Emerging use of isotope ratio mass spectrometry as a tool for discrimination of 3,4-methylenedioxymethamphetamine by synthetic route

Article abstract of DOI:10.1021/ac702559s

Drug profiling, or the ability to link batches of illicit drugs to a common source or synthetic route, has long been a goal of law enforcement agencies. Research in the past decade has explored drug profiling with isotope ratio mass spectrometry (IRMS). This type of research can be limited by the use of substances seized by police, of which the provenance is unknown. Fortunately, however, some studies in recent years have been carried out on drugs synthesized in-house and therefore of known history. In this study, 18 MDMA samples were synthesized in-house from aliquots of the same precursor by three common reductive animation routes and analyzed for ¹³C, ¹⁵N, and ²H isotope abundance using IRMS. For these three preparative methods, results indicate that ²H isotope abundance data is necessary for discrimination by synthetic route. Furthermore, hierarchical cluster analysis using ²H data on its own or combined with ¹³C and/or ¹⁵N provides a statistical means for accurate discrimination by synthetic route.

Full text of DOI:10.1021/ac702559s

Anal. Chem. 2008, 80, 3350-3356
Emerging Use of Isotope Ratio Mass
Spectrometry as a Tool for Discrimination of
3,4-Methylenedioxymethamphetamine by Synthetic
Route
Hilary A. S. Buchanan,† Niamh Nic Dae´ id,*,† Wolfram Meier-Augenstein,‡ Helen F. Kemp,‡
William J. Kerr,† and Michael Middleditch†
Centre for Forensic Science, Department of Pure & Applied Chemistry, University of Strathclyde, 204 George Street,
Glasgow G1 1WX, Stable Isotope Forensic Facility, Environmental Engineering Research Centre, Queen’s University
Belfast, David Keir Building, Belfast BT9 5AG
with this strategy have been suggested. For instance, the
clandestine laboratory may produce substances of high purity such
that very few (if any) impurities remain.9 Further, comparisons
based on impurity profiling are often not sufficiently conclusive
due to difficulty in achieving reproducible chromatograms as a
result of low level impurities. It has also been suggested that
chromatograms are often time and machine dependent, thereby
confounding the comparison of results across laboratories,9
although this type of problem may be encountered with other
instrumental techniques.
Drug profiling, or the ability to link batches of illicit drugs
to a common source or synthetic route, has long been a
goal of law enforcement agencies. Research in the past
decade has explored drug profiling with isotope ratio mass
spectrometry (IRMS). This type of research can be limited
by the use of substances seized by police, of which the
provenance is unknown. Fortunately, however, some
studies in recent years have been carried out on drugs
synthesized in-house and therefore of known history. In
this study, 18 MDMA samples were synthesized in-house
from aliquots of the same precursor by three common
reductive amination routes and analyzed for 13C, 15N, and
2H isotope abundance using IRMS. For these three
preparative methods, results indicate that 2H isotope
abundance data is necessary for discrimination by syn-
thetic route. Furthermore, hierarchical cluster analysis
using 2H data on its own or combined with 13C and/or
15N provides a statistical means for accurate discrimina-
tion by synthetic route.
Research in the past decade has explored isotope profiling with
isotope ratio mass spectrometry (IRMS).10-21 While the use of
authentic street drugs has its advantages, due to the unknown
(5) Swist, M.; Wilamowski, J.; Parczewski, A. Forensic Sci. Int. 2005, 152, 175-
184.
(6) Swist, M.; Wilamowski, J.; Zuba, D.; Kochana, J.; Parczewski, A. Forensic
Sci. Int. 2005, 149, 181-192.
(7) Swist, M.; Zuba, D.; Stanaszek, R.; Wilamowski, J.; Parczewski, A. Forensic
Sci. Int. 2003, 136, 102-103.
(8) van Deursen, M. M.; Lock, E.; Poortman-van der Meer A. J. Sci. Justice
2006, 46, 135-152.
(9) Besacier, F.; Chaudron-Thozet, H.; Rousseau-Tsangaris, M.; Girard, J.;
Lamotte, A. Forensic Sci. Int. 1997, 85, 113-125.
(10) Billault, I.; Courant, F.; Pasquereau, L.; Derrien, S.; Robins, R. J.; Naulet, N.
Anal. Chim. Acta 2007, 593, 20-29.
Drug profiling, or the ability to link batches of illicit drugs to
a common source or synthetic route, has long been a goal of law
enforcement agencies. Specifically, the use of synthetic drug
profiles has been targeted as a law enforcement goal in the
European Union Drugs Action Plan 2005-2008.1 Analysis of drugs
such as 3,4-methylenedioxymethamphetamine (MDMA or “ec-
stasy”) is often performed by organic impurity profiling utilizing
gas chromatography-mass spectrometry (GC-MS).2-8 Problems
(11) Casale, J.; Casale, E.; Collins, M.; Morello, D.; Cathapermal, S.; Panicker,
S. J. Forensic Sci. 2006, 51, 603-606.
(12) Makino, Y.; Urano, Y.; Nagano, T. Bull. Narc. 2005, LVII.
(13) Palhol, F.; Lamoureux, C.; Chabrillat, M.; Naulet, N. Anal. Chim. Acta 2004,
510, 1-8.
