IC ................................... ...

Reaction Product Imaging:The H + D2 Reaction

Theofanis N. Kitsopoulos, Mark A. Buntine, David P. Baldwin, Richard N. Zare, David W. Chandler*

The differential cross section for the H + D2 -> HD + D reaction has been measured usinga technique called reaction product imaging. In this experiment, a photolytically producedbeam of hydrogen (H) atoms crossed a beam of cold deuterium (D2) molecules. ProductD atoms were ionized at the intersection of the two particle beams and accelerated towarda position-sensitive detector. The ion images appearing on the detector are two-dimen-sional projections of the three-dimensional velocity distribution of the D atom products. Thereaction was studied at nominal center-of-mass collision energies of 0.54 and 1.29 electronvolts. At the lower collision energy, the measured differential cross section for D atomproduction, summed over all final states of the HD(vJ) product, is in good agreement withrecent quasi-classical trajectory calculations. At the higher collision energy, the agreementbetween the theoretical predictions and experimental results is less favorable.

Reaction dynamics can be said to haveemerged as a separate branch of chemistryin 1929 with the publication by London (1)of the quantum mechanical expression forthe potential energy of the hypotheticalreaction A + B-C -- A-B + C. Two yearslater, Eyring and Polanyi (2) published thefirst attempt to map out the potential ener-gy surface for the hydrogen exchange reac-tion, H + H2. Since then, a tremendousamount of theoretical and experimentaleffort has been directed toward determiningthe detailed nature of bimolecular interac-tions.

The study of bimolecular reactions un-der single-collision conditions has dramati-cally improved our understanding of chem-ical reactivity. Measurements of the na-scent product internal state distributions(integral cross sections) and the productvelocity distributions (differential cross sec-tions) have led to a fundamental under-standing of how chemical reactions pro-ceed. Information concerning the size andposition of energetic barriers, the geometryand lifetime of the reaction complex, andproduct branching ratios and energy parti-tioning can now be obtained from thesemeasurements. This information is neededfor adequately testing the theoretical poten-tial energy surfaces for chemical reactionsand the scattering calculations carried outwith them.

In our laboratory, we have implementedan ion-imaging technique for studying theT. N. Kitsopoulos and D. W. Chandler are in theCombustion Research Facility, Sandia National Labo-ratories, Livermore, CA 94551. M. A. Buntine is in theDepartment of Chemistry, Yale University, New Haven,CT 06511. D. P. Baldwin is in the Ames Laboratory,Iowa State University, Ames, IA 50011. R. N. Zare is inthe Department of Chemistry, Stanford University,Stanford, CA 94305.*To whom correspondence should be addressed.

differential cross section of a neutral bimo-lecular reaction. Beginning in 1987 (3),ion-imaging techniques have been used forthe measurement of photofragment velocitydistributions (speed and angle) from unimo-lecular dissociation processes. Those exper-

iments demonstrated the power of the tech-nique in several ways. First, if the experi-ment is designed such that the symmetryaxis of the velocity distribution is orientedparallel to the face of the imaging detector,one image is all that is required to defineuniquely the three-dimensional (3-D) an-

gular distribution. Second, the multiplex-ing advantage of measuring all angles atonce reduces the time necessary to deter-mine a velocity distribution. Third, thetechnique offers two modes of operation.One mode involves imaging of the atomicfragment or reaction product, which resultsin measurements that are characterized bymoderate energy resolution, comparable tothat of conventional time-of-flight (TOF)experiments, in which the internal energydistribution of the products is determinedby measuring their kinetic energies. Ionimages of atomic products are extremelyuseful in providing the overall appearanceof the differential cross section for a reac-

tion or photofragmentation process becausethey contain information concerning allproduct channels. The alternative moderelies on imaging of the molecular fragmentor reaction product in a quantum state-selective manner, thus enabling differentialcross section measurements for a singlerotational-vibrational (rovibronic) state ofthe molecular product. We report herepreliminary measurements of the differen-tial cross section (DCS) for the H + D2 --

HD + D reaction in which the D atomproduct is imaged. The reaction is studiedat relative collision energies of 0.54 and

SCIENCE * VOL. 260 * 11 JUNE 1993

1.29 eV, and we compare our results totheoretical quasi-classical trajectory (QCT)calculations of Aoiz et al. (4), which arebased on the LSTH (Liu-Siegbahn-Truhlar-Horowitz) (5) potential energy surface.

The H + H2 exchange reaction (and itsisotopic variants) is the simplest bimolecu-lar chemical reaction, yet experimentalcharacterization comparable to the level ofthe calculations has been possible only inthe last 7 years. Two types of experimentalefforts can be identified: those that deter-mine the integral cross sections for thereaction and those that measure DCSs.Almost simultaneously, rotationally re-solved state-to-state integral cross sectionswere measured by Zare and co-workers (6),using resonance-enhanced multiphotonionization (REMPI), and by Valentini andco-workers (7), using coherent anti-StokesRaman spectroscopy. The reactions studiedwere H+ para-H2, H + D2, and D + H2 atenergies between 0.5 and 2.2 eV. Overall,the results of these experiments agree fairlywell with theoretical predictions (8-12).

