Forensic Chemistry 42 (2025) 100627

Contents lists available at ScienceDirect

Forensic Chemistry
journal homepage: www.sciencedirect.com/journal/forensic-chemistry

Development and application of ion chromatography for analyzing

explosive residues in criminal Investigations: A study of ions in ATM and
residues from emulsion and pipe bomb explosions
Paula Ribeiro Junho a , Yuri Machado b, Ricardo M. Orlando a,*
a
Laboratory of Microfluidics and Separations, LaMS, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil
b
Institute of Criminalistics, Civil Police of Minas Gerais, Belo Horizonte, Minas Gerais, 30180-060, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: The criminal use of explosives, particularly in attacks on automated teller machines (ATMs), remains prevalent in
Residue Analysis many countries, including Brazil. Ion chromatography is a leading technique for analyzing residues from post-
Ion Chromatography blast materials. In this study, we developed methods to analyze 8 cations and 18 anions using suppressed ion
Pipe Bombs
chromatography with conductivity detection. Different collection materials and procedures were also evaluated.
ATM Explosions
Forensic Science
The chromatographic methods were assessed using various metrics and demonstrated analytical performance in
line with existing literature. The methods were applied to determine the ion content in three brands of swabs
commonly used for sample collection. Initial results showed significant levels of various ions, which were largely
removed through a simple washing process. Our findings indicate that certain swab brands of the same type
perfomed better than others, and the preferred swab was used in recovery studies. We concluded that using pre-
washed swabs moistened with water results in higher recovery rates and reduces interference. An ion stability
test conducted over 12 weeks suggested that swabs should be stored in a refrigerator and ideally analyzed within
the first week after collection to minimize analyte loss. Background samples collected from an ATM did not
contain ion levels that could be mistaken for those from explosive residues. The chromatographic and sampling
methods were applied to real samples from different types of explosive residues, including those from an
emulsion-based explosive and a pipe bomb containing black powder substitute. The analysis successfully iden­
tified key ions: NO3− and NH+ 4 in the emulsion, and ClO4 and K in the black powder substitute, allowing us to
− +

differentiate the explosives used. This work introduces significant methodological advancements in the analysis
of explosive residues, particularly through the development of an extensive chromatographic method, the
mitigation of ion interferences, improved sample preservation protocols, and the successful application to real-
world scenarios, distinguishing emulsion-based explosives from gunpowder.

Introduction mixtures commonly used by criminals are NH4NO3 emulsions and

gunpowder, whether black powder or chlorate-based [2].
Bank robberies are common in many countries, and in Brazil alone, Explosives may contain both fuels and oxidizers within a single
between 2021 and 2022, approximately 273 incidents per year were molecule, or they can be mixtures of combustible substances with
reported [1]. Many of these incidents involved the use of explosives to energetically oxidizing compounds [6,7]. The most common combus­
break into Automated Teller Machines (ATMs). In Brazil, criminals tible elements are carbon, hydrogen, and sulfur, while oxidizing groups
obtain explosives not only by stealing from military barracks or mining vary, but predominantly contain oxygen and nitrogen [6]. Molecular
sites but also by targeting the transport and storage stages of quarries groups with explosive properties are those with high internal energy,
and certain infrastructure projects that consume large amounts of ex­ such as peroxides, chlorates, perchlorates, nitrites, nitrates, azides, and
plosives [2,3]. The materials used by criminals can consist of various fulminates. When triggered by a stimulus, these groups react with fuels,
types of explosives, but are often mixtures of inorganic salts with fuels releasing energy in the form of heat and shock waves [7].
due to their relative ease of acquisition [4,5]. Two examples of explosive For the analysis of ionic or ionizable components that make up

* Corresponding author.
E-mail address: orlandoricardo@ufmg.br (R.M. Orlando).

https://doi.org/10.1016/j.forc.2024.100627
Received 3 June 2024; Received in revised form 29 November 2024; Accepted 29 November 2024
Available online 4 December 2024
2468-1709/© 2024 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

explosives, the most widely used method is ion chromatography, fol­ States) at a concentration of 1000 mg L− 1 were also used. Liquid re­
lowed by capillary electrophoresis [8]. Ion chromatography is consid­ agents such as methanol (JT Baker, Mexico) and acetone (Vetec, Brazil)
ered the preferred method for analyzing inorganic ions because it offers were of high-performance liquid chromatography (HPLC) grade or
high sensitivity and selectivity, in addition to being a robust and reliable higher. Stock solutions from salts were prepared at a concentration of
technique [5,9–11]. Identifying explosives used in criminal activities 1000 mg L− 1. Intermediate solutions containing mixed anions or cations
after an explosion is highly relevant in forensic chemistry, as it helps link were made at 50 mg L− 1 per analyte. Solutions containing divalent
suspects to crime scenes and guides investigative efforts [8,12]. Most cations were acidified with 0.1 M nitric acid. All solutions derived from
improvised explosives containing inorganic materials or emulsions can salts were filtered before injection. The prepared solutions were stored
be analyzed using ion chromatography [9]. A good chromatographic in polypropylene (PP) plastic bottles and kept refrigerated at 2 to 8 ◦ C.
method for analyzing explosive residues should have reproducible All glassware was designated exclusively for ion chromatography. After
retention times and the ability to clearly separate the ions present in the washing with diluted alkaline detergent (Extran, Merck, Brazil), the
residue, without significant interferences—features that ion chroma­ glassware was rinsed at least five times with DI. To filter samples and
tography can readily provide. Quantitative analysis is generally not standards, individually packaged sterile syringes (Becton Dickison,
required, as the conditions during the explosion are variable, resulting in Brazil), individually packaged sterile hypodermic needles (MedNeedle
differing amounts of intact and burnt materials. The mere presence or by MedGoldman), and Uniflo PP filters with PTFE membranes (0.45 µm,
absence of specific ions is often indicative of the explosive used [13,14]. 13 mm diameter) from Whatman (GE Healthcare, China) were used.
Table S1 (see Supplementary Material) lists the main ions found in the Stock solutions were diluted to a concentration of 10 mg L− 1 and
analysis of some inorganic explosives. injected 5 months after preparation, showing less than 5 % variation in
One of the most critical steps in the analysis of explosives using ion peak areas.
chromatography is sampling [15]. To ensure reliable analytical results,
the ions from the explosion should not be confused with ions already Equipment
present in the environment. One way to address this issue is to collect
samples from nearby locations to understand the environmental factors Chromatographic analyses were performed using a Dionex ICS-6000
that might affect the interpretation of results [16]. Additionally, the ions system (Thermo Scientific, United States). This system has two inde­
of interest must not be present in significant quantities in the materials pendent modules for cation and anion analysis, sharing a sample injector
used for sampling [12,17]. Depending on the distance between the local (Dionex AS-AP, United States) and a column oven. It features dual
of explosion and the analysis laboratory, as well as the volume of sam­ electrolytic eluent generators, two self-regenerating electrolytic sup­
ples collected, it is not uncommon for the time between sampling and pressors (CDRS and ADRS), and two conductivity detectors. The soft­
analysis to extend to two or three months [5,18]. Samples not stored in ware used for data acquisition and processing was Chromeleon (Thermo
sealed containers can suffer a loss of volatile components, especially in Scientific, 7.2.10, United States). Solid reagents were weighed with a
the case of organic explosives [19,20], which may also degrade due to Shimadzu AUW220D analytical balance (Japan). For cation analysis, a
exposure to light [21–23]. Dionex IonPac CS12A column (2 x 250 mm, United States) with a Dionex
In this work, an extensive chromatographic separation study using IonPac CG12A guard column (2 x 50 mm, United States) was used. The
ion chromatography was conducted to achieve rapid analysis and eluent generator was equipped with a methanesulfonic acid cartridge
adequate resolution for 26 ions (including cations and anions). The (Thermo Scientific, Dionex EGC 500 MSA, United States) (MSA). Anion
developed method was applied to study the preservation of explosive analysis was conducted with a Dionex IonPac AS19 column (2 x 250
samples under different storage conditions, as well as the presence and mm, United States) and a Dionex IonPac AG19 guard column (2 x 50
removal of ions naturally occurring in the sampling materials. Further­ mm, United States), using a Dionex EG 500 eluent generator with a
more, the method was used to detect the main ions found at sample sites potassium hydroxide cartridge (Thermo Scientific, Dionex EGC 500
(ATM and floor), as well as those identified in residues from a controlled KOH, United States) (KOH). Both systems used a fixed-volume injection
explosion of an emulsion-based explosive and a pipe bomb containing loop of 10 µL.
gunpowder.
Optimization of chromatographic conditions
Experimental
The following conditions were used: a flow rate of 0.250 mL/min,
Chemicals and solutions column and detector temperatures of 30 and 35 ◦ C, respectively, based
on previous studies [12,24–26] were used for cations and anions IC
Ultrapure water (DI) from an Elga system (Purelab Flex, United separation. For cation analysis, the initial condition involved isocratic
Kingdom) was used for preparing solutions and cleaning equipment. elution using MSA at 20 mmol/L, as employed by Mauricio et al. [12].
2+ 2+ 2+ 2+
Cation stock solutions (Li+; Na+; NH+ 4 ; K ; Mg ; Ca ; Sr ; Ba ) were
+
The effect of elution gradients on cation separation was evaluated by
prepared from lithium chloride (Sigma-Aldrich, United States), sodium varying the concentration range of the MSA eluent (20–60 mmol/L), the
chloride (Sigma, United States), ammonium formate (Vetec, Brazil), rate of concentration change, and the initial isocratic elution period. The
potassium chloride (Sigma, United States), magnesium chloride hexa­ conditions tested for cation analysis are summarized in Table 1 and
hydrate (Fmaia, Brazil), calcium chloride dihydrate (Synth, Brazil), Fig. S1.
strontium nitrate (Dinâmica, Brazil) and barium chloride dihydrate Table 1.
(Neon, Brazil). Stock solutions of anions (F− ; Ace− = acetate; For− = For the analysis of anions, a gradient concentration of KOH (10–45
formate; BrO−3 ; Br− ; Tar2− = tartrate; Oxa2− = oxalate; PO3− 3−
4 ; Cit = mmol/L) over 40 min was used, with an additional 5 min at this level, as
2− 2− 2−
citrate; Cl ; NO2 ; ClO3 ; NO3 ; CO3 ; SO4 ; S2O3 ; SCN ; ClO4 ) were
− − − − − −
described by Mauricio et al. [12]. Other final concentrations of KOH
made with sodium fluoride (Synth, Brazil), sodium acetate (Sigma- were evaluated with different time variations (as shown in Table 2 and
Aldrich, United States), ammonium formate (Vetec, Brazil), potassium Fig. S2).
bromate (Vetec, Brazil), potassium bromide (Pike, United States), so­ Table 2.
dium and potassium tartrate (Dinâmica, Brazil), sodium oxalate (Mal­
linckrodt, Brazil), sodium phosphate (Vetec, Brazil), sodium citrate Analytical performance evaluation
dihydrate (Synth, Brazil). Additionally, ion chromatography standard
solutions containing chloride, nitrite, chlorate, nitrate, carbonate, sul­ Linearity was evaluated by constructing an analytical curve in trip­
fate, thiosulfate, thiocyanate and perchlorate (Sigma-Aldrich, United licate (n = 3), with a blank and seven concentration levels (1.0; 2.5; 4.0;

