DETECTION OF CHEMICAL AGENTS IN WATER BY MIMS

M

Detection of Chemical Agents in Water by Membrane-Introduction Mass Spectrometry

John S. Morgan, Wayne A. Bryden, Rebecca F. Vertes, and Scott Bauer

embrane-Introduction Mass Spectrometry (MIMS) has evolved into a power-ful technique for the selective detection of volatile organic compounds in aqueousmatrices. Contaminants in water samples have been analyzed at the parts-per-trillionlevel using hydrophobic silicone membranes. Recent advances in new membranematerials will permit the examination of polar analytes such as chemical agents. Incombination with a small time-of-flight mass spectrometer, it will be possible toconstruct a portable detector for treaty verification, counterproliferation, and infrastruc-ture protection. Other developments include deconvolution algorithms for complexanalyte mixtures, improved understanding of analyte–water reactions, and thermaldesorption of analytes from the membrane into the vacuum. (Keywords: Chemical,Mass spectrometry, Membrane, Weapons.)

INTRODUCTIONIt is critically important to rapidly detect and iden-

tify chemicals during Chemical Weapons Convention(CWC) inspections. Miniaturized instrumentationwith this capability can also be used to protect person-nel. Small, transportable, analytical instrumentationcan assist in selecting which sites are of interest forfurther analysis using the more definitive, but slower,analytical techniques required for establishing compli-ance with the CWC.

Membrane-Introduction Mass Spectrometry (MIMS)is an emerging development in chemical analysis de-signed for speed, selectivity, sensitivity, and real-timemonitoring capabilities.1 The technique provides ananalyte separation method that can be performed in

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 20, NUMBER 3 (

real time with minimal sample preparation.2 WithMIMS, a liquid sample containing dissolved analyte(s)is drawn past a membrane material.3 The membraneselectively allows the diffusion of the analyte preferen-tially over the liquid matrix or other possible interferingcompounds.4 This selectivity enhances the sensitivityof the detection instrument because of the inherentseparation of the analyte from any background. TheMIMS sample introduction device is coupled to a massspectrometer for analyte detection and determination.

Mass spectrometry is recognized as the most defin-itive technique for the identification and quantifica-tion of chemical compounds in the laboratory. Untilrecently, the size, weight, and power demands of mass

1999) 381

J. S. MORGAN ET AL.

spectrometers, combined with ex-tensive sample preparation require-ments, made field measurements oftrace levels of materials difficult.APL is developing a small, field-por-table, time-of-flight mass spectrom-eter (Tiny TOF) with enhanced in-strument portability, flexible design,and broad detection capabilities.5

By integrating the simplicity andseparation capability of the MIMSsample introduction technique withthe portability and detection rangeof the Tiny TOF mass spectrometer,a powerful new technique will beavailable for on-site use. The tech-nique offers reduced sample prepara-tion requirements, excellent sensi-tivity, and compatibility withcurrent gas chromatography/massspectrometry databases needed forcompound identification.

MIMS: TECHNICAL BACKGROUNDMIMS has evolved as a successful analytical tech-

nique owing to its simplicity and sensitivity. In direct-injection MIMS (Fig. 1), water samples are injectedinto a silicone capillary membrane located near a massspectrometer source. An external heater elevates thetemperature of the water sample, which in turn heatsthe membrane. Typical membrane temperatures arebetween 50 and 80°C. At higher temperatures, more ofthe sample is passed through the membrane and intothe mass spectrometer source. However, more waterdiffuses through the membrane at high temperatures,which can increase the background pressure in the massspectrometer and thereby deteriorate performance andsensitivity.

Analyte crosses the membrane by a three-step pro-cess called pervaporation4 (Fig. 2), in which it

1. Adsorbs onto the surface of the membrane2. Diffuses through the bulk of the membrane3. Evaporates from the vacuum-side surface of the mem-

brane into the mass spectrometer source

The most often used membrane material is silicone,which tends to exclude polar molecules because of itshydrophobic nature, i.e., polar molecules are not sol-uble in silicone and therefore do not readily adsorb ontothe membrane surface. Also, higher molecular weightspecies tend to stick to the surface of the membrane anddo not evaporate as readily into the vacuum. Nonpolarcompounds with molecular weights under 200 amu,however, will pervaporate with high efficiency. Thus,

382

Figure 1. Direct-injection MIMS inlet with two-step analyte enrichment (closed circles =water molecules, open circles = analyte molecules). Shown are the required proximity tothe mass spectrometer source, the enrichment due to the membrane’s selectivity overwater, and a second enrichment, sometimes used, which draws excess water from theanalyte flow.