The DCS measurements that have beenmade were possible only through the use ofcrossed-beam TOF experiments (13). Atypical experiment involves TOF analysis ofthe products emitted at some angle withrespect to a laboratory fixed direction (oftenthis direction is chosen to be the directionof the reactant atom beam). The detectorposition can be rotated so that data can becollected at several angles. At each angle,only a small portion (typically 10-4) of theproducts is detected, which results in longhours of signal averaging. Subsequently,normalization of the data to a fixed angleand transformation of the data from thelaboratory frame to the center-of-massframe is necessary to obtain the DCS. Inspite of the experimental challenges, state-resolved DCS measurements for D + H2have been reported by Buntin et al. (14) fora few angles at collision energies of 0.8 and1.20 eV and by Continetti et al. (15) for awide distribution of angles at collision en-ergies of 0.53 and 1.01 eV. In both exper-iments TOF analysis of the HD product wassufficient to resolve individual vibrationalbut not individual HD rotational states. Inan alternative approach, Schnieder et al.(16) have used Rydberg atom TOF transla-tional spectroscopy to measure the vibra-tionally resolved DCS for the H + D2reaction at collision energies of 0.54 and

1605

on

Feb

ruar

y 9,

201

0 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

1.29 eV. More recently, by reducing theenergy spread in their D2 beam, Schniederand Welge (17) have reported the firstrotationally resolved DCS measurement forthis fundamental reaction.

As an evolutionary step toward the pre-sent reaction product imaging (RPI) experi-ments, Buntine et al. (18) studied the H +HI-H2(v= 1,J = 11,13, where v is thevibrational quantum number and J is therotational quantum number) + I(2p3/2,1/2)reaction at collision energies of 1.7 and 2.6eV with the ion-imaging technique. In thisexperiment, carried out in a single molecularbeam, it was possible to measure branchingratios for the various reaction channels thatwere observed; however, measurement ofthe DCS was prohibited because of the lackof a defined initial velocity for the reactants.More recently, Houston and co-workers (19)used a crossed-molecular beam ion-imagingtechnique to successfully measure the state-resolved DCS for the inelastic scattering ofNO from Ar. The approach was similar inpractice to the experiment presented hereand arose from collaborative work usingion-imaging techniques to study unimolecu-lar dissociation processes (3).

Experimental Apparatus

A schematic of the experimental apparatusused in this study is shown in Fig. 1. Theapparatus is a modified version of the single-beam, ion-imaging photofragment spec-trometer described in detail elsewhere (20).In brief, a photolytically produced beam ofH atoms is crossed with a supersonic mo-lecular beam of D2 molecules. Product Datoms are ionized and accelerated toward aposition-sensitive ion detector consisting ofmicrochannel plates coupled to a phosphorscreen. An image is created on the phos-phor, which is the 2-D projection of the3-D distribution of the D atom products.The image is recorded and averaged with acharged-coupled-device (CCD) camera.A 30% mixture of HI/He (backing pres-

sure, 20 psi) and pure D2 (backing pressure,55 psi) is expanded supersonically into thesource vacuum chamber through a pair ofparallel solenoid valves (General Valve se-ries 9). Both gas expansions are skimmed1.0 cm from the nozzle orifices and colli-mated to diameters of 3 mm by means of apair of collimating holes in the repellerplate. In this manner a pair of parallelmolecular beams are produced in the reac-tion chamber, 2 cm apart, at a distance of 7cm from the nozzle orifices. The beam of Hatoms is produced by photolyzing HI mid-way through the extraction field (see Fig. 1)using the fourth harmonic (266 nm) of aNd:yttrium-aluminum-garnet (YAG) laseroperating at 30 Hz (Spectra Physics GCR-3, 8 to 10 mJ per pulse). To match the

1606

Fig. 1. Schematic of thereaction product imag-ing apparatus. A pho-tolytically producedbeam of H atoms iscrossed with a super-sonic molecular beamof D2 molecules. Prod-uct D atoms are ionizedand accelerated to-ward a position-sensi-tive ion detector con- Position-sensitivesisting of microchannel detector_plates coupled to a rCCD cameraphosphor screen. Ion I -positions appearing onthe phosphor screen \ ll.-are recorded with aCCD camera.

diameter of the molecular beam, the laserbeam is focused to 3 mm with a lens thathas a focal length of 1.0 m. We find thatthese conditions produce the maximum fluxof H atoms at the D2 beam.

