B American Society for Mass Spectrometry, 2016 J. Am. Soc. Mass Spectrom. (2017) 28:1048Y1059DOI: 10.1007/s13361-016-1562-2

FOCUS: HONORING R. G. COOKS' ELECTION TO THENATIONAL ACADEMY OF SCIENCES: RESEARCH ARTICLE

Analytical Validation of a Portable Mass SpectrometerFeaturing Interchangeable, Ambient Ionization Sourcesfor High Throughput Forensic Evidence Screening

Zachary E. Lawton,1 Angelica Traub,1 William L. Fatigante,1 Jose Mancias,1

Adam E. O’Leary,1 Seth E. Hall,1 Jamie R.Wieland,2 Herbert Oberacher,3 Michael C. Gizzi,4

Christopher C. Mulligan1

1Department of Chemistry, Illinois State University, Normal, IL 61790, USA2Department of Management and Quantitative Methods, Illinois State University, Normal, IL 61790, USA3Institute of Legal Medicine and Core Facility Metabolomics, Innsbruck Medical University, Innsbruck, Austria4Department of Criminal Justice Sciences, Illinois State University, Normal, IL 61790, USA

Abstract. Forensic evidentiary backlogs are indicative of the growing need for cost-effective, high-throughput instrumental methods. One such emerging technology thatshows high promise in meeting this demand while also allowing on-site forensicinvestigation is portable mass spectrometric (MS) instrumentation, particularly thatwhich enables the coupling to ambient ionization techniques. While the benefits ofrapid, on-site screening of contraband can be anticipated, the inherent legal implica-tions of field-collected data necessitates that the analytical performance of technol-ogy employed be commensurate with accepted techniques. To this end, comprehen-sive analytical validation studies are required before broad incorporation by forensicpractitioners can be considered, and are the focus of this work. Pertinent perfor-

mance characteristics such as throughput, selectivity, accuracy/precision, method robustness, and ruggednesshave been investigated. Reliability in the form of false positive/negative response rates is also assessed,examining the effect of variables such as user training and experience level. To provide flexibility toward broadchemical evidence analysis, a suite of rapidly-interchangeable ion sources has been developed and character-ized through the analysis of common illicit chemicals and emerging threats like substituted phenethylamines.Keywords:Analytical validation, Ambient ionization, Portablemass spectrometer, Forensics, Forensic evidence,Desorption electrospray ionization, Paper spray ionization, Atmospheric pressure chemical ionization,SWGDRUG

Received: 1 September 2016/Revised: 14 November 2016/Accepted: 16 November 2016/Published Online: 20 December 2016

Introduction

Controlled substance analysis plays a major role in both thetypical workload and the substantial evidence backlog that

burdens the public forensic laboratory system and impedesongoing criminal investigations and judicial processing [1, 2].A review of the 2009 Census for Publicly Funded Forensic

Labs from the Bureau of Justice Statistics [3] shows thatcontrolled substance requests were second only to forensicbiology (e.g., serological screening, DNA analysis) in magni-tude, accounting for 33% of all work requests. For county andmunicipal labs, drug evidence attributed to the highest work-load, highlighting the burden of this routine casework at thelocal level. Of the overall backlog of evidence reported (i.e., ~1.2 million requests), forensic biology accounted for ~75%,with controlled substance attributing to a more modest 12%.Though novel training [4] and funding initiatives [5] havehelped to contend with the demand, forensic biology has prov-en to be indispensable for suspect identification, leading to asignificant level of outsourcing to private labs [6, 7]. While the

Electronic supplementary material The online version of this article (doi:10.1007/s13361-016-1562-2) contains supplementary material, which is availableto authorized users.

Correspondence to: Christopher C. Mulligan; e-mail: mulligan@ilstu.edu

development of higher performance analytical methods is seenas a means of accommodating the increasing number of theserequests [8, 9], so too are techniques that reduce the impact ofother types of forensic evidence [5, 10], allowing the realloca-tion of resources towards the more laborious biologicalanalyses.

Given its implications on the typical lab workload, increas-ing the throughput of routine drug evidence analysis is ofinterest. While established laboratory-based methods for druginvestigation (e.g., hyphenated mass spectrometric (MS), spec-troscopy, etc.) are known for their accuracy, broad applicabil-ity, and court admissibility [8, 11, 12], throughput is hinderedby the required preparative steps and overall duty cycle. Am-bient MS techniques [13, 14], which increase sample through-put by forgoing extensive sample preparations, have shownproficiency in forensic chemical analysis [15–17], with recentreports involving the analysis of biofluids [18, 19], explosives[20], adulterated foodstuffs [21], mind-altering plant-basedevidence [22], and simultaneous molecular/elemental compo-sition [23]. Specific techniques like desorption electrosprayionization (DESI) [24], direct analysis in real time (DART)[25], and paper spray ionization (PSI) [26, 27] have shownutility in both trace and bulk drug evidence analysis, particu-larly in combatting the rise of Bdesigner drugs^ evidence andaccommodating associated paraphernalia [28].

Field testing offers an inherent increase in throughput byalleviating the need for evidence transport to off-site forensiclaboratories, while also providing timely information to lawenforcement and forensic practitioners for establishing criminalintent and expediting investigations. Field-based processing ofchemical evidence has been mostly relegated to presumptivecolorimetric testing [29, 30], and while stalwart examples likethe Scott (cocaine) and Marquis (opiates, amphetamines) testsare commonly implemented, they have elevated false positiverates due to circumstantial ambiguity in observed color changes[31], and lack of overall chemical specificity [32]. Severalportable detection technologies have been recently investigatedfor forensic analysis, including Raman spectroscopy [33], nearinfrared spectroscopy (NIR) [34], and X-ray fluorescence mi-croscopy (XRF). Several portable MS systems featuring mem-brane introduction or GC separations have been reported [35,36], including the novel application of vehicle-mounted systemsfor spatial location of clandestine methamphetamine labs viavolatile effluent mapping by Verbeck and co-workers [37], andalthough commercial products are available [38], they have yetto be broadly incorporated into forensic investigation and havepreparative constraints similar to their lab-based counterparts.

The pursuit of the high throughput drug analysis has led tothe coupling of ambient ionization methods to field-portableMS instrumentation [39, 40], allowing rapid and flexiblescreening of controlled substances and other contraband whileenjoying the discriminating power of MS analysis. Reportedapplications demonstrate the utility of specific ambient MStechniques in this arena, allowing the direct analysis of utilizedsurface swabs [24], trace residues [41], bulk evidence [42], andhighly complex sample matrices [26, 43]. Furthermore,

implementing automated spectral library searching and chem-ical identification on said systems can alleviate the need forfield-based data interpretation, allowing the use by nontechni-cal operators [44].

While the benefit of rapid, on-site screening of contrabandvia can be anticipated (e.g., expedited criminal investigations,decreased chain-of-custody and sample degradation issues,reduced evidence load on off-site laboratories), adoption of thistechnology will be contingent on its overall analytical perfor-mance given the inherent legal ramifications of collected data.To this end, validation studies that follow standard practices[45] and recommendations from gatekeepers like the ScientificWorking Group for the Analysis of Seized Drugs(SWGDRUG) [46] are required to ensure subsequent courtadmissibility. While broad validation efforts have been under-taken for ambient ionization methods conducted on lab-scaleMS instrumentation, such as the work of Gurdak and co-workers at National Physical Laboratory (NPL) in assessinginter-laboratory repeatability and constancy of DESI-MS [47],similar efforts utilizing portable MS systems for forensic ap-plications have been limited to categorical assessments likelimits of detection (LOD) and linearity [48, 49]. Cooks andco-workers have assessed LODs for explosives residues [50,51] and drugs of abuse [52] using their lineage of miniaturizedMS systems, as well as demonstrated quantitative toxicologicalanalysis of biofluids [53, 54]. Other reports have evaluatedselectivity and throughput of drug evidence screening usingcommercially available systems [24, 42].

