ASTROBIOLOGYVolume 4 Number 1 2004copy Mary Ann Liebert Inc

Research Paper

Hydrogen Emission in Meteors as a Potential Marker forthe Exogenous Delivery of Organics and Water

PETER JENNISKENS1 and AVRAM M MANDELL2

ABSTRACT

We detected hydrogen Balmer-alpha (Ha) emission in the spectra of bright meteors and in-vestigated its potential use as a tracer for exogenous delivery of organic matter We foundthat it is critical to observe the meteors with high enough spatial resolution to distinguish the65646 nm Ha emission from the 65746 nm intercombination line of neutral calcium whichwas bright in the meteor afterglow The Ha line peak stayed in constant ratio to the atmos-pheric emissions of nitrogen during descent of the meteoroid If all of the hydrogen origi-nates in the Earthrsquos atmosphere the hydrogen atoms are expected to have been excited at T5 4400 K In that case we measured an H2O abundance in excess of 150 6 20 ppm at 80ndash90km altitude (assuming local thermodynamic equilibrium in the air plasma) This compareswith an expected 20 ppm from H2O in the gas phase Alternatively meteoric refractory or-ganic matter (and water bound in meteoroid minerals) could have caused the observed Ha

emission but only if the line is excited in a hot T 10000 K plasma component that is uniqueto meteoric ablation vapor emissions such as Si1 Assuming that the Si1 lines of the Leonidspectrum would need the same hot excitation conditions and a typical [H][C] 5 1 in cometaryrefractory organics we calculated an abundance ratio [C][Si] 5 39 6 14 for the dust of comet55PTempel-Tuttle This range agreed with the value of [C][Si] 5 44 measured for comet1PHalley dust Unless there is 10 times more water vapor in the upper atmosphere than ex-pected we conclude that a significant fraction of the hydrogen atoms in the observed meteorplasma originated in the meteoroid Key Words Prebiotic moleculesmdashOrigin of lifemdashMete-orsmdashMolecular band emission Astrobiology 4 123ndash134

123

INTRODUCTION

ATOMIC HYDROGEN EMISSION may be the onlyreadily observable signature of organic mat-

ter in meteors during exogenous delivery Mete-ors represent a unique chemical pathway towards

prebiotic compounds on the early Earth (Jen-niskens et al 2000a 2004ab) Other pathwaysfrom cometary and asteroidal organics to prebi-otic compounds include giant impacts during thefinal stages of planet accretion and the steadyrain of interplanetary dust particles and micro-

1Center for the Study of Life in the Universe SETI Institute Mountain View California2NASA Ames Astrobiology Academy NASA Ames Research Center Moffett Field CaliforniaPresent address Astronomy and Astrophysics Department The Pennsylvania State University State College Penn-

sylvania

meteorites that survived passage through Earthrsquosatmosphere (Oroacute 1961 Chyba and Sagan 19921997 Delsemme 1992 Brack 1999 Maurette etal 2000) Meteors are significant because the bulkof infalling meteoroid mass is ablated in Earthrsquosatmosphere

A significant fraction if not all of the refrac-tory organic matter is expected to remain in themeteoroids after ejection from a comet acting asa glue that holds the silicate grains together Anopen question is how much organics are still thereat the time of impact with Earth

During ablation meteoric organic matter canundergo unique chemical processes in the heatedmeteoroid and in the rarefied high Mach numberflow of the meteor plasma the physical condi-tions of which are studied by remote sensing ofpresent-day meteors (Jenniskens et al 2000a) Or-ganic matter can lose hydrogen atoms if the ma-terial is completely dissociated or in a processcalled carbonization which is the loss of func-tional groups and simultaneous growth of thecarbon skeleton a process commonly observedwhen organic matter is exposed to energetic par-ticles and ultraviolet light (Jenniskens et al 1993)

We conducted high-resolution spectroscopicstudies of meteors at visible wavelengths duringthe recent November 18 2001 Leonid storm on-board the NASA and US Air Force-sponsoredLeonid Multi-Instrument Aircraft Campaign(Leonid MAC) (Jenniskens and Russell 2003)Highest rates were observed at 1040 universaltime (UT) when Earth crossed the dust ejectedfrom comet 55PTempel-Tuttle during its returnin 1767 (Kondratrsquoeva and Reznikov 1985)

Elsewhere in this issue we focused on thesearch for the CN radical a small diatomic or-ganic breakup product and found much less CNthan expected if the meteoric organic matter is ef-ficiently decomposed (Jenniskens et al 2004a) Ifon the other hand the organic matter is car-bonized by a loss of functional groups such as -OH and -H (Jenniskens and Stenbaek-Nielsen2004) then CN radicals would be rare but hy-drogen atoms might be detected

Hydrogen may also originate from water andhydrogen-containing minerals in the meteoroidAt the typical T 4400 K temperatures in meteorplasma (Jenniskens et al 2004b) water moleculesand OH radicals are quickly dissociated produc-ing hydrogen atoms The water in our oceans mayhave been delivered by meteoroids if they con-

tained significant amounts of water with a DHratio more similar to that of our ocean than cometice (Maurette et al 2000) Cometary meteoroidsare thought to lose much if not all of their icyvolatiles to the vacuum of space after release (Lel-louch et al 1998) It is not known if they containwater bound to minerals (see discussion in Jen-niskens et al 2004c) On the other hand OH A-X (00) emission at 310 nm has previously beenidentified in meteor spectra including those ofLeonids (Harvey 1977 Abe et al 2002 Jen-niskens et al 2002) We did not find ground-stateOH X2P Meinel band emission in Leonid spec-tra providing support to the hypothesis that theOH originates from the dissociation of water (Jen-niskens et al 2004c)

This suggested to us to look for hydrogen emis-sion in meteors Unfortunately the strongest op-tical line of the hydrogen Balmer series hydro-gen Balmer-alpha (Ha) has a high excitationenergy of 1209 eV and is expected to be weak ifthermally excited at 4400 K However it has beenhypothesized that a hot T 10000 K componentexists in meteor plasma which is responsible forthe strong Ca1 and Si1 lines seen in bright andfast meteors (Borovicka 1994a) This componentwould provide sufficient population of the upperstate for the Ha line to become visible for the ex-pected abundances

Indeed BorovicIuml ka and Jenniskens (2000) re-ported the detection of Ha emission from low-resolution spectra of the 213 magnitude ldquoY2KrdquoLeonid fireball observed during the 1999 LeonidMAC mission and derived an abundance ratioof [H][Fe] 5 10ndash20 The hydrogen line wasidentified only in the meteor spectrum itself notin the meteor afterglow Before that severalother authors reported Ha in low-resolutionspectra (eg Cook and Millman 1955) or in relatively confused high-resolution spectra ofbright fireballs (eg Ceplecha 1971) Theseidentifications are sometimes in doubt and thehydrogen abundance was not quantified Ha

was not detected in the high-quality (Dl 5 07nm) photographic spectrum of the slow 29thmagnitude fireball EN 151068 (BorovicIuml ka1994b)

Whether hydrogen is present in meteor spec-tra needs to be established before addressing thequestion how much hydrogen derives from themeteoroid (from organics or water) or from theambient atmosphere

JENNISKENS AND MANDELL124

METHODS

Until now the success rate of capturing good-quality meteor spectra with traditional photo-graphic techniques has been low typically onlyone good spectrum per year per camera (Mill-man 1980) The photographic cameras in useneed a rare very bright fireball (25 magnitude orbrighter) to provide a good spectrum Key to in-creasing the sensitivity to more common faintermeteors and still maintaining a high spectral res-olution is to use cooled charge coupled device(CCD) cameras Cooled CCD cameras are about5 magnitudes more sensitive than photographicplates The instrumental layout and responsefunctions of our two-stage thermoelectricallycooled slit-less CCD spectrograph is discussedelsewhere in this issue (Jenniskens et al 2004a)By purposely limiting the spectral range to about100 nm over the 1024 3 1024 pixel CCD rangea spectral resolution is obtained of Dl 5 054 nmin first order which is similar to or better thanthat of traditional photographic slit-less spectro-graphs