The multiphoton ionization of the prod-uct D atoms is accomplished by absorptionof a vacuum ultraviolet (VUV, 121.6 nm,Lyman-a wavelength) and UV (364.8 nm)photon. The VUV photon resonantly ex-cites the D(ls) atom to the 2p state. Sub-sequent absorption of the UV photon eitherionizes the D atom directly or excites it to avery high Rydberg level, at which it thenfield ionizes because of the presence of theextraction field (15). The UV light is gen-erated by frequency doubling of the outputof a Nd:YAG pumped dye laser system(Spectra Physics GCR-5, and PDL-1, re-spectively). The VUV Lyman-oa light isgenerated by frequency tripling of the UVlight in a Kr/Ar gas mixture (total pressure550 torr) adjusted to achieve optimumphase matching (21). A lens with a focallength of 20 cm is used to focus the 364.8-nm light in a cell containing the Kr/Armixture. The VUV and the residual UVlaser light pass through an LiF lens and ontothe point where the H atom beam crossesthe D2 molecular beam. The spot size of theVUV laser light at the position of the D2molecular beam is about 3 mm. Matchingthe diameter size of the VUV ionizationlaser beam to the diameter size of the D2beam (3 mm) and the width of the H atombeam (3 mm) ensures that all velocitycomponents of the D atom products inter-act with the ionization laser for the sameamount of time (5 ns) and thus are ionizedwith equal efficiency. Both the frequencyand the intensity of the generated Lyman-aphotons are optimized by monitoring theresonant ionization yield of background Dor H atoms produced by the thermal de-composition of D2 or H2 on the hot fila-

- Nd:YAG pumped tunable dye laser

ment of an ionization gauge. It is impera-tive that during data acquisition all iongauges are turned off; otherwise, the ther-mal D atoms produced in this fashion willdominate the observed image.

The propagation direction of the ioniza-tion laser beam is perpendicular to the planeformed by the two crossed beams. Conse-quently, the ls-2p transition of the D atomsscattering out of this plane experiences aDoppler shift (22), that is, a shift in theresonant frequency needed to excite and sub-sequently ionize them (a blue shift for thoseproceeding away from the laser propagationdirection and a red shift for those proceedingtoward the laser propagation direction). Forthe collision energies studied in this experi-ment, we estimate the maximum Dopplershift to be approximately 3 to 4 cm-'. Thebandwidth ofour VU laser light is about 1.5cm-1; therefore, to ensure that all velocitycomponents are equally detected, we scan thefrequency of the VV laser over the transi-tion as the image is recorded. A typical scancovers a 15-cm-1 range.An important parameter in this study is

the population of rotational states of the D2molecule in the molecular beam. To deter-mine the rotational populations, we use 2+ 1 REMPI (23) of D2(v" = 0, J") throughthe v' = 0 level of the E,F state, andmeasure the D2+ ion yield for each D2(v" =0, J") state. By comparing the relative D2+signal levels from the different rovibronictransitions with those obtained by the sametechnique on room-temperature D2, we de-termined that the relative rotational popu-lations in the D2 are 1.0, 0.5, and 0.5 forJ"= 0, 1, and 2, respectively.

Results of the RPI Measurements

Two product channels are open when HI isphotodissociated with 266-nm photons;these correspond to formation of iodine in

SCIENCE * VOL. 260 * 11 JUNE 1993

`------` -111.111.11 1.1 -1111 1.111 11,111"""""""""""""""""-,""""""""," "" "" '"' 'I"'... -- 1...........I.'-.-.-,-"-.'......- .. ... -.'..11......I.............----------------------.,...-......,..,..................................l....- .. ..11.1.1. ... 1.1.1 I.'...'.. -,

on

Feb

ruar

y 9,

201

0 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Fig. 2. Data images for the product D atomsproduced from the reaction H + D2 -* HD + D

at nominal center-of-mass collision energies of(A) 0.54 eV and (B) 1.29 eV. The image intensi-ties are scaled between 0 and 255, and theintensity scale of the image color map is indicat-ed in the figure. The open circles represent thecalculated position of scattered D atoms corre-sponding to HD(v = 0, J = 0). The direction ofthe H atom beam is indicated by the arrow, andthe positions of the D2 beam and the center ofmass of the reaction are indicated by the solidcircles labeled D2and CM, respectively.

the 2P312 ground state and in the 2P1/2excited spin-orbit state:

HI + 266 nm -- H + I(2P3/2),

K.E.Hlab = 1.59 eV (fast channel)HI + 266 nm -- H + I(2P1/2),

and

K.E.Hlab = 0.66 eV (slow channel)where K.E.Hlab is the kinetic energy of thehydrogen in the laboratory frame of refer-ence. In the center-of-mass frame, the col-lision energies for reaction of H atoms withD2 that correspond to these two channels are

ECM = 1.29 and 0.54 eV, respectively. As

VHI

Fig. 3. Newton diagram for the reaction H + D2-* HD(v = 0, 1, 2; J = 0) + D at nominal

center-of-mass collision energy of 1.29 eV. Thevectors VHI and VD2 represent the HI and D2beam velocities, respectively (vHI VD2); VH is

the H atom velocity obtained from the HI dis-sociation; vHIab is the velocity of the H atoms inthe laboratory frame (vHIab = VH + VHI); 9 is therelative velocity of the collision (vH g); UD is

the D atom velocity in the center-of-mass frame;CM represents the position of the center ofmass of the system, and the circles indicate thecalculated D atom Newton circles correspond-ing to formation of HD(v= 0, 1, 2; J = 0)product states.

reported elsewhere (24), the angular distri-bution of the H atoms for the two channelsis strongly coupled to the polarization of thedissociation laser. The transition associatedwith dissociation into the slow channel is a

parallel one and leads to a cos2O distribution(0 is the scattering angle) of H atoms (Hatoms preferentially moving along the laserpolarization axis with a node in the distribu-tion perpendicular to the axis). For the fastchannel, H atoms are produced by excita-tion through a perpendicular transition lead-ing to a sin20 distribution (H atoms prefer-entially traveling perpendicular to the laserpolarization axis with a node in the distribu-tion along the laser polarization axis). Thisseparation of the two H atom channelsallows us to choose cleanly the velocity ofthe reactant H atoms by simply choosing thepolarization direction of the linearly polar-ized light (266 nm) that photolyzes the HImolecular beam.

