room dark temperature at room temperature and were replacedand were replacedapproximately approximately every three every threemonths. months. Solutions Solutions of of2-chloroethyl 2-chloroethyl ethyl ethyl sulfide, sulfide, diethylcyanophosphonate, dimeth dimethylphosphite,ylphosphite, and and diethyl (methylthiomethyl)phosphonate(methylthiomethyl)phosphonate were ◦ portioned into ~10 mL aliquots, stored at −−6060 °CC to to minimize decomposition [18], [18], and replaced approximately everyevery 33 months.months. Solutions Solutions were were allowed allowed to to thaw thaw at roomat room temperature temperature for for 1 h before1 h before use. use. Sensor (100 µL) was added to all the wells of a flat bottomed 96-well plate using a 12-channel electronicSensor pipet. (100 µL) One was hundred added microlitersto all the wells of each of a analyte flat bottomed were applied 96-well to plate rows using of the a 12-channel well plate. Thirty-twoelectronic pipet. samples One of waterhundred and microliters 16 samples of of each the 7 analytesanalyte were were applied tested to to make rows the of total the samplewell plate. size 144.Thirty-two The plate samples was scanned of water as and detailed 16 samples below. of the 7 analytes were tested to make the total sample size 144.All arrayThe plate images was (24-bit scanned color, as detailed 400 dpi) below. were collected using an Epson Perfection V700 desktop scannerAll inarray transparency images (24-bit mode. color, To eliminate 400 dpi) interferenceswere collected from using stray an light,Epson the Perfection scanner wasV700 draped desktop in scannerblack cloth. in transparency The images mode. were analyzed To eliminate with interfer ImageJences [19], andfrom the stray extraction light, the of scanner mean RGB was valuesdraped for in blackeach wellcloth. was The automated images were with analyzed a macro with [20]. ImageJ No attempts [19], and were the made extraction to correct of mean for image-to-imageRGB values for each well was automated with a macro [20]. No attempts were made to correct for image-to-image

SensorsSensors 20182018, ,1818, ,x 4291FOR PEER REVIEW 3 3of of 9 8 variation by subtracting a control row, as one sensor was tested on each well plate, and our previous workvariation found by correcting subtracting images a control to be row, unnecessary as one sensor [12]. was tested on each well plate, and our previous workAll found statistical correcting analysis images was toperformed be unnecessary using the [12]. statistical programming language R [21]. PCA was performedAll statistical using analysis the function was performed prcomp. using The data the statistical was mean-centered, programming but language was not R scaled [21]. PCA to unit was varianceperformed because using theall of function the data prcomp. were on The a dataconsiste wasnt mean-centered, scale of 0 to 255 but RGB was notunits. scaled Score to unitplots variance of the resultingbecause alldata of were the data constructed were on a using consistent ggbiplot scale library of 0 to [22]. 255 RGB units. Score plots of the resulting data were constructed using ggbiplot library [22]. 3. Results 3. Results

3.1.3.1. Visible Visible Colorimetric Colorimetric Changes Changes AnAn actual actual image ofof thethe color color changes changes of of a warfarea warfare mimic mimic is depicted is depicted in Figure in Figure2, where 2, anwhere excerpt an excerpta 96-well a 96-well plate of plate the colorimetricof the colorimetric array isarray depicted. is depicted. The first The row firstshows row shows the analyte the analyte 2-chloro 2-chloro ethyl ethylethylsulfide ethylsulfide in the in presence the presence of 4 different of 4 different sensors. sensor Sensorss. Sensors A, B, C,A, DB, areC, D Ellman’s are Ellman’s reagent, reagent, fast blue fast B, bluealizarin B, alizarin yellow yellow R, and R, eosin and Y.eosin In comparison Y. In comparison to the to control, the control, all sensors all sensors have ahave different a different color color in the inpresence the presence of the of analyte. the analyte.