(14) Kurashima, N.; Makino, Y.; Sekita, S.; Urano, Y.; Nagano, T. Anal. Chem.
2004, 76, 4233-4236.
(15) Carter, J. F.; Titterton, E. L.; Murray, M.; Sleeman, R. Analyst 2002, 127,
830-833.
(16) Carter, J. F.; Titterton, E. L.; Grant, H.; Sleeman, R. Chem. Commun. 2002,
2590-2591.
(17) Palhol, F.; Chabrillat, M.; Naulet, N. FIRMS 2002 Conference Presentations
& Posters; FIRMS: Kent, U.K., 2002; http://www.forensic-isotopes.rdg.ac.uk/
conf/conf2002.htm.
* To whom correspondence should be addressed. Phone: +44-141-548-4700.
Fax: +44-141-548-2532. E-mail: n.nicdaeid@strath.ac.uk.
† Centre for Forensic Science, Department of Pure & Applied Chemistry,
University of Strathclyde.
‡ Stable Isotope Forensic Facility, Environmental Engineering Research
Centre, Queen’s University Belfast.
(1) Council of the European Union. EU Drugs Action Plan (2005-2008); 8652/
1/05 Rev 1.
(18) Ehleringer, J. R.; Casale, J. F.; Lott, M. J.; Ford, V. L. Nature 2000, 408,
311-312.
(2) Gimeno, P.; Besacier, F.; Bottex, A.; Dujourdy, L.; Chaudron-Thozet, H.
Forensic Sci. Int. 2005, 155, 141-157.
(19) Ehleringer, J. R.; Cooper, D. A.; Lott, M. J.; Cook, C. S. Forensic Sci. Int.
1999, 106, 27-35.
(3) Gimeno, P.; Besacier, F.; Chaudron-Thozet, H. Forensic Sci. Int. 2003, 132,
182-194.
(4) Gimeno, P.; Besacier, F.; Chaudron-Thozet, H.; Girard, J.; Lamotte, A.
Forensic Sci. Int. 2002, 127, 1-44.
(20) Besacier, F.; Guilluy, R.; Brazier, J. L.; ChaudronThozet, H.; Girard, J.;
Lamotte, A. J. Forensic Sci. 1997, 42, 429-433.
(21) Mas, F.; Beemsterboer, B.; Veltkamp, A. C.; Verweij, A. M. A. Forensic Sci.
Int. 1995, 71, 225-231.
3350 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008
10.1021/ac702559s CCC: $40.75 © 2008 American Chemical Society
Published on Web 03/21/2008

batch of safrole. The 18 batches of MDMA‚HCl were subsequently
synthesized from this single “pot” of PMK and then subjected to
2
IRMS analysis for 13C, 15N, and H isotope abundance.
EXPERIMENTAL SECTION
Eighteen batches of MDMA‚HCl were synthesized, six from
each reductive amination route. Reagents and solvents were
sourced from commercial suppliers (such as Sigma Aldrich and
Fisher Scientific, U.K.) and used without further purification.
Precursor Synthesis. Safrole to Isosafrole.2 To safrole (20 mL,
135 mmol) were added solid potassium hydroxide (KOH) pellets
(9.0 g, 161 mmol) and Aliquat 336 (3.7 mL, 8.1 mmol, 6.0 mol %)
with stirring for 20 min. The mixture was then heated to 80 °C
for 75 min. After cooling to room temperature, the mixture was
filtered through celite with dichloromethane (DCM) rinsing, and
dried with magnesium sulfate (MgSO4). Solvent was removed in
vacuo. The mixture was filtered through a plug of silica with petrol.
Removal of solvent gave isosafrole (19.2 g, 87% yield) as a colorless
oil. On a larger scale, the silica purification step was not carried
out, resulting in isosafrole as a yellow oil. Analysis was in
agreement with that disclosed in the literature.24-26 IR vmax
(CH2Cl2)/cm-1: 1504, 1438, 1249 (C-O), 1193, 1041 (C-O), 965,
1
938. H NMR (400 MHz, CDCl3): δ 1.86 (dd, 3H, J 6.6 and 1.6
Hz, CH3), 5.94 (s, 2H, OCH2O), 6.07 (dq, 1H, J 15.6 and 6.6 Hz,
CH), 6.32 (dq, 1H, J 15.6 and 1.6 Hz, CH), 6.73 - 6.75 (m, 2H,
ArH), 6.89 ppm (s, 1H, ArH). 13C NMR (100 MHz, CDCl3):
δ 18.5, 101.1, 105.6, 108.4, 120.3, 124.2, 130.8, 132.8, 146.7,
148.2 ppm.
