Characterization of M4 carbine rifle emissions with three ammunition types

Abstract

Muzzle emissions from firing an M4 carbine rifle in a semi-enclosed chamber were characterized for an array of compounds to provide quantitative data for future studies on potential inhalation exposure and rangeland contamination. Air emissions were characterized for particulate matter (PM) size distribution, composition, and morphology; carbon monoxide (CO); carbon dioxide (CO₂); energetics; metals; polycyclic aromatic hydrocarbons; and methane. Three types of ammunition were used: a “Legacy” (Vietnam-era) round, the common M855 round (no longer fielded), and its variant, an M855 round with added potassium (K)-based salts to reduce muzzle flash. Average CO concentrations up to 1500 ppm significantly exceeded CO₂ concentrations. Emitted particles were in the respirable size range with mass median diameters between 0.33 and 0.58 μm. PM emissions were highest from the M855 salt-added ammunition, likely due to incomplete secondary combustion in the muzzle blast caused by scavenging of combustion radicals by the K salt. Copper (Cu) had the highest emitted metal concentration for all three round formulations, likely originating from the Cu jacket on the bullet. Based on a mass balance analysis of each round’s formulation, lead (Pb) was completely emitted for all three round types. This work demonstrated methods for characterizing emissions from gun firing which can distinguish between round-specific effects and can be used to initiate studies of inhalation risk and environmental deposition.

Graphical abstract

1. Introduction

Limited information is available on emissions from gun firing and the resulting inhalation exposure and environmental deposition. This is likely due to the difficulties in quantitatively sampling emissions from an unconfined source. Changes in ammunition composition, such as the use of lead-free bullets, remain minimally characterized. Recent work has begun to evaluate warfighter exposure to particle inhalation due to repeated, enclosed range firings (Grabinski et al., 2017; Voie et al., 2014; Sikkeland et al., 2018; Borander et al., 2017), raising health concerns about prolonged exposure. The toxicity hazard from energetic material testing may result from skin contact with solid or liquid residues, contact with contaminated water, contact with unconsumed energetic material, breathing aerosolized particulate matter, or breathing toxic permanent gases produced during explosive energy release.

Likewise, ground deposition of emissions, particularly on training ranges, presents an unknown subsequent exposure and environmental contamination. Some work has begun to characterize deposited residues from a variety of weapon systems (Brochu et al., 2011; Walsh et al., 2012). Particles containing metals, organics, and salts are likely to deposit on the area around the range or even be transported off-site. Depending on the chemical and the properties of the soil, solubilization of contaminants, and subsequent ground water contamination may be a significant concern (Fuller et al., 2014).

Emissions of interest from gun firings include metals and organic combustion byproducts. Sources of metal in the ammunition round include the projectile or projectile, the casing material, the primer, and the propellant. Spalling of the gun barrel lining itself is an additional metal source. Depending on the composition of the ammunition, metals may include lead (Pb), chromium (Cr), nickel (Ni), zinc (Zn) and copper (Cu) (Grabinski et al., 2017), among others. The propellant ignitor (primer) can contain metals such as Pb, barium (Ba), and antimony (Sb) compounds (Dalby et al., 2010) while other metallic compounds such as bismuth (Bi) and antimony (Sb) are designed to mitigate Cu deposits in the rifled barrel. The double base (nitrocellulose, NC, and nitroguanidine, NG) propellant may also contain Ba/K salts as a flash suppressant. Other neat energetic materials containing carbon-nitrogen (C–N) and nitrogen-hydrogen (N–H) bonds may also produce toxic permanent gases (e.g., cyanide (CN), ammonia (NH₃)) which are hazardous to humans.

Sampling gun emissions presents significant measurement challenges. The rapid dispersion of the emissions must be accounted for to be quantitative in deriving total emissions per firing. While the literature in this area is sparse, recent work by Wingfors et al. (2014) constructed a polymethylmethacrylate (PMMA) chamber into which multiple rounds of lead-free munitions were fired. Emission factors were calculated by multiplying the time-averaged concentration (mass/sample volume) of the compound sampled by the chamber volume, then dividing by the number of firings (ten). The emissions were found to be rich in particles at 29–30 mg/round. Metals and organics were also reported in relation to their reported composition in the ammunition. More historic work with M16 rifle firings (Ase et al., 1985) has identified relatively high levels of CO (over 30% by mass compared to the propellant mass), hydrogen cyanide (HCN), carbon disulfide (CS₂), polycyclic aromatic hydrocarbons (PAHs), and metals, most notably Ba, Cu, Pb, Sb, and Zn.

We build on previous small arms emissions measurements (Brochu et al., 2011; Wingfors et al., 2014; Ase et al., 1985) by developing and testing methods to quantify and characterize a range of pollutants to determine if varying ammunition formulations result in differences that may impact personal exposure and environmental contamination. For this work, we used the common M4 carbine rifle with three versions of the M855 ammunition. Targeted pollutants included CO, CO₂, volatile organic compounds (VOCs), particulate matter < 2.5 μm (PM_2.5), PM < 10 μm (PM₁₀), PM elemental composition, particle size distribution, PAHs, and nitroaromatics. Results are expressed in terms of emission factors, or the mass of emissions per gun firing and mass of emissions per mass of fuel (propellant plus primer), the latter to more easily relate to different propellant formulations. These emission factors are the critical values necessary for informed prediction of risk for human exposure and range contamination.