FigureFigure 2. 2. TheThe top top row row shows shows the the colorimetric colorimetric change changess of of the the mustard mustard gas gas mimic mimic 2-chloro 2-chloro ethyl ethyl ethylsulfideethylsulfide in in the the presence presence of of Ellm Ellman’san’s reagent, reagent, fast fast blue blue B, B, alizarin alizarin yellow, yellow, and and eosin eosin Y Y(A, (A, B, B, C, C, D). D). TheThe bottom bottom row row shows shows the the color color changes changes in in the the presence presence of of the the control. control. 3.2. Principle Component Analysis (PCA) 3.2. Principle Component Analysis (PCA) Principle component analysis (PCA) has been successfully used in the identification of explosives Principle component analysis (PCA) has been successfully used in the identification of with colorimetric arrays [23]. Thus, images of the sensors using different analyte exposure were explosives with colorimetric arrays [23]. Thus, images of the sensors using different analyte captured, and their red green blue (RGB) values were extracted for PCA analysis. This classification exposure were captured, and their red green blue (RGB) values were extracted for PCA analysis. method has been established as a viable tool to interpret data generated by sensor arrays [11,14,24]. This classification method has been established as a viable tool to interpret data generated by sensor Figure3 demonstrates the first two principle components of the RGB data that was obtained when arrays [11,14,24]. Figure 3 demonstrates the first two principle components of the RGB data that was all warfare mimics were reacted with the most effective 188 colorimetric sensors. It is seen that obtained when all warfare mimics were reacted with the most effective 188 colorimetric sensors. It is 2-chloro ethyl ethylsulfide, diethylcyanophosphonate, trifluoroacetic acid, methylphosphonic acid, seen that 2-chloro ethyl ethylsulfide, diethylcyanophosphonate, trifluoroacetic acid, and dimethylphosphite show clearly distinct clusters that are well separated from each other. Diethyl methylphosphonic acid, and dimethylphosphite show clearly distinct clusters that are well (methylthiomethyl)phosphonate and 2-20-thiodiethanol are well clustered in each group, but not as separated from each other. Diethyl (methylthiomethyl)phosphonate and 2-2′-thiodiethanol are well well separated. However, PCA was still able to distinguish these two analytes from the control. clustered in each group, but not as well separated. However, PCA was still able to distinguish these two analytes from the control.

Sensors 2018, 18, x FOR PEER REVIEW 4 of 9 Sensors 2018, 18, x FOR PEER REVIEW 4 of 9 Sensors 2018, 18, 4291 4 of 8

Figure 3. PCA plot showing the clustering of two mustard gas mimics and four mimics. Figure 3. PCA plot showing the clustering of two mustard gas mimics and four nerve agent mimics. FigureAll analytes 3. PCA are plot clearly showing clustered the clustering into indivi ofdual two groups mustard and gas distinguished mimics and fromfour nervewater. agent mimics. All analytes are clearly clustered into individual groups and distinguished from water. All analytes are clearly clustered into individual groups and distinguished from water. The PCA loading plot for the first component, shown in Figure 4, was used to determine which The PCA loading plot for the first component, shown in Figure4, was used to determine which sensorsThe and PCA color loading channels plot forcontributed the first component, most to classifying shown inthe Figure analytes 4, was that used were to screened determine with which 188 sensors and color channels contributed most to classifying the analytes that were screened with sensorssensors. andThe colorsensor channels channels contributed that are centered most to around classifying the 0.00 the principalanalytes that component were screened loadings with do 188not 188 sensors. The sensor channels that are centered around the 0.00 principal component loadings do sensors.contribute The effectively sensor channels to classify that the are analytes, centered whil arounde the thesensors 0.00 principalthat are loca componentted between loadings +0.1 to do +0.15 not notcontributeor contribute −0.1 to 0.2 effectively effectivelyloadings, to are toclassify effectively classify the theanalytes, contributing analytes, whil whilee to the the sensors theanalyte sensors that classification. are that loca areted located between between +0.1 to +0.15 +0.1 to +0.15or − or0.1− to0.1 0.2 to loadings, 0.2 loadings, are effectively are effectively contributing contributing to the toanalyte the analyte classification. classification.