Herein, we report an extensive analytical validation of aportable MS system featuring interchangeable, ambient ioniza-tion sources for on-site drug evidence screening. FollowingSWGDRUG recommendations [46], specific performancecharacteristics assessed included selectivity, accuracy/preci-sion, method robustness, ruggedness, and detection limit. Re-liability in the form of false positive/negative response ratesdetermined from large datasets are reported, examining theeffect of user training, experience level, and environmentalfactors stemming from field usage. Furthermore, the utilityand throughput afforded from the implementation of rapidlyinterchangeable, ambient ionization sources is investigated as ameans to combat the rigor and variable nature of crime sceneprocessing

ExperimentalSamples and Sample Preparation

For these studies, analytical standards of target analytes werepurchased from Cerilliant Corp. (Round Rock, TX, USA) in1000 ppm (1.0 mg/mL) concentrations, and stock solutions ofknown concentration were prepared via serial dilution in meth-anol. To produce surface residues of known mass, 1 μL ali-quots of these known composition solutions were spotted ontosubstrates of interest and allowed to dry. Limit of detection(LOD) studies were performed from four substrates of forensicinterest: glass (e.g., microscope slide), a brass key, a plastic bag

Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening 1049

(polyethylene), and textured laminate countertop purchasedfrom local distributors.

Solutions were analyzed as-is for ESI-MS analysis or spot-ted onto porous Teflon well slides (Prosolia Inc., IndianapolisIN, USA), dried, and subsequently analyzed for DESI-MS. ForPSI analysis, MQuant paper-based testing strips (EMDMillipore Corp., Billerica, MA, USA) with one end cut intoan isosceles triangle served as the ionization substrate. Thesetesting strips feature a plastic backing behind the paper sub-strate, which provides rigidity for direct, dip sampling of con-densed phases or surface swabbing with minimal deformationof the triangular shape, which is critical for spray quality andduration during analysis. For surface swabbing, the PSI sub-strate is pre-wetted with 2 μL of methanol to enhance recovery,used to probe the surface of interest, and then directly analyzed.All spray-based ionization sources utilized a spray solvent of1:1 methanol/water with 0.1% formic acid.

Portable MS System and BPlug and Play^ StyleIonization Sources

For these studies, a FLIR Systems AI-MS 1.2 cylindrical iontrap (CIT) mass spectrometer (FLIR Mass Spectrometry, WestLafayette, IN, USA) ruggedized for field use was implementedfor all data collection and validation studies. As previouslyreported [24, 26, 44], this system features a capillary-basedatmospheric pressure inlet for coupling to ambient ionizationmethods, while providing analyte confirmation via MS/MS.Cartridge-based helium CID damping gas, syringe pump sol-vent delivery, and the high voltage supply needed for allinvestigated ionization sources are incorporated into the instru-mental design. Nebulizing gas for ESI and DESI is supplied bya small self-contained breath apparatus (SCBA) tank duringfield use. Dimensions (60 × 50 × 40 cm, L × W × H) andweight (~45 kg) of this field-ready system are well-suited foron-site criminal investigations and CSI applications.

For maximum flexibility towards the broad forensic evi-dence screening, a centralized mounting/positioning systemthat allows rapid interchangeability of a suite of traditionaland ambient ionization sources was constructed for use on theAI-MS 1.2; further detail on the design and construction of therail system and associated ionization sources can be found inthe Supplementary Material. Each modular ionization sourcewas designed for quick coupling/disconnection to instrumentalvoltage and solvent lines and minimal user manipulation tosimplify overall operation for nontechnical users. Ion sourceswere selected to enable the rapid screening of condensedphases, residues, and gas-phase species, including ESI andDESI via a factory-mounted, dual function spray chamber,PSI, paper cone spray ionization (PCSI) [55], and direct airsampling-atmospheric pressure chemical ionization (APCI).The included sources required no external power requirementsother than that built-in to AI-MS 1.2 system. For brevity,explicit experimental detail for each ionization method can befound in the Supplemental Material.

Comparison to a Standard Mass Spectral ReferenceLibrary

To assess selectivity of the AI-MS 1.2 in identifying forensicanalytes, base MS and MS/MS spectra data were comparedwith both recent literature and a widely accepted referencelibrary, the ‘Wiley Registry of Tandem Mass Spectral Data,MSforID’ (Wiley: Hoboken, NJ, USA) [56]. The library wasdeveloped on QqTOF instrumentation (Qstar XL; AB Sciex)using ESI in positive and negative ion mode, with the detailedinstrumental parameters utilized reported previously [57, 58].To date, the published version of the library contains 12,122spectra of 1208 compounds. The library version used in thispublication covered more than 1700 entries.

Pertinent MS/MS spectral information (i.e., precursor ionm/z, fragment ion(s)m/z, relative ion intensities) was submittedfor compound identification with the MSforID library searchprogram [58, 59]. The search algorithm determines the degreeof similarity between a sample spectrum and library spectra,expressed as a ‘relative average match probability’ (ramp).High compound-specific ramp-values indicate high similaritybetween the unknown and the reference compound, and thesubstance with the highest ramp-values is considered to repre-sent the unknown compound. The correctness of each puta-tively positive match was then checked by expert reviewing.

Results and DiscussionThe validation plan implemented for the FLIR AI-MS 1.2 wasconstructed to include performance characteristics delineatedin recent SWGDRUG recommendations [46] for seized druganalysis methodologies. Specific categories incorporated werethroughput, selectivity of analyte identification, accuracy/pre-cision (i.e., repeatability, inter-user reproducibility, and errorrate), method robustness, environmental ruggedness, anddetection limit. Further description of each validation categorycan be found in Figure S-1 of the Supplemental Materials.

Throughput and Utility of BPlug and Play^ StyleIonization Sources

Ambient MS methods reported to date are diverse in terms ofthe analyte desorption and ionization mechanisms and minorpreparative steps that are employed [60]; therefore, certaintechniques are inherently better suited for specific analysisscenarios (e.g., analytes present, complexity and phase of ma-trix, surface geometry). For instance, the paper substrateemployed in PSI-MS makes it naturally applicable to surfaceswabbing [26], and flexibility of positioning for DART-MS isconvenient for rapid introduction of solid samples [42]. Tomaximize the utility of portable MS screening of forensicevidence, a centralized mounting/positioning system that al-lows rapid interchanging and operation of a suite of traditionaland ambient ion sources (i.e., ESI, DESI, PSI, PCSI, and APCI)was developed and assessed in terms of utility and throughput;the mounting system and constructed ion sources are detailed

1050 Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening

in Supplementary Figure S-1. The interchangeable aspect ofthese sources allows the user to select the optimal ionizationmethod for the evidence in question, while allowing broadcoverage of solid, liquid, and gas-phase evidence types.