2. Materials and methods

2.1. Ammunition and rifle

Gun firings were conducted at the U.S. Army Research Laboratory (ARL), Aberdeen Proving Ground, MD (U.S.A.). An M4 carbine rifle was selected for study due to its prevalence with U.S. military and North American Treaty Organization (NATO) forces for the last 20 years. The barrel length is 36.8 cm plus the length of “A2” muzzle, or “birdcage” device which is used to block/suppress light emission from incandescing gases/particles immediately following bullet exit. The M4 weapon was tested in operational configuration and was not modified for the testing. It was capable of being switched between semiautomatic and 3-shot burst mode but was operated in single-shot mode. M855 ammunition was selected because it is available in multiple compositions with the same base propellant formulations (nitrocellulose ((C₆H₇(NO₂)₃O₅)_n)/nitroglycerin (C₃H₅N₃O₉) “double-base”) and contains Pb in the projectile, affording a chance to study the fate of metals. Three types of M855 ammunition were tested, each of which used a M41 primer containing Pb, aluminum (Al), Ba, Sb and a Cu jacketed lead bullet, 5.56 mm in diameter. The ammunition consisted of a 1990s era M855 round, its “salted” counterpart which contains K for flash suppression, and an older M855 Vietnam-era (1960s) “legacy” round. While the M855 is in common sport use, this is not the same formulation as in the current military-fielded M855A1 “green” lead-free ammunition. A total of 20 shots were analyzed for the three ammunition types. The composition of the ammunition used here is described more fully in Supplemental Information (SI) Tables SI–1. The metal and carbon fraction for these rounds are found in Tables SI–2.

2.2. Test chamber and test setup

A PMMA rectangular enclosure (132 cm across, 76 cm high, and 66 cm in depth for a total volume of 0.3 m³) surrounded the M4 carbine muzzle to contain the rifle firing exhaust (Fig. 1). A 5 cm diameter hole, later enlarged to 15 cm to mitigate the shock wave stress on the enclosure, allowed for exit of the bullet. The clear side of the box facilitated imaging measurements using high speed cameras and spectrographs. The enclosure had a hinged side wall that could be opened for insertion of samplers and cleaning of the wall surfaces (Windex®) between firings (Fig. 1). The time between each round fired was approximately 20 min with 1 h between different ammunition types for cleaning of test chamber and resetting of instruments. Multiple sampler inlets were positioned approximately 15 cm below the center of the muzzle. Bulk PM_2.5, PM₁₀, and PAH sampler inlets were inside the enclosure. All other samples were extracted from the enclosure through stainless steel sample lines.

Fig. 1.

Schematic of test chamber and test setup superimposed on shadowgraph image (not to scale).

2.3. Target analytes and sampling methods

The target analytes, their sampling techniques/instruments, and methods are listed in Table 1. The CO₂ (LICOR-820, Lincoln, NE, U.S.A.) and the CO (E2vEC4–500-CO, SGX Sensortech, United Kingdom) sensors were calibrated in accordance with U.S. EPA Method 3A (U.S. EPA Method 3A, 2017), undergoing a daily three point, zero, and calibration drift test. A precision dilution calibrator, Serinus Cal 2000 (American ECOTECH L.L.C., Warren, RI, U.S.A.), was used to dilute the high-level span gases to appropriate levels for the CO₂ and CO calibration curves. The E2v CO sensor has a CO detection range of 1–500 ppm with resolution of 1 ppm. The LICOR-820 has an analytical range of 0–20,000 ppm with an accuracy specification of less than 3% of reading. The LICOR-820 calibration range for CO₂ was set to 0–3,000 ppm.

Table 1.

Target analytes, instrumentation, and methods.

Analyte	Technique/Instrument	Frequency/Sampling rate	Method reference
CO2	LICOR-820, NDIRa	Continuous 1 Hz	U.S. EPA Method 3A (U.S. EPA Method 3A, 2017)
CO	e2VEC4–500-CO, Electrochemical cellb	Continuous 1 Hz	U.S. EPA Method 3A (U.S. EPA Method 3A, 2017)
PM2.5	SKC Impactorc, 47 mm Teflon filter 2 μm pore size	Batch – 10 L/min constant flowd	40 CFR 50, Appendix L (40 CFR Part 50Appendix, 1987)
PM10	SKC Impactorc, 47 mm Teflon filter 2 μm pore size	Batch – 10 L/min constant flowd	40 CFR 50, Appendix J (40 CRF Part 50Appendix, 1987)
Nitrocellulose	47 mm Quartz fiber filter	Batch – 10 L/min constant flowd	U.S. EPA Method 353.2 (U.S. EPA, 1993)
Nitroaromatics	47 mm Quartz fiber filter	Batch – 10 L/min constant flowd	U.S. EPA Method 8330b (U.S. EPA Method 8330B, 2006)
PAHs	Quartz fiber filter, PUFe, XAD-2	Batch – 5 L/min constant flowd	Modified U.S. EPA Method TO-9A (U.S. EPA Compendium, 1999b)
Elements	Teflon filter from PM2.5 batch filter, XRFf	Batch – 10 L/min constant flowd	EPA Compendium Method IO-3.3 (>U.S. EPA Compendium, 1999a)
Elemental, Organic Carbon	Quartz filter	Batch - 10 L/min constant flow	Panteliadis et al. (2015)
VOCs	6 L SUMMA Canister	Batch – 0.5 L/min	U.S. EPA Method TO-15 (U.S. EPA, 1999)
CO, CO2, CH4	6 L SUMMA Canister	Batch – 0.5 L/min	Modified U.S. EPA Method 25C (U.S. EPA, 2017)
PN, Size Distribution, Composition	Electrical Low Pressure Impactorg	Continuous & Batch	(Kero and Jorgensen, 2016) and (Sowards et al., 2008)
PM Morphology	Scanning Electron Microscope Energy Dispersive XRayh	Batch	Casuccio et al. (2004)

^aLI-COR Biosciences, USA. NDIR - Non-dispersive infrared.

