Talanta 67 (2005) 262–268

A case study in forensic chemistry: The Bali bombings

David Royds a,∗ , Simon W. Lewis b , Amelia M. Taylor b
a Australian Federal Police Forensic Service, Box 401, Canberra 2601, Australia
b School of Biological and Chemical Sciences, Deakin University, Geelong, Vic. 3217, Australia

Received 21 June 2004; accepted 10 March 2005

Available online 25 May 2005

Abstract

The Bali bombings on 12 October 2002 killed 202 people and caused international outrage. The police investigation referred to as “Operation
Alliance” involved the Indonesian National Police, the Australian Federal Police and others, resulted in the arrests of key personnel and exposed
the link between Jemaah Islamiah (JI) and al-Queda.
This paper describes aspects of the investigation from a forensic chemists perspective. The recovery of water soluble residues from blast
scenes and suspect’s residences is a challenging task that requires a meticulous and methodical approach. The concept of deploying forensic
chemists and setting up a “mobile laboratory” from the outset proved to be a highly effective and efficient way to process large numbers of
samples and assist investigators with relevant information at the time that it was most needed. It also identified the need for a new generation
of field portable instruments.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Case study; Improvised explosives; Operation Alliance; Mobile laboratory; Bali bombing

1. Introduction reported massive structural damage, burning motor vehicles

and buildings, emergency personnel struggling to cope and
The bombings that occurred in Bali on the 12 October a large crater in the roadway outside the Sari Club. They
2002, were the most callous act of terrorism in Australian immediately recognised the scene for what it was and initiated
history. It was a typical busy Saturday night in the main tourist a full police investigation, which became known as Operation
area of Kuta when, shortly after 11.00 pm a bomb exploded in Alliance.
Paddy’s Bar. Moments later a second and much larger bomb
exploded on the roadway outside the popular Sari night club,
and then a third bomb, which was small in comparison to the 2. Operation Alliance
first two devices, exploded approximately 10 km away, near
the United States and Australian Consulates. Two hundred The Australian Federal Police (AFP) sent a team of inves-
and two people were killed of which 88 were Australians, tigative and forensic officers to Bali to assist the Indonesian
and many hundreds more were injured. National Police (INP) in their investigation. Operation Al-
Initial reports from that night started with a suggestion that liance became the largest investigation ever undertaken by
gas cylinders had exploded. Stories of confusion, panic and any Australian police force and was truly an “alliance” be-
destruction were relayed back to Australia. Two Australian tween the Australian, Indonesian and international law en-
Federal Police (AFP) forensic officers travelling to Jakarta forcement communities. Although there were many forensic
were redirected to the scene to make an assessment. They facets to the operation, such as the involvement of odontol-
ogists, pathologists, and police disaster victim identification
∗ Corresponding author. Tel.: +61 2 6287 0383; fax: +61 2 6287 0270.
specialists, this article only relates to some aspects of the
E-mail address: [email protected] (D. Royds). chemistry component of the criminal investigation.

0039-9140/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2005.03.026
 D. Royds et al. / Talanta 67 (2005) 262–268 263