Figure 1a depicts the throughput of interchanging betweenthe simplified ion sources in order to process differing evidencetypes. Increases in the total ion count denote the usage of eachion source for a minimum of 30 s, and a return to baselinedepicts the time required for source switching, coupling ofsolvent and voltage connections, and needed positioning.Overall, five distinct sample/source combinations were ableto be processed in less than 6 min total, generating character-istic spectral data for each scenario, as seen in Figure 1bthrough f. Ambient MS data collected upon the FLIR AI-MS1.2 commonly exhibits significant in-source fragmentation(i.e., inadvertent fragmentation prior to initiating MS/MS scanmodes) [24, 44], similar to other portable [61] and lab-scaleMSinstrumentation [62] capable of sampling externally-generatedions. This in-source fragmentation can be varied, but rarelyeliminated, by manipulating ion optic potentials in the highpressure regions of the vacuum system (e.g., tube lens/skimmerassembly) [63] and minimizing thermal dissociation (e.g., inletcapillary temperature) [64]. Specific detail regarding the in-source fragmentation pathways observed can be seen in theSupplemental Materials. Of note, the high spectral intensity forthe PSI analysis required a 30× demagnification to scale cor-respondingly with the other data. The higher sensitivity of PSI-MS is attributed to more efficient collection of analyte ions dueto its coaxial positioning relative to the MS inlet capillary; thisis further supported by the comparative detection limit studiessummarized in Table 4. Given the rigor and variable nature ofcrime scene processing, the throughput and applicability dem-onstrated could be well suited for rapidly establishing probativevalue of evidence at hand.

Selectivity of Analyte Identification

The FLIR AI-MS 1.2 and other trap-based portable MS sys-tems [41, 42] have demonstrated high spectral congruencywhen compared with lab-grade instrumentation for most tradi-tional forensic chemicals [24] and, so, enjoy the selectivityafforded to MS analysis. Incorporation of tandem MS data inanalyte identification even allows discernment from complexmatrices [26, 43]. Previous work towards establishing spectralcongruence and selectivity compared blind portable MS/MSdata to a widely accepted reference library, the ‘Wiley Registryof Tandem Mass Spectral Data, MSforID’, returning the cor-rect identification for over 30 forensic analytes and evidentiarysamples [44]. While this shows selectivity towards traditionaldrugs of abuse, extension towards novel psychoactive sub-stances (NPS) [28], particularly those that are structurally sim-ilar and even isomeric, must be assessed.

To determine spectral congruence and selectivity towardsnovel illicit drugs, MS/MS spectra for a selection of substitutedphenethylamines and N-(2-methoxy)benzyl (NBOMe) deriva-tiveswas compared to reference spectra contained in theMSforID

database. Library search results returned the most probable iden-tification, reported in Table 1, and a representative visual com-parison of sample and reference spectra for 2C-T2 can be seen inSupplementary Figure S-7. Although a majority of analytes werecorrectly identified, a few discrepancies are observed that serve toexemplify potential issues with NPS analysis and the comparisonof spectral data between mass analyzer types. The incorrectmatches for both 2C-D (Supplementary Figure S-8A) and 2C-T4 (Supplementary Figure S-8B) are other isomeric psychoactivecompounds that have proven difficult to discern even on highperformance LC-MS instrumentation [65], as they yield simplis-tic MS/MS data that is distinguishable only by minor differencesin relative abundance. For instance, MS/MS for chain isomers2C-T4 and 2C-T7 is marked only by the loss of an amidogenradical (17 Da).

Ion trap instrumentation is known to produce a low numberof compound-specific fragments, as resonant excitation-collision induced dissociation (CID) commonly produces prod-uct ions too cool to undergo further fragmentation and low-mass fragment ions are inefficiently trapped [66]. The combi-nation of inter-analyzer differences, variable CID energies anda low number of compound-specific transitions has beenshown to hinder accurate identification via library searching[67], which led to the misidentification of 25I-NBOMe.

Of note, MS/MS spectra obtained fromNBOMe derivativesusing the CIT of the AI-MS 1.2 demonstrated much highercomplexity compared to that contained in theWiley Registry ofTandem MS Data (Supplementary Figure S-8C), althoughthere have been similar reports in recent literature [28, 68,69]. It has been asserted that the addition of the N-(2-methoxy)benzyl group to the 2C-phenethylamine structure(another electron donating aromatic ring, as seen in Supple-mentary Figure S-8C) creates more favorable sites for dissoci-ation that can occur along the –C-C-N-C- linkage chain, spe-cifically C–C bond cleavage between the α- and β-carbonatoms on the ethylene bridge [68, 69]. The complexity andintensity to which these fragmentation patterns are seen ishighly dependent on the MS/MS conditions utilized [44],which leads to the inter-instrument variations reported [66,70]. While NBOMe derivatives may represent a special casedue to the number of low abundance transitions seen,instrument-specific spectral databases [71] may be more pru-dent in scenarios where automated chemical identification isneeded, but variability is a concern.

Assessment of Accuracy and Precision

Overall Instrument and User-Specific Error Rates To ensurea statistically-relevant population of data for determination oferror rate, a dataset consisting of over 1400 replicates of bothpositive and negative control samples was collected and ana-lyzed. For these studies, cocaine was selected as the analyte ofinterest; Bpositive^ control samples were comprised of 200 ngof cocaine, deposited directly onto a MQuant testing strip (forPSI analysis) or a glass slide (for DESI analysis) and directlyanalyzed, whereas Bnegative^ control samples involved the

Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening 1051

same substrate with no cocaine present. A blank sample (iden-tical substrate) was tested between each control to assess car-ryover and hygiene. In these studies, a cocaine Bdetection^signified the presence of both the precursor ion for cocaine,[M + H]+ (m/z 304), in base MS spectra and characteristicfragment (m/z 182) in subsequent MS/MS spectra observed at

an intensity of at least three times the signal-to-noise level. Them/z 182 fragment, although characteristic, was the only transi-tion observed for the cocaine precursor, serving as the identi-fying ion signature. Table 2 provides the detection and falsepositive rates attained in this study, controlling for the ioniza-tion source utilized. PSI was selected for a more intensive

0

70000

140000

0.00 1.00 2.00 3.00 4.00 5.00 6.00

Source Swap = *

* * * *÷ 30

yti

sn

et

nI.

sb

A

Time (min)

Desorption

Electrospray

Ionization (DESI)

Paper Spray

Ionization (PSI)

Paper Cone Spray

Ionization (PCSI)

Electrospray

Ionization (ESI)

Atmospheric

Pressure Chemical

Ionization (APCI)

Coleman Fuel

(Vapor)

MDMA

(400 ng residue)

25I-NBOMe

(400 ng swabbed

from surface)

Amphetamine tablet

(~10 mg crushed)

Cocaine

(1 ppm in MeOH)

0

20

40

60

80

100

50 125 200 275 350 425

304

182

150

[Cocaine + H]+

0

20

40

60

80

100

50 125 200 275 350 425 500

121

428

[25I-NBOMe + H]+

0

20

40

60

80

100

50 100 150 200 250

113

127

141

0

20

40

60

80

100

50 100 150 200 250

91

119

136

[Amphetamine + H]+

0

20

40

60

80

100

50 125 200 275 350 425

163

194

[MDMA + H]+

yti

sn

et

nIe

vit

ale

R

(a)

(b) (d)

(c) (e)

(f)

m/z

Figure 1. (a) Total ion chromatogram depicting five successive, yet discrete, ion source/sample combinations, with the rise in ionsignal corresponding to the onset and duration of said analysis. A return to baseline represents the time needed to physicallyinterchange between sources, including reattachment of voltage and solvent lines, if needed. Completing five diverse experiments inless than 6 min eludes to the throughput and utility of the instrumental platform; (b)–(f) corresponding MS data collected during saidanalyses

1052 Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening

study, as its source design incorporates flexibility in user posi-tioning and, consequently, has a higher susceptibility for usererror over the static nature of the DESI source.