RESULTS

The observation of meteor storms further in-creases the detection rate During the November2001 Leonid storm the instrument was deployedbehind an optically flat BK7-type glass windowonboard the US Air Force418th Flight TestSquadron-operated NKC-135 ldquoFISTArdquo researchaircraft from an elevation of 37000 ft This was a

single-plane mission en route from Alabama toCalifornia supported by ground stations in Ari-zona (Jenniskens and Russell 2003) Additionalspectra were obtained during a ground-based ob-serving run at Fremont Peak Observatory in Cal-ifornia at the time of the December 12 and 132001 Geminid shower

At 105232 UT (November 18 2001) a partic-ularly strong emission line close to the expectedHa line was detected in the spectrum of a 28 mag-nitude Leonid fireball known by the data ana-lyzers as ldquoHardquo after it was recognized that Ha

emission was present This was the brightest me-teor detected with this instrument during thatmission The ldquoHardquo fireball was recorded too bya low-resolution intensified video camera (Fig 1)After a gradual brightening a strong flare wasobserved and a small fragment continued tolower altitudes in a pattern that was similar to thewell-documented ldquoY2Krdquo fireball (BorovicIuml ka andJenniskens 2000 Jenniskens and Rairden 2000)A persistent train was seen for 12 min The me-teor appeared in the constellation Draco andpeaked at an elevation of 21deg above the horizon10deg east from north when the plane was at thecoordinates 10804 W 3500 N south of GallupNM Unfortunately this was just out of reach ofour ground stations in Arizona and no trajectorydata are available From the altitude of the ldquoY2Krdquopersistent train (92ndash79 km with a peak at 84 km)we surmise that the ldquoHardquo meteor peaked also ator just above 84 km and penetrated to an altitudebetween 79 and 84 km

The first order spectrum is shown in Fig 2 Justafter the sodium line intensity peaked there was

HYDROGEN EMISSION IN METEORS 125

FIG 1 The 105232 UT Leonid meteor as recorded by an intensified video camera The field of view is about 15 315deg From left to right (a) the meteor at its peak brightness (t 5 0 s) (b) the afterglow (bottom) and the 557 nm O Iwake (top) immediately following the meteor in an average of four frames (t 5 016 s) and (c) the persistent train 6s after the meteor in an average of 8 frames

an abrupt change of the spectrum and high ex-citation lines appeared This was the position ofthe meteoroid when the camera shutter openedand the meteor itself came in view At positionswhere the meteoroid was prior to this point therecorded spectrum was that of the meteor after-glow (top part of Fig 2) Although afterglow andmeteor were integrated for 0800 s the afterglowat positions earlier along the trajectory had de-cayed more than positions closer to the meteoroidat the start of the exposure The integration timewas not long enough to detect any chemilumi-nescence from a persistent train At positionswhere the meteoroid was seen while the shutterwas open the spectrum is dominated by the me-teor itself which persists for 002 s with a mi-nor contribution from the afterglow significantonly in the sodium lines and the strong inter-combination line of Ca I at 657 nm The onset ofthe meteor exposure was slightly earlier at thecenter of the plate because the meteoroid was inmotion while the shutter opened starting fromthe center There were periodic variations in thelight intensity along the trajectory most likelydue to a modulation of vapor production with therotation of the meteoroid at a rate of about 46 cycless

The spectrum of meteor and afterglow are dif-ferent as described in detail for the ldquoY2Krdquo Leonidfireball in BorovicIuml ka and Jenniskens (2000) Inshort the meteor spectrum consists predomi-nantly of allowed transitions of atoms and ionsof both low excitation (eg lines of Na I Fe I CaII) and high excitation (eg O I N I H I Si II)The afterglow only has low excitation lines butthis includes also excitation lines with low tran-sition probability (eg Ca I at 657 nm Fe I at 636nm and many others) which are very faint in thespectrum of the fireball itself

The late start of the exposure was fortunate Itenabled the separation of the fireball and after-glow spectrum This is normally possible only forvideo spectra (low resolution) and photographicspectra with a rotating shutter Moreover thefireball spectrum would have been heavily over-exposed at the position of the flare

Line identification

Because of the small detector size the wave-length scale is almost linear and wavelength cal-ibrations can be performed by matching knownlines in the spectrum Lines were identified bycalibrating the vacuum wavelength of the sodium

JENNISKENS AND MANDELL126

FIG 2 First order spectrum of the 105232 UT ldquoHardquo Leonid meteor in the range 588ndash720 nm The spectrum hasbeen expanded by a factor of 4 in vertical direction to undo the effect of binning and the image is shown reversedso that the wavelength scale runs left to right The meteor moved from top to bottom The main atomic lines andbands are identified InsetAn enlargement of the center part of the spectrum showing the hydrogen emission in theafterglow and along the length of the meteor spectrum

doublet to the theoretical wavelength scale (Jen-niskens et al 2004a) Any Doppler shift fromresidual meteor motion is negligibly small andsystematic Subsequently all of the emissions ex-pected from neutral metal atoms and singly ion-ized elements were matched to the observed line

positions For the first order there is no orderoverlap below about 740 nm due to a fast dropof optical transmittance below 370 nm

We found that many strong emission lines inthis wavelength range are due to neutral calciumWith that information in hand many other lineswere also identified (Table 1) The wavelength

HYDROGEN EMISSION IN METEORS 127

FIG 3 Extracted Ha and calcium emission linesShown are average Leonid spectra from afterglow andmeteor The comparison spectra are the Geminid and Per-seid meteors discussed in the text Dashed lines depict thebackground emission from molecular nitrogen Dottedlines in the middle two spectra are the fitted line profilesfor H and Ca emission respectively which are based onthe line positions and the instrumental line profile andare matched to fit the observed peak intensity of the lines

FIG 4 Hydrogen emission in the Gem-inid of 071528 UT (December 14 2001)from a ground-based observation fromFremont Peak Observatory in CaliforniaThe horizontal line is a star spectrum anddots are first order star images

FIG 5 Comparison of the Leonid Geminid and Per-seid meteor spectraLeonid and Geminid spectra are dis-placed upward to facilitate comparison Note the strongtelluric absorption of oxygen and water in the ground-based spectrum of the Geminid Also note the absence ofmetal atom ablation in the fainter Perseid

JENNISKENS AND MANDELL128

TABLE 1 LINE IDENTIFICATIONS

^ intensity

Leonid Leonid GeminidObserved Theory vacuum Element afterglow meteor meteor

58909 5891584 Na I (D) 1036 mdash mdash58967 5897558 Na I (D) 671 mdash mdash59193 5915753 Fe I low E 5 mdash mdash59588 5958342 Fe I low E 12 mdash mdash61047 6104412 Ca I 6 3 4861251 6123912 Ca I 15 9 5861382 6138313 Fe I 2 2 4861408 6139393 Fe I 4 3 4061599 615767561584596159876 O I 7 86 1761646 6163878x3001x5462x8146 Ca I 32 16 40061712 61707506171272 Ca I 21 11 4661758 6175049 Fe I 2 2 2361909 6193271 Fe I 2 2 4262236 6223391 Fe I low E 2 2 1562325 6232450 Fe I 2 2 3562372 mdash Unidentified 4 2 2362467 6248045 Fe I 3 2 2962558 6254284 Fe I 2 2 2962831 6282353 Fe I low E 24 2 3763028 63032416304237 Fe I 21 2 4363201 6317052976524438 (19856) Fe I (Ca I) 2 5 2663386 63370806338575 Fe I 2 2 1963499 6348849 Si II 2 14 163566 6355592 Fe I low E mdash 2 3563612 6360451 Fe I low E 117 0 7863630 6363545 Ca I mdash 2 3563736 6373125 Si II 3 7 263834 mdash Unidentified 6 2 1763871 mdash Unidentified 6 2 264030 6402083 Fe I low E 18 2 1264127 64097876413419 Fe I 2 2 3264218 6423124 Fe I 2 2 2664406 6440855 Ca I 28 20 3264428 6442718 N I 2 10 mdash64525 6451592 Ca I 10 8 9564579 645776456264553 O I 5 43 6964652 6464351 Ca I 30 18 2064738 6473450 Ca I 7 4 2964857 648660064844906485545 N I 6 42 264955 6495576 Ca I 19 8 2065020 6500733 Fe I low E 17 2 9065297 mdash Unidentified 2 2 3565459 6549381 Fe I low E 2 2 1565653 6564636 H I 6 15 9065757 6574595 Ca I 760 90a 2065955 65947346595692 Fe I 2 2 2366115 6611502 Fe I low E 2 3 2067194 6719536 Ca I 11 10 1467260 67244666725309 N I 7 12 1071512 7150120 Ca I 15 15 2373277 7328163 Ca I mdash mdash 1774250 7425686 N I mdash mdash 1474433 7444348 N I mdash mdash 4074700 7470369 N I mdash mdash 36