3. Detail of the events of October 12 made from a relatively small amount of TNT and was initi-
ated remotely using a mobile phone about 45 s after the Sari
It is estimated that the Paddy’s Bar explosion involved be- Club bomb. No injuries were caused by this bomb, however
tween 1 and 5 kg of TNT contained within five lengths of this site was important as it was rich in physical evidence.
PVC pipe of 50 mm diameter sewn into a tartan lined black
vest worn by the suicide bomber. Tiny fragments of tartan
fabric were recovered from sites surrounding the blast epi- 4. The breakthrough
centre. Numerous pieces of metal were also found amongst
the debris and these were found to match fragments recov- The initial breakthrough came about through forensic ex-
ered from living and deceased victims. Connective tissue and aminers from the INP finding impressed numbers in chassis
spatter marks were visible on the ceiling above the epicentre. fragments that survived the main blast. Although both chas-
The common organic high explosive (HE) known as trinitro- sis and engine numbers appeared to have been filed away
toluene (TNT) was identified in this material. The notable in an attempt to avoid detection, another number otherwise
absence of a crater and extrapolation of spatter indicated the unknown to the general public had been stamped on the chas-
position of the bomb to be between 80 and 120 cm above the sis rail during a routine vehicle inspection by the Indonesian
floor. This was the first evidence to indicate the involvement authorities. Through the finding this number, the INP inves-
of a suicide bomber. Later, analysis of this biological mate- tigators were able to trace the vehicle back to a Javanese
rial revealed that it had all come from the one individual. A man named Amrozi (referred to in the media as “the smil-
dismembered head and two lower legs were found amongst ing assassin”) who was a key member of the terrorist gang.
various body parts at the morgue. As samples for DNA analy- It has subsequently been determined that he purchased the
sis were being collected from all the significant body parts as van and chemicals and attended meetings to plan the crime.
a routine part of the Disaster Victim Identification (DVI) pro- Amrozi confessed and named his co-conspirators Samudra,
cess, it was a simple matter to compare profiles and link those Idris, Dulmatin, Muklas and Ali Imron.
parts with the ceiling spatter. These body parts turned out to
be a rich reservoir of precious trace evidence. Investigators,
now aware of this DNA database and suspecting the involve- 5. The chemistry
ment of a suicide bomber, sought out that person’s parents in
order to establish his true identity. To our surprise, the DNA An explosion is merely an energetic reaction that results in
profile of that union did not match. A further search of the the rapid release of gas. The rate of the reaction determines the
database however, did in fact reveal the presence of this pro- destructive consequences [1,2]. Improvised explosives using
file and it was in many tiny body parts that were found near inorganic components such as ammonium nitrate with fuel
the Sari Club. This observation implicated the involvement oils (ANFO) usually have relatively low velocities of det-
of a second suicide bomber. onation and are suitable for “pushing” applications such as
The second and largest device was detonated at the nearby in earthworks and mining [3]. ANFO is the most common
Sari Club at 11:08 pm, 15 s after the Paddy’s Bar explosion. explosive used in Australia with about 650,000 tonnes used
It is estimated that over 350 people were in the club at this annually [4]. Organic explosives such as TNT, PETN and
time. The force of the blast was so strong that it registered on RDX are characterised by high velocities of detonation (up
seismographs in neighbouring countries. It is estimated that to 9000 m/s) and have applications in detonators, boosting
an improvised explosive mixture with an estimated power chargers and military ordinance [5]. Although high explo-
equivalent of 150 kg of TNT had been used. Exhaustive anal- sives are used to initiate and boost low explosives, they are
yses revealed the presence of the chlorate ion in and around usually totally consumed in the ensuing reaction and rarely
the crater. When arrested the offenders stated the bomb was leave traces for the forensic chemist to find.
made from a mixture of potassium chlorate, sulphur and alu- Contrary to intuition, but accepted by most experienced
minium, boosted by TNT. It is now believed that the im- practitioners, large explosions usually leave less tell-tale
provised explosive mixture, more than 1 tonne in total, was residue than smaller ones. Most large bombs are improvised
contained in 12 four-drawer plastic filing cabinets intercon- and made of inorganic oxidizers such as nitrates, chlorates
nected with detonating chord which is made from another and perchlorates and fuels such as diesel oil, sugar, sulphur
organic high explosive known as pentaerythritol tetranitrate or reactive metals such as aluminium dust [2]. The release
(PETN). This device was placed in the back of a Mitsubishi of such tremendous energy often causes a fire which is in-
L300 van and parked outside the Sari Club. It was also initi- variably fought with water. The bomb outside the Sari Club
ated by a suicide bomber. Remarkably, some fragments from caused the thatched roofs over the courtyards to catch on fire
this individual survived the blast and were used to establish and the blast broke a water main under the roadway caus-
his identity through DNA analysis. ing the crater to fill. The quantity of post blast residual mat-
The third bomb, referred to in the media as the “USA Con- ter was reduced through dissolution and being washed away
sulate Bomb”, was detonated about 10 km away on a public down gutters and drains. The warm and humid climate of the
street near the Unites States and Australian Consulates. It was tropics compounded the problem as atmospheric moisture
264 D. Royds et al. / Talanta 67 (2005) 262–268