User-specific variables are also of interest, as educationalbackground and level of on-instrument training could varydepending on which first responder groups utilize said instru-mentation. An educationally diverse selection of users waschosen for this study, including M.S. in Chemistry candidates,B.S. in Chemistry undergraduates (with and without formalanalytical chemistry coursework), Ph.D. (Analytical Chemis-try) candidates, high school students with only introductorychemistry knowledge, and a recent police academy graduatewith little coursework in the general physical sciences.Instrument-specific experience was controlled by examiningerror rates during and after the initial training period.

When examining the error rate breakdowns shown in Ta-ble 2, better outcomes can be realized by users following the

formal training period and after instrument-specific experienceis gained. This can be observed in Figure 2a and b, whichdepict the average and range of maximum peak heights ob-served for replicate experiments obtained during and afterformal training sessions, as well as the daily sample throughput(i.e., min/sample); for this, the observed datapoint representsthe average value from replicate analyses, with the providedbar depicting the maximum and minimum intensity observedfor that day, and daily sample throughput (min/sample) isindicated above each range. The effect of cumulative experi-ence can be seen herein, particularly in regards to samplethroughput, but also to a lesser extent for the daily averageand maximum spectral intensity obtained. As seen in Figure 2a,the high school student user struggled to reproducibly acquirehigh intensity data shortly after the training period, but didtrend to better outcomes after several days of continued usage.The benefit of past experience with chemical instrumentation

Table 1. Summary of Library Searching Results Obtained for FLIR AI-MS 1.2 MS/MS Data

Spectral Data Submitted MSforID Match

2C-D 2,5-DimethoxyamphetamineE-C2E-C2P-C2P-C2C-C2C-C2B-C2B-C2I-C2I-C2

2C-T2 2C-T2 2C-T4 2C-T4 2C-T7 2C-T4

25C-NBOMe Nybomycin 25I-NBOMe 25I-NBOMe

Table 2. FLIR AI-MS 1.2 Error Rates, Controlling for Ion Source, User Experience, and Education Level

Detection rate (%)a False positive rate (%)b

Overall FLIR AI-MS 98.87 (1412) 0.14 (1412)PSI-MS 99.01 (1212) 0.17 (1212)DESI-MS 98.80 (200) 0.00 (200)

Breakdown of PSI-MS error rateTrained user operationGraduate students 99.84 (610) 0.00 (610)Undergraduate students 99.20 (250) 0.00 (250)High School graduate 97.96 (245) 0.41 (245)

During training periodUndergraduate and High School students 94.81 (77) 0.00 (77)

Other users (untrained)Ph.D. candidate (Analytical Chemistry) 100.00 (10) 0.00 (10)Undergraduate w/o AC coursework 100.00 (10) 0.00 (10)Police Academy graduate 100.00 (10) 10.00 (10)

Breakdown of DESI-MS error rateTrained user operationGraduate students 98.00 (200) 0.00 (200)

aDetection rate = percentage of positive samples analyzed that were accurately detected (i.e., true positive rate).bFalse positive rate = percentage of negative samples that returned an incorrect detection.Sample size, n, given in the parentheses.

Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening 1053

through pertinent coursework and hands-on usage is seen bythe comparable data obtained during and after training for anundergraduate Chemistry major user (Figure 2b). This data, incontrast, is marked by fairly reproducible average and maxi-mum signal intensities obtained directly after formal training;similar observations were made for other users in this demo-graphic. Sample throughput can also be considered as a metricof overall comfort with the methodology, and both user groupsexhibited higher throughputs as the level of method-specificexperience increased. Further comparison of user experience ispresented in Figure 2c, which compares the performance of agraduate student highly experienced with the AI-MS 1.2 plat-form (i.e., over 1000 uses) and an undergraduate in the post-training phase. The performance of other untrained users seen inTable 2 does show that decent outcomes can still be realizedwithout rigorous training experiences, though, alluding to theoverall simplicity of ambient MS methods. Of note, even whenrelatively low signal intensities are acquired (as observed inFigure 2a and c), a positive detection of cocainewas still obtained.

The PSI-MS methodology produced a true positive detectionrate of 99.01% and false positive detection rate of 0.17%.Although the samples investigated in this study were simplistic

in nature compared to traditional forensic evidence, the errorrates obtained suggest that reliable field-based chemical evi-dence screening can be accomplished even when operated bynontechnical users. Overall system reliability, incorporating allPSI andDESI-MS error rate investigations, modestly alters theserates to 98.87% and 0.14%, respectively. Specific errors ob-served throughout the study for each user are reported in Sup-plementary Table S-2, with a majority pertaining to rectifiablesystematic errors (e.g., improper positioning/sample loading,poor preparation of the PSI substrate, neglecting hygiene proto-cols and method blanks), suggesting that even better outcomescould be realized after extensive experience is gained.

Repeatability, Inter-User Reproducibility, and Ion SourceComparison Figure 3 depicts average and range trends forall cocaine positive controls analyzed by trained users (n =1070), broken down by date of analysis and user (color-codedfor visualization) in order to examine both inter-day and intra-user trends. While data collected via PSI-MS is marked bysubstantial variation in spectral intensity, even for highly expe-rienced users (i.e., graduate students A and B), some trends can

0

50000

100000

150000

200000

250000

300000

350000

2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM

Time of Analysis (May 17th, 2016)

n = 12

n = 3

n = 12

n = 7

Graduate Student

Average: 144584

Max: 303449

Min: 2799

n = 30

n = 6

n =10

Undergraduate Student

Average: 80933

Max: 203235

Min: 383

n = 20

0

100K

200K

300K

400K

0 2 4 6 8 10

Date of Analysis (Month/Day/2016)

Training

4.0

4.3

4.1

3.7

5.8

3.8

4.1 4.1

2.8

2.6

6/17 6/20 6/28 6/30 7/5 7/8 7/12 7/19 7/22 7/28

0

100K

200K

300K

400K

0 2 4 6

)U

A,

40

3z

/m

(t

hgi

eH

ka

eP

Date of Analysis (Month/Day/2016)

Training

3/17 4/21

3.62.5

3.3

3.6

3.5

1.6

1.8

2.8

4/28 5/17 5/18 5/24 6/14 6/17

Ave. P

eak H

eig

ht (m

/z 304, A

U)

Peak H

eig

ht (m

/z 304, A

U)

Av

e. P

ea

k H

eig

ht (m

/z

3

04

, A

U)

(b)

(c)

(a)