Isosafrole to PMK.27 To 35% aqueous hydrogen peroxide (3.1
mL, 100 mmol) was added 88% formic acid (12.3 mL, 326 mmol)
with stirring for 2 min. In a separate flask, isosafrole (2.9 mL, 20
mmol) was dissolved in acetone (12 mL) at room temperature,
and the solution swirled. The isosafrole/acetone solution was then
added dropwise to the hydrogen peroxide/formic acid solution
with cooling (ice + NaCl bath) to keep the temperature from
exceeding 40 °C. The mixture was allowed to warm to room
temperature with overnight stirring. The solvent was removed in
vacuo to reveal an orange oil. To the crude oil was added methanol
(6.0 mL) and then 15% sulfuric acid (36 mL, 676 mmol). The
mixture was heated to reflux for 3 h, producing a dark reddish-
brown oil. After cooling, the mixture was extracted with ether (3
× 75 mL) and the organic layer washed with water until neutral
pH.28 The organic layer was then washed with 1M sodium
hydroxide (50 mL) and water again until a neutral pH was realized,
dried with MgSO4, and the solvent removed in vacuo to reveal a
brown oil. The crude oil was distilled under vacuum (3 mbar, 120-
140 °C) to yield PMK as a yellow oil (1.6 g, 46% yield). Analysis
Figure 1. Synthetic pathway from safrole to PMK and the three
reductive amination routes from PMK to MDMA‚HCl.
provenance of the samples, this type of research can be limited.
Thus, linkages and conclusions drawn from research on seized
samples are tentative; although seized tablets can be “grouped”
according to IRMS data, the verification of the groupings is not
possible if the provenance of the samples is unknown. Fortunately,
however, some studies in recent years have been carried out on
drugs synthesized in-house and therefore of known history
(methamphetamine,12,14,16 cocaine,18 heroin,20 MDMA10,17). The
work herein is the first study correlating 13C, 15N, and 2H
abundance of MDMA samples of known provenance.
The 18 MDMA samples in this study were synthesized via
three commonly used reductive amination routes accessible to
the clandestine chemist. The three reductive aminations of “PMK”
(piperonyl methyl ketone) utilize different reducing agents: Al/
Hg amalgam, NaBH4, and Pt/H2. Since this ketone has limited
access without the appropriate licenses and thus cannot be easily
diverted from trade,22 clandestine synthesis of MDMA must start
from pre-precursors such as safrole or isosafrole (see Figure 1).
Often these pre-precursors are converted to PMK in Asia and then
shipped to a separate location for MDMA synthesis.23
was in agreement with published data.29,30 IR vmax (CH2Cl2)/cm-1
:
1710 (CdO), 1607, 1490, 1249 (C-O), 1040 (C-O). 1H NMR (400
MHz, CDCl3): δ 2.15 (s, 3H, CH3), 3.61 (s, 2H, CH2), 5.95
(24) Buss, A. D.; Warren, S. J. Chem. Soc., Perkin Trans. 1985, 1, 2307-2325.
(25) Joshi, B. P.; Sharma, A.; Sinha, A. K. Tetrahedron 2005, 61, 3075-3080.
(26) Rao, S. A.; Periasamy, M. J. Organomet. Chem. 1988, 342, 15-20.
(27) Shulgin, A. T.; Shulgin, A. PiHKAL - A Chemical Love Story; Transform
Press: Berkeley, CA, 1991.
(28) Besacier, F. Personal communication, Jun 21, 2006.
(29) Niwa, M.; Noda, H.; Kobayashi, H.; Yamamura, S. Chem. Lett. 1980, 85-
88.
To mimic clandestine synthesis as closely as possible, a large
batch (approximately 200 g) of PMK was synthesized from a single
(22) Europol. European Union, Situation Report on Drug Production and Drug
Trafficking; 2003-2004.
(23) United Nations Office on Drugs and Crime. 2006 World Drug Report, Volume
1: Analysis; 2006.
(30) Zappala, M.; Grasso, S.; Micale, N.; Polimeni, S.; De Micheli, C. Synth.
Commun. 2002, 32, 527-533.
Analytical Chemistry, Vol. 80, No. 9, May 1, 2008 3351

(s, 2H, OCH2O), 6.65 (d, 1H, J 7.9 Hz, ArH), 6.69 (s, 1H, ArH),
6.77 ppm (d, 1H, J 7.9 Hz, ArH). 13C NMR (100 MHz, CDCl3): δ
29.1, 50.6, 101.1, 108.5, 109.8, 122.5, 127.8, 146.7, 147.9, 206.5 ppm.
Reductive Amination of PMK. Al/Hg Amalgam (Route 1).27
To aluminum foil (4.5 g) cut into 2 cm squares was added water
(159 mL) containing mercuric chloride (0.11 g, 0.41 mmol, 1.2
mol %). The amalgamation was allowed to proceed with swirling
for 3 min, and fine bubbles and a light gray precipitate were
observed. The water was decanted and the foil washed with fresh
water (317 mL). The water from the last wash was removed from
the foil as thoroughly as possible by shaking. Added to the
remaining foil in succession were methylamine hydrochloride (6.8
g, 100 mmol, 3 equiv) dissolved in water (6.8 mL), isopropyl
alcohol (20 mL), 25% aqueous NaOH (16 mL, 588 mmol), PMK
(6 g, 33 mmol), and isopropyl alcohol (39 mL). The mixture was
immersed in an ice bath as necessary to keep the temperature
below 50 °C. Once thermally stable, the mixture was left stirring
overnight. The resultant mixture was then filtered through celite,
rinsing with methanol, and the volatiles removed in vacuo. The
residue was suspended in water (272 mL), and 32% aqueous
hydrochloric acid added to make the phase acidic (pH 1). Excess
PMK was extracted with DCM (3 × 75 mL). The remaining
aqueous layer was made basic (pH 9) with 25% aqueous NaOH
and extracted with DCM (3 × 75 mL). A further 10 drops of 25%
aqueous NaOH were added before the second and third DCM
extractions, respectively. The combined organic layers were dried
over magnesium sulfate and the solvent removed in vacuo to
reveal MDMA base as an amber oil (4.2 g, 65% yield). Analysis
NaBH4 (Route 2).2 To a well stirred solution of methylamine
hydrochloride (3 g, 44 mmol, 1.7 eq) in methanol (25 mL) at -20
°C was added a solution of solid sodium hydroxide (1.8 g, 45
mmol, 1.7 equiv) dissolved in methanol (35 mL). Addition was
portionwise so as to keep the temperature of the solution at -20
°C in an acetone bath containing chips of dry ice. PMK (4.6 g, 25
mmol) was then added and the mixture allowed to warm to 0 °C.