Product Images

Reaction product images of the D atomsproduced from the H + D2 reaction, atcollision energies of 0.54 and 1.29 eV, are

SCIENCE * VOL. 260 * 11 JUNE 1993

shown in Fig. 2, A and B, respectively. Theorientation of both images is such that theH atom beam lies in the plane of the imageand is directed from the top of the imagetoward the bottom; the D2 beam is directedout of the plane toward the viewer. Bothimages are obtained by averaging 180,000laser shots and subtracting a backgroundimage. Background images are obtained byadjusting the time delay of the HI photol-ysis laser such that the H atoms and the D2beam do not intersect during the VUV laserpulse. Typical D' count rates for this ex-periment are 0.3 to 1.0 ion per laser shot.Superimposed on the data images of Fig. 2are circles representing the maximum speedthat the D atoms can attain from thereaction under the assumption that the HDproduct is formed in its ground vibrationaland rotational state (v = 0 and J = 0).

The velocity vector (Newton) diagramrelevant to our experiment is shown in Fig.3. Newton diagrams are used to transformproduct velocities measured in the labora-tory frame into center-of-mass velocities.For bimolecular scattering, the angular dis-tribution of the products is cylindricallysymmetric with respect to the relative ve-locity (g) of the reactants. Consequently,an inverse Abel transform (20, 25) can beused to reconstruct the 3-D velocity distri-bution when g is oriented parallel to theimage plane (the face of the detector). Inour experimental setup we make the HIvelocity comparable to that of the D2 beam(VHI vD2) by seeding the HI beam in He.Hence, in the laboratory reference frame, g

VH (see Fig. 3), where VH is the H atomvelocity obtained from the HI dissociation.The direction of vH is defined by the cross-ing point of the 266-nm photolysis laserbeam with the HI beam and the crossingpoint of the VUV ionization laser beamwith the D2 beam. This direction is main-tained parallel to our position-sensitive de-tector (that is, the imaging plane) duringthe experiments.

In the data images of Fig. 2, structuraldetails representing individual HD rovibra-tional states are not observed for severalreasons. First, our ion images represent bytheir nature a 2-D projection of a 3-Dvelocity distribution. In three dimensions,scattered D atoms corresponding to a spe-cific HD(vJ) lie on a sphere (known as ascattering surface, or a Newton sphere),whose radius is proportional to their recoilspeed. Consequently, for an ensemble ofHD final states, each having its own scat-tering surface, the D atom velocity distri-bution will be represented by a series ofconcentric spheres. If the product angulardistribution on each scattering surface issimilar, then the projections of thesespheres onto the image plane would overlap(that is, the spheres would be crushed on

1607

on

Feb

ruar

y 9,

201

0 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

top of each other). Thus, the intensity ofeach point in the data image would ingeneral have contributions from several HDproduct states; as a result, observable struc-ture would be expected to be diminished.A second reason why individual rings

are not observed in the data images isbecause we have "geometric smearing,"which is a direct consequence of the finitedimensions of the reaction volume. Thesize of our reaction volume [about (0.3cm)3], defined by the overlap of the H atombeam, the D2beam, and the ionization laserbeam, is sufficiently large to cause the New-ton sphere of the D atoms associated withneighboring HD quantum states to overlap.In the extreme limit, where the particlebeams and laser beam would be sufficientlysmall, such that the products emanatedfrom a point-like source, individual quan-tum states of the HD product would beobserved. Geometric smearing in RPI is notinsurmountable and can be "remedied" byproper data analysis in situations where thenature of the smearing is well defined (im-age restoration) (26, 27). Because the exactnature of the blurring in the present exper-iment has not been independently mea-sured (that is, we have not measured aknown distribution and therefore experi-mentally determined our blurring), we can-not deconvolve the data for the particlebeam sizes and the laser ionization volume.

Reconstructed Product Images

The 2-D intensity profile through the 3-Dvelocity distributions for the D atoms, ob-tained from the data through an inverseAbel transform (20, 27), is shown in Fig. 4,A and B, for the two collision energies of0.54 and 1.29 eV, respectively. We refer tothese as the reconstructed images; theseimages should not be confused with the dataimages of Fig. 2, which are 2-D projectionsof the 3-D velocity distribution. The 3-Dvelocity distribution of the D atoms can begenerated by rotating the reconstructed im-ages about their symmetry axis, in this casethe direction of the relative velocity g. Todraw the analogy to the more conventionalcrossed-beam TOF scattering experiments,these images are equivalent to velocity con-tour plots of the scattering process. The"noisiness" down the centers of both imagesin Fig. 4 is introduced by the reconstructionalgorithm. To reduce some of the noisecaused by the inverse Abel transformation,we "smoothed" the data images beforehandby fitting a Gaussian (5 pixels in half-width)to every 9 sequential image pixels. Theextent of the "smoothing" Gaussian is small-er than the resolution of the present exper-iment (about 17 pixels), thus ensuring thatany structure present in the data has notbeen lost by this smoothing procedure. The

1608

data images shown in Fig. 2 have not beensmoothed. Overlaid on the images are sev-eral Newton circles, indicating the calculat-ed positions of scattered D atoms, corre-sponding to HD(vJ) product states.