2
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

Table 1 first day’s concentrations were calculated from the constructed linearity
Gradients evaluated for cation separation by IC. curve, and a new curve was generated on the second day to estimate
Method MSA concentration (mmol/L) along Pre-equilibration time those concentrations. Results were calculated using the coefficient of
time (min) variation (CV). The accuracy of recovery was calculated from sextupli­
Isocratic 20 (0–20 min) − cate extractions (n = 6) at three concentration levels (2.0; 5.0; 8.0 mg
Gradient 20 (0–6.1 min); 8 L− 1), except for Ba2+ (10.0; 25.0 and 40.0 mg L− 1) and Sr2+ (5.0; 12.5
1 linear increase to 55 (6.1–30 min) and 20.0 mg L− 1). Based on the analytical curve constructed on the
Gradient 20 (0–6.1 min); 10 second day, the concentration for each extraction was estimated, and the
2 linear increase to 55 (6.1–20 min);
hold at 55 (20–25 min)
recovery percentage (R%) was calculated by the ratio of the average
Gradient linear increase from 20 to 55 (0–15 10 found concentration to the theoretical concentration. Limit of detection
3 min); (LOD) and limit of quantitation (LOQ) were determined by performing
hold at 55 (15–20 min) 10 extractions (n = 10) of the lowest level in the analytical curve (1.0 mg
Gradient linear increase from 20 to 45 (0–10 7
L− 1), except for Ba2+ (5.0 mg L− 1) and Sr2+ (2.5 mg L− 1). The LOD was
4 min);
hold at 45 (10–13 min) calculated as three times the standard deviation of the lowest concen­
Gradient linear increase from 20 to 60 (0–10 7 tration level of the linearity range for each analyte (analyzed ten times),
5 min); divided by the square root of the number of replicates, while the LOQ
hold at 60 (10–13 min) was calculated as ten times this ratio.

Studies with collection material (tubes and/or swabs and/or porcelain spot
Table 2 plates)
Gradients evaluated for anion separation by IC.
Method KOH concentration (mmol/L) along Pre-equilibration time a) Evaluation of ions naturally present in the colletion material and
time (min) washing procedures.
Gradient linear increase from 10 to 45 (0–40 8 The presence of inorganic ions in materials used for sample collec­
1 min); tion was assessed in various types of collection tubes and flexible swabs.
hold at 45 (40–45 min) The procedures were adapted from published articles [10, 14, 68]. Six
Gradient linear increase from 10 to 45 (0–30 8
different 15 mL PP tubes from Gene (Zhejiang Renon Medical Instru­
2 min);
hold at 45 (30–45 min) ment Co., Ltd., China,), Sarstedt (Germany), TPP (Switzerland), Cral­
Gradient linear increase from 10 to 55 (0–30 8 plast (Brazil), and Perfecta (Brazil) were evaluated, as depicted in
3 min); Fig. S3. These tubes were numbered 1 through 6. PP tubes were filled
hold at 55 (30–40 min) with 2 mL of DI, capped, and vortexed for 1 min (n = 3). Subsequently, 1
Gradient linear increase from 10 to 60 (0–20 8
4 min);
mL of the DI from the tubes was transferred to a vial for chromato­
hold at 60 (20–35 min) graphic analysis. To evaluate the presence of ions in the swabs used for
Gradient linear increase from 10 to 50 (0–20 8 sampling, three different brands (rayon-made Global Swab; cotton-made
5 min); Cotonetes from Johnson & Johson: and cotton-made Mio, all from
linear increase from 50 to 70 (20–25
Brazil) were tested by placing them in 15 mL conical PP tubes (brand
min);
hold at 70 (25–35 min) Gene). To fit the Global Swab in the tube, 4 cm of its shaft was cut and for
Gradient linear increase from 10 to 70 (0–30 8 flexible shafts, the unused cotton portion at the other end was removed.
6 min); To each tube containing the swabs, 2 mL of DI was added with a volu­
hold at 70 (30–31 min) metric pipette, and the tube was vortexed for 1 min. Then, 1 mL was
transferred to a 2 mL microtube, filtered through a 0.45 μm syringe
5.5; 7.0; 8.5; 10.0 mg L− 1) in a swab matrix. Separate linearity ranges filter, and placed in a chromatography vial. All procedures were per­
were assessed for barium (5.0; 12.5; 20.0; 27.5; 35.0; 42.5; 50.0 mg L− 1) formed in triplicate (n = 3) for each swab type and also in triplicate for
and strontium (2.50; 6.25; 10.00; 13.75; 17.50; 21.25; 25.00 mg L− 1), the PP tubes without swabs, the microtubes, the syringes, and the filters.
due to the varying intensity of their analytical responses. To prepare the The effect of sequential washing of swabs for removing naturally
samples, 1 mL of cation or anion solution at the tested concentrations occurring ions was also evaluated, using a procedure adapted from
was transferred to PP tubes, followed by the addition of a swab previ­ Mauricio et al. [14]. Each swab was placed in a 15 mL tube with 2 mL of
ously washed four times (section 2.5. b). These were vortexed for 1 min, DI and vortexed for 1 min. Then, 1 mL of water was collected, and 1 mL
filtered into chromatography vials, and analyzed using the optimized of the extract was filtered through a 0.45 μm syringe filter into a chro­
method (Table 3). Linearity was assessed using the ordinary least matography vial, after which the residual water in the tube was dis­
squares method. carded. This procedure was repeated three more times, totaling four
Table 3. sequential washes, in three different tubes (n = 3) for each swab brand
Intraday precision (repeatability) and interday precision (interme­ evaluated.
diate precision) were evaluated at three concentration levels (2.0; 5.0; b) Evaluation of recovery.
8.0 mg L− 1), except for Ba2+ (10.0; 25.0 and 40.0 mg L− 1) and Sr2+ (5.0; Procedures adapted from previous works [17,27,28] were followed
12.5 and 20.0 mg L− 1), each in sextuplicate (n = 6) over two days. The to study the recovery of ions from explosive residues on collection sur­
faces. A porcelain spot plate cavity was filled with 100 μL of anion or
cation solutions. The amount deposited was 5 μg for each ion, except for
Table 3 strontium (12.5 μg) and barium (25 μg). The plates were left at room
Summary of conditions used in chromatographic analysis after optimization. temperature in a desiccator until the droplets dried completely (mini­
Parameter Cation analysis Anion analysis mum of 24 h). Collection was performed in triplicate (n = 3) with four
Pre-equilibration 7 min 8 min different groups of swabs, all from the brand Global Swab. The first
time group consisted of swabs washed four times using the protocol described
Eluent linear increase from 20 to 45 linear increase from 10 to 70 in the previous section and used while still wet for collection. The second
concentrations (0–10 min); (0–30 min); group comprised new swabs moistened by submerging them in 2 mL of
hold at 45 (10–13 min) hold at 70 (30–31 min)
DI just before collection. The third group consisted of new swabs used