Figure 4. PCA loading plot demonstrating which sensors and color channels perform most Figure 4. PCA loading plot demonstrating which sensors and color channels perform most effectively in analyte classification. The sensors and channels are (1) blue—2,4 dinitrophenol, effectively in analyte classification. The sensors and channels are (1) blue—2,4 dinitrophenol, (2) (2)Figure red—alizarin 4. PCA red loading S, (3) green—alizarin plot demonstrating red S,(4) which red—alizarin sensors yellowand color R, (5) channels green—alizarin perform yellow most R, red—alizarin red S, (3) green—alizarin red S, (4) red—alizarin yellow R, (5) green—alizarin yellow R, (6) red—aurintricarboxyliceffectively in analyte classification. acid, (7) red—bromocresol The sensors and green, channels (8) red—chlorophenol are (1) blue—2,4 red,dinitrophenol, (9) red—eosin (2) Y, (6) red—aurintricarboxylic acid, (7) red—bromocresol green, (8) red—chlorophenol red, (9) (10)red—alizarin red—erythrosin red S, B, (3) (11) green—alizarin red—fluorescein, red S, (4) (12) red—alizarin red—glycine yellow cresol R, red,(5) green—alizarin (13) red—methyl yellow orange, R, red—eosin Y,(10) red—erythrosin B, (11) red—fluorescein, (12) red—glycine cresol red, (13) (14)(6) red—orange red—aurintricarboxylic IV, (15) red—phloxine acid, (7) red—brom B, (16)ocresol red—thymol green, blue,(8) red—chlorophenol (17) red—xylenol red, orange, (9) red—eosinred—methyl Y,(10) orange, red—erythrosin (14) red—orange B, (11) IV, red—fl(15) red—phloxineuorescein, (12) B, red—glyc(16) red—thymoline cresol blue,red, (13)(17) (18) red—yellow A2, (19) red—alizarin yellow GG, (20) green—alizarin yellow GG, (21) red—iodophenol red—methylred—xylenol orange,orange, (18)(14) red—yellow red—orange A2, IV, (19) (15) red—alizarin red—phloxine yellow B, GG, (16) (20) red—thymol green—alizarin blue, yellow (17) blue, (22) red—rosolic acid, (23) blue—4-(4-diethylaminophenylazo), (24) blue—4-nitrocatechol, red—xylenol orange, (18) red—yellow A2, (19) red—alizarin yellow GG, (20) green—alizarin yellow (25) red—nitrazine yellow, (26) red—Ellman’s reagent, (27) green—Ellman’s reagent, (28) blue—Ellman’s

reagent, (29) red—metanil yellow, (30) blue—amanil fast yellow, (31) blue—thymolphthalein, (32) red—4-bromo-2,6-xylenol, (33) red—xylenol blue, (34) red—Rose bengal, (35) green—eosin B. Sensors 2018, 18, 4291 5 of 8

Table1 identifies the sensors, the color channels of the sensors, and the PCA 1 and 2 loadings for each sensor. The PCA 1 and 2 loadings for each sensor indicated that alizarin yellow had 5 and the most loading contributions by the red and green color channels. This is followed by 3 contributions of the blue, green, and red color channels from the Ellman’s reagent, 3 contributions by the green and red channels from alizarin red, and 3 loadings by the red, green, and blue color channels from fast blue B. All other sensors only contributed one loading in one color channel.

Table 1. The best performing sensors and color channels to classify two mustard gas mimics and four nerve agent mimics.