Stirring was continued for 1 h. To the solution was added sodium
borohydride (0.46 g, 12 mmol), portionwise, so as to keep the
temperature at 0 °C. The mixture was stirred overnight at 5 °C in
a cryogenic bath. Methanol was removed in vacuo and the residue
dissolved in water (330 mL). The mixture was acidified with 32%
aqueous HCl (pH 1) and extracted with DCM (3 × 75 mL) to
remove excess PMK. The aqueous phase was basified with 25%
aqueous NaOH (pH 11) and the organics extracted with DCM (3
× 75 mL). A further 20 drops of 25% aqueous NaOH were added
before the second and third DCM extractions, respectively. The
combined organic layers were dried over MgSO4 and concen-
trated, resulting in clean MDMA (3.4 g, 68% yield). The MDMA
base was converted to the HCl salt according to the procedure
detailed above. Analyses were as described previously.
Pt/H2 (Route 3).35 To ethanol (5 mL) was added a solution of
40% methylamine in water (2 mL, 58 mmol, 4.4 equiv). PMK (2
mL, 13 mmol) was slowly added to the mixture and allowed to
stir at room temperature for 1 h. To the mixture was added PtO2
(0.05 g, 0.2 mmol, 1.7 mol %) under a stream of nitrogen. The
resultant mixture was placed under a hydrogen atmosphere at
56 psi for 3.5 h. The pressure in the device was returned to 56
psi at the 1 and 2 h marks. After filtration through celite with
ethanol to remove PtO2, the volatiles were removed in vacuo. The
residue was dissolved in water (165 mL) and acidified with 32%
aqueous HCl (pH 1) and extracted with DCM (3 × 75 mL). The
aqueous phase was basified with 25% aqueous NaOH (pH 10) and
the organics extracted with DCM (3 × 75 mL). A further 10 and
20 drops of 25% aqueous NaOH were added before the second
and third DCM extractions, respectively. The combined organic
layers were dried over MgSO4 and concentrated to give clean
MDMA as an amber oil (2.0 g, 78% yield). The MDMA base was
converted into its HCl salt according to the procedure above.
Analyses were as described previously.
was in agreement with published data.31,32 IR vmax (CH2Cl2)/cm-1
:
1
1489, 1444, 1249 (C-O), 1041 (C-O), 938. H NMR (400 MHz,
CDCl3): δ 1.05 (d, 3H, J 6.2 Hz, CH3), 2.40 (s, 3H, NCH3), 2.52 -
2.65 (m, 2H, CH2), 2.69 - 2.77 (m, 1H, CH), 5.93 (s, 2H, OCH2O),
6.63 (dd, 1H, J 7.8 and 1.4 Hz, ArH), 6.68 (d, 1H J 1.5 Hz, ArH),
6.73 ppm (d, 1H, J 7.8 Hz, ArH). 13C NMR (100 MHz, CDCl3): δ
19.9, 34.2, 43.4, 56.7, 101.0, 108.4, 109.7, 122.4, 133.5, 146.1, 147.9
ppm.
Conversion of MDMA base to the HCl salt was achieved by
dissolving the base (1.5 g, 7.7 mmol) in isopropyl alcohol (2.5
mL) and acidifying with 37% aqueous HCl (30 drops). The solvent
was removed in vacuo with heat to reveal a precipitate (ranging
in color from white to pink to brown) which was homogenized
and washed with a 2:1 isopropyl alcohol:ether solution before a
final wash with ether. The resulting white MDMA hydrochloride
was dried under high vacuum (1.7 g, 95% yield). Analysis was in
agreement with published data.31,33,34 IR vmax (CH2Cl2)/cm-1: 1491,
Stable Isotope Analysis by Isotope Ratio Mass Spectrom-
etry. 13C and 15N Isotope Analysis by EA-IRMS: Carbon and
nitrogen isotope abundance analyses were carried out using an
automated nitrogen-carbon analyzer (ANCA) coupled to an
automated breath carbon analyzer (ABCA) isotope ratio mass
spectrometer (SerCon Ltd, Crewe, U.K.). Typically 0.4 mg of
sample material was weighed into tin capsules (Elemental Mi-
croanalysis, Devon, U.K.) and introduced via a solid Costech Zero-
Blank autosampler (Pelican Scientific Ltd, Alford, U.K.). The
Elemental Analyzer (EA) reactor tubes were comprised of two
quartz glass tubes filled with chromium(III) oxide / copper oxide
and reduced copper, held at 1020 °C and 620 °C for combustion
and reduction, respectively. A water trap filled with magnesium
perchlorate was used to remove water from combustion gases
thus generated, and a postreactor GC column was kept at 65 °C
1
1447, 1246 (C-O), 1040 (C-O), 932. H NMR (400 MHz, D2O):
δ 1.32 (d, 3H, J 6.6 Hz, CH3), 2.74 (s, 3H, NCH3), 2.85-2.91 (m,
1H, CH2) 3.00-3.05 (m, 1H, CH2), 3.53 (m, 1H, CH), 6.01 (s, 2H,
OCH2O), 6.83 (dd, 1H, J 7.9 and 1.4 Hz, ArH), 6.89 (d, 1H, J 1.4
Hz, ArH), 6.93 ppm (d, 1H, J 8.0 Hz, ArH). 13C NMR (100 MHz,
D2O): δ 15.4, 30.5, 39.0, 57.1, 101.7, 109.3, 110.3, 123.4, 130.3,
147.0, 148.4 ppm.