3
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

dry for collection. The fourth group comprised swabs washed four times, still damp, were rubbed across five different regions of the ATM,
dried at room temperature in a desiccator and used dry for collection. In covering approximate rectangular areas of 15 cm2, and then stored in 15
all cases, ions were extracted from the swab fibers using 2 mL of DI, with mL tubes. Extraction was performed with 2 mL of DI, followed by vortex
vortexing for 1 min. For injection, 1 mL was filtered into a chromatog­ agitation for 1 min, with 1 mL filtered for injection into a vial. The five
raphy vial. For each type of swab tested, a blank check of the porcelain different sampling locations on the ATM are marked in Fig. S4.
plate was conducted by collecting and analyzing an area of the plate not
exposed to the ion solution. Sampling and analysis of pre- and post-blast explosive residues in
c) Study of stability. controlled field detonations
A stability study of ions in swabs was conducted based on references
from the literature [19,21–23,29]. For this study, triplicate of swabs (n With assistance from specialists from the Special Police Operations
= 3) previously washed (see section 2.5. b) and dried at room temper­ Battalion (BOPE) of the Military Police of Minas Gerais (Belo Horizonte,
ature in a desiccator with silica gel were treated with 100 μL of a 50 mg Minas Gerais, Brazil), samples of residues were collected after two ex­
L− 1 solution of ammonium formate and then stored while still damp in plosions conducted in an open field at the Military Police Training
15 mL conical PP tubes, properly capped, either in a refrigerator (4 ◦ C) Center in Ribeirão das Neves (Minas Gerais, Brazil). The first detonation
or at room temperature (approximately 22 ◦ C) for later extraction and involved a 540 g ammonium nitrate emulsion stick, while the second
analysis. used a 210 g rectangular steel pipe bomb filled with potassium
A second group of swabs was dried with a hair dryer for 3 min (at perchlorate-based gunpowder. A metal rod and a wire were used to
medium speed and minimum temperature; model SP 3300 ion, Britânia, suspend the explosives about 90 cm off the ground to simulate the
Brazil) after adding the standards, then stored in PP tubes in a refrig­ typical height at which explosive devices are inserted during ATM
erator or at room temperature, also in triplicate (n = 3). Swabs that robberies. Lead azide detonators and fuses were used for the detona­
received only DI and were stored either damp or after hair-drying served tions. For residue collection, ceramic floor tiles (31 x 44 cm) and metal
as blanks. Swab extractions were performed by adding 2 mL of DI and plates (25 x 25 cm, 1 mm thick carbon steel) were placed around the
vortexing for 1 min. The extract was then filtered and transferred to a blast point as shown in Fig. S5. This experimental setup was adapted
chromatography vial for injection. Extractions from damp swabs were from several studies [15,5,8,30], aiming to simulate residue collection
carried out on days 0, 1, 7, 14, 21, 28, 42, 56, and 84, totaling 12 weeks. on crime scene surfaces. Ceramic tiles were positioned on the ground in
Extractions from dry swabs were done on days 0, 7, 28, 56, and 84. the four cardinal directions, with their centers at distances of 0.5 m, 1.0
m, and 1.5 m from the explosive. Metal plates were placed in the
Analysis of pre- and post-blast explosive residues in controlled laboratory southeast direction at 0.5 m and 1.0 m from the explosive. At the site,
detonations previously washed swabs (section 2.5. b), dried in a desiccator, were
moistened with two drops of DI from a Pasteur pipette and used to
Five gunpowder samples were received from the Technical Section sample rectangular areas of approximately 10 cm2 on the surfaces of the
for Ballistics and Firearm and Ammunition Identification at the Institute tiles and plates. Blank samples were collected from the metal and
of Criminalistics, Civil Police of Belo Horizonte, and numbered from 1 to ceramic surfaces before the explosions. After the explosions, samples
5. All samples were collected from materials seized by law enforcement were collected from the center of each tile and plate. If visible explosive
authorities. Gunpowder 1 was a nitrocellulose-based gunpowder (CBC residues were found elsewhere on the surfaces, they were also collected.
219, Brazil) but no information was provided on the source or compo­ All swabs were individually stored in sealed and labeled PP tubes, with
sition of gunpowders 2 to 5. The basic composition of these gunpowders all collection and storage performed by the same person. New plates
(2 to 5) was partially estimated using FTIR-ATR (Nicolet iZ10, Thermo were used for each of the two explosions.The extraction of ions from the
Scientific, United States), and their spectra were compared with those in swabs was carried out at the Institute of Criminalistics (Civil Police of
forensic FTIR libraries. Gunpowders 2, 3, and 4 were identified as po­ Minas Gerais, Belo Horizonte, Minas Gerais, Brazil) by adding 2 mL of DI
tassium nitrate, with spectral matches of 85.68 %, 82.25 %, and 81.54 to each swab and vortexing for 1 min. The extracts were filtered and
%, respectively. Gunpowder 5 exhibited low correlations, with its stored in a refrigerator until dilution and injection. Chromatographic
highest match (58.71 %) being to a smokeless powder listed in the analyses began on the same day as the collection.
’common materials’ library. For the pre-combustion analysis, 10 mg
samples of each gunpowder were added to 1 mL of DI in microtubes and Results and discussion
vortexed for 1 min. Gunpowders 2, 3, and 4 dissolved readily in water,
were diluted to a final concentration of 50 mg L− 1, and filtered before Optimization of chromatographic conditions
injection. Gunpowders 1 and 5 only fully dissolved in 1 mL of acetone
with vortexing for 1 min. For these two gunpowders, the acetone was In the chromatographic methods, ions potentially present in pre-
allowed to evaporate, and the residue was re-dissolved in 1 mL of DI, explosion devices, those generated after detonation, as well as other
filtered, and injected without dilution. For the post-combustion analysis, ions with retention times similar to the ions of interest, were included.
10 mg samples of each gunpowder were placed in cavities of a porcelain The goal was to achieve appropriate selectivity while minimizing anal­
spot plate that had been previously washed with DI. Using a propane- ysis time. For the cation analysis, five different gradients and one iso­
butane gas torch, each gunpowder was burned individually, and the cratic method were evaluated. For the anion analysis, six different
residues were collected using a Global Swab that had been washed four gradients were tested. In both cases, the column temperature was
times as described in section 2.5. b. The residue was extracted from the maintained at 30 ◦ C in a shared oven, with flow rates set at 0.250 mL
swab by adding 2 mL of DI and vortexing for 1 min. From this extract, 1 min− 1. The parameters used to choose the best separation methods were
mL was filtered and injected, with dilution applied when appropriate. based on the resolution of critical pairs of analytes, the asymmetry of
critical peaks, and the total analysis time. The results of these studies, for
Analysis of background ion levels in ATMs both cations and anions, are summarized in Table 4. These results are
also presented graphically in Fig. S6 and Fig. S7.
To assess the concentration of ions on ATM surfaces, a sampling was Table 4.
conducted, based on the work of Lahoda et al. [16] and the collection Based on the resolution of critical pairs, the asymmetry of the most
methodology of Johns et al. [10], at a ATM located in the Institute of asymmetric peaks, and the retention time of the most retained analytes,
Criminalistics building (Belo Horizonte, Minas Gerais, Brasil). Global Gradient 4 (Table 1) was selected for cations, and Gradient 6 (Table 2)
Swab swabs were washed four times with DI (section 2.5. b). The swabs, for anions. Although Gradient 4 for anions had the shortest retention