Figure 2. Inter-day repeatability plots for (a) high school and (b) undergraduate chemistry major users showing PSI-MS resultsobtained during training (shaded blue) and post-training. Each date-stamped datapoint expresses the average of themaximumpeakheights obtained frommultiple analyses of cocaine positive controls, and bars visualizing the range of peak heights obtained on thatday are included. Average throughput (in min/sample) for each period of operation is included above each datapoint. (c) Intra-dayreproducibility plot assessed between two users (graduate versus undergraduate), plotting themaximumpeak height obtained for allcocaine positive standards analyzed. Zones corresponding to historically low, medium, and high intensity are denoted by dashedred lines, and the number of samples falling within for each user is reported. Of note, even when relatively low signal intensities areobtained, a positive detection of cocaine was still obtained

1054 Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening

be observed. In examining the user-specific plots found in Sup-plementary Table S-9, maximum signals obtained and repeat-ability (i.e., test-retest variability of a single user) are lower forusers with minimal instrument-specific experience. The effect ofexperience level, in terms of education, instrument familiarity,and training, on user-to-user reproducibility can be insinuatedfrom Figure 2c and Figure 3. Ultimately, while repeatability ismodest and user-to-user reproducibility can be impacted byexperience and training, there is only a minor effect of overallreliability of qualitative assessments. Source-specific variabilitywas also examined, reported in Figure S-10.While overall signalwas shown to be highly variable for both DESI and PSI, averageandmaximum signals obtained by PSI are markedly higher. Thisis attributed to the desorption/ionization mechanism employedby DESI-MS, which requires the collection of secondary analyteions liberated from a surface of interest [60]. Portable MSsystems commonly employ miniaturized vacuum systems, andthe reduced conductance of the inlet capillary makes efficientcollection of non-proximity ion plumes desorbed from surfacesmore problematic compared with PSI and traditional ESI sourcesthat employ coaxial alignment.

Method Robustness for PSI-MS

Although PSI-MS is a flexible technique, several variableshave been identified as affecting the quality of spectral datacollected and subsequent instrumental hygiene, including sub-strate positioning in reference to the inlet capillary, paperimperfections stemming from manual cutting of the triangulartip, and, to a lesser extent, the amount of solvent and voltageapplied to induce spray ionization. To assess the robustness ofthese method-specific variables, a systematic study of spatialpositioning and ionization parameters was undertaken. Figure 4provides a graphical representation of the three-dimensionalpositioning of variables examined in reference to the inletcapillary of the AI-MS 1.2.

For this study, maximum ion signal, signal duration, andsample-to-sample hygiene was determined for triplicate analy-ses of 200 ng cocaine samples at each variable increment.Table 3 provides both the operational and optimal values sub-sequently established for PSI analysis; Boperational^ values

represent those that can hinder spectral characteristics, but stillprovide data that is satisfactory for analyte detection/identifica-tion. Supplementary Figure S-11 and the corresponding Supple-mental Material show representative experimental data regard-ing the effect of X, Y, Z positioning on spray duration, as well asa detailed experimental design. Given the variability of spectralintensity observed via PSI-MS (as seen in Figure 2 and Figure 3),spray duration served as a better indicator of proper positioning.

Spatial positioning studies showed that satisfactory spectraldata can be collected as long as the pinnacle of the PSI substratealigns within the outer diameter of the inlet capillary (Fig-ure 4b), with more optimal signals stemming from alignmentwithin the boundaries of the inner diameter. The only exceptionobserved arises during horizontal positioning (Y-axis), as theplastic backing of the PSI substrate interferes with the genera-tion of an effective Taylor cone when it resides in the line ofsight between the paper triangle and inlet capillary (0.4 to0.8 mm). Placement outside of the capillary face (i.e., past0.80 mm on X, Y axes) did not produce reliable signals and,so, is not recommended for operation.

0

200K

400K

600K

4/17/16 5/7/16 5/27/16 6/16/16 7/6/16 7/26/16 8/15/16

Av

era

ge

P

ea

k H

eig

ht

(m

/z 3

04

, A

U)

Date of Analysis (Month/Day/2016)

Graduate Student A

Graduate Student B

Undergraduate Student

High School Student

Figure 3. Variability plot of all PSI-MS analyses of cocaine positive control samples completed by trained users (n = 1070), showinginter-day and intra-user trends. Each date-stamped datapoint expresses the average of the maximum peak heights obtained frommultiple analyses, color-coded to identify the specific user. Bars visualizing the range of peak heights obtained on that day areincluded. Although the inherent variability of PSI-MS is apparent, the effect of education level and cumulative experience on averageand maximum peak height can be seen

-0.8

0.8

-0.8 0.8

-0.4

-0.4 0.4

0.4

MS InletPSI Substrate

Side View

(a)

(c)

(b)

Z-axis

X- Axis

Y-Axis

Paper

Plastic Backing

mm

Figure 4. Graphical depictions of spatial positioning variablesexamined for PSI-MS method robustness. (a) 3-Dimensionalrepresentation of the spatial (X, Y, Z) axes in relation to MS inletcapillary. (b) Positioning increments (in mm) for the X- and Y-axis. (c) Side view representation of Z-axis PSI substrate posi-tioning in relation to the MS inlet

Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening 1055

Optimization of voltage applied to induce ionization providedvalues that coincide well with those reported in literature [60].Interestingly, spray solvent volume was shown to highly influ-ence experimental outcomes, which is of note given that PSI inthe current embodiment requires user deposition of solvent.Small volumes (e.g., 1.50 μL) provided successful detection,but with markedly lower spray duration and higher spray insta-bility due to incomplete saturation of the paper substrate. This, inturn, could increase the frequency of false negative errors in thefield setting. Aliquots greater than 2.50 μL produced consider-able solvent build-up on the paper substrate, which had thepredilection to sputter larger droplets directly into the inlet cap-illary, increasing the chance of sample carryover from condensedphases like bulk powder and false positives errors. An optimalvalue of 2.00 μL was found to provide reliable spray durationand ion intensity for routine experimentation.

The factory ESI/DESI combination source employed on theAI-MS 1.2 has been previously characterized in terms of meth-od variables and reported elsewhere [24], and its static naturemostly alleviates source positioning issues. Sample positioning,however, still remains critical to effectively desorb and subse-quently sample analyte ions into capillary-based inlet systems.

Ruggedness Towards Field Implementation

Ruggedness towards external factors naturally entails the consid-eration of field implementation, which is especially pertinent forapplication to routine law enforcement (e.g., traffic control stops)and crime scene processing. An initial assessment of environmen-tal factors affecting on-site analysis was conducted during lateSpring/Summer 2016, tracking meteorological data (i.e., ambienttemperature and humidity) with an AcuRite weather station

(model VN1TXCA2; Chaney Instrument Co., Lake Geneva,WI, USA) in tandem with the analysis of positive cocaine controlsamples (n = 140). Although only a modest range of seasonally-dependent temperatureswas examined, data collectedwasmarkedby slight reductions in both minimum and average durationsobtained as temperature increased, as shown in SupplementaryFigure S-12. Of note, a significant portion of the false negativeerrors obtained on the AI-MS 1.2 (S-2) coincided with the in-creased temperatures, potentially stemming from poor spray char-acteristics. The overall effect of humidity was inconclusive. Giventhat PSI-MS is commonly operated as an open-air source, otherfactors such as wind direction/speed and dew point (for nightoperations) may be found relevant upon further investigation.