(31) Dal Cason, T. A. J. Forensic Sci. 1989, 34, 928-961.
(32) Dal Cason, T. A.; Meyers, J. A.; Lankin, D. C. Forensic Sci. Int. 1997, 86,
15-24.
(33) Bailey, K.; By, A. W.; Legault, D.; Verner, D. J. Assoc. Off. Anal. Chem. 1975,
58, 62-69.
(34) Lee, G. S. H.; Craig, D. C.; Kannangara, G. S. K.; Dawson, M.; Conn, C.;
Robertson, J.; Wilson, M. A. J. Forensic Sci. 1999, 44, 761-771.
(35) Uncle Fester. Secrets of Methamphetamine Manufacture: Including Recipes
for MDA, Ecstasy, & Other Psychedelic Amphetamines, 5th ed.; Loompanics
Unlimited: Port Townsend: Washington, 1999.
3352 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008

for separation of evolved N2 and CO2. Data were processed using
proprietary software (SerCon Ltd, Crewe, U.K.). Measured isotope
ratios are expressed in the δ notation [‰] (eq 1) relative to the
appropriate international isotope standard material anchoring the
2
isotope scale (e.g., VPDB for 13C or VSMOW for H).
RSample - RStrd
δ [‰] )
× 1000
(1)
[
]
RStrd
Isotopic Calibration and Quality Control of EA-IRMS Measure-
ments: Each batch of samples was bracketed by two blanks (empty
tin capsules) and two sets of laboratory certified standards of
known isotopic composition (Iso-Analytical, Crewe, U.K.). This
standard was leucine (δ13CVPDB ) -30.52‰, δ15NAIR ) +10.77‰).
Quality of isotope abundance measurement was monitored by
participation in annual inter-laboratory exercises organized by the
Forensic Isotope Ratio Mass Spectrometry Network (FIRMS).
2H Isotope Analysis by TC/EA-IRMS. A DeltaPlus-XP isotope ratio
mass spectrometer (IRMS) coupled to a high-temperature conver-
sion / elemental analyzer (TC/EA; both Thermo-Fisher Corpora-
tion, Bremen, Germany) was used for 2H/1H isotope ratio
measurement of synthesized MDMA samples. Typically, 0.2 mg
of solid sample was weighed into a silver capsule and placed in a
desiccator for one week before the samples were introduced into
the TC/EA by means of a solid Costech Zero-Blank solid
autosampler (Pelican Scientific Ltd, Alford, U.K.). The reactor tube
was self-packed and comprised of an Alsint ceramic tube contain-
ing a glassy carbon tube filled with glassy carbon granulate, silver
and quartz wool (SerCon, Crewe, Cheshire). The reactor temper-
ature was set to 1425 °C while the postreactor GC column was
maintained at 85 °C. Helium (99.99% purity, Air Products plc,
Crewe, Cheshire) pressure was set to 1.45 bar. Data were
processed using proprietary Isodat NT software, version 2.0
(Thermo-Fisher Corporation, Bremen, Germany). The run time
per analysis was 350 s. Measured 2H/1H isotope ratios are
expressed as δ values in ‰ relative to VSMOW as per eq 1.
Isotopic Calibration and Quality Control of TC/EA-IRMS
Measurements: The working reference gas, H2 (BOC, Guilford,
Surrey, U.K.), was calibrated against VSMOW (δ2H ) 0.0‰) and
checked against the international reference material (IRM), IAEA-
CH-7 polyethylene (δ2HVSMOW ) -100.3‰; IAEA, Vienna, Austria).
Cross-checking the δ2HVSMOW-value obtained for the working
reference gas H2 against a further international reference material
(IRM) for H2 isotope analysis, namely GISP, yielded a measured
Figure 2. δ13C results for the 18 batches: (b) Al/Hg amalgam; (9)
NaBH4; (2) Pt/H2. Visual discrimination of the Pt/H2 set is possible.
Error bars indicate one standard deviation around the mean for each
individual sample (n ) 3).
Figure 3. δ15N results for the 18 batches: (b) Al/Hg amalgam; (9)
NaBH4; (2) Pt/H2. Visual discrimination of the Al/Hg amalgam set is
possible, although the variation within that route is large. Error bars
indicate one standard deviation around the mean for each individual
sample (n ) 3).