4
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

Table 4 resolution of the S2O2− 3−

3 /PO4 pair was significantly degraded. Addi­
Chromatographic separation parameters evaluated for critical analytes in the tionally, the asymmetry of the ClO−4 peak under these conditions was the
cation and anion groups. Values in parentheses represent the standard deviation most intense, leading to the conclusion that Gradient 6 was the most
(n = 2). suitable. Flow rate and column temperature changes were also evalu­
Method Resolution Retention time Asymmetry ated (results not shown). Although a higher flow rate reduced the
(min) retention time of the last analytes, it increased their asymmetry, and for
Mg2þ/ S2O2−
3 / Ba2þ ClO4− Ba2þ ClO4−
this reason the flow rate was maintained at 0.250 mL min− 1. The tem­
Ca2þ PO3−
4
perature effect was minimal and did not justify an increase, considering
Isocratic 2.53 − 11.28 − 1.64 − that lower temperatures contribute to a longer column lifetime. Fig. 1
(0.002) (0.007) (0.006)
Gradient 2.59 5.65 16.22 41.46 1.47 1.38
shows a typical chromatogram for cations (A) and anions (B) at the
1 (0.024) (0.031) (0.031) (0.129) (0.007) (0.009) optimized chromatographic conditions employed throughout the rest of
Gradient 2.59 4.48 15.33 37.31 1.43 1.39 the study.
2 (0.016) (0.002) (0.002) (0.000) (0.007) (0.013)
Gradient 2.23 3.46 11.39 34.14 1.34 1.17
3 (0.004) (0.011) (0.018) (0.028) (0.003) (0.027) Analytical performance evaluation
Gradient 2.21 1.46 11.20 28.36 1.33 1.87
4* (0.001) (0,015) (0.017) (0.009) (0.004) (0.033)
The performance evaluation was not conducted for CO2− 3 , as this ion
Gradient 2.03 2.37 9.94 29.88 1.34 1.37
5 (0.002) (0.002) (0.014) (0.057) (0.017) (0.083)
is predominantly converted to carbonic acid in the ion chromatography
Gradient − 2.67 − 29.69 − 1.31 suppressor, which is weakly dissociated and therefore exhibits low
6** (0,003) (0.009) (0.071) conductivity and analytical signal [10]. Given that the goal of the
methods analyzing explosives through ion chromatography is semi-
quantitative or purely qualitative determination, it is common to eval­
time for ClO−4 , suggesting a potentially shorter total run time, the
uate only a few performance parameters to estimate the reliability of the

Fig. 1. Typical chromatogram for cation analysis (Dionex Ion Pac CS12A column) with Gradient 4 as shown in Table 1 (A). Typical chromatogram for anion analysis
(Dionex Ion Pac AS19 column) with Gradient 6 as shown in Table 2 (B). Analytes at 5 mg L− 1, except for Sr2+ (12.5 mg L− 1) and Ba2+ (25 mg L− 1).

5
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

methods [12,17]. Although the method was developed for qualitative use, as the presence of this ion is only considered significant for explo­
purposes, linearity was estimated to provide a preliminary assessment of sive detection when present in much higher concentrations, on the order
this parameter for future studies involving full validation and quanti­ of several milligrams per liter [16,31]. The CO2− 3 concentration is not
tative purposes. The parameters evaluated for the 25 ions included in indicative of any specific explosive. There was no statistically significant
this article are detailed in Table 5. It can be observed that the vast difference at the 95 % confidence level between the average (n = 3) NO−3
majority of ions were adequately determined (R2 > 0.98) with sufficient concentrations across different brands of PP tubes (results not shown),
precision and accuracy. so the Gene brand was chosen due to its ample availability in the lab.
Selectivity was evaluated using DI and swab extract without fortifi­ These tubes were used without prior washing. Regarding the swabs,
cation (n = 3), obtained after washing as described in section 2.5 a. Fig. S9 shows images before and after the four sequential washes. A
Fig. S8A and B show the overlaid chromatograms from injections of noticeable change in the compactness of the cotton wool fibers was
unfortified swab extract, DI, and the first point of the evaluated linearity observed for Cotonetes and Mio brands. For the Global Swab brand,
range. It can be seen that for anion analysis (Fig. S8B), the water had however, almost no physical alteration was noted. The initial extracts
only one major signal at a retention time close to CO2− 3 , due to carbon from the swabs contained various ions, with the concentration and type
dioxide adsorption in the water, and the swab extract showed signals varying across the brands (Fig. 2, Fig. S10 A, and Fig. S10 B), as also
from ions that remain in residual quantities after washing. For cation noted by Mauricio et al. [12] and Szomborg et al. [17]. Overall, these
and anions analysis, no significant signals were observed in the water, ions were satisfactorily removed after the four washing stages, but
indicating that the DI used does not contain ion concentrations at levels Global Swab exhibited the best ion removal while maintaining fiber
that could interfere with the developed method (result not shown). On compactness (Fig. 2). This swab is made of rayon fibers with a plastic
the other hand, the most significant signal in the swab extract came from stem and is individually packaged in sterile conditions. After the four
Na+. However, for these less predictable ions like Na+, appropriate washes, all ions initially found in the Global Swab samples were reduced
selectivity is far more critical than precise quantification, given that this to below 0.25 mg L− 1.
ion is naturally present in the environment. The other two evaluated brands (Cotonetes and Mio) were purchased
Table 5. from local pharmacies and, according to the manufacturer, may contain,
in addition to cotton fibers, hydroxyethylcellulose, sodium benzoate,
Evaluation of ions naturally present in the collection material and washing citric acid, and germicide residue. Since the rayon swab from Global
procedures Swab had no significant concentrations of the ions of interest after
washing and maintained its shape and fiber structure, it was chosen for
The surface of the collection material (swabs) and the walls of the subsequent analyses. This material is also inexpensive and readily
transport and storage tubes were evaluated for the presence of potential available in large quantities in the laboratory. The four sequential
interfering ions and naturally occurring ions. Chromatograms from the washes with DI were used throughout the rest of the study.
DI that was into contact with syringes and filters showed no appreciable
signals (result not shown). Subsequently, the analyses were performed Evaluation of recovery
on the DI that was vortexed inside of 15 mL PP tubes. The only ions
found in concentrations above the lowest level of the calibration curve The study evaluated the recovery of ions from a porcelain surface
were NO−3 (approximately 0.1 mg L− 1) and CO2− 3 (approximately 1.0 mg using four groups of swabs: dry (dS), wet (wS), dry after washing (dwS),
L− 1). This concentration of NO−3 in the PP tubes does not preclude their and wet after washing (wwS). The washing process followed the

Table 5
Analytical performance parameters of each analyte evaluated.
Analyte Linear regression r2 Precision (CV%) Accuracy(d) (%) LOD LOQ
equation(a) 2.0 mg L− 1 5.0 mg L− 1
8.0 mg L− 1
2.0 mg 5.0 mg 8.0 mg mg mg
Intraday Interday Intraday Interday Intraday Interday L− 1 L− 1 L− 1 L− 1 L− 1