These are vacuum wavelengths Integrated line intensities are in arbitrary units after correction for instrument response The line in boldface type is the Ha line discussed in the text

aContaminated by afterglow

Wavelength

scale is accurate to about half a pixel 6007 nmAt the altitude of the observations (37000 ft) therewas no detectable atmospheric water vapor ab-sorption Many of the weaker features seen in thespectrum are due to vibrational and rotationalstructure in the underlying First Positive band ofN2

We found that the expected position of Ha

(65646 nm) did not align with the brightest emis-sion line identified as the intercombination65746 nm line of calcium (Poole 1979) but witha fainter emission line just next to it Figure 3shows the line profiles from row averages at po-sitions where the meteor was before and afteropening of the shutter The dashed line shows theline positions of Ha and Ca I convolved with theinstrumental line profile A good match is ob-tained after scaling the line intensities to matchthe observed lines The line intensity ratio HCaequalled 017

Corroborating evidence for the presence ofboth Ha and the Ca I line comes from the spec-trum of an intrinsically slower Geminid (363 606 kms rather than 716 6 04 kms) shown inFig 4 A comparison of the details of the raw datanear the Ha line in Figs 3 and 4 reveal many sim-ilarities Though the meteor was not observed ina wide-field imager it must have been locatednear the horizon judging from the presence of thestrong telluric oxygen O2 B-band (686ndash688 nm)absorption in this spectrum (Fig 5) The inte-grated intensity over the final part of the Gemi-nid spectrum is about 11 magnitude fainter thanthat for the Leonid fireball The line intensity ra-tio HCa equalled 041 (Fig 3)

The Geminid spectrum is rich in metal atomemission lines Air plasma emissions were alsodetected The exposure was started before the me-teorrsquos peak brightness At positions just before theposition of the meteor at the onset of exposure(upper part of Fig 3) afterglow emission was ob-served Geminids are not known to have flareslike the Leonids but some more-rapid-than-ex-pected variations of the light curve can be recog-nized shortly after the beginning of the exposureThe Geminid produced a repetitive series of darkbands that are less well separated than in theldquoHardquo meteor indicative of rotation of the mete-oroid at a rate of about 40 cycless

No such calcium or hydrogen emission was ob-served in a fainter 23 magnitude Perseid meteorobserved at 103526 UT (August 13 1999) asshown in Fig 5 (Jenniskens et al 2000a) The Per-

seid meteor was detected too by a second cam-era and appeared high in the sky As expected ithas less telluric O2 B-band absorption The spec-trum has a lower spectral resolution due to a lessfavorable orientation of the meteor trajectory rel-ative to the dispersion direction of the gratingBecause of the much weaker metal atom ablationlines this spectrum illustrates the thermal emis-sions from oxygen and nitrogen atoms Note thatthis spectrum too has an Ha line as strong as thatin the Leonid and Geminid spectra as well as afaint calcium intercombination line with inten-sity ratio HCa equal to 069 (Fig 3)

Temporal variation

The temporal variation of various representa-tive emission lines in the ldquoHardquo Leonid is shownin Fig 6 In the afterglow phase neutral metalatoms of sodium iron and calcium which derivefrom widely varying volatile minerals but havesimilar excitation energies follow initially thesame exponential increase in intensity Sodiumremains in the same proportion to iron and cal-cium during the afterglow implying that therewas no differential ablation in this part of the tra-jectory Minerals containing sodium did not ab-late earlier than the less volatile minerals con-taining calcium The light curve reflects theexponential increase in brightness of the meteorleading up to a flare that occurred just beforeopening of the shutter

The ldquoHardquo meteorrsquos spectrum is dominated byair plasma emissions from nitrogen moleculesand atoms but metal atom ablation lines are alsoobserved (Table 1) Low excitation lines of iron(Ei 4 eV) are comparatively weak but calciumand sodium lines are stronger than in the after-glow earlier in the trajectory This suggests thata significant part of the calcium emission is fromthe meteor phase For T 5 4400 K we calculatedan abundance ratio of [Ca][Fe] equal to 013 Ingood agreement with this finding is the result ofRietmeijer and Nuth (2000) who calculated thatcomet 1PHalleyrsquos dust has a [Ca][Fe] ratio of012 Hence in this particular meteor the calciumwas ablated as efficiently as iron The decrease oflight from the strong sodium and calcium lines isexponential and follows the same slope as the airplasma emissions reflecting a weakening of themeteor after the earlier flare The ionized siliconemission has an interesting flare at row 15 (Fig7) that was not seen in other lines including the

HYDROGEN EMISSION IN METEORS 129

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

meteorites that survived passage through Earthrsquosatmosphere (Oroacute 1961 Chyba and Sagan 19921997 Delsemme 1992 Brack 1999 Maurette etal 2000) Meteors are significant because the bulkof infalling meteoroid mass is ablated in Earthrsquosatmosphere

A significant fraction if not all of the refrac-tory organic matter is expected to remain in themeteoroids after ejection from a comet acting asa glue that holds the silicate grains together Anopen question is how much organics are still thereat the time of impact with Earth

During ablation meteoric organic matter canundergo unique chemical processes in the heatedmeteoroid and in the rarefied high Mach numberflow of the meteor plasma the physical condi-tions of which are studied by remote sensing ofpresent-day meteors (Jenniskens et al 2000a) Or-ganic matter can lose hydrogen atoms if the ma-terial is completely dissociated or in a processcalled carbonization which is the loss of func-tional groups and simultaneous growth of thecarbon skeleton a process commonly observedwhen organic matter is exposed to energetic par-ticles and ultraviolet light (Jenniskens et al 1993)

We conducted high-resolution spectroscopicstudies of meteors at visible wavelengths duringthe recent November 18 2001 Leonid storm on-board the NASA and US Air Force-sponsoredLeonid Multi-Instrument Aircraft Campaign(Leonid MAC) (Jenniskens and Russell 2003)Highest rates were observed at 1040 universaltime (UT) when Earth crossed the dust ejectedfrom comet 55PTempel-Tuttle during its returnin 1767 (Kondratrsquoeva and Reznikov 1985)

Elsewhere in this issue we focused on thesearch for the CN radical a small diatomic or-ganic breakup product and found much less CNthan expected if the meteoric organic matter is ef-ficiently decomposed (Jenniskens et al 2004a) Ifon the other hand the organic matter is car-bonized by a loss of functional groups such as -OH and -H (Jenniskens and Stenbaek-Nielsen2004) then CN radicals would be rare but hy-drogen atoms might be detected

Hydrogen may also originate from water andhydrogen-containing minerals in the meteoroidAt the typical T 4400 K temperatures in meteorplasma (Jenniskens et al 2004b) water moleculesand OH radicals are quickly dissociated produc-ing hydrogen atoms The water in our oceans mayhave been delivered by meteoroids if they con-

tained significant amounts of water with a DHratio more similar to that of our ocean than cometice (Maurette et al 2000) Cometary meteoroidsare thought to lose much if not all of their icyvolatiles to the vacuum of space after release (Lel-louch et al 1998) It is not known if they containwater bound to minerals (see discussion in Jen-niskens et al 2004c) On the other hand OH A-X (00) emission at 310 nm has previously beenidentified in meteor spectra including those ofLeonids (Harvey 1977 Abe et al 2002 Jen-niskens et al 2002) We did not find ground-stateOH X2P Meinel band emission in Leonid spec-tra providing support to the hypothesis that theOH originates from the dissociation of water (Jen-niskens et al 2004c)