Sample Preparation: Upon receipt by the Stable Isotope
Forensics Facility at Queen’s University, sample ID and weight
of synthesized MDMA samples were entered into a chain of
custody sheet as well as into the lab’s drug book. Aliquots
sufficient for stable isotope analysis were weighed out and dried
in a desiccator to remove any traces of moisture (in vacuo over
P4O10). Weights of the remaining drug samples were recorded in
the lab’s drug book and the samples locked into the drug safe.
To prepare samples for isotope analysis, drug aliquots were
removed from the desiccator and approximately 0.2 and 0.4 mg
samples were weighed out in triplicate into silver and tin capsules
(SerCon Ltd, Crewe, U.K.) for analysis by TC/EA-IRMS and EA-
IRMS, respectively. Capsules were subsequently crimped and
placed into 96 well-plates already prepared with blanks and
appropriate reference materials. Batch run ready well-plates were
placed into another desiccator, where they were kept in vacuo
over P4O10 until analysis.
δ2HVSMOW-value for GISP of -194.6‰ (accepted δ2HVSMOW
)
-189.73‰; IAEA, Vienna, Austria). The H3+ factor was determined
on reference H2 gas pulses of different signal size and was found
to be 4.43‰/nA. A typical batch analysis comprised 10 samples
run in triplicate, preceded and followed by a set of standards as
reported previously.36 This consisted of in-house standards (cou-
marin, δ2HVSMOW ) +62.56‰) and one IRM (IAEA-CH-7) as
calibration controls at the beginning and end of the set. Each batch
was preceded and followed by a blank silver capsule. Precision
of 2H isotope analysis as monitored by the IRMs and lab standards
was (1.15‰ or better. Measured δ2H-values were normalized
according to the method described by Sharp et al.37 with stretch
factors typically being of the order of 1.027 to 1.053.
RESULTS AND DISCUSSION
A large stock of PMK was first synthesized from one bottle of
safrole, according to the scheme in Figure 1. Once a sufficient
stock of PMK had been attained, MDMA‚HCl was synthesized
according to the three reaction pathways in Figure 1. The δ13C,
δ15N and δ2H data for the 18 MDMA samples are graphically
represented in Figures 2-4. Each data point represents the mean
of triplicate analyses, and the error bars indicate one standard
deviation around the mean for each individual sample. The δ13C,
δ15N, and δ2H data of the 18 batches are given in Table 1. These
results are in line with published IRMS values of MDMA.10,13,15-17,21
(36) Farmer, N. L.; Meier-Augenstein, W.; Kalin, R. M. Rapid Commun. Mass
Spectrom. 2005, 19, 3182-3186.
(37) Sharp, Z. D.; Atudorei, V.; Durakiewicz, T. Chem. Geol. 2001, 178, 197-
210.
Analytical Chemistry, Vol. 80, No. 9, May 1, 2008 3353

Figure 5. Origin of the marked carbon and nitrogen atoms on the
MDMA‚HCl molecule synthesized by three reductive amination routes
are as follows: (/) safrole; (9) methylamine hydrochloride or aqueous
methylamine.
Figure 4. δ2H results for the 18 batches: (b) Al/Hg amalgam; (9)
NaBH4; (2) Pt/H2. Visual discrimination of all routes is possible, but
the separation between the Al/Hg amalgam and NaBH4 sets is small.
Error bars indicate one standard deviation around the mean for each
individual sample (n ) 3).
Figure 6. Origin of the marked hydrogen atoms on the MDMA‚HCl
molecule synthesized by three reductive amination routes are as
follows: (/) safrole; (9) a proton source throughout the synthetic
process such as safrole, KOH, H2O2, HCOOH, MeOH, H2SO4, H2O,
CH3NH2, HCl, H2, etc.
Table 1. δ13C, δ15N, and δ2H Values for the MDMA‚HCl
Samples Synthesized from the Same Precursor (PMK)
Using Three Reductive Amination Routesa
other two sets, although the variation within that set is large.
Notably, MDMA synthesis with the Al/Hg amalgam was the
hardest to replicate as the temperature of the exothermic reaction
is difficult to control exactly, and a series of reagents must be
added by hand in quick succession; thus, temperature reached
and rate of addition of reagents are likely to vary slightly from
batch to batch. From this study, the δ15N data appears to be the
most sensitive to these inadvertent differences in preparative
method, confirming the observations by Carter et al.16 during the
synthesis of methamphetamine. It should be noted, however, that
results were published by Billault et al.10 in which a different Al/
Hg amalgam method produced MDMA with consistent δ15N
values.