Ace− 0.0869 x + 0.1662 0.9832 6.7 23.7 2.5 5.7 1.7 4.0 97.3 110.7 104.1 0.534 1.781
Ba2+(b) 0.1367 x – 1.6154 0.9802 1.9 6.0 9.6 8.9 3.8 12.9 126.9 92.9 101.2 1.236 4.120
Br− 0.2354 x – 0.0914 0.9961 3.1 18.0 1.2 1.8 0.6 3.4 108.6 102.4 101.0 0.015 0.052
BrO−3 0.1257 x – 0.0521 0.9959 0.4 17.6 0.7 1.2 0.6 3.7 107.1 101.9 100.8 0.019 0.065
Ca2+ 0.4931 x – 0.5346 0.9574 7.6 19.8 10.6 12.2 4.7 10.7 87.5 93.7 102.1 0.152 0.508
Cit3− 0.1115 x – 0.0367 0.9949 1.1 25.6 0.8 6.7 0.6 5.9 108.2 103.8 101.9 0.089 0.298
Cl− 0.5027 x – 0.0347 0.9959 3.3 22.5 1.9 2.4 1.0 3.3 105.7 101.3 100.8 0.120 0.401
ClO−3 0.2017 x – 0.0927 0.9957 0.8 17.2 0.6 1.4 0.4 3.5 107.5 101.6 100.7 0.014 0.048
ClO−4 0.1389 x – 0.0786 0.9943 0.6 13.8 0.8 3.6 0.3 3.8 107.2 98.7 99.8 0.024 0.081
F− 0.8484 x – 0.0180 0.9963 0.5 19.7 0.7 1.1 0.6 3.6 107.7 104.4 101.1 0.061 0.204
For− 0.3572 x + 0.0344 0.9967 2.7 3.8 1.1 9.0 1.1 3.4 105.9 104.0 101.2 0.325 1.082
K+ 0.3236 x – 0.0402 0.9972 5.2 10.3 1.3 2.9 0.5 10.1 102.2 105.1 101.2 0.084 0.282
Li+ 1.2656 x – 0.7305 0.9947 6.3 7.1 1.9 1.9 0.7 11.6 99.7 102.3 101.4 0.090 0.300
Mg2+ 0.9899 x – 1.9960 0.9950 3.6 8.9 6.6 5.9 2.1 10.5 119.5 100.2 101.4 0.021 0.072
Na+ 0.9763 x + 3.9628 0.8652 19.5 45.1 20.5 21.6 11.8 20.4 168.7 137.3 97.3 0.579 1.932
NH+ 4 − 0.01x2 + 0.40x + 0.9911 8.6 15.1 2.7 3.4 2.1 12.1 108.7 101.4 100.2 0.204 0.681
0.27
NO−2 0.3477 x – 0.0522 0.9964 1.3 18.2 1.1 1.4 0.6 3.2 106.1 103.2 101.1 0.029 0.096
NO−3 0.2836 x + 0.1434 0.9940 9.9 78.7 8.2 14.0 4.3 6.7 98.9 107.5 103.9 0.289 0.963
Oxa− 2 0.3523 x + 0.0833 0.9967 1.9 20.8 0.9 5.3 0.8 4.5 103.4 104.1 101.2 0.111 0.371
PO43− 0.1621 x – 0.0593 0.9940 3.7 18.7 1.1 1.9 0.7 5.9 105.0 100.8 101.0 0.125 0.415
S2O2−
3 0.3122 x – 0.1121 0.9955 3.0 19.3 0.7 2.8 0.7 4.3 106.0 101.2 100.4 0.082 0.274
SCN− 0.2670 x – 0.1234 0.9902 6.5 16.2 3.6 2.9 2.1 4.4 107.2 100.5 102.0 0.030 0.099
2−
SO4 0.3679 x + 0.1987 0.9950 3.3 26.6 1.1 3.7 1.0 4.3 106.2 102.0 101.0 0.276 0.921
Sr2+(c) 0.2506 x – 1.3780 0.9869 2.4 8.9 8.2 7.5 3.2 11.7 122.6 95.2 101.4 0.638 2.128
Tar2− 0.1755 x + 0.0207 0.9958 1.8 23.3 0.7 2.9 0.9 5.1 106.3 103.9 101.7 0.101 0.335
(a)
Linearity range: 1.0 to 10.0 mg L− 1, except for Ba2+ (5.0 to 50.0 mg L− 1) and Sr2+ (2.5 to 25.0 mg L− 1); (b)Ba2+ concentrations for precision and recovery: 10.0; 25.0
and 40.0 mg L− 1; (c)Sr2+ concentrations for precision and recovery: 5.0; 12.5 and 20.0 mg L− 1; (d)Accuracy assessed through spiking and recovery testing.

6
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

Fig. 2. Percentages of ion concentrations in the wash extracts (n = 4) from Global Swab brand swabs. Error bars represent standard deviation (n = 3). * = initial
concentration (mg L− 1).

procedure described in section 2.5 b. Extracts from new and clean swabs Fig. 4 shows the concentrations of formate observed for each storage
taken from the porcelain surface showed no significant amounts of any condition over the 12-week period to illustrate the behavior of ions
ions. The efficiency of the collection process can be difficult to ensure undergoing degradation.
due to various factors such as applied pressure, number and direction of Therefore, to preserve this ion, it is recommended that analyses be
movements, and contact time, which makes consistency challenging in conducted within the first week after collection and that samples be
manual collections. Additionally, analyte behavior is influenced by stored in a refrigerator. No significant difference was observed between
electrostatic, van der Waals forces, and other chemical forces, especially dry and wet swab storage, indicating that the decision to use a wet swab
in solutions with high salt concentrations, leading to variability in the should be based more on ease of sampling and less on concerns about
results across different swab groups, as observed in Fig. 3A and B. stability. Most studies on the stability of explosive samples in collection
Despite this, most swabs showed good recovery for almost all ions, materials focus on organic explosives, which tend to be more volatile
except the dry swabs without prior washing (dS), which had the lowest [19,21–23,29]. The general recommendation from these studies is to
recovery rates. The use of water as a solvent is often preferred for col­ store them at low temperatures, in a refrigerator or freezer. Song-Im and
lecting inorganic residues, as it can solubilize analytes, break hydrogen colleagues [21] also analyzed NaClO3 and NH4NO3, finding no signifi­
bonds, solvate charges, and reduce electrostatic repulsion. This effect is cant losses of these compounds over a 30-day period in swab extracts,
supported by the literature, where swabs humidified with water had alcohol wipes, and glass surfaces at different temperatures (20 ◦ C, 10 ◦ C,
better performance in collecting cations and anions, as reported by and − 21 ◦ C), consistent with the results for most analytes in this study.
Song-Im et al. [27].
However, some anions, such as BrO−3 , S2O2− 3 , SCN , For , and NO2 ,
− − −

had very low recovery across almost all swab groups, suggesting possible Analysis of pre- and post-blast explosive residues in controlled laboratory
degradation or irreversible sorption on the surface or the swab. The best detonations
results for anion recovery were obtained with washed and wet swabs
(wwS), highlighting the importance of using water for effective collec­ Five different batches of gunpowder, numbered 1 to 5, were analyzed
tions. Based on these results, we have chosen to use wwS because these before and after combustion. The initial appearance of the gunpowder
procedures enhance collection efficiency, minimize variability, and in­ on a porcelain plate and its appearance after burning are shown in
crease the likelihood of successfully recovering a variety of ions. The Fig. S11. The concentrations of the main ions found in pre-combustion
recovery values significantly higher or lower than the expected 100 % and post-combustion samples are presented in Table 6, respectively. In
suggest that using materials not specifically designed for the collection samples 2, 3, and 4, which had dark granules and were easily solubilized
of ions at low concentrations can lead to erroneous conclusions. These in DI, NO−3 and K+ were predominant, suggesting they were likely
materials may release ions from particles or compounds naturally pre­ traditional black powders based on KNO3.
sent in the fabricated material (even after washing) and/or cause ion Table 6.
losses through irreversible adsorption. Gunpowder 1 was the only one with a known composition as indi­
cated by the manufacturer, coming from a package labeled “CBC 219″.
Its composition included nitrocellulose, ethyl centralite, and K2SO4,
Study of stability classifying it as a single-base powder (only nitrocellulose). The
morphology of gunpowders 1 and 5 was also characteristic, with more
In forensic analyses, preserving samples is crucial because, in many defined granules, indicating they were smokeless powders. These two
cases, re-collection is not an option. For swabs containing explosive gunpowders could only be solubilized with acetone, even after burning.
residues, it is important to determine how long the collected material On the other hand, gunpowders 2, 3, and 4 continued to exhibit high K+
can be stored before laboratory analysis and under what conditions the content even after burning. The high concentration of sulfate and S2O2−3
swabs should be stored [32,33]. To investigate this, swabs were fortified observed in the post-combustion extracts, which was not detected before
with ammonium formate solution in four different conditions: wet in a burning, likely results from the oxidation of sulfur, a primary component
refrigerator (about 4 ◦ C), wet at room temperature (about 22 ◦ C), dry in in traditional black powder formulation. Residues from black powder
a refrigerator, and dry at room temperature. Analyses were conducted in can also exhibit other characteristic ions such as S2− , SCN− , and OCN−
triplicate (n = 3) for each group over a 12-week period. Ammonium [14]. It is interesting to note that gunpowders 1 and 5 displayed a
formate showed a significant reduction in concentration when stored at distinct post-combustion profile compared to gunpowders 2, 3, and 4,
room temperature, particularly with wet swabs and from day 7 onwards. pointing toward smokeless powders. The presence of Oxa2− , though in

7
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

Fig. 3. Recovery of cations (A) and anions (B) by different groups of swabs on a porcelain surface. Error bars represent the standard deviation (n = 3).

small quantities, could indicate the replacement of sulfur with ascorbic was from the lateral area of the ATM. The background concentrations of
acid, which can produce threonic and oxalic acids upon degradation K+, NO−3 , and SO2− 4 , in the analyzed ATM could interfere with the
[33]. The small amounts of sulfate detected and the absence of thio­ detection of explosive traces in areas where fires occurred or where
sulfate confirm the use of compounds other than sulfur. firefighting activities caused leaching [5]. However, in most cases, the
concentration of explosive-related ions is expected to be significantly
higher than those in Table 7 [17].
Analysis of background ion levels in ATMs Table 7.
Lahoda and colleagues [16] found higher concentrations when they
The ion profile analysis of ATM surfaces involved injecting swab assessed environmental levels of explosive-related compounds in 28
extracts without dilution, and the concentrations found are presented in cities across the United States, noting that the collections were per­
Table 7. The cation analysis of these samples primarily revealed Na+, formed by different individuals, potentially affecting the recovered
K+, and Ca2+. In terms of anions, Cl− , NO−3 , SO2− 2− 3−
4 , Oxa , and PO4 were amounts. Of the 286 locations analyzed, 10 were ATMs. The ions found
detected. Extracts 1 to 3 were obtained from samples collected near the 2+ 2−
in ATMs were NH+ 4 , K , Na , Ca , Cl , NO2 , NO3 , and SO4 , with Cl
+ + − − − −

banknote exit, extract 4 was from the side of the display, and extract 5

8
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

Fig. 4. Formate concentration in swabs stored over 12 weeks under different conditions: wet and refrigerated (A); dry and refrigerated (B); wet at room temperature
(C); and dry at room temperature (D). Error bars represent the standard deviation (n = 3).