Fig. 1. Water filled crater outside the Sari Club.

combined with the fuels and carbonaceous materials (traffic forward command post for security and effective communi-
soot) to further reduce the amount of reactive chemicals to cations, and as close to the scene as to be convenient, yet
be found (see Fig. 1). far enough away to minimise the risk of contamination. A
nearby motel room was cleared and prepared for the installa-
tion of such instruments as a microscope with camera, an ion
6. The incident scene mobility spectrometer (IMS), a portable infra-red spectrom-
eter (FT-IR), and a range of reagents for various presump-
Good forensic science starts at the crime scene where care- tive and other spot tests. Of particular interest were those for
ful observation is required in order to collect important ev- detecting strong oxidizers (diphenylamine reagent, modified
idence and reconstruct events [6]. In the case of explosion Griess test (nitrites/nitrates), aniline hydrochloride for chlo-
scenes the information that can be gained concerning the rates, and others such as Nessler’s for the ammonium ion,
type of explosive, quantity and bomb architecture is vital and so on, see Table 1). From lessons learnt in Bali the AFP
for successful apprehension of the perpetrators [2]. It is im- has since purchased a portable ion analyser, which is used
portant to ensure details such as the location of the sample for the instrumental analysis of these inorganic ions that are
and the identity of every person involved in the chain of pos- commonly found in improvised explosives.
session is accurately recorded and it is also imperative that The rationale behind a mobile laboratory is to produce
the risk of contamination be minimised from the outset [6]. timely, albeit tentative findings for the investigators whilst
A professional law enforcement organisation has procedures reducing the number and volume of samples to be sent
in place to ensure this occurs and that the integrity of sam- to the main laboratory for more exhaustive confirmatory
ples is maintained from scene to court [6,7]. Crime scenes analyses. In such an important case as Operation Alliance,
where explosions have occurred are particularly challenging which involved victims from 22 countries and presented
to process, due to their complexity. Microscopic fragments the possibility of a prosecution in the International Court,
of explosive and its residues will be spread throughout a large it was considered prudent to have findings corroborated
amount of post blast debris [2]. by other reputable laboratories. On occasions, up to four
samples were collected; one for the main AFP laboratory in
Canberra, a duplicate for the host country, and one each for
7. The mobile laboratory the Victoria Police Forensic Services Centre (VPFSC) and
the Forensic Explosives Laboratory (FEL), Britain.
Provision of information at an early stage in an inves- The contribution of the scientists from Britain was partic-
tigation is extremely helpful in rapid identification of sus- ularly praiseworthy, for not only did they bring a wealth of
pects. The AFP deployed part of its mobile laboratory to experience to the table during the early stages of the investi-
Bali to assist with the collection of trace evidence and ex- gation, they also undertook an environmental survey to estab-
plosive residues. AFP practice is to set it up as close to the lish the background levels of various ions (including chlorate,
 D. Royds et al. / Talanta 67 (2005) 262–268 265

Table 1
Colorimetric screening methods for explosives [1,8,9]
Test Target explosives Colour change
Alcoholic KOH: 3% KOH in ethanol TNT Purple/brown
2,4-DNT Yellow
2,6-DNT Yellow
Tetryl Violet
Griess reagent: methanolic N-(1-napthyl)ethylenediamine in the presence of sulphanilic acid and zinc dust Nitrates and nitrites Purple
Nitrate Pink to red
Nitrocellulose Pink
Nitroglycerin Pink to red
PETN Pink to red
RDX Pink to red
Tetryl Pink to red
Diphenylamine: reaction with diphenylamine in concentrated sulphuric acid Chlorate Blue
Nitrates Blue
Nitrocellulose Blue-black
Nitroglycerin Blue
PETN Blue
RDX Blue
Tetryl Blue
TATP Blue
Aniline hydrochloride: aniline acidified with hydrochloric acid and activated with potassium chlorate Chlorate Blue
Hypochlorite Blue
Chlorite Blue
Nessler’s reagent: mercury-iodide solution (commercially available) Ammonium ions Yellow/brown
Nitroglycerin Black
TNT Red