This suggested to us to look for hydrogen emis-sion in meteors Unfortunately the strongest op-tical line of the hydrogen Balmer series hydro-gen Balmer-alpha (Ha) has a high excitationenergy of 1209 eV and is expected to be weak ifthermally excited at 4400 K However it has beenhypothesized that a hot T 10000 K componentexists in meteor plasma which is responsible forthe strong Ca1 and Si1 lines seen in bright andfast meteors (Borovicka 1994a) This componentwould provide sufficient population of the upperstate for the Ha line to become visible for the ex-pected abundances

Indeed BorovicIuml ka and Jenniskens (2000) re-ported the detection of Ha emission from low-resolution spectra of the 213 magnitude ldquoY2KrdquoLeonid fireball observed during the 1999 LeonidMAC mission and derived an abundance ratioof [H][Fe] 5 10ndash20 The hydrogen line wasidentified only in the meteor spectrum itself notin the meteor afterglow Before that severalother authors reported Ha in low-resolutionspectra (eg Cook and Millman 1955) or in relatively confused high-resolution spectra ofbright fireballs (eg Ceplecha 1971) Theseidentifications are sometimes in doubt and thehydrogen abundance was not quantified Ha

was not detected in the high-quality (Dl 5 07nm) photographic spectrum of the slow 29thmagnitude fireball EN 151068 (BorovicIuml ka1994b)

Whether hydrogen is present in meteor spec-tra needs to be established before addressing thequestion how much hydrogen derives from themeteoroid (from organics or water) or from theambient atmosphere

JENNISKENS AND MANDELL124

METHODS

Until now the success rate of capturing good-quality meteor spectra with traditional photo-graphic techniques has been low typically onlyone good spectrum per year per camera (Mill-man 1980) The photographic cameras in useneed a rare very bright fireball (25 magnitude orbrighter) to provide a good spectrum Key to in-creasing the sensitivity to more common faintermeteors and still maintaining a high spectral res-olution is to use cooled charge coupled device(CCD) cameras Cooled CCD cameras are about5 magnitudes more sensitive than photographicplates The instrumental layout and responsefunctions of our two-stage thermoelectricallycooled slit-less CCD spectrograph is discussedelsewhere in this issue (Jenniskens et al 2004a)By purposely limiting the spectral range to about100 nm over the 1024 3 1024 pixel CCD rangea spectral resolution is obtained of Dl 5 054 nmin first order which is similar to or better thanthat of traditional photographic slit-less spectro-graphs

RESULTS

The observation of meteor storms further in-creases the detection rate During the November2001 Leonid storm the instrument was deployedbehind an optically flat BK7-type glass windowonboard the US Air Force418th Flight TestSquadron-operated NKC-135 ldquoFISTArdquo researchaircraft from an elevation of 37000 ft This was a

single-plane mission en route from Alabama toCalifornia supported by ground stations in Ari-zona (Jenniskens and Russell 2003) Additionalspectra were obtained during a ground-based ob-serving run at Fremont Peak Observatory in Cal-ifornia at the time of the December 12 and 132001 Geminid shower

At 105232 UT (November 18 2001) a partic-ularly strong emission line close to the expectedHa line was detected in the spectrum of a 28 mag-nitude Leonid fireball known by the data ana-lyzers as ldquoHardquo after it was recognized that Ha

emission was present This was the brightest me-teor detected with this instrument during thatmission The ldquoHardquo fireball was recorded too bya low-resolution intensified video camera (Fig 1)After a gradual brightening a strong flare wasobserved and a small fragment continued tolower altitudes in a pattern that was similar to thewell-documented ldquoY2Krdquo fireball (BorovicIuml ka andJenniskens 2000 Jenniskens and Rairden 2000)A persistent train was seen for 12 min The me-teor appeared in the constellation Draco andpeaked at an elevation of 21deg above the horizon10deg east from north when the plane was at thecoordinates 10804 W 3500 N south of GallupNM Unfortunately this was just out of reach ofour ground stations in Arizona and no trajectorydata are available From the altitude of the ldquoY2Krdquopersistent train (92ndash79 km with a peak at 84 km)we surmise that the ldquoHardquo meteor peaked also ator just above 84 km and penetrated to an altitudebetween 79 and 84 km

The first order spectrum is shown in Fig 2 Justafter the sodium line intensity peaked there was

HYDROGEN EMISSION IN METEORS 125

FIG 1 The 105232 UT Leonid meteor as recorded by an intensified video camera The field of view is about 15 315deg From left to right (a) the meteor at its peak brightness (t 5 0 s) (b) the afterglow (bottom) and the 557 nm O Iwake (top) immediately following the meteor in an average of four frames (t 5 016 s) and (c) the persistent train 6s after the meteor in an average of 8 frames

an abrupt change of the spectrum and high ex-citation lines appeared This was the position ofthe meteoroid when the camera shutter openedand the meteor itself came in view At positionswhere the meteoroid was prior to this point therecorded spectrum was that of the meteor after-glow (top part of Fig 2) Although afterglow andmeteor were integrated for 0800 s the afterglowat positions earlier along the trajectory had de-cayed more than positions closer to the meteoroidat the start of the exposure The integration timewas not long enough to detect any chemilumi-nescence from a persistent train At positionswhere the meteoroid was seen while the shutterwas open the spectrum is dominated by the me-teor itself which persists for 002 s with a mi-nor contribution from the afterglow significantonly in the sodium lines and the strong inter-combination line of Ca I at 657 nm The onset ofthe meteor exposure was slightly earlier at thecenter of the plate because the meteoroid was inmotion while the shutter opened starting fromthe center There were periodic variations in thelight intensity along the trajectory most likelydue to a modulation of vapor production with therotation of the meteoroid at a rate of about 46 cycless

The spectrum of meteor and afterglow are dif-ferent as described in detail for the ldquoY2Krdquo Leonidfireball in BorovicIuml ka and Jenniskens (2000) Inshort the meteor spectrum consists predomi-nantly of allowed transitions of atoms and ionsof both low excitation (eg lines of Na I Fe I CaII) and high excitation (eg O I N I H I Si II)The afterglow only has low excitation lines butthis includes also excitation lines with low tran-sition probability (eg Ca I at 657 nm Fe I at 636nm and many others) which are very faint in thespectrum of the fireball itself

The late start of the exposure was fortunate Itenabled the separation of the fireball and after-glow spectrum This is normally possible only forvideo spectra (low resolution) and photographicspectra with a rotating shutter Moreover thefireball spectrum would have been heavily over-exposed at the position of the flare

Line identification

Because of the small detector size the wave-length scale is almost linear and wavelength cal-ibrations can be performed by matching knownlines in the spectrum Lines were identified bycalibrating the vacuum wavelength of the sodium

JENNISKENS AND MANDELL126

FIG 2 First order spectrum of the 105232 UT ldquoHardquo Leonid meteor in the range 588ndash720 nm The spectrum hasbeen expanded by a factor of 4 in vertical direction to undo the effect of binning and the image is shown reversedso that the wavelength scale runs left to right The meteor moved from top to bottom The main atomic lines andbands are identified InsetAn enlargement of the center part of the spectrum showing the hydrogen emission in theafterglow and along the length of the meteor spectrum

doublet to the theoretical wavelength scale (Jen-niskens et al 2004a) Any Doppler shift fromresidual meteor motion is negligibly small andsystematic Subsequently all of the emissions ex-pected from neutral metal atoms and singly ion-ized elements were matched to the observed line

positions For the first order there is no orderoverlap below about 740 nm due to a fast dropof optical transmittance below 370 nm

We found that many strong emission lines inthis wavelength range are due to neutral calciumWith that information in hand many other lineswere also identified (Table 1) The wavelength

HYDROGEN EMISSION IN METEORS 127

FIG 3 Extracted Ha and calcium emission linesShown are average Leonid spectra from afterglow andmeteor The comparison spectra are the Geminid and Per-seid meteors discussed in the text Dashed lines depict thebackground emission from molecular nitrogen Dottedlines in the middle two spectra are the fitted line profilesfor H and Ca emission respectively which are based onthe line positions and the instrumental line profile andare matched to fit the observed peak intensity of the lines

FIG 4 Hydrogen emission in the Gem-inid of 071528 UT (December 14 2001)from a ground-based observation fromFremont Peak Observatory in CaliforniaThe horizontal line is a star spectrum anddots are first order star images