Table 6
Estimated concentrations (mg L− 1) of ions in aqueous extracts of gunpowder.
Ion Gunpowder 1 Gunpowder 2 Gunpowder 3 Gunpowder 4 Gunpowder 5

Pre-burning
NO−3 1.0 19.4 26.0 20.4 17.4
SO2−4 1.4 0.8 0.3 1.2 −
K+ 1.3 12.0 15.4 12.8 8.1
Post-burning
NO−3 7.5 9.6 7.5 20.0 4.2
SO2−4 1.4 376.0 166.6 632.4 0.4
K+ 1.0 409.2 268.4 1008.0 1.9
For− 20.2 − − − 3.2
NO−2 4.1 2.8 7.2 8.0 0.5
Oxa2− 1.7 − − − 0.2
Na+ 4.7 8.4 7.8 10.4 2.7
S2O2−3 − 125.2 84.0 138.6 −
Cl− − 3.9 2.2 4.6 −

being the most common ion, detected in 7 ATMs. These ions were found
Table 7
in average quantities close to 30 mg L− 1, with Na+ being the only
Ion concentrations (mg L− 1) on extracts of samples collected from different re­
exception, with an average of 81 mg L− 1. Ions of interest for explosive
gions of ATM surface.
analysis, such as NO−2 , ClO−3 , ClO−4 , CNO− , and SCN− , were detected in
Ion Extract 1 Extract 2 Extract 3 Extract 4 Extract 5
only about 3 % of the 286 samples, suggesting these ions may be good
Na +
5.8 1.7 1.5 2.7 5.1 indicators of explosive residues. Although the estimated concentrations
K+ 0.6 − 0.2 0.3 0.4 of the identified ions in Table 7 are relatively low, their presence
Ca2+ 0.2 1.3 1.1 1.9
highlights the limitations of qualitative methods and emphasizes the
−
Cl− 7.1 0.3 0.6 2.0 3.8
NO−3 1.2 2.1 3.2 3.6 5.3 need for more accurate quantitative analyses to improve reliability in
SO2−4 0.6 1.0 1.5 1.1 1.1 similar studies.
Oxa2− − − − 0.1 0.5
PO3−4 0.2 0.3 − − 0.1

9
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

Sampling and analysis of pre- and post-blast explosive residues in explosion, all ceramic tiles and metal plates were replaced with new
controlled field detonations ones. Unlike NH4NO3, gunpowder is a low explosive and deflagrates.
When confined, as in a pipe bomb, the hot gases produced after ignition
In evaluations of explosive residues from controlled field detona­ cause a rapid increase in pressure, resulting in an explosion [40]. The
tions, samples were collected from the local soil, ceramic floor tiles, and likelihood of unreacted material in the explosion is higher with mate­
metal plates before detonations for determination of background ion rials that deflagrate, as the propagation is slower and decomposition on
concentration. These initial analyses showed low levels of all target ions exposed surfaces during the process is typically less efficient [35]. The
(e.g. Na+ = 4.3 mg L− 1; Ca2+ = 2.2 mg L− 1; NO−3 = 1.7 mg L− 1; and SCN− key frames capturing the explosion moment are in Fig. S13A, depicting
= 2.3 mg L− 1). the fireball and material projection. The shock waves were much less
The first detonation used an NH4NO3 emulsion. As a high explosive, intense for this explosive and not strong enough to break the ceramic
NH4NO3 undergoes detonation with a propagation velocity of up to tiles (Fig. S13B), unlike the NH4NO3 emulsion.
6000 m s− 1 [34,35]. The key frames capturing the detonation moment Samples collected from the center of the ceramic floors tiles, even
are shown in Fig. S12A. The shock wave caused significant ground where no visible residues were present, showed primarily ClO−4 and K+
disturbance, breaking all ceramic tiles (Fig. S12B). Sampling was ions, confirming the use of chlorate-based gunpowder. The concentra­
focused on the center of ceramic tiles and on fragments showing visible tions observed in these samples were lower than those found in the
explosive residues. Due to extensive fragmentation, samples were not NH4NO3 samples, but still indicative of explosive use. Fig. 6A and B
collected from the center at the West and South positions (both 50 cm show the ClO−4 and K+ concentrations, respectively.
away from center). In the central region samples, NO−3 and NH+ 4 were The highest concentrations were at 100 cm East of the detonation
found in higher concentrations. Despite the high detonation speed, the center (K+= 54 mg L− 1 and ClO−4 = 115 mg L− 1). Only in the West di­
kinetic process depends on factors like charge diameter, particle size, rection was there a clear trend of decreasing concentration with
moisture, and porosity [36], which can lead to unreacted explosive increasing distance, which may be due to the sampling process and non-
residues [34,35]. Fig. 5A and B show the NO−3 and NH+ 4 concentrations, uniform residue distribution. This heterogeneous behavior has been
corrected for dilution, based on the collection points. The highest con­ observed in other studies [37,41–43]. Samples from visible residues
centrations were at 50 cm East (NO−3 = 604 mg L− 1 and NH+ 4 = 103 mg showed significantly higher K+ and ClO−4 concentrations (up to 362.8
L− 1) and 50 cm North (NO−3 = 1013 mg L− 1 and NH+ 4 = 251 mg L ).
− 1
mg L− 1 and 782.8 mg L− 1, respectively) than those collected from the
Lower concentrations were found in samples from the South direction, center of tiles, suggesting that visible residue sites should be prioritized
yet even at 150 cm away, NO−3 = 74 mg L− 1 and NH+ 4 = 5.2 mg L
− 1
were in sampling. Although concentrations were below 10 mg L− 1, almost all
detected. samples collected after the gunpowder explosion contained Sr2+. No
The dispersion of explosive residues is complex, depending on factors NH+ 4 was detected in any of the samples after the gunpowder explosion,
like detonation efficiency, explosive charge shape, size and mass of but some samples contained NO−3 , with a maximum concentration of 18
fragments, detonator position, and surrounding arrangements (furni­ mg L− 1. Some visible residue samples also contained SO2− 4 and S2O3 ,
2−
− 1 − 1
ture, ATM type, room setup, etc.) [15,35]. While intuitively, an increase with maximum concentrations of 38 mg L and 34 mg L , respec­
in distance from the explosion center leads to lower detected residues tively, indicating the use of sulfur as a fuel [14,44–46], suggesting that
(the inverse square law), experimental observations do not always align some black powder might have been mixed with chlorate-based gun­
with this principle [15,30,35,37]. Besides NO−3 and NH+ 4 , some samples powder for the explosive device. These samples had lower ClO−4 and K+
showed high Na+ concentrations (data not shown). Na+ can be added to concentrations compared to samples without SO2− 2−
4 and S2O3 .
NH4NO3 explosives as NaCl or NaNO3, the former to lower the explosion Johns et al. [8] analyzed residues from perchlorate-sugar explosives
temperature [38] and the latter to prepare supersaturated NH4NO3 and found ClO−4 and K+ concentrations around 20 and 30 mg L− 1,
emulsions, increasing stability [39]. Johns et al. [8] also found signifi­ respectively, values considered indicative of explosive detection.
cant Na+ levels in a commercial ANFO sample, suggesting NaCl addition Lahoda et al. [16] found ClO−4 in only 1 % of 286 samples analyzed from
in its formulation. Samples from the center of metal plates showed much different surfaces, with most readings below 15 µg. K+ was more com­
higher NO−3 and NH+ 4 concentrations than those from ceramic tiles (e.g. mon, found in 44 % of samples with an average of only 25 µg. Thus, even
NO−3 = 5000 mg L− 1 and NH+ − 1
4 = 1095 mg L ). This aligns with studies the concentrations found in samples from the center of ceramic tiles
indicating different recovery rates based on surface type [17,27,28]. indicate the use of perchlorate-based gunpowder. The detailed compo­
After the detonation of NH4NO3 emulsion and before the pipe bomb sitions of the explosives were not supplied, and their inclusion could

Fig. 5. NO−3 (A) and NH+4 (B) concentrations found in samples collected from the center of the plates after the NH4NO3 explosion. Cardinal points indicated by letters
N (North), S (South), E (East) and W (West).