Sensor [Channel] PC1 Loading Sensor [Channel] PC2 Loading Methyl Orange [Red] −0.198 Eosin Y [Red] −0.274 Ellman’s Reagent [Blue] −0.194 Phloxine B [Red] −0.259 Thymol Blue [Red] −0.191 Methyl Orange [Red] 0.223 Xylenol Blue [Red] −0.188 Morin [Red] −0.203 Ellman’s Reagent [Green] −0.187 o-Dianisidine [Blue] −0.202 Ellman’s Reagent [Blue] −0.186 Bromocresol Purple [Red] −0.201 Orange IV [Red] −0.179 Bromothymol Blue [Red] −0.199 Metanil Yellow [Red] −0.179 Thymol Blue [Red] −0.196 Alizarin Yellow GG [Red] −0.174 o-Dianisidine [Green] −0.195 Eosin Y [Red] −0.165 o-Dianisidine [Red] −0.194 Bromocresol Green [Red] 0.154 Morin [Green] −0.187 Phloxine B [Red] −0.153 Bromophenol blue [Red] −0.180 Fluorescein [Red] −0.153 Alizarin Red S [Green] 0.174 Nitrazine Yellow [Red] 0.149 Orange IV [Red] 0.170 Hexammine Cobalt (III) [Red] 0.148 Alizarin Red S [Red] 0.163 Yellow A2 [Red] −0.147 Ferric Chloride [Blue] −0.154 Alizarin Yellow GG[Green] −0.146 Alizarin Yellow R [Red] 0.153 Erythrosin B [Red] −0.145 1,10-Diethyl-4,40-Cyanine Iodide [Red] −0.149 Alizarin Red S [Green] 0.143 Alizarin Yellow R [Green] 0.144 Alizarin Yellow R [Red] −0.141 Chlorophenol Red [Red] −0.135

For practical applications, the number of sensors should be minimized as much as possible, while still achieving appropriate selectivity. Four, eight, and twelve sensor arrays were simulated by making a subset of the RGB dataset, using sensors that had large contributions to the first and second components of the PCA analysis for the entire dataset. As shown in Figure5, the twelve-sensor array shows well-separated groups, similar to the analysis of the entire dataset, except for an overlap of diethyl (methylthiomethyl)phosphonate and 2-20-thiodiethanol with the water control. Similarly, the four- and eight-sensor arrays resulted in less separation between the organophosphorus mimics (data not shown). This suggests that at least twelve sensors would be needed in order to classify the warfare mimics most efficiently. Sensors 2018, 18, x FOR PEER REVIEW 6 of 9