FIG 5 Comparison of the Leonid Geminid and Per-seid meteor spectraLeonid and Geminid spectra are dis-placed upward to facilitate comparison Note the strongtelluric absorption of oxygen and water in the ground-based spectrum of the Geminid Also note the absence ofmetal atom ablation in the fainter Perseid

JENNISKENS AND MANDELL128

TABLE 1 LINE IDENTIFICATIONS

^ intensity

Leonid Leonid GeminidObserved Theory vacuum Element afterglow meteor meteor

58909 5891584 Na I (D) 1036 mdash mdash58967 5897558 Na I (D) 671 mdash mdash59193 5915753 Fe I low E 5 mdash mdash59588 5958342 Fe I low E 12 mdash mdash61047 6104412 Ca I 6 3 4861251 6123912 Ca I 15 9 5861382 6138313 Fe I 2 2 4861408 6139393 Fe I 4 3 4061599 615767561584596159876 O I 7 86 1761646 6163878x3001x5462x8146 Ca I 32 16 40061712 61707506171272 Ca I 21 11 4661758 6175049 Fe I 2 2 2361909 6193271 Fe I 2 2 4262236 6223391 Fe I low E 2 2 1562325 6232450 Fe I 2 2 3562372 mdash Unidentified 4 2 2362467 6248045 Fe I 3 2 2962558 6254284 Fe I 2 2 2962831 6282353 Fe I low E 24 2 3763028 63032416304237 Fe I 21 2 4363201 6317052976524438 (19856) Fe I (Ca I) 2 5 2663386 63370806338575 Fe I 2 2 1963499 6348849 Si II 2 14 163566 6355592 Fe I low E mdash 2 3563612 6360451 Fe I low E 117 0 7863630 6363545 Ca I mdash 2 3563736 6373125 Si II 3 7 263834 mdash Unidentified 6 2 1763871 mdash Unidentified 6 2 264030 6402083 Fe I low E 18 2 1264127 64097876413419 Fe I 2 2 3264218 6423124 Fe I 2 2 2664406 6440855 Ca I 28 20 3264428 6442718 N I 2 10 mdash64525 6451592 Ca I 10 8 9564579 645776456264553 O I 5 43 6964652 6464351 Ca I 30 18 2064738 6473450 Ca I 7 4 2964857 648660064844906485545 N I 6 42 264955 6495576 Ca I 19 8 2065020 6500733 Fe I low E 17 2 9065297 mdash Unidentified 2 2 3565459 6549381 Fe I low E 2 2 1565653 6564636 H I 6 15 9065757 6574595 Ca I 760 90a 2065955 65947346595692 Fe I 2 2 2366115 6611502 Fe I low E 2 3 2067194 6719536 Ca I 11 10 1467260 67244666725309 N I 7 12 1071512 7150120 Ca I 15 15 2373277 7328163 Ca I mdash mdash 1774250 7425686 N I mdash mdash 1474433 7444348 N I mdash mdash 4074700 7470369 N I mdash mdash 36

These are vacuum wavelengths Integrated line intensities are in arbitrary units after correction for instrument response The line in boldface type is the Ha line discussed in the text

aContaminated by afterglow

Wavelength

scale is accurate to about half a pixel 6007 nmAt the altitude of the observations (37000 ft) therewas no detectable atmospheric water vapor ab-sorption Many of the weaker features seen in thespectrum are due to vibrational and rotationalstructure in the underlying First Positive band ofN2

We found that the expected position of Ha

(65646 nm) did not align with the brightest emis-sion line identified as the intercombination65746 nm line of calcium (Poole 1979) but witha fainter emission line just next to it Figure 3shows the line profiles from row averages at po-sitions where the meteor was before and afteropening of the shutter The dashed line shows theline positions of Ha and Ca I convolved with theinstrumental line profile A good match is ob-tained after scaling the line intensities to matchthe observed lines The line intensity ratio HCaequalled 017

Corroborating evidence for the presence ofboth Ha and the Ca I line comes from the spec-trum of an intrinsically slower Geminid (363 606 kms rather than 716 6 04 kms) shown inFig 4 A comparison of the details of the raw datanear the Ha line in Figs 3 and 4 reveal many sim-ilarities Though the meteor was not observed ina wide-field imager it must have been locatednear the horizon judging from the presence of thestrong telluric oxygen O2 B-band (686ndash688 nm)absorption in this spectrum (Fig 5) The inte-grated intensity over the final part of the Gemi-nid spectrum is about 11 magnitude fainter thanthat for the Leonid fireball The line intensity ra-tio HCa equalled 041 (Fig 3)

The Geminid spectrum is rich in metal atomemission lines Air plasma emissions were alsodetected The exposure was started before the me-teorrsquos peak brightness At positions just before theposition of the meteor at the onset of exposure(upper part of Fig 3) afterglow emission was ob-served Geminids are not known to have flareslike the Leonids but some more-rapid-than-ex-pected variations of the light curve can be recog-nized shortly after the beginning of the exposureThe Geminid produced a repetitive series of darkbands that are less well separated than in theldquoHardquo meteor indicative of rotation of the mete-oroid at a rate of about 40 cycless

No such calcium or hydrogen emission was ob-served in a fainter 23 magnitude Perseid meteorobserved at 103526 UT (August 13 1999) asshown in Fig 5 (Jenniskens et al 2000a) The Per-

seid meteor was detected too by a second cam-era and appeared high in the sky As expected ithas less telluric O2 B-band absorption The spec-trum has a lower spectral resolution due to a lessfavorable orientation of the meteor trajectory rel-ative to the dispersion direction of the gratingBecause of the much weaker metal atom ablationlines this spectrum illustrates the thermal emis-sions from oxygen and nitrogen atoms Note thatthis spectrum too has an Ha line as strong as thatin the Leonid and Geminid spectra as well as afaint calcium intercombination line with inten-sity ratio HCa equal to 069 (Fig 3)

Temporal variation

The temporal variation of various representa-tive emission lines in the ldquoHardquo Leonid is shownin Fig 6 In the afterglow phase neutral metalatoms of sodium iron and calcium which derivefrom widely varying volatile minerals but havesimilar excitation energies follow initially thesame exponential increase in intensity Sodiumremains in the same proportion to iron and cal-cium during the afterglow implying that therewas no differential ablation in this part of the tra-jectory Minerals containing sodium did not ab-late earlier than the less volatile minerals con-taining calcium The light curve reflects theexponential increase in brightness of the meteorleading up to a flare that occurred just beforeopening of the shutter

The ldquoHardquo meteorrsquos spectrum is dominated byair plasma emissions from nitrogen moleculesand atoms but metal atom ablation lines are alsoobserved (Table 1) Low excitation lines of iron(Ei 4 eV) are comparatively weak but calciumand sodium lines are stronger than in the after-glow earlier in the trajectory This suggests thata significant part of the calcium emission is fromthe meteor phase For T 5 4400 K we calculatedan abundance ratio of [Ca][Fe] equal to 013 Ingood agreement with this finding is the result ofRietmeijer and Nuth (2000) who calculated thatcomet 1PHalleyrsquos dust has a [Ca][Fe] ratio of012 Hence in this particular meteor the calciumwas ablated as efficiently as iron The decrease oflight from the strong sodium and calcium lines isexponential and follows the same slope as the airplasma emissions reflecting a weakening of themeteor after the earlier flare The ionized siliconemission has an interesting flare at row 15 (Fig7) that was not seen in other lines including the

HYDROGEN EMISSION IN METEORS 129

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

METHODS

Until now the success rate of capturing good-quality meteor spectra with traditional photo-graphic techniques has been low typically onlyone good spectrum per year per camera (Mill-man 1980) The photographic cameras in useneed a rare very bright fireball (25 magnitude orbrighter) to provide a good spectrum Key to in-creasing the sensitivity to more common faintermeteors and still maintaining a high spectral res-olution is to use cooled charge coupled device(CCD) cameras Cooled CCD cameras are about5 magnitudes more sensitive than photographicplates The instrumental layout and responsefunctions of our two-stage thermoelectricallycooled slit-less CCD spectrograph is discussedelsewhere in this issue (Jenniskens et al 2004a)By purposely limiting the spectral range to about100 nm over the 1024 3 1024 pixel CCD rangea spectral resolution is obtained of Dl 5 054 nmin first order which is similar to or better thanthat of traditional photographic slit-less spectro-graphs