^bSGX Sensortech, United Kingdom.

^cSKC Inc., USA.

^dLeland Legacy sample pump, SKC Inc., USA.

^ePUF - Polyurethane foam.

^fXRF – X-ray fluorescence.

^gDekati, Finland.

^hMIRA3 Tescan, Czech Republic.

PM_2.5 and PM₁₀ were sampled with SKC impactors on 47 mm Teflon filters with a pore size of 2.0 μm using a Leland Legacy sample pump (SKC Inc., Eighty Four, PA, U.S.A.) with a constant airflow of 10 L/min. The Leland Legacy sample pump was calibrated with a Gilibrator Air Flow Calibration System 610 (Sensidyne LP, St. Petersburg, FL, U.S.A.). The sampling time for the PM impactors ranged from 2 to 9 min, representing composite samples from multiple shots and uncertainty over how much sample was sufficient.

The elemental composition of PM was measured by energy dispersive x-ray fluorescence spectrometry (ED-XRF) following U.S. EPA Compendium Method IO-3.3 (U.S. EPA Compendium, 1999a). Measurement of 13 standard reference materials resulted in recoveries (experimental value divided by given value) between 92 and 109%. The PM elemental carbon (EC) and organic carbon (OC) were measured following the NIOSH 930 protocol cited in Panteliadis et al. (2015) using a Sunset Laboratory Carbon Aerosol Analyzer.

Gas phase and particle-bound PAHs were collected using polyurethane foam, XAD-2, and a quartz filter, extracted together with toluene, concentrated, and then analyzed using a Modified U.S. EPA Method 8270D (U.S. EPA, 1998) on a Thermo GC Trace 1310/ISQ (ThermoScientific, Inc., Milan, IT/Austin, TX, U.S.A.) using a DB-5 chromatographic column (Agilent technologies, Santa Clara, CA, U.S.A.). The method modifications included addition of pre-sampling, pre-extraction, and recovery standards with quantification via isotope dilution as per U.S. EPA Method 23 (U.S. EPA, 1991). PAH analyses were conducted by a gas chromatograph-low resolution mass spectrometer (GC/LRMS) in selective ion monitoring mode, targeting the 16 PAH priorities of U.S. EPA (Keith, 2015). PAH emission factors were also evaluated using toxic equivalent factors (TEFs) relative to benzo[a]-pyrene toxicity equivalent (B[a]]P-TEQs) (Larsen and Larsen, 1998).

Energetics, including nitroaromatics, nitrocellulose, and their byproducts, were sampled using a Leland Legacy pump (SKC, Inc.) at a constant flow rate of 10 L/min onto a PM_2.5 impactor with a quartz fiber filter. Filter analyses followed U.S. EPA Method 8330b (U.S. EPA Method 8330B, 2006) for nitroaromatics and possible degradation products and U.S. EPA Method 353.2 (U.S. EPA, 1993) for nitrocellulose by a nitrate-nitrite colorimetric method. The filters were analyzed by APPL Inc. (Clovis, CA, USA).

VOCs were sampled using laboratory-supplied 6L SUMMA canisters via U.S. EPA Method TO-15 (U.S. EPA, 1999) at approximately 0.5 L/min. Analyses were done in accordance with U.S. EPA Method TO-15 (U.S. EPA, 1999) using full scan mode GC/LRMS by ALS Environmental (Simi Valley, California, USA). The SUMMA canisters were also analyzed for CO₂, CO, and CH₄ by a GC/flame ionization detector according to modified U.S. EPA Method 25C (U.S. EPA, 2017).

Emissions were diluted with nitrogen using a porous tube dilution probe (Lyyränen et al., 2004) and an eductor (DI-1000, Dekati Ltd., Kangasala, Finland) to reduce concentrations for continuous PM measurements. The dilution ratio was monitored continuously by measuring CO and CO₂ concentrations in the diluted sample. Continuous particle number concentration (PN) and size distributions were measured after the gun firing with an Electrical Low Pressure Impactor (ELPI, Dekati Ltd.). The ELPI continuously measures particle mass and number concentrations on a series of impactor plates supporting greased aluminum foils (10 bins from 7 nm–10 μm). The dilution was adjusted so that the current observed on any single stage was below the maximum current allowed by the instrument to ensure that all measurements were within range. The mass median distribution was determined using a density calculated from the elemental composition analysis of the bulk PM₁₀ aerosol. The resulting particle densities were 4.1, 3.0, and 4.2 g/cm³ for the M855, M855 Salted, and Legacy ammunition, respectively. The online ELPI sampled for approximately 2 ± 0.7 min for each shot. For some gun firings lacey carbon transmission electron microscope (TEM) grids were placed on the ELPI impactors to collect samples for morphological analysis. For other gun firings greased polycarbonate filters were placed on the ELPI impactors to collect PM for gravimetric and chemical analysis. Size resolved particle elemental composition was measured by SEM-EDX according to U.S. EPA Compendium Method IO-3.3 (U.S. EPA Compendium, 1999a). Select, individual particles of interest were imaged using Scanning Electron Microscopy (SEM, Tescan MIRA3, Brno, Czech Rep.) with the elemental composition measured by a SEM-EDX (Bruker AXS, Inc., Madison, WI, U.S.A.). Particles were collected with a lacey carbon-coated copper TEM grid affixed to an aluminum SEM stub with carbon tape, allowing half of the TEM grid to be suspended in the open space over the SEM stub. A background analysis was conducted off-particle and the beam was placed in the carbon matrix holes.