To rough pump

To rough pump

Sample in

Sample out

Secondarymembrane

Enrichedanalyte

Primarymembrane

Secondarymembrane

Excesswater

Massspectrometer

source

JO

Figure 2. The pervaporation process.

the silicone rubber membrane provides a useful selec-tivity that can greatly enhance the detection of com-pounds that cross the membrane. For volatile organiccompounds, for example, detection limits in thesub–part per billion range can be expected, as seen in

Adsorption

Diffusion

Evaporation

HNS HOPKINS APL TECHNICAL DIGEST, VOLUME 20, NUMBER 3 (1999)

Table 1. Of course, the sensitivity ofthe mass spectrometer also affectsthe detection limit. The data inTable 1 were obtained using anHP5973 quadrupole mass spec-trometer. Ion trap mass spectrome-ters, which can collect analyte overa period of time before spectralanalysis, can significantly extendthe detection limit.

The mass spectrometer acquiresa continuous set of spectra during aMIMS measurement interval. Forexample, it will scan every massfrom 40 to 300 amu every few sec-onds. Over time, an ion currentchromatogram is obtained for eachmass, which can then be extractedat the end of the experiment. Atypical data set is seen in Fig. 3.Since each compound has a uniquemass spectrum, one can readily ex-tract identification informationfrom the ion chromatograms by ex-amining the spectra integrated un-der each peak. Mixtures, as detailedin the following discussion, presenta more difficult problem.

CHEMICAL WEAPONSDETECTION FOR TREATYVERIFICATION

A current effort at APL and MIMS Technology, Inc.,of Palm Bay, Florida, focuses on the application ofMIMS for the detection of chemical weapons. Chem-ical weapons tend to be polar molecules with weightsof 100 to 200 amu. We used simulant molecules (e.g.,chemically related compounds, breakdown products,

Table 1. MIMS detection limits for volatile organiccompounds.

Molecular DetectionAnalyte weight limit (ppb)

Benzene 78 0.9

Diethyl ether 74 5.0Nitrobenzene 123 1.0o-Xylene 106 0.1Hexachlorobutadiene 258 0.5Chlorobenzene 112 0.1

JOHNS HOPKINS APL TECHNICAL DIGEST, VOL

DETECTION OF CHEMICAL AGENTS IN WATER BY MIMS

4,000,000

Rel

ativ

e ab

unda

nce

2,000,000

3,000,000

1,000,000

0

4,000,000

2,000,000

3,000,000

1,000,000

0

0 2 4 6 8 10 12 14 16 18 20Time (min)

60 70 80 90 100 110

9

61

71

74

85

87

89

100

(a)

(b)

Rel

ativ

e ab

unda

nce

Mass-to-charge ratio, m/z

Figure 3. Typical MIMS data showing (a) the total ion chromatogram, which indicates thatthe analyte was passing through the membrane and into the mass spectrometer sourceduring a 6-min interval, and (b) the mass spectrum of the analyte integrated over theinterval from 7.7 to 8.5 min. The spectrum matches 2-butoxyethyl acetate, a commonsolvent.

UME 20, NUMBER 3 (19

precursors of chemical agents) for testing purposes, aslisted in Table 2. These analytes were chosen for theirchemical similarity to common chemical warfareagents, in particular the nerve agents.

Nerve agents, which were first developed in Ger-many in the 1930s and 1940s, work by attacking thehuman nervous system. They are all organophosphoruscompounds and are closely related to insecticides,where the P(=O) is replaced by P(=S), and less reactivefunctional groups than fluoride or cyanide are used.These agents inhibit the ability of the enzymecholinesterase to prevent the buildup of excess acetyl-choline, which transmits nerve impulses.