RESULTS

The observation of meteor storms further in-creases the detection rate During the November2001 Leonid storm the instrument was deployedbehind an optically flat BK7-type glass windowonboard the US Air Force418th Flight TestSquadron-operated NKC-135 ldquoFISTArdquo researchaircraft from an elevation of 37000 ft This was a

single-plane mission en route from Alabama toCalifornia supported by ground stations in Ari-zona (Jenniskens and Russell 2003) Additionalspectra were obtained during a ground-based ob-serving run at Fremont Peak Observatory in Cal-ifornia at the time of the December 12 and 132001 Geminid shower

At 105232 UT (November 18 2001) a partic-ularly strong emission line close to the expectedHa line was detected in the spectrum of a 28 mag-nitude Leonid fireball known by the data ana-lyzers as ldquoHardquo after it was recognized that Ha

emission was present This was the brightest me-teor detected with this instrument during thatmission The ldquoHardquo fireball was recorded too bya low-resolution intensified video camera (Fig 1)After a gradual brightening a strong flare wasobserved and a small fragment continued tolower altitudes in a pattern that was similar to thewell-documented ldquoY2Krdquo fireball (BorovicIuml ka andJenniskens 2000 Jenniskens and Rairden 2000)A persistent train was seen for 12 min The me-teor appeared in the constellation Draco andpeaked at an elevation of 21deg above the horizon10deg east from north when the plane was at thecoordinates 10804 W 3500 N south of GallupNM Unfortunately this was just out of reach ofour ground stations in Arizona and no trajectorydata are available From the altitude of the ldquoY2Krdquopersistent train (92ndash79 km with a peak at 84 km)we surmise that the ldquoHardquo meteor peaked also ator just above 84 km and penetrated to an altitudebetween 79 and 84 km

The first order spectrum is shown in Fig 2 Justafter the sodium line intensity peaked there was

HYDROGEN EMISSION IN METEORS 125

FIG 1 The 105232 UT Leonid meteor as recorded by an intensified video camera The field of view is about 15 315deg From left to right (a) the meteor at its peak brightness (t 5 0 s) (b) the afterglow (bottom) and the 557 nm O Iwake (top) immediately following the meteor in an average of four frames (t 5 016 s) and (c) the persistent train 6s after the meteor in an average of 8 frames

an abrupt change of the spectrum and high ex-citation lines appeared This was the position ofthe meteoroid when the camera shutter openedand the meteor itself came in view At positionswhere the meteoroid was prior to this point therecorded spectrum was that of the meteor after-glow (top part of Fig 2) Although afterglow andmeteor were integrated for 0800 s the afterglowat positions earlier along the trajectory had de-cayed more than positions closer to the meteoroidat the start of the exposure The integration timewas not long enough to detect any chemilumi-nescence from a persistent train At positionswhere the meteoroid was seen while the shutterwas open the spectrum is dominated by the me-teor itself which persists for 002 s with a mi-nor contribution from the afterglow significantonly in the sodium lines and the strong inter-combination line of Ca I at 657 nm The onset ofthe meteor exposure was slightly earlier at thecenter of the plate because the meteoroid was inmotion while the shutter opened starting fromthe center There were periodic variations in thelight intensity along the trajectory most likelydue to a modulation of vapor production with therotation of the meteoroid at a rate of about 46 cycless

The spectrum of meteor and afterglow are dif-ferent as described in detail for the ldquoY2Krdquo Leonidfireball in BorovicIuml ka and Jenniskens (2000) Inshort the meteor spectrum consists predomi-nantly of allowed transitions of atoms and ionsof both low excitation (eg lines of Na I Fe I CaII) and high excitation (eg O I N I H I Si II)The afterglow only has low excitation lines butthis includes also excitation lines with low tran-sition probability (eg Ca I at 657 nm Fe I at 636nm and many others) which are very faint in thespectrum of the fireball itself

The late start of the exposure was fortunate Itenabled the separation of the fireball and after-glow spectrum This is normally possible only forvideo spectra (low resolution) and photographicspectra with a rotating shutter Moreover thefireball spectrum would have been heavily over-exposed at the position of the flare

Line identification

Because of the small detector size the wave-length scale is almost linear and wavelength cal-ibrations can be performed by matching knownlines in the spectrum Lines were identified bycalibrating the vacuum wavelength of the sodium

JENNISKENS AND MANDELL126

FIG 2 First order spectrum of the 105232 UT ldquoHardquo Leonid meteor in the range 588ndash720 nm The spectrum hasbeen expanded by a factor of 4 in vertical direction to undo the effect of binning and the image is shown reversedso that the wavelength scale runs left to right The meteor moved from top to bottom The main atomic lines andbands are identified InsetAn enlargement of the center part of the spectrum showing the hydrogen emission in theafterglow and along the length of the meteor spectrum

doublet to the theoretical wavelength scale (Jen-niskens et al 2004a) Any Doppler shift fromresidual meteor motion is negligibly small andsystematic Subsequently all of the emissions ex-pected from neutral metal atoms and singly ion-ized elements were matched to the observed line

positions For the first order there is no orderoverlap below about 740 nm due to a fast dropof optical transmittance below 370 nm

We found that many strong emission lines inthis wavelength range are due to neutral calciumWith that information in hand many other lineswere also identified (Table 1) The wavelength

HYDROGEN EMISSION IN METEORS 127

FIG 3 Extracted Ha and calcium emission linesShown are average Leonid spectra from afterglow andmeteor The comparison spectra are the Geminid and Per-seid meteors discussed in the text Dashed lines depict thebackground emission from molecular nitrogen Dottedlines in the middle two spectra are the fitted line profilesfor H and Ca emission respectively which are based onthe line positions and the instrumental line profile andare matched to fit the observed peak intensity of the lines

FIG 4 Hydrogen emission in the Gem-inid of 071528 UT (December 14 2001)from a ground-based observation fromFremont Peak Observatory in CaliforniaThe horizontal line is a star spectrum anddots are first order star images

FIG 5 Comparison of the Leonid Geminid and Per-seid meteor spectraLeonid and Geminid spectra are dis-placed upward to facilitate comparison Note the strongtelluric absorption of oxygen and water in the ground-based spectrum of the Geminid Also note the absence ofmetal atom ablation in the fainter Perseid

JENNISKENS AND MANDELL128

TABLE 1 LINE IDENTIFICATIONS

^ intensity

Leonid Leonid GeminidObserved Theory vacuum Element afterglow meteor meteor

58909 5891584 Na I (D) 1036 mdash mdash58967 5897558 Na I (D) 671 mdash mdash59193 5915753 Fe I low E 5 mdash mdash59588 5958342 Fe I low E 12 mdash mdash61047 6104412 Ca I 6 3 4861251 6123912 Ca I 15 9 5861382 6138313 Fe I 2 2 4861408 6139393 Fe I 4 3 4061599 615767561584596159876 O I 7 86 1761646 6163878x3001x5462x8146 Ca I 32 16 40061712 61707506171272 Ca I 21 11 4661758 6175049 Fe I 2 2 2361909 6193271 Fe I 2 2 4262236 6223391 Fe I low E 2 2 1562325 6232450 Fe I 2 2 3562372 mdash Unidentified 4 2 2362467 6248045 Fe I 3 2 2962558 6254284 Fe I 2 2 2962831 6282353 Fe I low E 24 2 3763028 63032416304237 Fe I 21 2 4363201 6317052976524438 (19856) Fe I (Ca I) 2 5 2663386 63370806338575 Fe I 2 2 1963499 6348849 Si II 2 14 163566 6355592 Fe I low E mdash 2 3563612 6360451 Fe I low E 117 0 7863630 6363545 Ca I mdash 2 3563736 6373125 Si II 3 7 263834 mdash Unidentified 6 2 1763871 mdash Unidentified 6 2 264030 6402083 Fe I low E 18 2 1264127 64097876413419 Fe I 2 2 3264218 6423124 Fe I 2 2 2664406 6440855 Ca I 28 20 3264428 6442718 N I 2 10 mdash64525 6451592 Ca I 10 8 9564579 645776456264553 O I 5 43 6964652 6464351 Ca I 30 18 2064738 6473450 Ca I 7 4 2964857 648660064844906485545 N I 6 42 264955 6495576 Ca I 19 8 2065020 6500733 Fe I low E 17 2 9065297 mdash Unidentified 2 2 3565459 6549381 Fe I low E 2 2 1565653 6564636 H I 6 15 9065757 6574595 Ca I 760 90a 2065955 65947346595692 Fe I 2 2 2366115 6611502 Fe I low E 2 3 2067194 6719536 Ca I 11 10 1467260 67244666725309 N I 7 12 1071512 7150120 Ca I 15 15 2373277 7328163 Ca I mdash mdash 1774250 7425686 N I mdash mdash 1474433 7444348 N I mdash mdash 4074700 7470369 N I mdash mdash 36