PM concentrations, CO₂, and CO chamber air background samples were measured before each shot was fired. One VOC chamber air background sample was collected after testing and one PAH trip blank was collected.

2.4. Calculations

Emission factors were calculated using the carbon balance method which takes the mass of the sampled target compound divided by the mass of the sampled carbon. With knowledge of the carbon composition of the source (propellant and primer), the total mass of the target compound per bullet firing can be calculated regardless of an unknown dilution amount. Emission factors thus can be expressed as mass of target compound per mass of carbon or mass of target compound per firing. Use of the carbon balance method allows one to sample a subset of the carbine box gases rather than the whole box atmosphere as the emission factor relies only on the ratio of the pollutant to carbon which is preserved regardless of sample volume. Emission factors calculated by this method are expressed here as mass of pollutant per single round firing, mass of pollutant per mass of “fuel”, where fuel is the mass of the propellant and primer, and mass of pollutant per mass of element present in the original propellant and primer (for example, g Pb emitted/g Pb in the original propellant and primer). Additional calculations allow determinations of the modified combustion efficiency (MCE) which is defined as the ratio of the CO₂ increase (above background) to the total amount of carbon emitted as CO₂ + CO. Trace carbon species, such as non-methane organic carbon and particulate carbon are commonly neglected in the total amount of carbon emitted as their inclusion has been reported to result in errors of less than a few percent (Sinha et al., 2003), well within the error range of the measurements.

Standard deviations (SD) and relative standard deviations (RSD) showed dispersion of three or more data values while relative percent difference (RPD) was used as a quality indicator when only two data values were obtained. One-way analysis of variance (ANOVA) was used to determine significant differences between ammunition formulations (p < 0.05 and F = F/F_crit > 1.0).

3. Results and discussion

3.1. CO₂, CO, and CH₄

Emission factors for major gaseous carbon species measured from SUMMA canister samples are listed in Table 2 for each of the ammunition types. Significant levels of incomplete CO oxidation resulted in MCE values less than or equal to 0.337, higher than those (0.165) from Wingfors et al. (2014). CO concentrations as high as 2,000 ppm were observed in SUMMA samples. Shot to shot variation was reflected by RSD and RPD values between 13% and 24%. The amount of CH₄ to CO₂ plus CO (ΔCH₄/(ΔCO₂+ΔCO)) was less than 1%, implying that the CH₄ has minimal impact on the MCE value. No literature values for CH₄ measurements from gun firing could be located. Continuous CO and CO₂ measurements indicated that the combined M855 salted and M855 unsalted rounds (n = 9) exhibited less complete CO oxidation than the Legacy rounds (n = 8) although the result is not statistically distinct (p > 0.26, F < 0.30). CO emission factors, 464–517 g/kg fuel, are higher than those of Ase et al. (1985), 337 g/kg fuel, for an M16 firing.

Table 2.

CO2, CO, and CH₄ emission factors with modified combustion efficiency.

Bullet type		Unit	CO2	CO	CH4a	MCEb
855 no salt	nc		4	4	2	4
	Mean	g/kg fuel	339	517	3.4	0.297
	Stand. Dev.	g/kg fuel	73	47	N/A	0.064
	RSDd	%	22	9	N/A	22
	RPDe	%	N/A	N/A	1.15	N/A
855 Salt	nc		6	6	3	6
	Mean	g/kg fuel	330	510	3.6	0.288
	Stand. Dev.	g/kg fuel	66	42	0.1	0.056
	RSDd	%	20	8	3.9	20
Legacy	nc		9	9	1	9
	Mean	g/kg fuel	414	460	4.1	0.345
	Stand. Dev.	g/kg fuel	84	54	N/A	0.080
	RSDd	%	20	12	N/A	23

^aCalculated from SUMMA Canister samples.

^bMCE = modified combustion efficiency (ΔCO₂ ppm/(ΔCO₂ ppm +ΔCO ppm)), unitless fraction.

^cn = Number of samples collected and number of shots (equivalent).

^dRSD = relative standard deviation, calculated when n = 3 or more.

^eRPD = relative percent difference, calculated when n = 2. N/A – not applicable.

3.2. Particulate matter

The PM_2.5 and PM₁₀ emission factor results for M855 and Legacy ammunition were nearly the same, which indicates that the majority of the emitted particles are of diameter less than or equal to 2.5 μm and is confirmed by the measured size distributions (Table 3 and Fig. 2). However, the M855-salted ammunition PM₁₀ emission factor was nearly 50% larger than the PM_2.5 emission factor. The RSD(10–15%) and RPD(6–38%) indicate the level of consistency for these rounds in the emissions as well as the testing and sampling method. The emission factors for PM₁₀ (Table 3) are consistent with those measured by Wingfors et al. (2014) for total suspended particles (0.029–0.030 g/round), albeit using different methods and different ammunition types. Total PM values for an M16 rifle were found to be 4 mg/kg fuel (Ase et al., 1985) versus 11–34 g/kg fuel reported in this current work for the M4 rifle.

Table 3.

Particulate matter and element emission factors.