It is also interesting to examine the δ15N data in relation to
the nitrogen contributing precursor used. The nitrogen atom on
the MDMA molecule is contributed, along with the N-methyl
carbon, by methylamine hydrochloride (Al/Hg amalgam and
NaBH4) or aqueous methylamine (Pt/H2). If the δ15N of the
nitrogen contributing reagent was solely responsible for the δ15N
of the MDMA‚HCl, then the δ15N values would be expected to
fall into two groups: Al/Hg amalgam and NaBH4 data points in
one group and Pt/H2 data points in the other. Instead (with the
exception of one Pt/H2 batch which is enriched in 15N relative to
the rest of the set) the NaBH4 and Pt/H2 data points fall within
the same range, thus indicating the synthetic process itself is
responsible for fractionation of the nitrogen isotopes. This
observation confirms previous studies suggesting that δ15N values
are largely the result of fractionation due to the synthetic process
rather than the precursor.10,16
The δ2H data points show the most variation (100‰); see
Figure 4. This is perhaps unsurprising due to the number of
potential proton contributors. Furthermore, hydrogen atoms at
select positions under certain conditions may be prone to
exchange. For the synthetic routes in this study, a few of the
possible hydrogen atom contributors are safrole, KOH, H2O2,
HCOOH, MeOH, H2SO4, H2O, CH3NH2, HCl, H2, etc; see Figure
6. The observed variation of the δ2H values is therefore expected
due to the large number of possible hydrogen contributors, but
fractionation due to synthetic process is also likely to contribute.
Interestingly, the δ2H data points corresponding to the Pt/H2 set
cluster together remarkably well relative to those of the other
two sets. Practically, the Pt/H2 synthesis was the easiest to
route
δ13C [‰]
δ15N [‰]
δ2H [‰]
Al/Hg amalgam
Al/Hg amalgam
Al/Hg amalgam
Al/Hg amalgam
Al/Hg amalgam
Al/Hg amalgam
NaBH4
-26.5
-27.1
-26.7
-27.1
-27.2
-27.0
-26.8
-26.7
-26.6
-26.5
-26.7
-26.7
-28.1
-27.9
-28.1
-27.8
-27.9
-27.9
8.9
-0.1
8.9
-40.7
-50.5
-36.5
-45.6
-43.8
-41.6
-18.8
-7.6
-3.6
-1.9
-2.3
17.4
16.5
15.9
16.4
17.0
17.3
16.3
16.6
23.8
16.6
18.1
16.5
NaBH4
NaBH4
6.3
NaBH4
-16.7
-20.6
-5.8
-93.4
-92.7
-92.5
-93.5
-92.6
-93.9
NaBH4
NaBH4
Pt/H2
Pt/H2
Pt/H2
Pt/H2
Pt/H2
Pt/H2
a Each value is the mean of triplicate analyses; standard deviation
around the mean ranged from 0.10 to 0.25‰, from 0.15 to 0.95 ‰, and
from 0.5 to 10.0 ‰ for δ13C, δ15N, and δ2H values, respectively.
Precision for measured δ13C, δ15N, and δ2H values of the reference
materials was ( 0.15‰, ( 0.20‰, and ( 1.0‰ or better.
When examining first the δ13C data, the least variation is shown
by this element (1.6‰); see Figure 2. The small variation in δ13C
values of the samples in this study is unsurprising given that 10
of the 11 carbon atoms on the final MDMA molecule are
contributed by the safrole starting material, and all of these
samples were synthesized using the same bottle of safrole. The
N-methyl carbon on the molecule is contributed by methylamine
hydrochloride (Al/Hg amalgam and NaBH4) or aqueous methyl-
amine (Pt/H2); see Figure 5. In general, the δ13C data points within
a synthetic route cluster together visually, indicating the fraction-
ation induced within the synthetic process is reproducible by the
same chemist using the same method and materials. By δ13C data,
the Pt/H2 route can be discriminated from the other two, but the
δ13C values of the Al/Hg amalgam and NaBH4 routes overlap.
Examination of the δ15N data points reveals variation over 27‰,
and the Al/Hg amalgam batches exhibit the widest variation within
any synthetic route with two of the batches showing considerable
15N enrichment relative to the remaining four (Figure 3). By δ15N
data, the Al/Hg amalgam route can be discriminated from the
3354 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008

Figure 7. Two variable plot of δ13C and δ2H data. Here, the Pt/H2
route can be clearly separated from the other two sets, but the Al/Hg
amalgam and NaBH4 sets cannot be accurately discriminated. Error
bars indicate one standard deviation around the mean for each
individual sample (n ) 3).
Figure 10. HCA using δ13C, δ15N, and δ2H data of 18 synthesized
MDMA batches. Accurate clustering by synthetic route is evident using
SPSS 14.0 (furthest neighbor and Euclidean distance).
Chemometric Analysis. Because visual discrimination may
lend itself to subjectivity and has not allowed irrefutable discrimi-
nation of the sets (note the marginal separation of the NaBH4
and Al/Hg batches in Figures 4 and 7), cluster analysis was
employed to find an algorithm that could accurately link the data
according to synthetic route. Hierarchical cluster analysis (HCA)
has been used in similar published works, but in this study, the
validity of the resulting clusters can be assessed because the
samples are of known provenance. Remarkably, HCA analysis of
the δ13C, δ15N, and δ2H data allows the samples to be grouped
correctly according to synthetic route (Figure 10). HCA analysis
was performed using SPSS 14.0, furthest neighbor clustering
method, and Euclidean distance measure. In the resulting den-
dogram, samples that are found to be most similar are linked by
a bracket near the 0 end of the scale. HCA generates three clusters
in the data, each one corresponding to a synthetic route. By this
algorithm, samples from the Pt/H2 route are more similar to the
Al/Hg amalgam set than the NaBH4 set.