These are vacuum wavelengths Integrated line intensities are in arbitrary units after correction for instrument response The line in boldface type is the Ha line discussed in the text

aContaminated by afterglow

Wavelength

scale is accurate to about half a pixel 6007 nmAt the altitude of the observations (37000 ft) therewas no detectable atmospheric water vapor ab-sorption Many of the weaker features seen in thespectrum are due to vibrational and rotationalstructure in the underlying First Positive band ofN2

We found that the expected position of Ha

(65646 nm) did not align with the brightest emis-sion line identified as the intercombination65746 nm line of calcium (Poole 1979) but witha fainter emission line just next to it Figure 3shows the line profiles from row averages at po-sitions where the meteor was before and afteropening of the shutter The dashed line shows theline positions of Ha and Ca I convolved with theinstrumental line profile A good match is ob-tained after scaling the line intensities to matchthe observed lines The line intensity ratio HCaequalled 017

Corroborating evidence for the presence ofboth Ha and the Ca I line comes from the spec-trum of an intrinsically slower Geminid (363 606 kms rather than 716 6 04 kms) shown inFig 4 A comparison of the details of the raw datanear the Ha line in Figs 3 and 4 reveal many sim-ilarities Though the meteor was not observed ina wide-field imager it must have been locatednear the horizon judging from the presence of thestrong telluric oxygen O2 B-band (686ndash688 nm)absorption in this spectrum (Fig 5) The inte-grated intensity over the final part of the Gemi-nid spectrum is about 11 magnitude fainter thanthat for the Leonid fireball The line intensity ra-tio HCa equalled 041 (Fig 3)

The Geminid spectrum is rich in metal atomemission lines Air plasma emissions were alsodetected The exposure was started before the me-teorrsquos peak brightness At positions just before theposition of the meteor at the onset of exposure(upper part of Fig 3) afterglow emission was ob-served Geminids are not known to have flareslike the Leonids but some more-rapid-than-ex-pected variations of the light curve can be recog-nized shortly after the beginning of the exposureThe Geminid produced a repetitive series of darkbands that are less well separated than in theldquoHardquo meteor indicative of rotation of the mete-oroid at a rate of about 40 cycless

No such calcium or hydrogen emission was ob-served in a fainter 23 magnitude Perseid meteorobserved at 103526 UT (August 13 1999) asshown in Fig 5 (Jenniskens et al 2000a) The Per-

seid meteor was detected too by a second cam-era and appeared high in the sky As expected ithas less telluric O2 B-band absorption The spec-trum has a lower spectral resolution due to a lessfavorable orientation of the meteor trajectory rel-ative to the dispersion direction of the gratingBecause of the much weaker metal atom ablationlines this spectrum illustrates the thermal emis-sions from oxygen and nitrogen atoms Note thatthis spectrum too has an Ha line as strong as thatin the Leonid and Geminid spectra as well as afaint calcium intercombination line with inten-sity ratio HCa equal to 069 (Fig 3)

Temporal variation

The temporal variation of various representa-tive emission lines in the ldquoHardquo Leonid is shownin Fig 6 In the afterglow phase neutral metalatoms of sodium iron and calcium which derivefrom widely varying volatile minerals but havesimilar excitation energies follow initially thesame exponential increase in intensity Sodiumremains in the same proportion to iron and cal-cium during the afterglow implying that therewas no differential ablation in this part of the tra-jectory Minerals containing sodium did not ab-late earlier than the less volatile minerals con-taining calcium The light curve reflects theexponential increase in brightness of the meteorleading up to a flare that occurred just beforeopening of the shutter

The ldquoHardquo meteorrsquos spectrum is dominated byair plasma emissions from nitrogen moleculesand atoms but metal atom ablation lines are alsoobserved (Table 1) Low excitation lines of iron(Ei 4 eV) are comparatively weak but calciumand sodium lines are stronger than in the after-glow earlier in the trajectory This suggests thata significant part of the calcium emission is fromthe meteor phase For T 5 4400 K we calculatedan abundance ratio of [Ca][Fe] equal to 013 Ingood agreement with this finding is the result ofRietmeijer and Nuth (2000) who calculated thatcomet 1PHalleyrsquos dust has a [Ca][Fe] ratio of012 Hence in this particular meteor the calciumwas ablated as efficiently as iron The decrease oflight from the strong sodium and calcium lines isexponential and follows the same slope as the airplasma emissions reflecting a weakening of themeteor after the earlier flare The ionized siliconemission has an interesting flare at row 15 (Fig7) that was not seen in other lines including the

HYDROGEN EMISSION IN METEORS 129

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

an abrupt change of the spectrum and high ex-citation lines appeared This was the position ofthe meteoroid when the camera shutter openedand the meteor itself came in view At positionswhere the meteoroid was prior to this point therecorded spectrum was that of the meteor after-glow (top part of Fig 2) Although afterglow andmeteor were integrated for 0800 s the afterglowat positions earlier along the trajectory had de-cayed more than positions closer to the meteoroidat the start of the exposure The integration timewas not long enough to detect any chemilumi-nescence from a persistent train At positionswhere the meteoroid was seen while the shutterwas open the spectrum is dominated by the me-teor itself which persists for 002 s with a mi-nor contribution from the afterglow significantonly in the sodium lines and the strong inter-combination line of Ca I at 657 nm The onset ofthe meteor exposure was slightly earlier at thecenter of the plate because the meteoroid was inmotion while the shutter opened starting fromthe center There were periodic variations in thelight intensity along the trajectory most likelydue to a modulation of vapor production with therotation of the meteoroid at a rate of about 46 cycless

The spectrum of meteor and afterglow are dif-ferent as described in detail for the ldquoY2Krdquo Leonidfireball in BorovicIuml ka and Jenniskens (2000) Inshort the meteor spectrum consists predomi-nantly of allowed transitions of atoms and ionsof both low excitation (eg lines of Na I Fe I CaII) and high excitation (eg O I N I H I Si II)The afterglow only has low excitation lines butthis includes also excitation lines with low tran-sition probability (eg Ca I at 657 nm Fe I at 636nm and many others) which are very faint in thespectrum of the fireball itself

The late start of the exposure was fortunate Itenabled the separation of the fireball and after-glow spectrum This is normally possible only forvideo spectra (low resolution) and photographicspectra with a rotating shutter Moreover thefireball spectrum would have been heavily over-exposed at the position of the flare

Line identification

Because of the small detector size the wave-length scale is almost linear and wavelength cal-ibrations can be performed by matching knownlines in the spectrum Lines were identified bycalibrating the vacuum wavelength of the sodium

JENNISKENS AND MANDELL126

FIG 2 First order spectrum of the 105232 UT ldquoHardquo Leonid meteor in the range 588ndash720 nm The spectrum hasbeen expanded by a factor of 4 in vertical direction to undo the effect of binning and the image is shown reversedso that the wavelength scale runs left to right The meteor moved from top to bottom The main atomic lines andbands are identified InsetAn enlargement of the center part of the spectrum showing the hydrogen emission in theafterglow and along the length of the meteor spectrum

doublet to the theoretical wavelength scale (Jen-niskens et al 2004a) Any Doppler shift fromresidual meteor motion is negligibly small andsystematic Subsequently all of the emissions ex-pected from neutral metal atoms and singly ion-ized elements were matched to the observed line

positions For the first order there is no orderoverlap below about 740 nm due to a fast dropof optical transmittance below 370 nm

We found that many strong emission lines inthis wavelength range are due to neutral calciumWith that information in hand many other lineswere also identified (Table 1) The wavelength