Although we are unable to resolve indi-vidual quantum states of the HD product inthe reconstructed images of the D atoms,both qualitative and quantitative informa-tion pertaining to the reaction can be ex-tracted from the images. On a qualitativelevel, inspection of Fig. 4A (ECM = 0.54eV) reveals that the D atoms are stronglypeaked along the beam direction of the H

Fig. 4. Reconstructed images for the product Datoms produced from the reaction H + D2 ->HD + D at nominal center-of-mass collisionenergies of (A) 0.54 eV and (B) 1.29 eV. Theimage intensity is scaled between 0 and 255,and the image color map is identical to thatused in Fig. 2. The open circles represent thecalculated positions of scattered D atoms cor-responding to HD(v = 0, J = 0) and HD(v = 0,

J = 4) in (A) and HD(v= 0, J = 0), HD(v = 1,J = 0) and HD(v = 2, J= 0) in (B). The positionof the center of mass for the reaction is indicat-ed by the solid circle labeled CM, and thedirection of the relative velocity by the vector g.

SCIENCE * VOL. 260 * 11 JUNE 1993

atoms and that their distribution is relative-ly narrow [there is not much intensity atvelocities that would correspond to HD(v= 1) being formed]. This observation sup-ports the theoretical predictions (4, 8-11)that at low collision energies HD will beformed predominantly in v = 0 and in lowrotational states and product D atoms willbe forward scattered with respect to thedirection of the H atom beam velocity.

Inspection of the reconstructed ion im-age of Fig. 4B (ECM = 1.29 eV) revealsinformation not readily observed in thecorresponding data image of Fig. 2B. Withthe visual aid offered by the overlaid New-ton circles for selected HD quantum states(v = 0,J = 0), (v = 1,J = 0), (v = 2,J =0), it is apparent that at wider scatteringangles formation of internally excited HD isthe dominant channel of the reaction. Al-though very wide angle scattering is ob-served, there is no evidence of direct Datom backscattering (D atoms movingagainst the direction of the incoming Hatoms), which would imply the formationof a long-lived complex near the transitionstate (28).

Accurate ab initio potential energy sur-faces are available for the hydrogen ex-change reaction (5, 29, 30). AlthoughDCSs for H + H2(HD) and D + H2 havebeen calculated with the use of fully con-verged 3-D quantum scattering calculations(12), researchers who have carried out sim-ilar calculations on the H + D2 reactionhave reported only on the integral crosssections (11). Recently, however, Aoiz etal. (4) have simulated the state-to-stateDCSs for the latter reaction with a QCTcalculation based on the LSTH (5) poten-tial energy surface. At the collision energiesstudied, 0.54 and 1.29 eV, and for theD2(" == 0," 0,1,2) initially populated,the calculation predicts that the D atomproduct is predominantly forward scattered,which is in qualitative agreement with ourexperimental observations.

Comparison with QCT Calculations

To compare quantitatively our experimen-tal angular distributions to those predictedby the QCT simulation, we have plotted inFig. 5, A and B, the total D atom yield fromthe reaction as a function of scatteringangle 0, where 0 is defined with respect tothe direction of the relative velocity vectorg (Fig. 4A). We have integrated the inten-sity of our reconstructed images along radialslices, centered at the center of mass of thereaction. Each radial slice is a 40 sectorwhose radius is equal to the radius of thelargest Newton circle [that is, the onecorresponding to formation of HD(v = J =0)1. For ECM = 0.54 eV (Fig. 5A), theagreement between theory and experiment

on

Feb

ruar

y 9,

201

0 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

---" -. "" '.' ` "' -- "" ` "' " "' "' ` ` "" 'I "" """,'. `. "I"."-"..".'.'?,' .1".- ."I.",- I..' ..` 11.1 I'.,....'

.......

is very good for 0 > 100, but the theoryoverestimates the intensity relative to theexperiment at small scattering angles. Thisdifference could be caused by the fact thatthe inverse Abel transformation introducessome intensity errors in the reconstructedimage near the symmetry axis (27). Thismay be a consequence of the nature of thereconstruction process, which works fromthe outside of the image toward the center.Propagation of the error during this processmeans that the relative error is largest at thesymmetry axis.