	855 Salt (n = 3, s = 6)a			855 (n = 2, s = 4)b			Legacy (n = 2, s = 10)b
	g/kg fuel	mg/round	g/kg element	g/kg fuel	mg/round	g/kg element	g/kg fuel	mg/round	g/kg element
Particulates
PM2.5	23 ± 2.3	40 ± 4.0	NA	15 ± 1.1	25 ± 2.0	NA	10 ± 1.9	17 ± 3.2	NA
PM10	34 ± 5.3	59 ± 9.1	NA	17 ± 0.5	29 ± 0.8	NA	11 ± 1.2	20 ± 2.1	NA
PN (#/kg fuel)	6.5 ± 2.0 × 1016			1.0 ± 0.4 × 1017			7.5 ± 2.3 × 1016
Elements in the PM2.5fraction
Al	0.072 ± 0.028	0.12 ± 0.048	51 ± 20	0.086 ± 0.0004	0.15 ± 0.0007	85 ± 0.4	0.077 ± 0.013	0.13 ± 0.02	76 ± 13
Ba	0.31 ± 0.039	0.53 ± 0.07	90 ± 12	0.33 ± 0.008	0.57 ± 0.01	441 ± 10	0.28 ± 0.04	0.48 ± 0.07	372 ± 52
Bi	2.91 ± 0.60	5.02 ± 1.04	14804 ± 307	2.85 ± 0.47	4.98 ± 0.82	1440 ± 238	0.07 ± 0.04	0.13 ± 0.08	38 ± 22
Cd	ND (0.0018)	ND (0.0031)	NA	0.003	0.0048	NA	0.002 ± 0.001	0.0042 ± 0.002	NA
Co	ND (0.0013)	ND (0.0022)	NA	ND (0.0012)	ND (0.0021)	NA	ND (0.0009)	ND (0.00016)	NA
Cr	ND (0.0009)	ND (0.0016)	NA	ND (0.0008)	ND (0.0014)	NA	ND (0.0005)	ND (0.0009)	NA
Cu	16.4 ± 1.0	28.3 ± 1.8	NA	15.6 ± 0.89	27.3 ± 1.5	NA	12.1 ± 1.7	21.1 ± 2.9	NA
Fe	0.17 ± 0.021	0.30 ± 0.04	NA	0.15 ± 0.003	0.26 ± 0.006	NA	0.50 ± 0.09	0.87 ± 0.16	NA
K	9.07 ± 0.69	15.7 ± 1.2	65 ± 7.7	0.57 ± 0.29	1.00 ± 0.51	NA	0.10 ± 0.001	0.17 ± 0.002	NA
Mg	0.28 ± 0.02	0.49 ± 0.03	NA	0.22 ± 0.005	0.39 ± 0.009	NA	0.17 ± 0.03	0.29 ± 0.05	NA
Mn	0.003 ± 0.0001	0.0057 ± 0.0002	NA	ND (0.0015)	ND (0.0026)	NA	0.004 ± 0.001	0.007 ± 0.003	NA
Mo	0.001 ± 0.0004	0.0019 ± 0.0007	NA	0.0009 ± 0.0003	0.0015 ± 0.0006	NA	0.002 ± 0.0005	0.004 ± 0.001	NA
Ni	ND (0.0014)	ND (0.0024)	NA	ND (0.0013)	ND (0.0023)	NA	ND (0.0009)	ND (0.002)	NA
Pb	5.77 ± 0.42	9.96 ± 0.73	1672 ± 123	6.07 ± 0.22	10.60 ± 0.39	1765 ± 64	4.95 ± 0.5	8.58 ± 0.87	1438 ± 146
S	0.76 ± 0.20	1.31 ± 0.34	884 ± 232	0.0005	0.0008	0.55	ND	ND	NA
Sb	1.56 ± 0.22	2.69 ± 0.37	720 ± 99	1.55 ± 0.42	2.70 ± 0.07	330 ± 9.0	1.30 ± 0.2	2.25 ± 0.35	280 ± 43
V	0.017 ± 0.002	0.029 ± 0.03	NA	0.09 ± 0.0004	0.15 ± 0.0007	NA	0.01 ± 0.001	0.02 ± 0.002	NA
W	0.055 ± 0.004	0.096 ± 0.006	NA	0.064 ± 0.004	0.11 ± 0.006	NA	0.01	0.015	NA
Zn	2.12 ± 0.14	3.66 ± 0.25	NA	2.04 ± 0.11	3.55 ± 0.19	NA	1.76 ± 0.26	3.05 ± 0.45	NA

^aRange of data equals ±1 standard deviation, n = number of samples, s = number of shots.

^bRange of data equals mean absolute deviation, n = number of samples, s = number of shots. NA = not applicable, element not in the propellant or in the primer. ND = not detected, uncertainty level within parentheses.

Fig. 2.

A) Representative initial mass normalized mass weighted PM size distributions from M4 carbine for different ammunition types. B) Mass normalized mass weighted PM size distributions from M4 carbine for M855 ammunition over the first 2 min after firing.

The M855-Salted ammunition had the largest PM_2.5/10 emission factors, likely due to K-based particles (Table 3). The PM emission factors for the two different ammunition compositions (M855-Salted versus M855-Unsalted and Legacy) are statistically distinct, (p<0.004, F>3.8). The Legacy ammunition resulted in emissions within 50% of its more recently manufactured M855 counterpart but these comparisons are limited by only two values for each ammunition type.

PN emission factors showed no statistically significant differences between the different ammunition types (Table 3). To our knowledge there are no other PN emission factors from gun firings to compare with; however, these factors are an order of magnitude larger than what has been measured from other combustion sources. For example, PN emission factors from heavy duty diesel trucks were ∼2 × 10¹⁶ #/kg (Ban-Weiss et al., 2009), from ships were 1.5 × 10¹⁶ #/kg (Beecken et al., 2015), and from woodstoves were in the range of 2–7 × 10¹⁵ #/kg (Wardoyo et al., 2006). The higher PN emission factors from gun firing as compared to other combustion sources may be due to the additional particles formed from non-combustion sources such as from the bullet fragments or from the barrel.