nitrate and so on). This is essential information from a scien- be argued (and was) that although chlorates are not normally
tific perspective. found in the environment, it may be possible in such a scene
to have detected residues of unburnt matches that fell from
victims during or after the blast! For this reason, elevated
8. The analytical approach surfaces were preferred sites for sampling.
Some forensic explosive laboratories favour liquid
The approach taken in analysing post debris starts with a chromatography in combination with mass spectrometry
number of presumptive tests (see Table 1). These presump- (LC–MS) as the principle technique [10], whilst others prefer
tive tests generally involve reactions resulting in a distinctive gas chromatography with either a thermal energy analyser
colour. These tests require careful use of blanks and pos- (GC–TEA) [11,12], which is specific for compounds con-
itive controls to ensure valid results, and while rapid they taining the nitro and nitroso functional groups; or positive
are not definitive [1,2,8,9]. Samples that yield promising re- and/or negative ion chemical ionisation mass spectrometry
sults are collected and packaged for further examinations (GC–nciMS) detectors [9,13].
and analysis in the laboratory (see Table 2) [1,9]. The level Time of flight mass spectrometry (TOF-MS) can get past
of confidence rises with each positive result. A satisfactory the problem of thermal instability by performing very rapid
level of confidence in establishing the identity of a ques- analyses using a short narrow bore column, and deconvolut-
tioned sample is generally achieved after corroboration from ing the resulting chromatogram by exceedingly rapid mass
two independent techniques. This is known as “orthogonal” spectral analysis. This however raises the question as to
testing. whether it is truly orthogonal testing for if the chromatogra-
Orthogonal testing is relatively easy to achieve when deal- phy does not resolve co-eluting compounds, then is it really
ing with bulk material. For example, a very high level of only the one technique?
confidence would be expressed in an identification based Capillary electrophoresis (CE) was viewed as the emerg-
on an infra-red spectrum and a chromatogram where reten- ing technology to displace ion chromatography (IC) as the
tion times have been shown to be precise and reproducible. preferred method for the analysis of inorganic ions [13]. It
The problem becomes more difficult when dealing with trace seemed to have the greatest promise for conversion to minia-
quantities, and especially so when they are from an inorganic turisation to produce field-portable units. Certainly the CE
source which may occur naturally in the environment, or if technique has acceptable detection limits and the capacity
they are volatile and unlikely to persist in the area. It could to resolve ions with an extraordinary number of theoretical
266 D. Royds et al. / Talanta 67 (2005) 262–268

Table 2
Analytical methods for explosive residue analysis [1,9–13]
Analytical method Target explosives Detection limit Advantages Limitations
TLC Organica ␮g to sub-␮g Simple inexpensive rapid Low resolution susceptible to contamination
GC–ECD Nitrogen-containingb pg Rapid selective highly sensitive for Requires volatile analytes insensitive to
nitrogen-containing compounds hydrocarbons
GC–TEA Nitro-containingc pg Sensitive and selective for nitro Requires volatile analytes expensive limited
compounds to nitro compounds
HPLC-UV or DA Organica ng Sensitive Low selectivity
IR Organica and inorganicd ␮g Universal organic and inorganic Residues require separation prior to analysis
compounds have characteristic spectra
MS Organica and inorganicd pg to ng Rapid selective sensitive reliable
GC–MS Organica (low bp) pg to ng Rapid selective reliable sensitive Limited to volatiles
plasticizerse stabilisersf
LC–MS Organica pg to ng Analysis of non-volatiles Requires interface between HPLC and MS
SEM/EDX Inorganicd Organic and inorganic samples Expensive, requires expert operators
a Organic explosives: DNT, TNT, EGDN, NG, PETN, NC, RDX, HMX, TATP, HMTD.
b Nitrogen-containing explosives: DNT, TNT, picric acid, EGDN, NG, PETN, NC, nitrate salts.
c Nitro explosives: DNT, TNT, picric acid, EGDN, NG, PETN, NC, RDX, HMX. RDX, HMX, nitrate salts, black powder.
d Inorganic explosives: nitrate and chlorate salts.
e Plasticizers: phthalate and sebacate esters.
f Stabilisers: diphenylamine (DPA), ethyl centralite (EC).