For the higher collision energy, 1.29 eV,it is clear from Fig. 5B that the agreementbetween the QCT calculation and the ex-periment is not as good as for the ECM =0.54 eV case. Qualitatively, a slight inflec-tion in the DCS slope at about 450 isobserved in both the theoretical and exper-imental results. Quantitatively, at smallscattering angles the observed DCS is largerthen the QCT prediction, but at higherscattering angles this trend is reversed. Onepossible experimental explanation for thisdeviation could be that our data are notnormalized to the power of the VU ion-ization laser. The VUV generation is opti-mized by monitoring D' yield from thermalD atoms, which corresponds to optimizingthe center of the Doppler profile for the Datom products. Because the optimum Kr/Ar

0 (dog)

0 (dog)Fig. 5. The DCS for D atom production from thereaction H + D2 -> HD + D at nominal center-of-mass collision energies of (A) 0.54 eV and(B) 1.29 eV, summed over all final states of theHD(v,J) product; 0 is the center-of-mass scat-tering angle; (K) this experiment; (solid line) theresults of the QCT calculation of Aoiz et al. (4).

pressures for maximum VUV generationvary with wavelength, it is possible that theionization laser power is not constant overthe entire product D atom Doppler profilethat is scanned to obtain complete ionimages. To study this effect, we addedacetone to the D2 beam and monitored the[acetoneJ' yield from the absorption of theVUV laser light (one-photon ionization)and observed only a 10% variation over thespectral range of the D atom Doppler pro-file. This intensity variation is too small tomake up for the observed discrepancy be-tween the QCT predictions and the exper-imental results.We can also compare our results to the

DCS measurements obtained by crossed-beam TOF experiments for the hydrogenexchange reactions. Schnieder et al. (16)have measured the angular distribution ofDatoms from the H + D2 -- HD + Dreaction. They found that at a collisionenergy of 0.54 eV the reaction producesHD(v = 0, low J) and is backscattered withrespect to the H atom beam direction. At acollision energy of 1.29 eV, they observedthat the scattering angle for the D atomsincreases with increasing vibrational androtational excitation of the HD product.Although these observations are qualita-tively similar to our results, because thecenter-of-mass DCS was not reported in thestudy of Schnieder et al., quantitative com-parison between the two studies is notpossible. The most detailed experimentalmeasurement concerning the DCS for thehydrogen exchange reaction has been madeby Continetti et al. (15) for the D + H2 --

HD + H. Their results show that, at acollision energy of 0.51 eV, the reactionproduces HD(v = 0) in low rotational statesand is scattered backward (1800) with re-spect to the D atom beam velocity, where-as, at the higher collision energy (0.98 eV),the DCS for the HD product peaks at asmaller angle (1250). The results comparedwell with the quantum mechanical trajec-tory calculations (12) based on the LSTHsurface. It is interesting that the experimen-tal DCS of Continetti et al. (15) alsoshowed an increased amount of HD back-scattering when compared to the theory, asdo our results.

Conclusions

We have applied an ion-imaging tech-nique, RPI, to the study of bimolecularreactive scattering, which we believe willbe a powerful method for studying reactiondynamics. RPI offers several advantagesover traditional methods: (i) The 4+r parti-cle collection efficiency leads to count ratesthat are capable of producing product an-gular distributions for the reaction within afew hours; (ii) the technique samples the

SCIENCE * VOL. 260 * 11 JUNE 1993

entire product angular distribution simulta-neously; and (iii) a single quantum state ofthe product, in this case ground-state Datoms, can be selected by utilizing REMPI.Also, instead of imaging the atomic frag-ment, the molecular product can be ionizedin a quantum state-selective manner and itsion image recorded. This will yield theDCS for a single rovibrational quantumstate of the molecular product, a measure-ment that will allow a most stringent test oftheory.

In the present experimental study on theH + D2-> HD + D reaction, our experimen-tal results are in fair agreement with a QCTcalculation based on the LSTH potentialenergy surface. Although we have suggestedsome experimental reasons for the differencesobserved between theoretical predictions andthe experimental results, this does not elimi-nate the possibility that improvements in thetheoretical treatment might be necessary tobridge the disagreement.

REFERENCES AND NOTES

1. F. London, Z. Elektrochem. 35, 552 (1929).2. H. Eyring and M. Polanyi, Z. Phys. Chem. Abt. B

12, 279 (1931).3. D. W. Chandler and P. L. Houston, J. Chem. Phys.

87, 1445 (1987).4. F. J. Aoiz, V. J. Herrero, 0. Puentedura, V. Saez

Rabanos, Chem. Phys. Lett. 198, 321 (1992).5. B. Liu, J. Chem. Phys. 58, 1925 (1973); P. Sieg-

bahn and B. Liu, ibid. 68, 2457 (1978); D. G.Truhlar and C. J. Horowitz, ibid., p. 2466.

6. K. D. Rinnen, D. A. V. Kliner, R. S. Blake, R. N.Zare, Chem. Phys. Lett. 153, 371 (1988); K. D.Rinnen, D. A. V. Kliner, R. N. Zare, J. Chem. Phys.91, 7514 (1989); D. A. V. Kliner, K. D. Rinnen, R.N. Zare, Chem. Phys. Lett. 166,107 (1990); D. A.V. Kliner and R. N. Zare, J. Chem. Phys. 92, 2107(1990); D. A. V. Kliner, D. E. Adelman, R. N. Zare,ibid. 94, 1069 (1991); ibid. 95, 1648 (1991); D.Neuhauser et al., Science 257, 519 (1992).