The agents of German origin are identified by codesbeginning with the letter “G.” The three main G agentsare tabun, sarin, and soman, although GF is also ofnote. Nerve agents of later origin include those desig-nated “V,” which tend to be more chemically stable andare roughly 10 times more toxic. The most importantof these is VX. Nerve agents are further divided intovolatile and persistent chemical weapon agents, de-pending on their tendency to evaporate. Sarin is themost volatile (17,000 mg/m3 at 25°C) and VX the mostpersistent (10 mg/m3 at 25°C) of the major nerveagents. In general, exposure to G agents is by inhalationof their volatile vapors, and exposure to V agents is

99) 383

J. S. MORGAN ET AL.

3

Table 2. Chemical weapon simulant compounds used in MIMS testing.

Associated chemical Boiling MolecularCompound weapon agent(s) point (°C) weight (amu)

Ethyl methylphosphonate (EMPA) G agents 109a 124Pinacolyl methylphosphonate (PMPA) G agents 96–106 180Thiodiglycol (TDG) VX 164–166 1222-Chloroethylethyl sulfide (CES) Sulfur mustard 156–157 124Dimethyl methylphosphonate (DMMP) G agents 181 124Triethanolamine Precursor,

Schedule 3B (17) 190–193 149Methylphosphonic acid (MPA) Sarin, soman, VX 107b 96Bis(2-chloroethyl)amine hydrochloride Nitrogen mustard 212b 178aFlash point.bMelting point.

through skin contact. Thus, G agent–like chemicals aremore easily detected by a MIMS-based technique,whereas V agents are less so, owing to their lowervolatility.

Both MIMS Technology and APL conducted flow-injection MIMS studies using standard, 0.28-mm-thicksilicone capillary membranes. At high concentrationsand membrane temperatures, someof these chemical agents can bedetected in aqueous solutions. Forexample, Fig. 4 presents a MIMSprofile and mass spectrum of DMMPin a 200-ppm solution with a mem-brane temperature of 95°C; diagnos-tic ions for this compound at mass-to-charge (m/z) ratios of 79, 94, 109,and 124 are shown. These fragmentscorrespond to the molecular ion at anm/z of 124 and successive methyl-group or formaldehyde-group losses.At lower membrane temperatures,the compound is not observed, pre-sumably because of a low diffusionrate through the silicone membraneand the low rate of evaporation ofDMMP into the vacuum system ofthe mass spectrometer.

Silicone membranes do not readi-ly adsorb polar compounds such asthe organophosphorus agents, so it isdesirable to substitute other mem-brane materials for optimal perfor-mance. The silicone capillary mem-brane can be replaced with a flatmembrane geometry, which enablesthe examination of a wide range of

35000

30000

25000

20000

15000

10000

5000

0

500

400

300

200

100

0

0 1

600

Rel

ativ

e in

tens

ityR

elat

ive

inte

nsity

40 50

47

(a)

(b)

Figure 4. MIMS prodiagnostic ions (m/zmass spectrum that Technology.

84 JOH

alternative materials as well as thinner silicone mem-branes. Johnson et al.6 introduced liquid membranetechniques in which low vapor pressure liquids are heldin place on a supporting frit by surface tension. Bothalternative polymer and liquid membranes were used inthis study, each having hydrophilic and hydrophobiccharacteristics. Among the alternatives used were

2 3 4 5 6 7 8 9 10 11Time (min)

60 70 80 90 100 110 120 130 140 150 160 170 180 190

57

70

79

94

109

124

137

162

185

UnknownDMMP

Mass-to-charge ratio, m/z

file of DMMP showing (a) an ion chromatogram of the three main= 94, black; m/z = 79, green; m/z = 109, red) and (b) an extractedmatches library data from the National Institute of Standards and

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 20, NUMBER 3 (1999)

• GFT Pervap 1170: A compositemembrane consisting of a sili-cone rubber coating on amicroporous cellulose substrate

• GFT Pervap 1060: Same as 1170,except that silicone coating ishalf the thickness

• GFT Pervap 1000: A compositemembrane consisting of a propri-etary, hydrophobic coating on amicroporous substrate

• Versapor 200H: A microporousmembrane with hydrophobiccoating

• SSP-M213: A silicone polymertreated with proprietary, hydro-philic chemicals

• SSP-M100: A proprietary, hydro-philic membrane similar to Saranwrap in appearance