Sensors 2018, 18, 4291 6 of 8

Figure 5. PCA plot of the reduced dataset using only the RGB data of twelve sensors. The sensors consistedFigure 5. ofPCA alizarin plot of red the S, reduced alizarin yellowdataset R,using Ellman’s only the reagent, RGB fastdata blue of twelve B, methyl sensors. orange, The eosinsensors Y, phloxineconsisted B, of thymol alizarin blue, red orangeS, alizarin IV, morin,yellow bromocresol R, Ellman’s purple,reagent, and fast xylenol blue B, blue. methyl orange, eosin Y, phloxine B, thymol blue, orange IV, morin, bromocresol purple, and xylenol blue. 4. Discussions 4. Discussions Our results showed that the screening of 188 sensors for the detection of warfare mimics can generateOur RGBresults values showed that that were the analyzed screening by of PCA 188 tosensors determine for the the detection top twenty of warfare sensors mimics and color can channelsgenerate thatRGB were values most that effective were analyzed in classifying by PCA 6 analytes to determine that represented the top twenty mustard sensors gas andand nerve color agentchannels warfare that mimics.were most Furthermore, effective in theclassifying entire dataset 6 analytes using that 188 represented sensors was mustard reduced gas to just and twelve nerve sensorsagent warfare showing mimics. very similar Furthermore, results as the the entire large dataset.dataset using This is 188 important sensors becausewas reduced the miniaturization to just twelve ofsensors colorimetric showing arrays very allows similar for less results sensor andas the analyte large consumption, dataset. This as well is asimportant quicker analysisbecause of thethe RGBminiaturization data. The first of colorimetric step of screening arrays potential allows sensorsfor less forsensor analytes and likeanalyte warfare consumption, mimics involves as well the as qualitativequicker analysis analysis of of the sensor–analyte RGB data. The interactions. first step Therefore, of screening only potential one analyte sensors for analytes between like 0.1warfare M and mimics 1.0 M wasinvolves used inthe this qualitative study, which analysis was of previously sensor–analyte described interactions. to give distinguishable Therefore, only results one foranalyte various concentration and bases between [25]. A0.1 quantitative M and 1.0 analysisM was used of warfare in this analytes study, willwhich be publishedwas previously soon, similardescribed to whatto give was distinguishable described for printedresults for sensors various [26]. acids and bases [25]. A quantitative analysis of warfareThe analytes color changes will be of published the sensors soon, in the similar array canto what been was attributed described to several for printed mechanisms, sensors including[26]. acid–baseThe color reactions, changes solvatochromic of the sensors dye interactions,in the array can reactions,been attributed –dipole to several interactions, mechanisms, and protonincluding transfer acid–base reactions reactions, [27–29]. solvatochromic dye interactions, redox reactions, dipole–dipole interactions,However, and not proton all of thetransfer color reactions changes are[27–29]. clearly understood. For example, one of the analytes, the mustardHowever, gas not mimic, all of 2-chloro the color ethyl changes ethylsulfide, are clearl reactedy understood. with 5,5 For0-dithiobis-(2-nitrobenzoicacid), example, one of the analytes, alsothe mustard known as gas the mimic, Ellman’s 2-chloro reagent ethyl (Figure ethylsulfide,2). The Ellman’s reacted reagentwith 5,5 typically′-dithiobis-(2-nitrobenzoicacid), creates a yellow color changealso known when as a the reaction Ellman’s occurs reagent with thiols(Figure by 2). the The nucleophilic Ellman’s reagent substitution typically on the creates disulfide a yellow bond color [30]. However,change when in our a reaction study, a occurs brown with color thiols change by wasthe inducednucleophilic by thesubstitution mustard gason the mimic, disulfide which bond is a sulfide,[30]. However, leading in to our a brown study, colora brown change color in change comparison was induced to the by control the mustard (water). gas Figure mimic,6 shows which a is potentiala sulfide, mechanism leading to a that brown may color still include change a in nucleophilic comparison substitution to the control of 2-chloro(water). ethylFigure ethylsulfide 6 shows a (1)potential on the disulfidemechanism in thethat Ellman’s may still regent include (2), a potentially nucleophilic forming substitution the brown of 2-chloro colored cationethyl ethylsulfide (3), and the 2-carboxy-3-nitro-benzenethiol(1) on the disulfide in the Ellman’s anion regent (4), in (2), comparison potentially to forming the control the(water). brown colored cation (3), and the 2-carboxy-3-nitro-benzenethiol anion (4), in comparison to the control (water).

Sensors 2018, 18, 4291 7 of 8 Sensors 2018, 18, x FOR PEER REVIEW 7 of 9

FigureFigure 6.6.A A possible possible mechanism mechanism of of the the reaction reaction of of mustard mustard gas gas mimics mimics 2-chloro 2-chloro ethyl ethyl ethylsulfide ethylsulfide in the in presencethe presence of Ellman’s of Ellman’s reagent, reagent, leading leading to a to dark a dark brown brown color color change. change. The The nucleophilic nucleophilic substitution substitution of theof the sulfur on the on disulfidethe disulfide leads le toads proposed to proposed cationic cationic product product1, and the1, and anionic the leavinganionic groupleaving forming group aforming brown-colored a brown-colored product. product.