7. D. L. Phillips, H. B. Levene, J. J. Valentini, J.Chem. Phys. 90,1600 (1989); J. C. Nieh and J. J.Valentini, ibid. 92, 1083 (1990).

8. N. C. Blais and D. G. Truhlar, Chem. Phys. Lett.162, 503 (1989); N. C. Blais etal., J. Chem. Phys.91, 1038 (1989); M. Zhao and D. G. Truhlar, J.Phys. Chem. 94, 6696 (1990).

9. W. J. Keogh et al., Chem. Phys. Lett. 195, 144(1992).

10. F. J. Aoiz, V. J. Herrero, 0. Puentedura, V. SaezRabanos, ibid. 169, 243 (1990); F. J. Aoiz, V. J.Herrero, V. Saez Rabanos, J. Chem. Phys. 97,7423 (1992).

11. D. E. Manolopoulos and R. E. Wyatt, Chem. Phys.Lett. 159, 123 (1989); M. D'Mello, D. E. Manolo-poulos, R. E. Wyatt, J. Chem. Phys. 94, 5985(1991).

12. J. Z. H. Zhang and W. H. Miller, Chem. Phys. Lett.153, 465 (1988); ibid. 159, 130 (1989); J. Chem.Phys. 91, 1528 (1989).

13. R. Goetting, H. R. Mayne, J. P. Toennies, J. Chem.Phys. 80, 2230 (1984); ibid. 85, 6396 (1986).

14. S. A. Buntin, C. F. Giese, W. R. Gentry, ibid. 87,1443 (1987); Chem. Phys. Lett. 168, 513 (1990).

15. R. E. Continetti, B. A. Balko, Y. T. Lee, J. Chem.Phys. 93, 5719 (1990).

16. L. Schnieder, K. Seekamp-Rahn, F. Liedeker, H.Steuwe, K. H. Welge, Faraday Discuss. Chem.Soc. 91, 259 (1991).

17. L. Schnieder and K. H. Welge, XlVth InternationalSymposium on Molecular Beams, Abstracts of invit-ed talks and contributed papers, Asilomar, CA, 7 to12 June 1992 (organized by G. Scoles et al.), p. 27.

1609

on

Feb

ruar

y 9,

201

0 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

18. M. A. Buntine, D. P. Baldwin, R. N. Zare, D. W.Chandler, J. Chem. Phys. 94, 4672 (1991).

19. A. L. Suits, L. S. Bontuyan, P. L. Houston, B. J.Whitaker, ibid. 96, 8618 (1992).

20. D. W. Chandler, J. W. Thoman, Jr., M. H. M.Janssen, D. H. Parker, Chem. Phys. Lett. 156,15(1989); D. P. Baldwin, M. A. Buntine, D. W. Chan-dler, J. Chem. Phys. 93, 6578 (1990); W. P. Hess,D. W. Chandler, J. W. Thoman, Jr., Chem. Phys.163, 277 (1992).

21. R. Wallenstein, Opt. Commun. 33,119 (1980); R.Hilbig and R. Wallenstein, IEEE J. Quantum Elec-tron. QE-17, 1566 (1981); ibid. QE-19, 194 (1983).

22. W. Demtroder, Laser Spectroscopy (Springer-Verlag, New York, 1982), p. 86.

23. E. E. Marinero, C. T. Rettner, R. N. Zare, Phys.

Rev. Lett. 48, 1323 (1982); E. E. Merinero, R.Vasudev, R. N. Zare, J. Chem. Phys. 78, 692(1983); K. D. Rinnen, M. A. Buntine, D. A. V.Kliner, R. N. Zare, ibid. 95, 214 (1991); W. M. Huo,K. D. Rinnen, R. N. Zare, ibid., p. 205; D. E.Adelman, N. E. Shafer, D. A. V. Kliner, R. N. Zare,ibid. 97, 7323 (1992).

24. R. D. Clear, S. J. Riley, K. R. Wilson, J. Chem.Phys. 63,1340 (1975); R. Schmiedl, H. Dugan, W.Meier, K. H. Welge, Z Phys. A 304, 137 (1982).

25. R. N. Bracewell, The Fourier Transform and ItsApplications (McGraw-Hill, New York, 1986).

26. G. C. Gonzalez and P. Wintz, Digital Image Pro-cessing (Addison-Wesley, London, 1977).

27. R. N. Strickland and D. W. Chandler, Appl. Opt.30,1811 (1991).

28. W. H. Miller and J. Z. H. Zhang, J. Phys. Chem.95,12 (1991).

29. A. J. Varandas, F. B. Brown, C. A. Mead, D. G.Truhlar, N. C. Blais, J. Chem. Phys. 86, 6258(1987).

30. A. I. Boothroyd, W. J. Keogh, P. G. Martin, M. R.Peterson, ibid. 95, 4343 (1991).

31. We thank M. J. Jaska for technical support, and F.J. Aoiz for providing the OCT results for the H +D2 reaction. T.N.K. thanks R. E. Continetti formany enlightening discussions. M.A.B. andR.N.Z. acknowledge support from the NationalScience Foundation (under grant CHE 90-7939).This work is supported by the U.S. Department ofEnergy, Office of Basic Energy Sciences, Divisionof Chemical Sciences.