Detection Limit Comparisons

While both DESI and PSI have been shown to be capable ofattaining low-level detection limits using portable instrumenta-tion, source-to-source comparisons could prove useful inselecting the appropriate method for the trace evidence ofinterest. Using a selection of forensically-relevant chemicals,LODs were established using PSI (surface swabbing) anddirect DESI analysis for residues deposited onto surfaces ofevidentiary value (shown in Supplementary Figure S-13). Asseen in Table 4, PSI LODs ranged in the low to mid-nanogram(ng) range, with DESI limits determined to be an order ofmagnitude or higher in most cases. Mild surface and ionizationefficiency effects across both sources can be seen, particularlyfor surfaces with more geometric complexity or those prone tomovement during analysis. Particularly for DESI, the staticnature of the ion source can hinder the investigation of non-flat surfaces, making surface swabbing PSI more appealing.Representative LODs stemming from APCI analysis of lowconcentration vapors of flammable organic solvents were alsocollected (Supplementary Table S-3), demonstrating trace gasanalysis capability of the AI-MS 1.2 system.

ConclusionsIn the present work, the novelty of an ambient sampling, field-portable mass spectrometer featuring interchangeable, ambientionization sources for high throughput forensic investigation

Table 3. Method Robustness for PSI-MS Positioning and Ionization Variables

Positioning parameters (mm)Operational Optimal

X - axis 0.00 to 0.80 0.00 ± 0.40Y - axis –0.80 to 0.40 – 0.40 to 0.00Z - axis 2.00 to 3.25 2.00

Ionization parametersOperational Optimal

Voltage (kV) 3.00 to 4.00 3.75 to 4.00Solvent (μL) 1.50 to 2.50 2.00

Table 4. LODs for PSI and DESI-MS Analysis of Surface-Bound Residues

Compound Glass slide Brass key Polyethylene bag Countertop

PSI Cocaine 20 25 20 35Codeine 25 35 25 45Methamphetamine 5.0 6.5 7.0 202C-B 55 60 65 8525B-NBOMe 300 330 335 345

DESI Cocaine 250 600 350 300Codeine 300 700 650 450Methamphetamine 300 700 400 4502C-B 250 550 400 40025B-NBOMe 800 1250 1050 950

1056 Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening

was demonstrated, while addressing the practical obstacles thatsurround field-based forensic evidence screening (e.g., appli-cability to diverse sample matrices and emerging threats, qual-ity and reliability of evidentiary data collected, and data col-lection by nontechnical operators). While recent advances inthe fields of ambient and portable mass spectrometry haveintrinsic value in combatting the growing forensic evidentiarybacklog, this work represents the first multi-category analyticalvalidation of a field-ready system. Validation categories exam-ined, particularly selectivity, error rate, and ruggedness, willassist in meeting the demands of the Daubert standard forfuture admissibility of field-collected MS data.

While collected data exhibited variability in regards to thespectral intensity observed, the reliability of detection wasshown to be relatively unaffected when investigating low-complexity samples. Although previous reports have shownthe AI-MS 1.2 platform to be capable of accurately identifyingstreet drugs [24] and clandestine drug manufacturing parapher-nalia [26], further work involving associated error rate is per-tinent. Interestingly, user-specific variables, such as educationand training, were shown to affect experimental outcomes. It isanticipated that the categorical breakdown of user-specificerror rates presented will prove to be beneficial in determiningthe most effective approach and duration of training for non-technical operators. Incorporation of internal quality controlstandards into the needed spray solvent and monitoring of ionsource spray current could enhance outcomes for novice users.Initial ruggedness studies regarding field usage scenarios sug-gest that meteorological variables could be of interest, such asambient temperature. Future investigations controlling for dai-ly meteorological phenomenon (i.e., dew point), extremeweather, harsh environments, and the potential of operatorduress in unsafe usage scenarios could be of value to the firstresponse community.

AcknowledgmentsThis project was supported by Award nos. 2011-DN-BX-K552and 2015-IJ-CX-K011, awarded by the National Institute ofJustice, Office of Justice Programs, U.S. Department of Justice.The opinions, findings, and conclusions or recommendationsexpressed in this publication are those of the authors and do notnecessarily reflect those of the Department of Justice. Molec-ular assignments for fragmentation spectra were made forselect compounds with assistance from high resolution MSinstrumentation acquired through support by the National Sci-ence FoundationMRI Program under Grant no. CHE 1337497.The authors would also like to thank Cerilliant Corp. forsupplying the substituted phenethylamine and NBOMe deriv-ative standards utilized in this work.

References

1. Strom, K.J., Hickman, M.J.: Unanalyzed evidence in law-enforcementagencies. Criminol. Public Policy 9, 381–404 (2010)

2. Peterson, J.L., Hickman, M.J., Strom, K.J., Johnson, D.J.: Effect offorensic evidence on criminal justice case processing. J. Forensic Sci.58, S78–S90 (2013)

3. Durose, M.R., Walsh, K.A., Burch, A.M.: Census of publicly fundedforensic crime laboratories, 2009. U.S. Department of Justice, Office ofJustice Programs, National Institute of Justice, Washington DC (2012)

4. Lodhi, K.M., Grier, R.L., Speaker, P.J.: A novel approach to forensicmolecular biology education and training: it’s impact on the criminaljustice system. Aust. J. Forensic Sci. 47, 182–193 (2015)

5. Sceince, N.A.: Strengthening forensic science in the United States: a pathforward. National Academies Press, Washington DC (2009)

6. Maguire, C., Houck, M.M., Williams, R., Speaker, P.J.: Efficiency andthe cost-effective delivery of forensic science services: insourcing,outsourcing, and privatization. Forensic Sci. Policy Manage. - Int. J. 3,62–69 (2012)

7. Ludwig, A.: E Bvalue^ ating forensic science. Forensic Sci. Policy Man-age. - Int. J. 7, 66–77 (2016)

8. Brettell, T.A., Butler, J.M., Almirall, J.R.: Forensic science. Anal. Chem.83, 4539–4556 (2011)

9. Butler, J.M.: The future of forensic DNA analysis. Philos. Trans. R. Soc.Lond. Ser. B Biol. Sci. 370, 577–579 (2015)

10. Policy, W.H.O. of S. and T.: Strengthening forensic science: a progressreport, February 2014. Available at: http://www.whitehouse.gov/sites/default/files/microsites/ostp/forensic_science_progress_2-14-14.Accessed 15 Aug 2016

11. Sampat, A., Lopatka, M., Sjerps, M., Vivo-Truyols, G., Schoenmakers,P., van Asten, A.: Forensic potential of comprehensive two-dimensionalgas chromatography. TrAC Trends Anal. Chem. 80, 345–363 (2016)

12. Meyer, M.R., Maurer, H.H.: Review: LC coupled to low- and high-resolution mass spectrometry for new psychoactive substance screeningin biological matrices –where do we stand today? Anal. Chim. Acta 927,13–20 (2016)

13. Cooks, R.G., Ouyang, Z., Takats, Z., Wiseman, J.M.: Detection technol-ogies. Ambient Mass Spectrometry. Science 311, 1566–1570 (2006)

14. Monge, M.E., Harris, G.A., Dwivedi, P., Fernández, F.M.: Mass spec-trometry: recent advances in direct open air surface sampling/ionization.Chem. Rev. 113, 2269–2308 (2013)

15. Correa, D.N., Santos, J.M., Eberlin, L.S., Eberlin, M.N., Teunissen, S.F.:Forensic chemistry and ambient mass spectrometry: a perfect coupledestined for a happy marriage? Anal. Chem. 88, 2515–2526 (2016)