• Silastic sheet (Dow Corning): A sheet version ofsilicone rubber tubing used in standard flow-injec-tion MIMI instruments and serving as a benchmark

In initial trials, the more hydrophilic membranes didallow more analyte through to the mass spectrometersource, but the higher flux of water masked the signaland eliminated any signal improvements. For example,Fig. 5 shows a typical ion current chromatogram of

Table 3. Detectmembrane mater

Compound

EMPAPMPATDGCESDMMPTriethanolamineMPABis(2-chloroethy

hydrochloride

94122

64 7545

106

252235270219203187138

700

100

200

300

400

500

600

00 1 2 3 4 5 6 7 8 9 10 11 12

Time (min)

Rel

ativ

e in

tens

ityR

elat

ive

inte

nsity

100200300400500600700800900

040 60 80 100 120 140 160 180 200 220 240 260 280 300

281 299

Mass-to-charge ratio, m/z

(a)

(b)

Figure 5. Ion current chromatogram of m/z = 75 (blue curve) and 124 (red curve) fromCES (20 ppm) in water using a GFT Pervap 1060 membrane (a) and a MIMS massspectrum (b).

JOHNS HOPKINS APL TECHNICAL DIGEST, V

DETECTION OF CHEMICAL AGENTS IN WATER BY MIMS

ion of eight chemical weapon simulants using alternativeials.

Alternatives

GFT 1170 GFT 1060 SSP-M213 Silastic

N N N NY Y Y YN N N NY Y Y YY Y N NN N N NN N N N

l)amineY Y Y Y

OLUME 20, NUMBER 3 (19

CES in water (20-ppm solution, 70°C membrane tem-perature) through a GFT Pervap 1060 membrane. Sev-eral of the compounds were detected using the Silasticmembrane, GFT 1170, GFT 1060, and SSP-M100 asdetailed in Table 3, although it is clear that a broaderresponse is required. The other membrane materialsshowed poor response. Initial studies with liquid mem-branes have also been completed, but further analysisto determine the optimum membrane chemistry is

needed. In these experiments, liq-uids such as high vacuum pump oilwere supported on porous polymers,which presented a tortuous path tothe analyte, thus preventing effi-cient transport through the mem-brane. Alternative supports em-ploying small, straight pores arecurrently being tested.

After studies using a quadrupolemass spectrometer have deter-mined the optimal membrane sys-tem to use for detection of chemicalagents, the membrane inlet will beintegrated with the APL-developedelectron-impact Tiny TOF massspectrometer. The system, as shownin Fig. 6, will be based on theWiley-McLaren two-stage designfirst introduced in 1954. Here, mol-ecules in the source are ionizedby low-energy electrons (typically70 eV).

In other Tiny TOF designs de-veloped at APL, ionization occursby a process known as matrix-assisted laser desorption/ionization

99) 385

J. S. MORGAN ET AL.

Figure 6. Block diagram of an electron-impact Tiny TOF mass spectrometer with membrane inlet sampling (LAN = localarea network).

He gasflow

Externalheatingsource

Flowinjectionmodule

Out

In

HV pulser

HV power supply

HV pulser

Electrongate

Filamentsupply

–70-V biassupplyHigh-voltage (HV)

power supply

PC withLabviewcontrol

SRS DG535pulser

AB out CD out

Interface

HPInfiniumscope

Channel A output

LAN

Trigger out

Trigger in

Vacuum chamber

Microchannelplate areadetector

Turbopump

Backingpump

HV power supply

(MALDI). With this technique, a laser pulse desorbsmaterial from a solid surface that has been co-crystal-lized with a matrix chemical that ionizes at the wave-length of the laser pulse. The sample is ionized bycharge exchange with the matrix. Because the laserpulse occurs at a precise interval, MALDI is well suitedto TOF mass spectrometry.

With an electron-impact source, the task is moredifficult, since an electron pulse will extend in bothspace and time. Thus, molecules are ionized at differenttimes and at different locations within the source re-gion. This issue is addressed in two ways. First, theelectron pulse is limited in space by magnetic focusingand gating in time. Thus, a small bunch of electronswill be introduced into the source in a time interval assmall as 30 ns. Second, the two-stage Wiley-McLarensource corrects for spatial spreads in ions. In the first(extraction) stage, a small electric field is applied, sothat ions originating in different locations in the sourceare not accelerated to widely different energies. In thesecond (acceleration) stage, a much stronger electricfield accelerates the ions into the mass spectrometer

386 JOH

flight tube. With modern electronics providing fast risetimes for the voltage pulses that drive the source elec-tric fields, the flight tube can be reduced to 13 cm, thuspermitting the construction of a highly mobile, versa-tile MIMS system.

