Analyst Identification of accelerants, fuels and post -combustion residues using a colorimetric sensor array Journal: Analyst Manuscript ID: AN-ART-04-2015-000806.R1 Article Type: Paper Date Submitted by the Author: 14-Jul-2015 Complete List of Authors: Li, Zheng; University of Illinois at Urbana-Champaign, Jang, Minseok; University of Illinois at Urbana-Champaign, Askim, Jon; University of Illinois at Urbana-Champaign, Suslick, Kenneth; University of Illinois at Urbana-Champaign, Chemistry Page 1 of 7 Please do notAnalyst adjust margins 1 2 3 4 Analyst 5 6 7 ARTICLE 8 9 10 11 Identification of accelerants, fuels and post-combustion residues 12 13 using a colorimetric sensor array 14 Received 00th January 20xx, a a a a Accepted 00th January 20xx Zheng Li, Minseok Jang, Jon R. Askim and Kenneth S. Suslick* 15 16 DOI: 10.1039/x0xx00000x A linear (1x36) colorimetric sensor array has been integrated with a pre-oxidation technique for detection and identification of a variety of fuels and post-combustion residues. The pre-oxidation method permits the conversion of fuel 17 www.rsc.org/ 18 vapor into more detectable species and therefore greatly enhances the sensitivity of the sensor array. The pre-oxidation 19 technique used a packed tube of chromic acid on an oxide support and was optimized in terms of the support and 20 concentration. Excellent batch to batch reproducibility was observed for preparation and use of the disposable pre- 21 oxidation tubes. Twenty automotive fuels including gasolines and diesel from five gasoline retailers were individually 22 identifiable with no confusions or misclassifications in quintuplicate trials. Limits of detection were at sub-ppm 23 concentrations for gasoline and diesel fuels. In addition, burning tests were performed on commonly used fire accelerants, 24 and clear differentiation was achieved among both the fuels themselves and their volatile residues after burning. 25 dog’s responses; in addition, training dogs requires substantial 26 Introduction time and effort. 22 Some commercialized hydrocarbon gas 27 analyzers can detect and quantify flammable accelerants by Fire incidents, both accidental and malicious, have become a 28 vapor sampling, but are unable to identify specifically which pressing issue in modern life due to their threat to human life, 29 accelerant is present. 23, 24 Numerous other detection methods property, and environmental safety. According to reports from 30 still suffer from high cost, low sensitivity, lack of the US Fire Administration, over 1.5 million fires occurred 31 reproducibility, interference from humidity, or changeable throughout the US in 2013 which caused over 3000 deaths, 25, 26 32 1 responses due to sensor aging. For these reasons, the 17,000 injuries and $10 billion in property damage. 33 11 Automotive fuels and other petroleum products such as development a high-performance portable sensor for the on- 34 gasoline, diesel, and kerosene are commonly employed as site analysis of fire accelerants or quality control of fuels 35 accelerants in case of arson; rapid discrimination among remains an important goal. 36 accelerants is therefore particularly important for fire scene In the past decade, the use of disposable colorimetric sensor 37 2-7 arrays (CSAs) has been developed for a variety of vapor investigation. Additionally, the need for simple field- 27-29 38 deployable quality control of automotive fuels has drawn great analyses. CSAs use strong chemical interactions between 39 attention because of the negative effects caused by the the analytes and a diverse set of cross-responsive 40 adulteration of gasoline or diesel (e.g., engine damage and air chemoresponsive dyes; digital imaging of the arrays permits 41 8, 9 identification of a composite pattern of response as the pollution), and the fuel oil industry has suffered from 27-33 42 fraudulent mixing of low-priced reagents with higher-priced “fingerprint” for a given odorant. These arrays take 43 fuels. 10 advantage of plasticized films or organically modified siloxanes 44 Currently, the detection of fire accelerants is generally (ormosils) as matrices for colorants whose color changes are affected by polarity/dipolarity, Br ønsted and Lewis acid-base 45 determined by standard analytical methods including 27-29 46 electrochemistry, 9, 11 fluorescence, 12 Fourier transform interactions, redox reactions, and π-π interactions. 13-15 2, 16, 17 Although colorimetric sensor arrays perform well for a 47 infrared spectroscopy (FTIR), Raman spectroscopy, 34-36 48 GC 13, 18 or GC–MS, 19-21 most of which demand non-portable variety of gases and volatile liquids, they have not shown high sensitivity to less-reactive analytes, such as aliphatic and 49 and expensive instrumentation. Canine teams offer a more aromatic hydrocarbons or halocarbons. 