Figure 8. Two variable plot of δ13C and δ15N data. The samples
fall into five groups because two Al/Hg amalgam samples and one
Pt/H2 sample are separated from their respective sets. Error bars
indicate one standard deviation around the mean for each individual
sample (n ) 3).
To ascertain the discriminating power of each of the elements,
HCA was performed on all combinations of δ13C, δ15N, and δ2H
taken one or two at a time. Accurate discrimination of the synthetic
routes was only possible when δ2H data was included, that is when
δ2H data was subjected to HCA on its own or in conjunction with
δ13C and/or δ15N (Figures 11 and 12). Clearly, this work indicates
that δ2H analysis of MDMA is sufficient and necessary for the
discrimination of samples by synthetic route, as δ13C and δ15N
data cannot discriminate all three routes accurately when exam-
ined visually or with HCA.
Unsurprisingly, carbon isotope data was shown to be relatively
poor for discrimination among the batches as the dendogram
produced from HCA of δ13C and δ2H data was found to be exactly
identical to that produced from HCA of δ2H data alone (Figure
11). Furthermore, the dendogram produced from HCA of δ13C,
δ15N and δ2H data (Figure 10) is nearly identical to that produced
from HCA analysis of δ15N and δ2H (Figure 12). The slight
difference in these two dendograms is evidenced by comparison
of the order of the first three Pt/H2 batches along the vertical
column, i.e., 16, 18, 13 vs 13, 16, 18. This “difference” is of course
meaningless as the level of similarity and, thus, grouping are the
Figure 9. Two variable plot of δ2H and δ15N data. The samples fall
into five groups because two Al/Hg amalgam samples and one Pt/H2
sample are separated from their respective sets. Error bars indicate
one standard deviation around the mean for each individual sample
(n ) 3).
replicate since once the imine is formed, the reduction occurs in
a hydrogenation device under controlled pressure. By δ2H data,
it is tentatively possible to discriminate between the three synthetic
routes, but the separation between the Al/Hg amalgam and NaBH4
routes is small given the spread of data within each group.
Two variable plots were then assessed to determine if two
elements might afford better visual discrimination than single
variable plots. The resulting groups, however, prove somewhat
misleading in terms of discrimination according to synthetic route.
It is clear from visual interpretation of Figures 7-9 that the 18
samples do not cluster into three groups according to synthetic
route, and therefore the single variable plot using δ2H data
provides the best visual discrimination of these MDMA‚HCl
samples according to synthetic route.
Analytical Chemistry, Vol. 80, No. 9, May 1, 2008 3355

route. For this reason, δ15N data may be of little use in grouping
batches synthesized by the same preparative route or even
laboratory; instead, δ15N data may be so affected by reaction
conditions that it is useful only for discriminating samples
according to the production batch.
If these MDMA‚HCl samples had been seized on the street
and subjected to IRMS analysis, δ2H data would have allowed
tentative visual discrimination into three groups corresponding
to the synthetic route used for manufacture, although the
separation between the NaBH4 and Al/Hg amalgam sets is narrow
and therefore disputable. It should be noted that each synthetic
route in this study was as carefully controlled as possible so that
it could be “exactly” repeated. In reality, however, clandestine
chemists are likely to work in a less controlled fashion during
the manufacture of MDMA‚HCl, although one might argue that
repetitive large scale syntheses over long periods of time would
develop a degree of “control-by-routine” on the part of the
clandestine chemist. It is not yet known if slight alterations to
the conditions of a synthetic route (such as temperature, stirring
time, pressure achieved in the hydrogenator, etc.) would change
the magnitude of fractionation to such a degree that data points
lie on a continuum rather than in discrete groups. This is the
subject of ongoing further studies within our laboratories.
Regardless, the present study shows that hierarchical cluster
analysis of δ2H data of MDMA‚HCl is powerful enough to
accurately discriminate between three common synthetic routes,
and the addition of carbon and/or nitrogen data does not diminish
or enhance this. Because visual discrimination may be subjective,
HCA has been shown as an effective method for accurately
discriminating among synthetic routes when δ2H data is consid-
ered on its own or in conjunction with δ13C and/or δ15N data.
Figure 11. HCA of δ2H data of 18 synthesized MDMA batches.
Accurate clustering by synthetic route is evident using SPSS 14.0
(furthest neighbor and Euclidean distance). HCA analysis of δ2H and
δ
13C data results in an identical dendogram.
ACKNOWLEDGMENT
N.N.D. and W.M.-A. gratefully acknowledge financial support
of this work through an Engineering and Physical Sciences
Research Council (EPSRC) research grant (EP/D04030345/1)
which provided a studentship for H.A.S.B. W.M.-A. also gratefully
acknowledges financial support for H.F.K. through an EPSRC
Platform Grant (GR/T25200/01).
Figure 12. HCA of δ15N and δ2H data of 18 synthesized MDMA
batches. Accurate clustering by synthetic route is evident using SPSS
14.0 (furthest neighbor and Euclidean distance).
same, again illustrating that the carbon isotope data do not add
discriminating power on the basis of the synthetic route.
Received for review December 18, 2007. Accepted
February 21, 2008.
CONCLUSIONS
Despite the careful control exercised during each synthesis,
the δ15N data varied between batches, even within a synthetic
AC702559S
3356 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008