FUTURE OF MEMBRANE-INTRODUCTION TECHNOLOGY

The examination of alternative membrane materialspotentially will allow for the detection of other com-pounds of interest, including those from biologicalspecimens. Several parallel developments must be pur-sued to achieve the performance required to extend thecapability of MIMS.

First, thermal desorption of analytes from the vac-uum side of membranes should permit much greatersensitivity for detection of high molecular weight, highboiling point compounds. Many of these compoundsdiffuse through silicone membranes, but do not vola-tilize in vacuo owing to their low vapor pressure attypical membrane temperatures. Several methods of

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 20, NUMBER 3 (1999)

DETECTION OF CHEMICAL AGENTS IN WATER BY MIMS

thermal desorption can be applied, including the use ofa continuous infrared source or laser desorption.7

Also, complex mixtures are difficult to completelyanalyze using MIMS. In gas chromatography/mass spec-trometry, mixtures are separated by the gas chromato-graph before introduction into the mass spectrometer.In comparison, analytes within mixtures will pervapo-rate through a membrane with very minimal separationin time, producing overlapping ion current peaks. Tosome extent, mixtures can be analyzed by comparingthe mass spectra of expected components, but this doesnot solve the general problem. However, each analyteshould diffuse through a membrane at a slightly differ-ent rate, so that the ion current chromatogram for eachcomponent in a mixture will reach a maximum atslightly different times. This does not provide thequality of separation found in gas chromatography,since the ion current profiles for different componentswill usually overlap significantly. Differing diffusionrates permit extraction of the components, but only bycareful examination of the time at which each ionchromatogram in a data set reaches its maximum. Bycollecting the ion chromatograms that peak at the sametime, it is possible to reconstruct the original massspectrum of a particular chemical in a mixture. Usingthis and other techniques, we are developing algo-rithms to separate complex mixtures in MIMS data sets.

Another priority in this area is the detection of awide range of polar chemical agents using novel mem-brane materials and inlet systems (e.g., trap-and-releaseMIMS, liquid membranes, jet separator inlets). Sili-cone membranes are inherently more sensitive to non-polar analytes, which adsorb onto the surface of siliconemore readily than polar chemicals. Therefore, past workin MIMS almost exclusively has involved the analysisof volatile organic compounds in various matrices,

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 20, NUMBER 3 (1

usually water. Now, semivolatile analytes, includingbiological compounds, can be examined with muchgreater sensitivity than previously possible.8 For exam-ple, liquid membranes can be tailored to specific appli-cations by varying the membrane chemistry. Any vis-cous, liquid solution can be used as a liquid membrane.Thus a solution designed to preferentially dissolve or-ganophosphorus compounds could be used to efficientlydetect the nerve agents. Since liquid membranes willallow rapid diffusion of higher molecular weight com-pounds, applications in biomedicine and process mon-itoring are also gaining interest.9

REFERENCES1Bauer, S., “Membrane Introduction Mass Spectrometry (MIMS): An OldMethod That Is Gaining New Interest Through Recent TechnologicalAdvances,” Trends Anal. Chem. 14(5), 202–211 (1997).

2Creaser, C. S., Stygall, J. W., and Weston, D. J., “Developments in MembraneInlet Mass Spectrometry,” Anal. Commun. 35, 9H–11H (Jun 1998).

3Lauritsen, F. R., and Kotiaho, T., “Advances in Membrane Inlet MassSpectrometry,” Rev. Anal. Chem. 15(4) 237–264 (1996).

4Srinivasan, N., Johnson, R. C., Kasthurikrishnan, N., Wong, P., and Cooks,R. G., “Membrane Introduction Mass Spectrometry,” Anal. Chim. Acta 350,257–271 (1997).