The mass (Fig. 2A) and particle number distributions (not shown) of the M855 and legacy ammunition types immediately after firing are dominated by the smallest particles (<30 nm), similar to distributions measured at firing ranges, where the count median diameters ranged from 30 to 80 nm (Grabinski et al., 2017). The size distribution evolved rapidly over the first few seconds (Fig. 2B) as the particles aggregated due to the very high concentrations inside the enclosure. Initially, the mass distribution for the M855 ammunition is entirely less than 1 μm and bimodal, with a peak at 21 nm and one at approximately 100 nm. Over the first 2 min the particles mass median diameter shifts to around 770 nm and approximately 30% of the mass is contained in particles larger than 1 μm. This aggregation process explains the slightly larger PM₁₀ emission factors versus PM_2.5 emission factors despite the initial particle distributions all being significantly smaller than 10 μm in diameter. This shift in sizes is consistent with results reported by Wingfors et al. (2014), where the median diameter increased from 0.2 μm at 2 min after firing to 1 μm after 12 min.

The initial size distribution was different for all three types of ammunition, with the M855 and the Legacy exhibiting a peak in the smallest size bin (<28 nm) and a second larger mode in the 200–500 range. The M855 salted lacked the small particle mode, and had a single mode at a mass median diameter of 0.570 μm. This may be due to the K-salt suppressing the nucleation of the smallest particles, however the mechanism by how this occurs is not known, because the composition of the nuclei mode is presently unknown. After several seconds the distributions were similar for all types, with mass median diameters of 0.389 ± 0.109 μm for the M855 Salted, 0.330 ± 0.124 μm for the M855, and 0.575 ± 0.130 μm for the Legacy ammunitions. PN emission factors calculated from the initial particle concentrations (before significant aggregation occurred) showed no statistically significant differences between the different ammunition types (Table 3) elements.

Elemental emissions from analysis of the sample filters are shown in Table 3, for the Salted, M855, and Legacy ammunition types, respectively. Data are presented in terms of g/kg fuel, mg/round and g/kg element, the latter being a measure of the amount of the original metal that ends up in the emissions. The results show relatively similar metal emission factors (Fig. 3) with the exception of higher K from the salted (KNO₃) ammunition. The relative standard deviations for the element concentrations are quite low, averaging 12%, despite only three replicates, indicating the robustness of the combined sampling and analytical methods.

Fig. 3.

Bullet element emission factors in A) g/kg fuel, where the denominator “fuel” accounts for the weight of the propellant and primer, and B) g/g element, where the denominator accounts for the weight of the element in the original propellant and primer composition; both PM_2.5 and PM₁₀ data shown.

Cu has the highest metal emission factor (Fig. 3A) despite the lack of Cu in the propellants and primers of the three tested bullets. Rather, the bullets (the projectile) are encased in a Cu jacket which is swaged by the rifled barrel to enable the bullet to spin, aiding in accuracy. While the barrel etches the copper jacket, Cu is deposited on the barrel interior and, as these results show, emitted from the barrel. In addition to Cu, Pb and Sb are added to the propellant to keep the rifle barrel clean. After Cu, Pb has the next highest emission factor for the three bullet types (other than K for the salted ammunition) at around 5 g/kg (Fig. 3A). The abundance of Cu and Pb is consistent with findings of Brochu et al. (2011). Elements not reported in the primer and propellant formulation but observed in the particle analyses include Na, Fe, Zn, and K (in the non-salted rounds). This may be due to trace amounts of these elements alloyed in the bullet (projectile) casing.

Emission factors presented in proportion to their reported amounts in the primer plus propellant (g/g element) are shown in Fig. 3B. Pb is consistently over 1,000 g Pb/kg_Pb for all three bullet types suggesting a discrepancy with the composition in the manufacturer’s Safety Data Sheet. While the projectile is a possible source of measured Pb, images of launch survival (not covered in this paper) saw no evidence of bullet breakup at uncorking. Bi also exceeds the known amount except for the Legacy ammunition despite its being reported as having the same amount of Bi in the propellant (Tables SI–1) as the other two bullet formulations. S and Sb both appear to near 100% emission but only for the Salted ammunition. The excess reported for Pb and Bi may come from barrel sloughing as the age of the barrel and its previous use are unknown. As well, since we don’t have unreported trace elements from the primer and propellent, these may also provide a source. Some trace elements are detected such as Cd, Co, Cr, Fe, Mn, Mo, Ni, V, and W that were not reported in product formulations provided by the manufacturer. Additionally, there may be other sources of elements in stead of the primer and propellant, such as the jacket material, projectile, muzzle, cleaning residues.

Wipes of the sampling chamber were taken after each round type was finished and analyzed by ICP for Pb, the most likely element to be observed. Between 1.3 and 1.9 μg of Pb were observed per shot for the three ammunition round types (the blank wipe was below detection limits at 0.4 μg) amounting to less than 3.1% of the total air emission of Pb.

The element emission factors for PM_2.5 and PM₁₀ (Fig. 3) are quite similar, indicating a similar mechanism of particle formation despite their size difference. The Legacy ammunition alone showed trace detectable levels of Cl contrary to its reported composition. The M855 unsalted ammunition showed detectable levels of S, unlike the Legacy ammunition.

The PM_2.5 element emission factors by ammunition type are compared with those reported for the respirable fraction (<4 μm D₅₀) from Wingfors et al. (2014) in Tables SI–3. There is general agreement on the magnitude of the emission factors with the exception of their low Pb values which are expected due to their testing of Pb-free ammunition. Elemental emission data from Ase et al. (1985) with the M16 rifle are not compared due to the substantially different propellant and primer composition.