10
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

Fig. 6. ClO−4 (A) and K+ (B) concentrations found in samples collected from the center of the plates after the gunpowder explosion. Cardinal points indicated by
letters N (North), S (South), E (East) and W (West).

further enrich the results and conclusions of this work. Nevertheless, the CRediT authorship contribution statement
main components declared by BOPE—ammonium nitrate (for the
emulsion) and potassium perchlorate (for the pipe bomb)—are consis­ Paula Ribeiro Junho: Writing – review & editing, Validation,
tent with the results presented in Fig. 5 and Fig. 6. Methodology, Investigation, Formal analysis, Conceptualization. Yuri
Quantitative determination of ions in residues is generally not crit­ Machado: Writing – review & editing, Validation, Supervision, Re­
ical in post-explosion analysis, as the percentage of original material sources, Methodology, Investigation, Formal analysis, Conceptualiza­
consumption depends on many factors [13,14]. What matters is that the tion. Ricardo M. Orlando: Writing – review & editing, Validation,
detected concentration is high enough to differentiate from environ­ Supervision, Resources, Methodology, Investigation, Formal analysis,
mental background. Overall, the proposed sampling and analysis Conceptualization.
method was capable of identifying residues from NH4NO3 and chlorate-
based gunpowder explosions, allowing differentiation between these Declaration of competing interest
two explosive materials.
The authors declare that they have no known competing financial
Conclusions interests or personal relationships that could have appeared to influence
the work reported in this paper.
The development of the method for separating cations and anions
proved crucial to achieving adequate resolution and selectivity in the Acknowledgments
shortest possible analysis time with suitable chromatographic condi­
tions. During the development of the sample collection procedure, it was The authors are grateful to the financial support from Rede Mineira
observed that the choice and preparation of swabs are key to avoiding de Ciências Forenses (project RED-00120-23), INCTAA/CNPq/FAPESP
false-positive results due to contamination from naturally occurring (projects 465768/2014-8 and 2014/50951-4), PROCAD Segurança
ions. Sequential washes with DI were found to remove most of these Pública e Ciências Forenses (process 88881.516313/2029-01, public
ions. Additionally, it was noted that using a humidified swab at the time notice n. 16/2020). And also thanks to the Superintendência de Polícia
of collection increases ion recovery, especially for anions. When exam­ Técnico-Científica da Polícia Civil de Minas Gerais for providing access
ining the stability of ions in swabs, it was found that the degradation of to samples, sharing space and equipment; to the commander and spe­
For− , Ace− , Oxa− , and Tar2− during the first few days of storage at room cialists from the Batalhão de Operações Policiais Especiais (BOPE) da
temperature may be a problem to consider. However, this issue can be Polícia Militar de Minas Gerais; and to the technical support provided by
effectively mitigated by refrigerating the swabs (approximately 4 ◦ C) for Police Officer Frederico Nunes from the Seção de Balística da Polícia
up to 12 weeks. Therefore, it’s concluded that swabs collected from Técnico-Científica da Polícia Civil de Minas Gerais.
potential explosion sites should ideally be analyzed within the first week
and, whenever possible, stored at lower temperatures. The analysis of Appendix A. Supplementary data
samples collected after two controlled explosions allowed for the iden­
tification of the primary ions from the explosives used, as well as their Supplementary data to this article can be found online at https://doi.
differentiation. NO−3 and NH+ 4 were detected after the detonation of org/10.1016/j.forc.2024.100627.
NH4NO3 emulsion, and ClO−4 and K+ were found after the explosion of a
device containing chlorate-based gunpowder. Although each controlled Data availability
explosion was conducted only once for each material analyzed, it was
observed that ion concentrations decreased as the distance from the The authors do not have permission to share data.
central point of the explosion increased. Therefore, it is suggested that
sampling be done as close as possible to the center of the explosion and References
from any visible residues to maximize the recovery of ions. This infor­
mation can aid forensic investigations involving the analysis of explo­ [1] Fórum Brasileiro de Segurança Pública, Anuário Brasileiro de Segurança Pública
sive residues and guide future studies, including a greater number of 2019 <https://www.forumseguranca.org.br/wp-content/uploads/2019/10/
Anuario-2019-FINAL_21.10.19.pdf>, (2019), accessed 2024-05-02.
ATMs and sampling points, as well as a larger number of controlled [2] L.P.L. Logrado, M.N. Silva, J.C.A. Laboissiere, J.W.B. Braga, Profile of explosives’s
explosions across various surface types and environments. use in ATMs/cash safes robberies in Brazil, J. Forensic. Sci. 67 (2022) 1441–1449.