5Bryden, W. A., Benson, R. C., Ecelberger, S. A., Phillips, T. E., Cotter, R. J.,and Fenselau, C., “The Tiny-TOF Mass Spectrometer for Chemical andBiological Sensing,” Johns Hopkins APL Tech. Dig. 16(3), 296–310 (1995).

6Johnson, R. C., Koch, K., Kasthurikrishman, I. N., Plass, W., Patrick, J. S.,and Cooks, R. G., “An Evaluation of Low Vapor Pressure Liquids forMembrane Introduction Mass Spectrometry,” J. Mass Spectrom. 32, 1299–1304 (1997).

7Soni, M. H., Callahan, J. H., and McElvaney, S. W., “Laser Desorption-Membrane Introduction Mass Spectrometry,” Anal. Chem. 70, 3103–3113(1998).

8Beck, H. C., Lauritsen, F. R., Patrick, J. S., and Cooks, R. G., “Metabolismof Halogenated Compounds in the White Rot Fungus Bjerkandera adustaStudied by Membrane Inlet Mass Spectrometry and Tandem Mass Spectrom-etry,” Biotechnol. Bioeng. 51, 23–32 (1996).

9Lauritsen, F. R., and Gylling, S., “On-Line Monitoring of BiologicalReactions at Low Parts-per-Trillion Levels by Membrane Inlet Mass Spec-trometry,” Anal. Chem. 67, 1418–1420 (1995).

ACKNOWLEDGMENT: The support of the Defense Threat Reduction Agencyand DARPA under contract MDA972-96-D-0002 is gratefully acknowledged.

THE AUTHORS

JOHN S. MORGAN is a senior engineer in the APL Research and TechnologyDevelopment Center. He received a B.S. in physics from Loyola College inMaryland in 1984 and an M.S.E. and Ph.D. in materials science and engineeringfrom JHU in 1988 and 1990, respectively. His research included high-temperature superconductivity and high-bandgap optoelectronic semiconductorsbefore he joined the APL Materials Laboratory in 1991. While there, heworked on spacecraft contamination monitoring, nondestructive evaluation,and thin-film batteries. Dr. Morgan’s current interests include technology de-veloped to improve the detection of agents of mass destruction, includingexplosives and chemical and biological warfare agents. His e-mail address isjohn.morgan@jhuapl.edu.

999) 387

J. S. MORGAN ET AL.

WAYNE A. BRYDEN is a Principal Professional Staff chemist in the APLResearch and Technology Development Center. He obtained a B.S. degree inchemistry from Frostburg State University in 1977, and M.S. and Ph.D. degreesin physical chemistry from JHU in 1982 and 1983, respectively. In 1983, hejoined APL as a Senior Staff chemist and was promoted to the PrincipalProfessional Staff in 1993. Dr. Bryden is a member of the American ChemicalSociety, American Physical Society, American Vacuum Society, MaterialsResearch Society, and Sigma Xi. He is listed in American Men and Women ofScience and is the author of over 60 scientific publications. His current researchinterests include materials physics, mass spectrometry, magnetic resonance,miniaturized sensor technology, and chemical and biological detection. His e-mail address is wayne.bryden@jhuapl.edu.

REBECCA F. VERTES is a chemist in the Sensor Science Group of the APLResearch and Technology Development Center. She received her B.S. degree inchemistry from Emory University and an M.S. degree in analytical chemistryfrom Virginia Commonwealth University. Currently, Ms. Vertes is involved inthe use of mass spectrometry in counterproliferation efforts. Her e-mail addressis rebecca.vertes@jhuapl.edu.

SCOTT BAUER, President of MIMS Technology, Inc., received his B.S. inchemistry with a minor in physics from the University of Central Florida.Following a 1-year experience managing an environmental organic analysislaboratory, Mr. Bauer went on to earn his M.S. in chemistry from PurdueUniversity, where he specialized in the design and implementation of membraneinlet systems for mass spectrometers. Mr. Bauer designed and patented a two-stage MIMS device which extends the utility of membrane inlets to includemicroporous as well as pervaporation membrane use. He founded MIMSTechnology in 1993 to develop commercial membrane inlet systems for a widevariety of mass spectrometers. The company has become a world leader in thesale of high-quality MIMS devices designed for the environmental water qualitymarket. His e-mail address is mims@digital.net.

388 JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 20, NUMBER 3 (1999)