3.3. PM composition by size

The smaller particles, which can penetrate deep into the lung, were enriched with Pb. Particles less than 1 μm were composed of approximately 14% Pb, while larger particles were only 6% Pb (Table 4). Similar trends were observed for K, Sb, and Bi, which may have been derived from combustion products. In contrast, Zn was distributed uniformly across all size ranges. A sizable fraction of the PM was not detected by XRF and was likely carbonaceous soot formed from combustion. Almost all of the M855 PM₁₀ mass not detected by XRF was carbonaceous, and most was in the form of organic carbon (11.7%) as opposed to elemental carbon (2.4%). The carbon content was not measured for the M855 salted ammunition but is likely to be a major constituent of the particle mass not accounted for by XRF. This is due to the function of the K-based salt acting as a muzzle flash suppressant by inhibiting combustion of unburned gases (the secondary muzzle flash) after bullet exit (K acts as an H, O, OH scavenger). The unburned gases then contribute to soot formation, resulting in a relative increase of carbon for the salted rounds.

Table 4.

PM percent elemental mass composition by particle size determined using XRF.

Bullet type	PM Size Range (μm)	Cu	Pb	K	Sb	Zn	Bi	OC	EC	Oestb	Trace elements	Balancec
M855	0.14–0.27	16.8	12.8	1.5	5.7	3.3	7.1			12.1	7.3	33.4
	0.27–0.37	20.5	12.9	0.8	4.9	2.5	5.0			10.5	3.5	39.3
	0.37–0.63	33.4	14.3	0.8	4.9	3.1	5.1			14.5	4.0	19.9
	0.63–0.94	36.8	11.9	0.7	3.7	3.1	4.4			14.2	3.1	22.1
	0.94–1.55	38.8	8.5	0.5	2.5	3.1	4.0			14.0	3.2	25.5
	1.55–2.4	42.4	6.8	0.5	1.3	3.3	4.6			15.0	4.3	21.9
	PM2.5	34.5	14.8	0.7	3.5	4.5	5.6			14.5	3.9	16.5
	PM10	36.7	12.7	0.5	3.1	4.3	5.5	11.7	2.4	14.2	3.1	5.8
M855 Salted	0.27–0.37	20.8	14.8	14.7	4.8	2.4	5.1			16.8	8.6	11.9
	0.37–0.63a
	0.63–0.94	27.0	9.9	5.7	2.2	2.3	2.9			12.3	3.4	34.4
	0.94–1.55	30.6	8.9	4.9	1.7	2.6	2.7			13.3	4.6	30.7
	1.55–2.4	25.5	6.3	3.1	1.3	2.1	2.4			12.4	6.8	40.0
	PM2.5	22.5	8.7	12.1	1.9	2.9	3.3			11.7	2.4	34.6
	PM10	23.5	7.5	8.7	1.7	2.7	3.2			10.6	1.8	40.3
Legacy	PM2.5	38.5	15.8	0.3	4.1	5.6	0.3			17.1	6.5	11.9
	PM10	37.0	12.7	0.3	3.4	4.9	0.2	10.0	1.0	15.3	5.1	10.3

^aThe gravimetric mass was less than the sum of the elements detected by XRF, which may have occurred due to contamination deposited on the filter after gravimetric determination but before XRF analysis.

^bOxygen estimated by assuming most common metal oxide form as described in Reff et al., 2009) (Reff et al., 2009).

^cMass unable to be analyzed by XRF comprised of atomic number less than 12.

3.4. PM morphology

The particle morphology was mostly spherical, although some particles appeared to be aggregates of smaller spherical particles. Individual particles were composed of primarily Cu, and some containing Pb and lesser amounts of other elements (e.g., Bi, O, C, Mg, Al) (Figure SI-2). These results suggest the formation of a metal alloy upon condensation, which may have restructured into a larger spherical particle in the cooling exhaust. Microscopy samples were only obtained of the larger particles (>1 μm) but considering the similarity of composition across all size ranges, the morphology and the individual particle composition may be similar for the smaller particles.

Similar aggregates of much smaller spherical particles (diameter < 100 nm) were observed at a firing range where M4 rifles were being used with lead free ammunition (Grabinski et al., 2017). These particles emitted from lead free ammunition were also composed of copper and various other elements (Bi, Ni, Zn, Na, C), but no lead was detected. The smaller particle size may have been due to the higher ventilation rates at the firing range that suppressed particle aggregation. Additionally, in the present study the presence of Pb in the emissions can form alloys with other metals, reducing the melting point, which may promote the formation of larger spheres upon aggregation.

3.5. Volatile organic compounds

The VOC emission factors were background corrected and data from the three bullet types are shown in Tables SI–4 in both mg/kg fuel and μg/round. For most compounds, the variation of the individual emission factors was low, averaging 34%, despite the limited number (n = 1 to 3) of trials. For comparison with published emission factors, the sole overlap with data from Wingfors et al. (2014) is acrolein for which they obtained 8 ± 1 μg/round as compared to this work at a 6-firing average of 3.8 ± 1.6 μg/round. The benzene emission factor levels, 93.8–120 mg/kg are in same range as benzene emissions from open detonation of munitions (89–264 mg/kg (Aurell et al., 2011; Aurell et al., 2015)), although higher than from open burning of propellant (2.5–4.7 g/kg (Aurell et al., 2011; Aurell et al., 2015)).