16. Hoffmann,W.D., Jackson, G.P.: Forensic mass spectrometry. Annu. Rev.Anal. Chem. 8, 419–440 (2015)

17. Ifa, D.R., Jackson, A.U., Paglia, G., Cooks, R.G.: Forensic applications ofambient ionization mass spectrometry. Anal. Bioanal. Chem. 394, 1995–2008 (2009)

18. Manicke, N.E., Bills, B.J., Zhang, C.: Analysis of biofluids by paperspray MS: advances and challenges. Bioanalysis 8, 589–606 (2016)

19. Damon, D.E., Davis, K.M., Moreira, C.R., Capone, P., Cruttenden, R.,Badu-Tawiah, A.K.: Direct biofluid analysis using hydrophobic paperspray mass spectrometry. Anal. Chem. 88, 1878–1884 (2016)

20. Kauppila, T.J., Flink, A., Pukkila, J., Ketola, R.A.: Analysis of nitrogen-based explosives with desorption atmospheric pressure photoionizationmass spectrometry. Rapid Commun.Mass Spectrom. 30, 467–475 (2016)

21. Black, C., Chevallier, O.P., Elliott, C.T.: The current and potential appli-cations of ambient mass spectrometry in detecting food fraud. TrACTrends Anal. Chem. 82, 268–278 (2016)

22. Lesiak, A.D., Cody, R.B., Ubukata, M., Musah, R.A.: Directanalysis in real time high resolution mass spectrometry as a toolfor rapid characterization of mind-altering plant materials andrevelation of supplement adulteration – the case of Kanna. Foren-sic Sci. Int. 260, 66–73 (2016)

23. Evans-Nguyen, K.M., Gerling, J., Brown, H., Miranda, M., Windom, A.,Speer, J.: Towards universal ambient ionization: direct elemental analysisof solid substrates using microwave plasma ionization. Analyst 141,3811–3820 (2016)

24. Vircks, K.E., Mulligan, C.C.: Rapid screening of synthetic cathinones astrace residues and in authentic seizures using a portable mass spectrom-eter equipped with desorption electrospray ionization. Rapid Commun.Mass Spectrom. 26, 2665–2672 (2012)

25. Pavlovich, M.J., Musselman, B., Hall, A.B.: Direct analysis in real time-mass spectrometry (DART-MS) in forensic and security applications.Mass Spectrom Rev. (2016). doi:10.1002/mas.21509.

Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening 1057

26. O’Leary, A.E., Hall, S.E., Vircks, K.E., Mulligan, C.C.: Monitoring theclandestine synthesis of methamphetamine in real-time with ambient sam-pling, portable mass spectrometry. Anal. Methods 7, 7156–7163 (2015)

27. Damon, D.E., Maher, Y.S., Yin, M., Jjunju, F.P.M., Young, I.S., Taylor,S., Maher, S., Badu-Tawiah, A.K.: 2D wax-printed paper substrates withextended solvent supply capabilities allow enhanced ion signal in paperspray ionization. Analyst 141, 3866–3873 (2016)

28. Carvalho, T.C., Oliveira, I.F., Tose, L.V., Vanini, G., Kill, J.B., Neto,A.C., Machado, L.F., Ambrosio, J.C.L., Lacerda, V., Vaz, B.G., Romão,W.: Qualitative analysis of designer drugs by paper spray ionization massspectrometry (PSI-MS). Anal. Methods 8, 614–620 (2016)

29. Namera, A., Nakamoto, A., Saito, T., Nagao, M.: Colorimetric detectionand chromatographic analyses of designer drugs in biological materials: acomprehensive review. Forensic Toxicol. 29, 1–24 (2011)

30. Philp, M., Shimmon, R., Tahtouh, M., Fu, S.: Development and valida-tion of a presumptive color spot test method for the detection of syntheticcathinones in seized illicit materials. Forensic Chem. 1, 39–50 (2016)

31. Tsumura, Y., Mitome, T., Kimoto, S.: False positives and false negativeswith a cocaine-specific field test and modification of test protocol toreduce false decision. Forensic Sci. Int. 155, 158–164 (2005)

32. Krauss, S.T., Remcho, T.P., Lipes, S.M., Aranda, R., Maynard, H.P.,Shukla, N., Li, J., Tontarski, R.E., Landers, J.P.: Objective method forpresumptive field-testing of illicit drug possession using centrifugalmicrodevices and smartphone analysis. Anal. Chem. A–H (2016)

33. Vandenabeele, P., Edwards, H.G.M., Jehlička, J.: The role of mobileinstrumentation in novel applications of Raman spectroscopy:archaeometry, geosciences, and forensics. Chem. Soc. Rev. 43, 2628–2649 (2014)

34. Tsujikawa, K., Yamamuro, T., Kuwayama, K., Kanamori, T., Iwata,Y.T., Miyamoto, K., Kasuya, F., Inoue, H.: Application of a portablenear infrared spectrometer for presumptive identification of psychoactivedrugs. Forensic Sci. Int. 242, 162–171 (2014)

35. Eckenrode, B.: Environmental and forensic applications of field-portableGC-MS: an overview. J. Am. Soc. Mass Spectrom. 12, 683–693 (2001)

36. Leary, P.E., Dobson, G.S., Reffner, J.A.: Development and applicationsof portable gas chromatography-mass spectrometry for emergency re-sponders, the military, and law-enforcement organizations. Appl.Spectrosc. 70, 888–896 (2016)

37. Mach, P.M., McBride, E.M., Sasiene, Z.J., Brigance, K.R., Kennard,S.K., Wright, K.C., Verbeck, G.F.: Vehicle-mounted portable mass spec-trometry system for the covert detection via spatial analysis of clandestinemethamphetamine laboratories. Anal. Chem. 87, 11501–11508 (2015)

38. Giannoukos, S., Brkić, B., Taylor, S., Marshall, A., Verbeck, G.F.:Chemical sniffing instrumentation for security applications. Chem. Rev.116, 8146–8172 (2016)

39. Chen, C.-H.H., Chen, T.-C.C., Zhou, X., Kline-Schoder, R., Sorensen, P.,Cooks, R.G., Ouyang, Z.: Design of portable mass spectrometers withhandheld probes: aspects of the sampling and miniature pumping sys-tems. J. Am. Soc. Mass Spectrom. 26, 240–247 (2015)

40. Ma, X., Ouyang, Z.: Ambient ionization andminiaturemass spectrometrysystem for chemical and biological analysis. TrAC Trends Anal. Chem.1–10 (2016)

41. Snyder, D.T., Pulliam, C.J., Ouyang, Z., Cooks, R.G.: Miniature andfieldablemass spectrometers: recent advances. Anal. Chem. 88, 2–29 (2016)

42. Brown, H., Oktem, B., Windom, A., Doroshenko, V., Evans-Nguyen, K.:Direct analysis in real time (DART) and a portable mass spectrometer forrapid identification of common and designer drugs on-site. ForensicChem. 1, 66–73 (2016)

43. Ma, Q., Bai, H., Li, W., Wang, C., Li, X., Cooks, R.G., Ouyang, Z.:Direct identification of prohibited substances in cosmetics and foodstuffsusing ambient ionization on a miniature mass spectrometry system. Anal.Chim. Acta 912, 65–73 (2016)