plates; however it has fallen out of favour with many analysts of samples (six in total), however this is a remarkable ob-
due to difficulties in obtaining reproducible migration times. servation in itself as such a reactive ion is unlikely to persist
In order to mount an argument that an inorganic ion in the environment. Notably, it was recovered from crevices
was part of the explosive mixture, the forensic analyst must in the bitumen surface of the blast crater, and on elevated
demonstrate its concentration to be significantly higher in and surfaces of lamp posts facing the blast.
around the crater than in the background. A large number of Scientists from the FEL searched the pitting in the soft
analyses may be required to achieve this objective. At this aluminium of street signs that had been retrieved from nearby
stage of technological development there are no instruments roof tops using an SEM EDX. They reported elevated levels
available on the market for such applications. of chloride, low levels of chlorate and but only marginally
The total number of samples analysed at the AFP labora- higher than background levels of potassium. Sulphur was not
tory was in excess of 2000 (ca. 2400). Many of these were detected.
items of clothing recovered from victims as they returned to Soils from in and around the blast seat were anal-
Australia. The chlorate ion was identified in a small number ysed by scientists from the AFP using X-ray fluorescence

Fig. 2. The newspaper with a barely visible footprint on its surface.

 D. Royds et al. / Talanta 67 (2005) 262–268 267

Fig. 3. The footprint in aluminium dust on a black gelatin lift.

spectroscopy. No evidence was found to indicate higher than progressed, confessions from the accused added further cor-
background levels of any of the elements implicated. roboration with TNT, potassium chlorate, sulphur and alu-
The organic explosive TNT was detected using IMS at minium being volunteered as the ingredients for the main
the two other scenes and confirmation was achieved by both Sari Club bomb, and TNT being the main charge in both the
GC–TEA and GC–nciMS. other bombs.
The findings of the AFP in relation to chlorate and TNT It should be borne in mind that these bomb scene analy-
were corroborated by the VPFSC and the FEL using differ- ses formed part of wider investigation. About 4 months after
ent procedures and techniques. As the police investigation the incident the police investigators led the forensic team to a

Fig. 4. A microscopic particle of potassium chlorate, sulphur and aluminium found in vacuumings from suspect premises.
268 D. Royds et al. / Talanta 67 (2005) 262–268

suburban dwelling in Denpasar where it was alleged the bomb In hindsight, it was a privilege to be involved in Operation
was constructed. Although an initial cursory inspection pro- Alliance with friendships being forged between like-minded
vided little encouragement, a mass of physical, chemical and people from around the world. The unifying effect of a world
trace evidence was harvested from this scene. To mention but outraged by the senseless acts of a religious fundamentalist
two, minute particles of the explosive mixture were recovered minority will ensure those who use terrorism will not win.
from vacuumings taken from the gaps in the tiled floor and
a barely visible footprint in aluminium dust was found on a
sheet of newspaper (see Figs. 2–4). Acknowledgements