TheThe futurefuture of these sensors includes includes the the testing testing with with real real warfare warfare agents agents and and the the deposition deposition on onpaper-like paper-like substrates substrates that that is followed is followed by imaging by imaging with withsmartphones. smartphones. Colorimetric Colorimetric arrays arrays that are that on arepaper on papersubstrates substrates may mayprovide provide a low-cost a low-cost alternative alternative to toother other available available sensors sensors or lab-basedlab-based instrumentation,instrumentation, making making them them a usefula useful primary primary or substitute or substitute detection detection method method when portablewhen portable means aremeans necessary. are necessary. Thus, this Thus, technology this technology has great potentialhas great for potential field deployment, for field deployment, where light, inexpensive,where light, andinexpensive, easy-to-use and technology easy-to-use is criticaltechnology for identification is critical for of identification hazardous materials, of hazardous like warfare materials, agents. like Additionalwarfare agents. applications Additional for these applications versatile sensorsfor these include versatile the analysis sensors of include street drugs the analysis by police of officers street ordrugs the detectionby police ofofficers potentially or the dangerousdetection of compounds potentially dangerous by first responders compounds in emergency by first responders situations. in Theemergency expedited situations. process wouldThe expedited provide process these individuals would provide with anthese effective individuals method with for testingan effective and triagingmethod threateningfor testing and chemicals triaging when threatening their safety chemicals may rely when on the their speed safety of detection. may rely on the speed of detection. Supplementary Materials: The following are available online at http://www.mdpi.com/1424-8220/18/12/4291/ s1, Table S1: List of sensors. Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Table S1: List of Authorsensors. Contributions: Conceptualization was done by M.J.K., A.E.H.; methodology by M.J.K., A.E.H.; software by M.J.K.; validation by M.J.K..; formal analysis by M.J.K.; investigation by M.J.K..; A.E.H.; resources by M.J.K., A.E.H.;Author dataContributions: curation by A.E.,Conceptualization R.L., A.V.M., M.P.,was done N.K., A.Q.by M.J.K., and M.J.K.; A.E.H.; writing—original methodology by draft M.J.K., preparation A.E.H.; by M.J.K. and A.E.H.; writing—review and editing by M.J.K. and A.E.H.; visualization by M.J.K. and A.E.H.; supervisionsoftware by by M.J.K.; A.E.H.; validation project administration by M.J.K..; formal by A.E.H.; analysis and by fundingM.J.K.; invest acquisitionigation by by A.E.H. M.J.K..; A.E.H.; resources by M.J.K., A.E.H.; data curation by A.E., R.L., A.V.M., M.P., N.K., A.Q. and M.J.K.; writing—original draft Funding: This publication was made possible by the US Army W911SR-15-C-0027 SBIR Phase I -Chemical Biologicalpreparation Radiological by M.J.K. and Nuclear A.E.H.; and writing—review Explosives (CBRNE) and editing Reconnaissance by M.J.K. and Sampling A.E.H.; visualization Kit (A15-048), by M.J.K. SBIR Phaseand A.E.H.; II -Chemical supervision Biological by A.E.H.; Radiological project administra Nuclear andtion Explosivesby A.E.H.; and (CBRNE) funding Reconnaissance acquisition by SamplingA.E.H. Kit (W911SR-16-C-0051), and the National Science Foundation (Grant No. 1459838). Funding: This publication was made possible by the US Army W911SR-15-C-0027 SBIR Phase I -Chemical ConflictsBiological of Radiological Interest: The Nuclear authors an declared Explosives no conflict (CBRNE) of interest. Reconnaissance Sampling Kit (A15-048), SBIR Phase II -Chemical Biological Radiological Nuclear and Explosives (CBRNE) Reconnaissance Sampling Kit References(W911SR-16-C-0051), and the National Science Foundation (Grant No. 1459838).