30, 37 A typical gasoline 50 easily fielded approach for detecting accelerants in fire debris, consists of 30-50% alkanes, 5-10% alkenes and 20-40% 51 though results are less reliable than traditional analytical methods as they are subject to human interpretation of a aromatics and therefore does not respond to a sensor array 52 designed for strong chemical interactions. We have previously 53 shown that substantial improvements in the detection, 54 identification, and quantitation of less-reactive volatiles can be 55 made by employing a pre-treatment technique in which the 56 analyte gas stream is subjected to partial oxidation and thus 57 58 59 60 This journal is © The Royal Society of Chemistry 20xx J. Name ., 2013, 00 , 1-3 | 1 Please do not adjust margins Please do notAnalyst adjust margins Page 2 of 7 ARTICLE Journal Name 1 2 converted into more easily detected oxidation products (e.g., 3 aldehydes and carboxylic acids). 37 We demonstrate here that 4 this technique can be extended for the identification of 5 complex fuel mixtures and have examined discrimination 6 among a large number of commercial fuels, differentiating 7 among both the fuels themselves and their volatile residues 8 after burning. 9 10 11 Experimental Fig. 1 Photographs of the colorimetric sensor array used for fuel detection. (a) Linear 12 Chemicals and Materials 13 36-spot colorimetric sensor array containing metalloporphyrins, acid- or base-treated 14 For the gasolines used in these experiments, we provide their pH indicators, solvatochromic/vapochromic and metal-containing dyes; (b) colorimetric brand name and average octane number (ON = (R+M)/2, sensor array mounted in an aluminum holder with an o-ring placed in a groove and a 15 glass slide cover in place; this provides a nearly ideal flow path for the analyte stream where R is the research octane number and M is the motor with a flow volume of ~85 µL. 16 38 17 octane number). Three different grades of gasoline (i.e., regular, ON87; plus, ON89; and premium, ON93) and diesel 18 Analyte Vapor Generation fuel were purchased from five local gasoline distributors (i.e., 19 Mobil, Marathon, Shell, BP and Schnucks). Ethanol, i-propanol, Analyte flow streams were produced by bubbling dry nitrogen 20 kerosene, mineral oil, aluminum oxides (Brockmann I, Sigma- through the liquid fuels (ESI†, Fig. S2A), or by flushing dry 21 Aldrich), silica gels (Davisil, Sigma-Aldrich) and all other nitrogen over a carpet sample (2.5x2.5 cm nylon carpet 22 reagents were of analytical-reagent grade and used without samples loaded with 1 mL of accelerant with or without 23 further purification unless otherwise specified. Lubricant (WD- burning for 1 min, as shown in ESI†, Fig. S2B). The resulting 24 40 type 110071), vegetable oil (Great Value) and nylon carpet vapor streams were then mixed with a diluting stream of dry 25 (Guardian, platinum series) were purchased from a local and wet nitrogen to attain the desired concentrations at 50% 26 supermarket. relative humidity (RH) by using MKS digital mass flow 27 controllers (MFCs). For all the experiments performed in this 28 Formulations, Preparation and Sensor Array Printing study, the flow rate was 500 sccm. The response of the sensor 29 array essentially reaches equilibrium during the first minute Sol-gel pigments were prepared as previously described. 39, 40 30 and is not dependent (after equilibration) on flow rate or dose. Sol-gel formulations were obtained via the acid catalyzed All data was compared after equilibration after 1 min exposure 31 hydrolysis of silane precursors (e.g., mixtures of to the analyte flow. 32 tetraethoxysilane (TEOS), methyltriethoxysilane (MTEOS), 33 octyltriethoxysilane (octyl-TEOS)). The resulting ormosil Digital Imaging and Data Analysis 34 formulations after hydrolysis were added to the 36 selected 35 dyes (ESI†, Table S1) and then loaded into a 36-hole Teflon For all sensing experiments, sensor arrays were imaged on a 36 inkwell. Sensor arrays were printed on a robotic microarray flatbed scanner (Epson Perfection V600). The array was 37 printer (Arraylt Co., Mountain View, CA) by dipping slotted equilibrated with 50% RH nitrogen for 1 min at a flow rate of 38 pins into the inkwell and delivering the formulation (~100 nL) 500 sccm to capture the before-exposure image, and after- 39 to a polyvinylidene difluoride (PVDF) membrane (Fig. 1a). Once exposure image was acquired after 1 min exposure to the fuel or post-combustion vapor at 500 sccm. Difference maps were 40 printed, the arrays were stored in a N 2-filled glove bag for 41 three days. Each array was then cut into strips and mounted in obtained by subtracting the red, green, and blue (RGB) values 42 a custom made aluminum flow cell, with channel dimensions of before-exposure images from those of after-exposure 43 of 3.0 × 0.5 × 57 mm (Fig. 1b). A Viton o-ring is placed in a images; each sensor spot was ~100 pixels, the values of which 44 groove around the channel and a standard glass microscope were averaged.