44. O’Leary, A.E., Oberacher, H., Hall, S.E., Mulligan, C.C.: Combining aportable, tandem mass spectrometer with automated librarysearching—an important step towards streamlined, on-site identificationof forensic evidence. Anal. Methods 7, 3331–3339 (2015)

45. ASTM E2549-14: Standard practice for validation of seized-drug analyt-ical methods. ASTM International, West Conshohocken, PA (2014)

46. Scientific Working Group for the Analysis of Seized Drugs(SWGDRUG) Recommendations. United States Department of Justice,Drug Enforcement Administration, Washington DC (2011)

47. Gurdak, E., Green, F.M., Rakowska, P.D., Seah, M.P., Salter, T.L.,Gilmore, I.S.: VAMAS interlaboratory study for desorption electrospray

ionization mass spectrometry (DESI MS) intensity repeatability andconstancy. Anal. Chem. 86, 9603–9611 (2014)

48. Wells, J.M., Roth, M.J., Keil, A.D., Grossenbacher, J.W., Justes, D.R.,Patterson, G.E., Barket, D.J.: Implementation of DART and DESI ioni-zation on a fieldable mass spectrometer. J. Am. Soc. Mass Spectrom. 19,1419–1424 (2008)

49. Wolf, J.-C., Etter, R., Schaer, M., Siegenthaler, P., Zenobi, R.: Direct andsensitive detection of CWA simulants by active capillary plasma ioniza-tion coupled to a handheld ion trap mass spectrometer. J. Am. Soc. MassSpectrom. 27, 1197–1202 (2016)

50. Sanders, N.L., Kothari, S., Huang, G., Salazar, G., Cooks, R.G.: Detec-tion of explosives as negative ions directly from surfaces using a minia-ture mass spectrometer. Anal. Chem. 82, 5313–5316 (2010)

51. Wiley, J.S., Shelley, J.T., Cooks, R.G.: Handheld low-temperature plas-ma probe for portable Bpoint-and-shoot^ ambient ionization mass spec-trometry. Anal. Chem. 85, 6545–6552 (2013)

52. Mulligan, C.C., Talaty, N., Cooks, R.G.: Desorption electrospray ioniza-tion with a portable mass spectrometer: in situ analysis of ambientsurfaces. Chem. Commun. 74, 1709 (2006)

53. Cooks, R.G., Manicke, N.E., Dill, A.L., Ifa, D.R., Eberlin, L.S., Costa,A.B., Wang, H., Huang, G., Ouyang, Z.: New ionization methods andminiature mass spectrometers for biomedicine: DESI imaging for cancerdiagnostics and paper spray ionization for therapeutic drug monitoring.Faraday Discuss. 149(247), 67–56 (2011)

54. Li, L., Chen, T.-C., Ren, Y., Hendricks, P.I., Cooks, R.G., Ouyang, Z.:Mini 12, Miniature mass spectrometer for clinical and otherapplications—introduction and characterization. Anal. Chem. 86, 2909–2916 (2014)

55. Kim, P., Cha, S.: Paper cone spray ionization mass spectrometry (PCSIMS) for simple and rapid analysis of raw solid samples. Analyst 140,5868–5872 (2015)

56. Oberacher, H.M.: Wiley registry of tandem mass spectral data. Wiley,Hoboken (2011)

57. Pavlic, M., Libiseller, K., Oberacher, H.: Combined use of ESI–QqTOF-MS and ESI–QqTOF-MS/MS with mass-spectral library search for qual-itative analysis of drugs. Anal. Bioanal. Chem. 386, 69–82 (2006)

58. Pavlic, M., Schubert, B., Libiseller, K., Oberacher, H.: Comprehensiveidentification of active compounds in ts by flow-injection data-dependenttandem mass spectrometry combined with library search. Forensic Sci.Int. 197, 40–47 (2010)

59. Oberacher, H., Pavlic, M., Libiseller, K., Schubert, B., Sulyok, M.,Schuhmacher, R., Csaszar, E., Köfeler, H.C.: On the inter-instrumentand the inter-laboratory transferability of a tandemmass spectral referencelibrary: 2. Optimization and characterization of the search algorithm. J.Mass Spectrom. 44, 494–502 (2009)

60. Venter, A.R., Douglass, K.A., Shelley, J.T., Hasman, G., Honarvar, E.:Mechanisms of real-time, proximal sample processing during ambientionization mass spectrometry. Anal. Chem. 86, 233–249 (2014)

61. Laughlin, B.C., Mulligan, C.C., Cooks, R.G.: Atmospheric pressureionization in a miniature mass spectrometer. Anal. Chem. 77, 2928–2939 (2005)

62. Nyadong, L., Hohenstein, E.G., Johnson, K., Sherrill, C.D., Green, M.D.,Fernández, F.M.: Desorption electrospray ionization reactions betweenhost crown ethers and the influenza neuraminidase inhibitor oseltamivirfor the rapid screening of Tamiflu. Analyst 133, 1513–1522 (2008)

63. Nefliu, M., Smith, J.N., Venter, A., Cooks, R.G.: Internal energy distri-butions in desorption electrospray ionization (DESI). J. Am. Soc. MassSpectrom. 19, 420–427 (2008)

64. Chen, H., Eberlin, L.S., Cooks, R.G.: Neutral fragment mass spectra viaambient thermal dissociation of peptide and protein ions. J. Am. Chem.Soc. 18, 5880–5886 (2007)

65. Burrai, L., Nieddu, M., Carta, A., Trignano, C., Sanna, R., Boatto, G.:Validated LC-MS-MS method for multiresidual analysis of 13 illicitphenethylamines in amniotic fluid. J. Anal. Toxicol. 3, 194–200 (2016)

66. Oberacher, H., Pitterl, F., Siapi, E., Steele, B.R., Letzel, T., Grosse, S.,Poschner, B., Tagliaro, F., Gottardo, R., Chacko, S.A., Josephs, J.L.: Onthe inter-instrument and the inter-laboratory transferability of a tandemmass spectral reference library. 3. Focus on ion irap and upfront CID. J.Mass Spectrom. 47, 263–270 (2012)

67. Würtinger, P., Oberacher, H.: Evaluation of the performance of a tandemmass spectral library with mass spectral data extracted from literature.Drug Test. Anal. 4, 235–241 (2012)

1058 Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening

68. Zuba, D., Sekuła, K.: Analytical characterization of three hallucinogenicN -(2-methoxy)benzyl derivatives of the 2C-series of phenethylaminedrugs. Drug Test. Anal. 5, 634–645 (2013)

69. Zuba, D., Sekuła, K., Buczek, A.: 25C-NBOMe—new potent hallucinogenicsubstance identified on the drug market. Forensic Sci. Int. 227, 7–14 (2013)

70. Oberacher, H., Pavlic, M., Libiseller, K., Schubert, B., Sulyok, M.,Schuhmacher, R., Csaszar, E., Köfeler, H.C.: On the Inter-instrument

and inter-laboratory transferability of a tandem mass spectral referencelibrary: 1. Results of an Austrian multicenter study. J. Mass Spectrom. 44,485–493 (2009)

71. Mulligan, C.C., O’Leary, A.E.: Accessing the probative value of physicalevidence at crime scenes with ambient mass spectrometry and instrumen-tation. National Criminal Justice Reference Service, Washington, DC(2015)

Z. E. Lawton et al.: Validation of Portable MS for Forensics Evidence Screening 1059