1.ConflictsKangas, of Interest: M.J.; Burks, The authors R.M.; Atwater, declare J.;no Lukowicz,conflict of interest. R.M.; Williams, P.; Holmes, A.E. Colorimetric Sensor Arrays for the Detection and Identification of Chemical Weapons and Explosives. Crit. Rev. Anal. Chem. References2017, 47 , 138–153. [CrossRef][PubMed] 2. Sidell, F.R. Chemical Warfare Agents. In Military Preventive Medicine: Mobilization and Deployment; Kelley, P.W., 1. Ed.;Kangas, Textbooks M.J.; Burks, of Military R.M.; Medicine; Atwater, Borden J.; Lukowicz, Institute, R. WalterM.; Williams, Reed Army P.; Holmes, Medical A.E. Center: Colorimetric Washington, Sensor DC, USA,Arrays 2003; for Volumethe Detection 1, pp. 611–625,and Identificati ISBN 0-16-050500-3.on of Chemical Weapons and Explosives. Crit. Rev. Anal. Chem. 3. Witkiewicz,2017, 47, 138–153, Z.; Neffe, doi:10.1080/10408347.2016.1233805. S.; Sliwka, E.; Quagliano, J. Analysis of the Precursors, Simulants and Degradation 2. ProductsSidell, F.R. of Chemical WarfareWarfare Agents.Agents.Crit. In Military Rev. Anal. Preventive Chem. 2018Medicine, 48,: 337–371. Mobilization [CrossRef and Deployment][PubMed; ]Kelley, 4. Selby,P.W., R.C.Ed.; DRSKOTextbooks Project of RampingMilitary UpMedicine; Quickly Borden to Production. Institute, Available Walter online: Reed https://www.army.mil/Army Medical Center: article/131773/DRSKO_project_ramping_up_quickly_to_production/Washington, DC, USA, 2003; Volume 1, pp. 611–625, ISBN 0-16-050500-3. (accessed on 17 June 2016). 5.3. Veerabuthiran,Witkiewicz, Z.; S.; Neffe, Razdan, S.; Sliwka, A.K. LIDAR E.; Quagliano, for detection J. An ofalysis chemical of the and Precursors, biological Simulants warfare agents. and DegradationDef. Sci. J. 2011Products, 61, 241–250. of Chemical [CrossRef ] Warfare Agents. Crit. Rev. Anal. Chem. 2018, 48, 337–371, 6. Hill,doi:10.1080/10408347.2018.1439366. H.H., Jr.; Martin, S.J. Conventional analytical methods for chemical warfare agents. Pure Appl. Chem. 4. 2002Selby,, 74 R.C., 2281–2291. DRSKO Project [CrossRef Ramping] Up Quickly to Production. Available online: https://www.army.mil/ article/131773/DRSKO_project_ramping_up_quickly_to_production/ (accessed on 17 June 2016). 5. Veerabuthiran, S.; Razdan, A.K. LIDAR for detection of chemical and biological warfare agents. Def. Sci. J. 2011, 61, 241–250, doi:10.14429/dsj.61.556.