David Royds would like to acknowledge his colleagues

9. Conclusion Ben Kwok, Sarah Benson and Dr. Naomi Jones who worked
selflessly toward our common goal. He would also acknowl-
The “mobile laboratory” concept proved to be effective in edge and thank Dr. Jim Pearson, John Kelleher and George
providing valuable information to the investigators at a time Xydias from the Victoria Police Forensic Service Centre for
when it was most needed. bringing their knowledge and expertise to the fore in this com-
A multi-jurisdictional response is the most appropriate plex investigation. They made many important contributions
way to effectively deal with an incident involving mass ca- that assured the success of Operation Alliance.
sualties from many nationalities because it allowed the inde-
pendent forensic strengths of each jurisdiction to be shared,
whilst reducing the workload on individuals. A high level of References
confidence can be expressed when different laboratories from
different jurisdictions arrive at the same conclusion using [1] R. Saferstein, Criminalistics: An Introduction to Forensic Science,
their own independently developed and validated methods. Pearson Education Inc, New Jersey, USA, 2004.
The forensic chemist’s contribution to Operation Alliance [2] H.J. Yallop, Explosion Investigation, Scottish Academic Press Ltd,
was significant and helped lead to the early arrests of offend- Edinburgh, Scotland, 1980.
[3] S.G. Murray, in: J.A. Siegel, P.J. Saukko, G.C. Knupfer (Eds.), Ex-
ers, thus thwarting another similar incident from occurring plosives/Commercial in Encyclopaedia of Forensic Science, vol. 2,
and saving countless unknown lives and suffering. Academic Press, UK, 2000, pp. 750–758.
[4] Queensland Government Department of Natural Resources and
Mines, Key points for discussion regarding the regulation of secu-
10. Personal comments (David Royds) rity sensitive ammonium nitrate (SSAN), 2004. http://www.nrm.qld.
gov.au/mines/explosives/pdf/ssan discussion.pdf (accessed 7 March
Participation in an investigation of this magnitude de- 2005).
[5] S.G. Murray, in: J.A. Siegel, P.J. Saukko, G.C. Knupfer (Eds.), Ex-
mands an intensity of purpose that normally protects one from plosives/Military in Encyclopaedia of Forensic Science, vol. 2, Aca-
joining in the bereavement journey of the relatives mourning demic Press, UK, 2000, pp. 764–771.
their loved ones. But in its enormity there were periods of [6] N. Weston, in: P.C. White (Ed.), The Crime Scene in Crime Scene
frustration caused largely by instrument failure, and tedium to Court—The Essentials of Forensic Science, The Royal Society of
that made it difficult to manage until the results of our work Chemistry, Cambridge, UK, 2004, pp. 21–55.
[7] P. Meikle, in: J.A. Siegel, P.J. Saukko, G.C. Knupfer (Eds.),
turned those feelings of hopelessness to exhilaration. Explosives/Bomb-scene Management in Encyclopaedia of Forensic
The Bali bombings confirmed my deeply held belief that Science, vol. 2, Academic Press, UK, 2000, pp. 745–750.
forensic scientists should aspire to develop a broad under- [8] S. Palenik, in: J.A. Siegel, P.J. Saukko, G.C. Knupfer (Eds.), Micro-
standing of a wide range of evidence types in order to be chemistry in Encyclopaedia of Forensic Science, vol. 3, Academic
most effective in such situations. Scientists must be able to Press, UK, 2000, pp. 1111–1116.
[9] T. Tamiri, in: J.A. Siegel, P.J. Saukko, G.C. Knupfer (Eds.), Explo-
work in the morgue to recover evidence from bodies, examine sives/Analysis in Encyclopaedia of Forensic Science, vol. 2, Aca-
scenes and interpret the damage so as to collect material that demic Press, UK, 2000, pp. 729–745.
is most likely to produce evidence, search suspect premises [10] J. Yinon, in: J. Yinon (Ed.), Analysis of Explosives by LC/MS in Ad-
whilst minimising the potential for cross contamination, and vances in Forensic Applications of Mass Spectrometry, CRC Press,
conduct a wide range of analyses in the laboratory, including Boca Raton, 2004.
[11] P. Kolla, A. Sprunkel, J. Forensic Sci. 40 (1995) 406.
fibre examinations, paint and plastic analysis as well as post [12] A.M. Jiménez, M.J. Navas, J. Hazard. Mater. A 106 (2004) 1.
blast explosive residue analysis, and interpret all the scientific [13] J. Yinon, in: M.C. Bogusz (Ed.), Explosives in Handbook of Ana-
observations in an objective manner. lytical Separations, vol. 2, Elsevier, Amsterdam, 2000, pp. 603–616.