3.6. Polycyclic aromatic hydrocarbons

PAH emission factors from the combined gas phase and particle-bound phase are shown in Table 5 (the same table, expressed in units of mg/kg of fuel is in Tables SI–5). PAHs were about 50% higher in the Legacy round emissions than that of the M855 and M855 Salt. Naphthalene, pyrene, acenaphthylene, and phenanthrene are the most predominant PAHs for all three bullet types. The most toxic PAH, common to all three bullet types, is benzo(a)pyrene, which accounts for over 55% of the 16-PAH toxic equivalency value (Larsen and Larsen, 1998). These tests sampled for and analyzed both gas phase and particle bound PAHs, finding values that are often three orders of magnitude higher than particle phase (only) data from Wingfors et al. (2014) indicating the predominantly volatile nature of the PAHs. For instance, the average emission factor for pyrene was 1.2 mg/kg or 2200 ng/round in this study which is three times higher than derived by Wingfors et al. (2014) (620 ng/round) and a hundred times higher than from open burning of propellant (Aurell et al., 2015) (0.012 mg/kg fuel).

Table 5.

PAH emission factors from firing of M4 carbine.

Targets	855-Salt, n = 2			855, n = 1		Legacy, n = 2
	Average	Average	RPD	mg/round	mg B[a]P TEQ/round	Average	Average	RPD
	mg/round	mg B[a]P TEQ/round	%			mg/round	mg B[a]P TEQ/round	%
Naphthalene	1.59E-02	N/A	25	1.56E-02	N/A	2.75E-02	N/A	42
Acenaphthylene	1.61E-03	N/A	25	1.44E-03	N/A	2.27E-03	N/A	49
Acenaphthene	ND	N/A	N/A	ND	N/A	ND	N/A	N/A
Fluorene	ND	N/A	N/A	ND	N/A	ND	N/A	N/A
Phenanthrene	1.56E-03	7.78E-07	4	1.48E-03	7.40E-07	2.18E-03	1.09E-06	49
Anthracene	2.55E-04	1.27E-07	37	2.33E-04	1.17E-07	3.05E-04	1.52E-07	34
Fluoranthene	7.08E-04	3.54E-05	7	6.43E-04	3.22E-05	8.82E-04	4.41E-05	62
Pyrene	1.84E-03	1.84E-06	4	1.67E-03	1.67E-06	2.81E-03	2.81E-06	99
Benzo(a)anthracene	1.22E-04	6.09E-07	6	1.11E-04	5.54E-07	1.10E-04	5.49E-07	57
Chrysene	2.08E-04	6.24E-06	4	2.02E-04	6.05E-06	2.64E-04	7.92E-06	66
Benzo(b)fluoranthene	3.31E-04	3.31E-05	5	2.76E-04	2.76E-05	3.62E-04	3.62E-05	62
Benzo(k)fluoranthene	1.45E-04	7.27E-06	37	1.41E-04	7.05E-06	1.61E-04	8.04E-06	91
Benzo(a)pyrene	2.35E-04	2.35E-04	6	2.23E-04	2.23E-04	2.33E-04	2.33E-04	64
Indeno(1,2,3-cd)pyrene	3.46E-04	3.46E-05	11	2.70E-04	2.70E-05	3.16E-04	3.16E-05	76
Dibenz(a,h)anthracene	4.70E-05	5.17E-05	88	ND	ND	ND	ND	N/A
Benzo(ghi)perylene	1.08E-03	2.17E-05	0.04	8.88E-04	1.78E-05	1.89E-03	3.79E-05	114
SUM 16-EPA PAHs	2.44E-02	4.29E-04	18	2.32E-02	3.44E-04	3.93E-02	4.03E-04	51

PAH emission factors are compared to those reported elsewhere (Wingfors et al., 2014) in Tables SI–6. Reasonable agreement is noted given the limited number of trials, two each for the three ammunition types in this work and five for Wingfors et al. (2014) However, our data were gas and particle phase whereas those of Wingfors et al. (2014) were particle phase only.

3.7. Energetics

Energetics were sampled by collecting three samples of the Salted round, consisting of 2, 2, and 3 cumulative shots, two samples of the 855 round, consisting of 4 and 2 cumulative shots, and two samples of the Legacy round, each sample consisting of four shots each. None of the 17 nitroaromatics and energetic byproducts examined in these seven samples resulted in levels above the detection limit (Tables SI–7). The analysis method (U.S. EPA Method 8330 B (U.S. EPA Method 8330B, 2006)) in our work had a sample detection limit of 2 μg/sample which amounts to an emission factor of <0.039 mg/round for all three bullet types. However, Wingfors et al. (2014) found that nitrogen-containing heterocyclic compounds and aromatic nitrogen compounds accounted for a relatively large proportion of the particulate-bound organics and environmental sampling. Similarly, open range sampling by Brochu et al. (2011) found a nitroglycerin emission mass of less than 0.9 mg/round for a number of ammunition and gun types.

4. Conclusion

This work demonstrated methods to comprehensively characterize air emissions from gun firing, providing emission factor data that can be used for environmental and health assessments. Distinctions between bullet/primer formulations could be discerned, potentially allowing for optimization of propellant, primer, and casing formulations to reduce risk of environmental contamination and human exposure. PM emissions were highest from the M855 salt-added ammunition, likely due to incomplete secondary combustion in the muzzle blast caused by scavenging of combustion radicals by the K salt. CO levels higher than CO₂ resulted in MCE values lower than 0.35, indicating incomplete carbon oxidation. Some metals, such as Pb, were completely released in the emissions. The majority of particles were in the respirable range, < 1 μm, and were enriched in Pb, among other metals. Toxics including acrolein and benzo(a)pyrene were among the volatiles and PAHs sampled. The emission of metals and organics could provide an environmental contamination concern for outdoor firing ranges and an inhalation risk for indoor ranges without sufficient ventilation. These data can be used in deposition and exposure models to estimate potential environmental contamination and health toxicity assessments.