Sensors 2018, 18, 4291 8 of 8

7. Zhang, S.-W.; Swager, T.M. Fluorescent Detection of Chemical Warfare Agents: Functional Group Specific Ratiometric Chemosensors. J. Am. Chem. Soc. 2003, 125, 3420–3421. [CrossRef][PubMed] 8. Chen, L.; Wu, D.; Yoon, J. Recent Advances in the Development of Chromophore-Based Chemosensors for Nerve Agents and Phosgene. ACS Sens. 2018, 3, 27–43. [CrossRef][PubMed] 9. Suslick, K.S. An Optoelectronic Nose: “Seeing” Smells by Means of Colorimetric Sensor Arrays. MRS Bull. 2004, 29, 720–725. [CrossRef][PubMed] 10. Musto, C.J.; Suslick, K.S. Differential sensing of sugars by colorimetric arrays. Curr. Opin. Chem. Biol. 2010, 14, 758–766. [CrossRef] 11. Feng, L.; Musto, C.J.; Kemling, J.W.; Lim, S.H.; Suslick, K.S. A colorimetric sensor array for identification of toxic below permissible exposure limits. Chem. Commun. 2010, 46, 2037–2039. [CrossRef] 12. Kangas, M.J.; Burks, R.M.; Atwater, J.; Lukowicz, R.M.; Garver, B.; Holmes, A.E. Comparative chemometric analysis for classification of acids and bases via a colorimetric sensor array. J. Chemom. 2018, 32, e2961. [CrossRef] 13. Royo, S.; Martínez-Máñez, R.; Sancenón, F.; Costero, A.M.; Parra, M.; Gil, S. Chromogenic and fluorogenic reagents for chemical warfare nerve agents’ detection. Chem. Commun. 2007, 46, 4839–4847. [CrossRef] 14. Chulvi, K.; Gaviña, P.; Costero, A.M.; Gil, S.; Parra, M.; Gotor, R.; Royo, S.; Martínez-Máñez, R.; Sancenóna, F.; Vivancos, J.-L. Discrimination of nerve gases mimics and other organophosphorous derivatives in gas phase using a colorimetric probe array. Chem. Commun. 2012, 48, 10105–10107. [CrossRef][PubMed] 15. Costero, A.M.; Gil, S.; Parra, M.; Mancini, P.M.E.; Martínez-Máñez, R.; Sancenón, F.; Royo, S. Chromogenic detection of nerve agent mimics. Chem. Commun. 2008, 45, 6002–6004. [CrossRef][PubMed] 16. Dale, T.J.; Rebek, J., Jr. Fluorescent Sensors for Organophosphorus Nerve Agent Mimics. J. Am. Chem. Soc. 2006, 128, 4500–4501. [CrossRef][PubMed] 17. Foster, L.S.; Gruntfest, I.J. Demonstration experiments using universal indicators. J. Chem. Educ. 1937, 14, 274–276. [CrossRef] 18. Nomeir, A.A.; Burka, L.T.; Matthews, H.B. Analysis of Dimethyl Hydrogen Phosphite and Its Stability under Simulated Physiological Conditions. J. Anal. Toxicol. 1988, 12, 334–338. [CrossRef][PubMed] 19. Abramoff, M.D.; Magalhaes, P.J.; Ram, S.J. Image Processing with ImageJ. Biophotonics Int. 2004, 11, 36–42. 20. Soldat, D.J.; Barak, P.; Lepore, B.J. Microscale Colorimetric Analysis Using a Desktop Scanner and Automated Digital Image Analysis Douglas. J. Chem. Educ. 2009, 86, 617–620. [CrossRef] 21. R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2017. 22. Available online: http://github.com/vqv/ggbiplot (accessed on 5 December 2011). 23. Salles, M.O.; Meloni, G.N.; de Aaujo, W.R.; Paixão, T.R.L.C. Explosive colorimetric discrimination using a smartphone, paper device and chemometrical approach. Anal. Methods 2014, 6, 2047–2052. [CrossRef] 24. Bang, J.H.; Lim, S.H.; Park, E.; Suslick, K.S. Chemically Responsive Nanoporous Pigments: Colorimetric Sensor Arrays and the Identification of Aliphatic Amines. Langmuir 2008, 24, 13168–13172. [CrossRef] 25. Kangas, M.J.; Wilson, C.L.; Burks, R.M.; Atwater, J.; Lukowicz, R.M.; Garver, B.; Mayer, M.; Havenridge, S.; Holmes, A.E. An Improved Comparison of Chemometric Analyses for the Identification of Acids and Bases with Colorimetric Sensor Arrays. Int. J. Chem. 2018, 10, 36–55. [CrossRef] 26. Kangas, M.J.; Lukowicz, R.; Atwater, J.; Pliego, A.; Al-Shdifat, Y.; Havenridge, S.; Burks, R.; Garver, B.; Mayer, M.; Holmes, A.E. Printed Colorimetric Arrays for the Identification and Quantification of Acids and Bases. Anal. Chem. 2018.[CrossRef][PubMed] 27. Askim, J.R.; Mahmoudi, M.; Suslick, K.S. Optical sensor arrays for chemical sensing: The optoelectronic nose. Chem. Soc. Rev. 2013, 42, 8649–8682. [CrossRef][PubMed] 28. Okuom, M.O.; Wilson, M.V.; Jackson, A.; Holmes, A.E. Intermolecular interactions between eosin Y and caffeine using 1H-NMR spectroscopy. Int. J. Spectrosc. 2013, 2013, 245376. [CrossRef][PubMed] 29. Okuom, M.O.; Holmes, A.E. Developing a Color-Based Molecular Sensing Device: DETECHIP®. Sens. Transducers 2014, 183, 30–33. 30. Ellman, G.L. Tissue sulfhydryl groups. Arch. Biochem. Biophys. 1959, 82, 70–77. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).