11
P.R. Junho et al. Forensic Chemistry 42 (2025) 100627

[3] F.G.M. Mauricio, A.Z. Pralon, M. Talhavini, M.O. Rodrigues, I.T. Weber, [24] C. Pohl, N. Avdalovic, K. Srinivasan (Eds.), Ion Chromatography: Instrumentation,
Identification of ANFO: Use of luminescent taggants in post-blast residues, Forensic Techniques and Applications, first ed., Academic Press, 2021.
Sci. Int. 275 (2017) 8–13, https://doi.org/10.1016/j.forsciint.2017.02.029. [25] Thermo Fisher Scientific Inc., Product manaual for IonPac CS12A and IonPac
[4] F. Zapata, M.Á.F. Ossa, E. Gilchrist, L. Barron, C. García-Ruiz, Progressing the CG12A Columns. <https://tools.thermofisher.com/content/sfs/manuals/4340-
analysis of Improvised Explosive Devices: Comparative study for trace detection of Man-031132-09-IonPac-CS12A-May10.pdf>, (2010), accessed 2024-05-02.
explosive residues in handprints by Raman spectroscopy and liquid [26] Thermo Fisher Scientific Inc., IonPac AS19 Product Manual. <https://www.
chromatography, Talanta 161 (2016) 219–227, https://doi.org/10.1016/j. thermofisher.com/document-connect/document-connect.html?url=https://assets.
talanta.2016.05.057. thermofisher.com/TFS-Assets%2FCMD%2Fmanuals%2FMan-065003-IonPac-
[5] J.P. Hutchinson, C. Johns, G.W. Dicinoski, L. Jones, M.C. Breadmore, P.R. Haddad, AS19-Columns-Man065003-EN.pdf>, (2013), accessed 2024-05-02.
Identification of Improvised Inorganic Explosive devices by analysis of postblast [27] N. Song-Im, S. Benson, C. Lennard, Evaluation of different sampling media for their
residues using ion chromatography and capillaty electrophoresis, Encycl. Anal. potential use as a combined swab for the collection of both organic and inorganic
Chem. (2020) 1–14, https://doi.org/10.1002/9780470027318.a9031.pub2. explosive residues, Forensic Sci, Int. 222 (2012) 102–110, https://doi.org/
[6] F.R. Aragão, Explosivos e explosões químicas e mecânicas, in: R. F. Aragão (Ed.), 10.1016/j.forsciint.2012.05.006.
Incêndios e Explosivos: uma Introdução à Engenharia Forense, second ed., [28] L.E. DeGreeff, H.P.H. Liddell, W.R. Pogue III, M.H. Merrill, K.J. Johnson, Effect of
Millennium (2020). re-use of surface sampling traps on surface structure and collection efficiency for
[7] J. Akhavan, The Chemistry of Explosives, third ed., The Royal Society of Chemistry trace explosive residues, Forensic Sci. Int. 297 (2019) 254–264, https://doi.org/
(2011). https://doi.org/10.1039/9781839168802. 10.1016/j.forsciint.2019.02.002.
[8] C. Johns, R.A. Shellie, O.G. Potter, J.W. O’Reilly, J.P. Hutchinson, R.M. Guijt, M. [29] R.V. Taudte, C. Roux, A. Beavis, Stability of smokeless powder compounds on
C. Breadmore, E.F. Hilder, G.W. Dicinoski, P.R. Haddad, Identification of collection devices, Forensic Sci. Int. 270 (2017) 55–60, https://doi.org/10.1016/j.
homemade inorganic explosives by ion chromatographic analysis of post-blast forsciint.2016.11.027.
residues, J. Chromatogr. a. 1182 (2008) 205–214, https://doi.org/10.1016/j. [30] N. Abdul-Karim, C.S. Blackman, P.P. Gill, K. Karu, The spatial distribution patterns
chroma.2008.01.014. of condensed phase post-blast explosive residues formed during detonation,
[9] G.W. Dicinoski, R.A. Shellie, P.R. Haddad, Forensic identification of inorganic J. Hazard. Mater. 316 (2016) 204–213, https://doi.org/10.1016/j.
explosives by ion chomatography, Anal. Lett. 39 (2006) 639–657, https://doi.org/ jhazmat.2016.04.081.
10.1080/00032710600609735. [31] L. Barron, E. Gilchrist, Ion cromatography-mass spectrometry: a review of recent
[10] J. Weiss, Handbook of Ion Chromatography, third ed., Wiley-VHC (2004). https:// technologies and applications in forensic and environmental explosives analysis,
doi.org/10.1002/9783527619610. Anal. Chim. Acta 806 (2014) 27–54, https://doi.org/10.1016/j.aca.2013.10.047.
[11] K. Evans-Nguyen, An introduction to instrumentation used in fire debris and [32] S.E. Klassen, T.M. Massis, E.P. Boespflug, B.M. Montoya, J.L. Reif, Ion
explosive analysis, in: K. Evans-Nguyen, K. Hutches (Eds.), Forensic Analysis of Fire chromatography of energetic materials at Sandia National Laboratories,
Debris and Explosives, first ed., Springer Nature (2019). Thermochim. Acta 384 (2002) 329–341, https://doi.org/10.1016/S0040-6031(01)
[12] F.F.G.M. Mauricio, V.R.M. Abritta, R.L. Aquino, J.C.L. Ambrósio, L.P.L. Logrado, I. 00790-0.
T. Weber, Evaluation of interferers in sampling materials used in explosive residue [33] G.L. Lang, K. Boyle, The analysis of black powder substitutes containing ascorbic
analysis by ion chromatography, Forensic Sci. Int. 307 (2020) 109908, https://doi. acid by ion chromatography/mass spectrometry, J. Forensic. Sci. 54 (2009)
org/10.1016/j.forsciint.2019.109908. 1315–1322, https://doi.org/10.1111/j.1556-4029.2009.01144.x.
[13] D.J. Klapec, G. Czarnopys, J. Pannuto, Interpol review of detection and [34] S. Doyle, Improvised explosives, in: J.A. Siegel, P.J. Saukko, M.M. Houck (Eds.),
characterization of explosives and explosive residues 2016-2019, Forensic Sci, Int.: Encyclopedia of Forensic Sciences, second ed., Academic Press (2013).
Synergy 2 (2020) 670–700, https://doi.org/10.1016/j.fsisyn.2020.01.020. [35] N. Abdul-Karim, C.S. Blackman, P.P. Gill, E.M.M. Wingstedt, B.A.P. Reif, Post-blast
[14] M. Hargadon, Explosive residue analysis by capillary electrophoresis and ion explosive residue – a review of formation and disperson theories and experimental
chromatography, J. Chromatogr. a. 602 (1992) 241–247, https://doi.org/10.1016/ research, RSC Adv. 4 (2014) 54354–54371, https://doi.org/10.1039/
0021-9673(92)80087-B. C4RA04195J.
[15] R. Borusiewicz, G. Zadora, J. Zieba-Palus, Chemical analysis of post explosion [36] A. Nazarian, C. Presser, Forensic methodology for the thermochemical
samples obtained as a result of model field experiments, Talanta 116 (2013) characterization of ANNM and ANFO homemade explosives, Thermochim. Acta
630–636, https://doi.org/10.1016/j.talanta.2013.07.050. 608 (2015) 65–75, https://doi.org/10.1016/j.tca.2015.04.006.
[16] K.G. Lahoda, O.L. Collin, J.A. Mathis, H.E. LeClair, S.H. Wise, B.R. McCord, [37] P. Kolla, A. Sprunkel, Identification of dynamite explosives in post explosion
A survey of background levels of explosives and related compounds in the residues, J. Forensic Sci. 40 (1995) 406–411, https://doi.org/10.1520/JFS13795J.
environment, J. Forensic Sci. 53 (2008) 802–806, https://doi.org/10.1111/j.1556- [38] R. Meyer, J. Köhler, A. Homburg, Explosives, seventh ed., Wiley-VHC, (2015).
4029.2008.00743.x. https://doi.org/10.1002/9783527689583.
[17] K. Szomborg, F. Jongekrijg, E. Gilchrist, T. Webb, D. Wood, L. Barron, Residues [39] P. Yu, N.N.A. Ling, E.O. Fridjonsson, M.L. Johns, The effect of inert salts on
from low-order energetic materials: the comparative performance of a range of explosive emulsion thermal degradation, Propellants Explos. Pyrotech. 4 (2021)
sampling approaches prior to analysis by ion chromatography, Forensic Sci. Int. 360–367, https://doi.org/10.1002/prep.202000258.
233 (2013) 55–62, https://doi.org/10.1016/j.forsciint.2013.08.018. [40] M. Marshall, J.C. Oxley, Explosives: the threats and the materials, in: M. Marshall,
[18] C. Black, T. D’Souza, J.C. Smith, N.G.R. Hearns, Identification of post-blast J.C. Oxley (Eds.), Aspects of Explosives Detection, Elsevier, Amsterdam, 2009,
explosive residues using direct-analysis-in-real-time and mass spectrometry pp. 11–26.
(DART-MS), Forensic Chem. 16 (2019) 100185, https://doi.org/10.1016/j. [41] T.F. Jenkins, Use of snow-covered ranges to estimate explosives residues from high-
forc.2019.100185. order deetonations of army munitions, Thermochim. Acta 384 (2002) 173–185,
[19] D.A. DeTata, P.A. Collins, A.J. McKinley, A comparison of common swabbing https://doi.org/10.1016/S0040-6031(01)00803-6.
materials for the recovery of organic and inorganic explosive residues, J. Forensic. [42] S. Taylor, A. Hewitt, J. Lever, C. Hayes, L. Perovich, P. Thorne, C. Daghlian, TNT
Sci. 58 (2013) 757–763, https://doi.org/10.1111/1556-4029.12078. particle size distributions from detonated 155-mm howitzer rounds, Chemosphere
[20] C. Calhoun, R.F. Mothershead, Explosive analysis: introduction to post-blast 55 (2004) 357–367, https://doi.org/10.1016/j.chemosphere.2003.11.031.
analysis, in: K. Evans-Nguyen, K. Hutches (Eds.), Forensic Analysis of Fire Debris [43] R.Q. Thompson, D.D. Fetterolf, M.L. Miller, R.F. Mothershead, Aqueous recovery
and Explosives, Springer Nature, (2019). from cotton swabs of organic explosive residue followed by solid phase extraction,
[21] N. Song-Im, S. Benson, C. Lennard, Stability of explosive residues in methanol/ J. Forensic Sci. 44 (1999) 795–804, https://doi.org/10.1520/JFS14555J.
water extracts, on alcohol wipes and on a glass surface, Forensic Sci. Int. 226 [44] J.P. Agrawal, High Energy Materials, first ed., Wiley-VCH, (2010). https://doi.org/
(2013) 244–253, https://doi.org/10.1016/j.forsciint.2013.01.037. 10.1002/9783527628803.
[22] N.N. Daeid, A.Y. Holly, M.S. Beardah, Investigatin TNT loss between sample [45] O.N. Neto, Análises periciais envolvendo explosivos, in: R. F. Aragão (Ed.),
collection and analysis, Sci. Justice 57 (2017) 95–100, https://doi.org/10.1016/j. Incêndios e Explosivos: uma Introdução à Engenharia Forense, second ed.,
scijus.2016.10.007. Millennium (2020).
[23] E. Sisco, M. Najarro, D. Samarov, J. Lawrence, Quantifying the stability of trace [46] J. Cummins, J. Hull, K. Kittsa, J.V. Goodpaster, Separation and identification of
explosives under different environmental conditions using electrospray ionization anions using porous graphitic carbon and electrospray ionization mass
mass spectrometry, Talanta 165 (2017) 10–17, https://doi.org/10.1016/j. spectrometry: Application to inorganic explosives and their post blast residues,
talanta.2016.12.029. Anal. Methods 3 (2011) 1682–1687, https://doi.org/10.1039/C1AY05143A.

12