The Biological and SocialPhenomenon of Lyme Disease

Alan G. Barbour and Durland FishLyme disease, unknown in the United States two decades ago, is now the most commonarthropod-borne disease in the country and has caused considerable morbidity in severalsuburban and rural areas. The emergence of this disease is in part the consequence ofthe reforestation of the northeastern United States and the rise in deer populations.Unfortunately, an accurate estimation of its importance to human and animal health hasnot been made because of difficulties in diagnosis and inadequate surveillance activities.Strategies for prevention of Lyme disease include vector control and vaccines.

burgdorferi infection (9, 10). Some of theseresources would better benefit the commu-nity if directed toward methods of diseaseprevention, such as vector control. Azoonosis can be characterized with respectto the microbiology of the agent, the ecol-ogy in relation to vectors and reservoirhosts, and the epidemiology of human dis-ease. Full description of the phenomenon ofLyme disease will also require considerationof behavioral and economic factors in theresponse to the disease's emergence. Thesesocial factors are still poorly understood.

The Origins of Lyme Disease inNorth America

Lyme disease is a zoonosis, an inadvertentinfection of humans with an animal patho-gen. In temperate regions of the NorthernHemisphere, ticks transmit the etiologicbacterium Borrelia burgdorfen from its usualwildlife reservoirs to humans and domesticanimals (1, 2). In the United States, Lymedisease occurs primarily in suburban andrural areas (3). Early infection of humans isusually a self-limited, flu-like illness with a

skin rash where the tick imbeds itself (4)(Fig. 1). After a few weeks to severalmonths, as many as 70% of untreated,infected patients suffer the effects of bacte-rial invasion of one or more distant organsor systems, including the brain, nerves,eyes, joints, and heart (5). These latemanifestations, particularly the dysfunctionof the central nervous system and chronicarthritis, are disabling but rarely fatal (5).

In the United States during 1991, 9465cases of Lyme disease were formally report-ed, making it by far the most common

arthropod-borne disease (6). The rising in-cidence and geographic spread of thiszoonosis have interested the general public(7). Lyme disease probably ranks only be-

A. G. Barbour is in the Departments of Microbiologyand Medicine, University of Texas Health ScienceCenter, San Antonio, TX 78284. D. Fish is in theMedical Entomology Laboratory, Department of Com-munity and Preventive Medicine, New York MedicalCollege, Valhalla, NY 10595.

1610

hind acquired immunodeficiency syndromein media coverage of infectious diseases inthe United States over the last decade.News, public health programs, and patientadvocacy groups have informed the publicof the symptoms of Lyme disease and ofways to avoid infection (8). A less salubri-ous consequence of the attention has beenthe attribution to B. burgdorferi of a numberof ills, only a fraction of which are likely tobe Lyme disease (9, 10). Wisconsin had545 reported cases of Lyme disease in 1989;in the same year, 94,000 serum sampleswere received by reference laboratories inthe state for Lyme disease testing (11).Georgia reported hundreds of cases of Lymedisease until it was documented that therewere few ticks bearing B. burgdorferi in thestate (12).A provisional diagnosis of Lyme disease

is often acceptable to patients with vexing,undefined illnesses, not only because thereis hope for a cure with antibiotics but alsobecause Lyme disease is acquired throughwhat are generally perceived to be whole-some activities, such as hiking and workingout-of-doors (7). The full extent to whichpeople are being inappropriately treatedwith antibiotics cannot be estimated atpresent, but it is likely that a large minor-ity, if not a majority, of the health care

dollars expended on therapy for Lyme dis-ease are for inaccurate diagnoses of B.

SCIENCE * VOL. 260 * 11 JUNE 1993

The clinical syndrome of B. burgdorferi in-fection had been described in Europe (13)more than six decades before Steere andcolleagues in 1975 investigated an unusualcluster of childhood arthritis in the coastalcommunity of Lyme, Connecticut (14).Soon after the Connecticut investigation,the relation between the arthritis and aprior episode of the characteristic skin rash,erythema migrans (Fig. 1), common inEurope, was noted (15). The search for anetiologic agent implicated a tick, Ixodesscapularis (I. dammini), as the vector onepidemiological grounds (16). The bite of arelated species, I. ricinus, was known tocause erythema migrans in Europe (17).Identity between the two tick-borne condi-tions was established when B. burgdorferiwas first isolated from I. scapularis and I.ricinus and then from patients in the UnitedStates and Europe with Lyme disease anderythema migrans (18).

The events leading to an epidemic ofarthritis in residents of Lyme began severalcenturies earlier. Infections from B. burg-dorferi probably occurred in North Americabefore the first waves of European coloniza-tion. Erythema migrans, the hallmark of B.burgdorferi infection, was already present inmidwestern and Pacific states at the timeLyme disease was first described in Con-necticut (19). Early descriptions of colonialforests, the abundance of deer, and ticksannoying explorers suggest that the condi-

................................

on

Feb

ruar

y 9,

201

0 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from