HYDROGEN EMISSION IN METEORS 127

FIG 3 Extracted Ha and calcium emission linesShown are average Leonid spectra from afterglow andmeteor The comparison spectra are the Geminid and Per-seid meteors discussed in the text Dashed lines depict thebackground emission from molecular nitrogen Dottedlines in the middle two spectra are the fitted line profilesfor H and Ca emission respectively which are based onthe line positions and the instrumental line profile andare matched to fit the observed peak intensity of the lines

FIG 4 Hydrogen emission in the Gem-inid of 071528 UT (December 14 2001)from a ground-based observation fromFremont Peak Observatory in CaliforniaThe horizontal line is a star spectrum anddots are first order star images

FIG 5 Comparison of the Leonid Geminid and Per-seid meteor spectraLeonid and Geminid spectra are dis-placed upward to facilitate comparison Note the strongtelluric absorption of oxygen and water in the ground-based spectrum of the Geminid Also note the absence ofmetal atom ablation in the fainter Perseid

JENNISKENS AND MANDELL128

TABLE 1 LINE IDENTIFICATIONS

^ intensity

Leonid Leonid GeminidObserved Theory vacuum Element afterglow meteor meteor

58909 5891584 Na I (D) 1036 mdash mdash58967 5897558 Na I (D) 671 mdash mdash59193 5915753 Fe I low E 5 mdash mdash59588 5958342 Fe I low E 12 mdash mdash61047 6104412 Ca I 6 3 4861251 6123912 Ca I 15 9 5861382 6138313 Fe I 2 2 4861408 6139393 Fe I 4 3 4061599 615767561584596159876 O I 7 86 1761646 6163878x3001x5462x8146 Ca I 32 16 40061712 61707506171272 Ca I 21 11 4661758 6175049 Fe I 2 2 2361909 6193271 Fe I 2 2 4262236 6223391 Fe I low E 2 2 1562325 6232450 Fe I 2 2 3562372 mdash Unidentified 4 2 2362467 6248045 Fe I 3 2 2962558 6254284 Fe I 2 2 2962831 6282353 Fe I low E 24 2 3763028 63032416304237 Fe I 21 2 4363201 6317052976524438 (19856) Fe I (Ca I) 2 5 2663386 63370806338575 Fe I 2 2 1963499 6348849 Si II 2 14 163566 6355592 Fe I low E mdash 2 3563612 6360451 Fe I low E 117 0 7863630 6363545 Ca I mdash 2 3563736 6373125 Si II 3 7 263834 mdash Unidentified 6 2 1763871 mdash Unidentified 6 2 264030 6402083 Fe I low E 18 2 1264127 64097876413419 Fe I 2 2 3264218 6423124 Fe I 2 2 2664406 6440855 Ca I 28 20 3264428 6442718 N I 2 10 mdash64525 6451592 Ca I 10 8 9564579 645776456264553 O I 5 43 6964652 6464351 Ca I 30 18 2064738 6473450 Ca I 7 4 2964857 648660064844906485545 N I 6 42 264955 6495576 Ca I 19 8 2065020 6500733 Fe I low E 17 2 9065297 mdash Unidentified 2 2 3565459 6549381 Fe I low E 2 2 1565653 6564636 H I 6 15 9065757 6574595 Ca I 760 90a 2065955 65947346595692 Fe I 2 2 2366115 6611502 Fe I low E 2 3 2067194 6719536 Ca I 11 10 1467260 67244666725309 N I 7 12 1071512 7150120 Ca I 15 15 2373277 7328163 Ca I mdash mdash 1774250 7425686 N I mdash mdash 1474433 7444348 N I mdash mdash 4074700 7470369 N I mdash mdash 36

These are vacuum wavelengths Integrated line intensities are in arbitrary units after correction for instrument response The line in boldface type is the Ha line discussed in the text

aContaminated by afterglow

Wavelength

scale is accurate to about half a pixel 6007 nmAt the altitude of the observations (37000 ft) therewas no detectable atmospheric water vapor ab-sorption Many of the weaker features seen in thespectrum are due to vibrational and rotationalstructure in the underlying First Positive band ofN2

We found that the expected position of Ha

(65646 nm) did not align with the brightest emis-sion line identified as the intercombination65746 nm line of calcium (Poole 1979) but witha fainter emission line just next to it Figure 3shows the line profiles from row averages at po-sitions where the meteor was before and afteropening of the shutter The dashed line shows theline positions of Ha and Ca I convolved with theinstrumental line profile A good match is ob-tained after scaling the line intensities to matchthe observed lines The line intensity ratio HCaequalled 017

Corroborating evidence for the presence ofboth Ha and the Ca I line comes from the spec-trum of an intrinsically slower Geminid (363 606 kms rather than 716 6 04 kms) shown inFig 4 A comparison of the details of the raw datanear the Ha line in Figs 3 and 4 reveal many sim-ilarities Though the meteor was not observed ina wide-field imager it must have been locatednear the horizon judging from the presence of thestrong telluric oxygen O2 B-band (686ndash688 nm)absorption in this spectrum (Fig 5) The inte-grated intensity over the final part of the Gemi-nid spectrum is about 11 magnitude fainter thanthat for the Leonid fireball The line intensity ra-tio HCa equalled 041 (Fig 3)

The Geminid spectrum is rich in metal atomemission lines Air plasma emissions were alsodetected The exposure was started before the me-teorrsquos peak brightness At positions just before theposition of the meteor at the onset of exposure(upper part of Fig 3) afterglow emission was ob-served Geminids are not known to have flareslike the Leonids but some more-rapid-than-ex-pected variations of the light curve can be recog-nized shortly after the beginning of the exposureThe Geminid produced a repetitive series of darkbands that are less well separated than in theldquoHardquo meteor indicative of rotation of the mete-oroid at a rate of about 40 cycless

No such calcium or hydrogen emission was ob-served in a fainter 23 magnitude Perseid meteorobserved at 103526 UT (August 13 1999) asshown in Fig 5 (Jenniskens et al 2000a) The Per-

seid meteor was detected too by a second cam-era and appeared high in the sky As expected ithas less telluric O2 B-band absorption The spec-trum has a lower spectral resolution due to a lessfavorable orientation of the meteor trajectory rel-ative to the dispersion direction of the gratingBecause of the much weaker metal atom ablationlines this spectrum illustrates the thermal emis-sions from oxygen and nitrogen atoms Note thatthis spectrum too has an Ha line as strong as thatin the Leonid and Geminid spectra as well as afaint calcium intercombination line with inten-sity ratio HCa equal to 069 (Fig 3)

Temporal variation

The temporal variation of various representa-tive emission lines in the ldquoHardquo Leonid is shownin Fig 6 In the afterglow phase neutral metalatoms of sodium iron and calcium which derivefrom widely varying volatile minerals but havesimilar excitation energies follow initially thesame exponential increase in intensity Sodiumremains in the same proportion to iron and cal-cium during the afterglow implying that therewas no differential ablation in this part of the tra-jectory Minerals containing sodium did not ab-late earlier than the less volatile minerals con-taining calcium The light curve reflects theexponential increase in brightness of the meteorleading up to a flare that occurred just beforeopening of the shutter

The ldquoHardquo meteorrsquos spectrum is dominated byair plasma emissions from nitrogen moleculesand atoms but metal atom ablation lines are alsoobserved (Table 1) Low excitation lines of iron(Ei 4 eV) are comparatively weak but calciumand sodium lines are stronger than in the after-glow earlier in the trajectory This suggests thata significant part of the calcium emission is fromthe meteor phase For T 5 4400 K we calculatedan abundance ratio of [Ca][Fe] equal to 013 Ingood agreement with this finding is the result ofRietmeijer and Nuth (2000) who calculated thatcomet 1PHalleyrsquos dust has a [Ca][Fe] ratio of012 Hence in this particular meteor the calciumwas ablated as efficiently as iron The decrease oflight from the strong sodium and calcium lines isexponential and follows the same slope as the airplasma emissions reflecting a weakening of themeteor after the earlier flare The ionized siliconemission has an interesting flare at row 15 (Fig7) that was not seen in other lines including the

HYDROGEN EMISSION IN METEORS 129

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

doublet to the theoretical wavelength scale (Jen-niskens et al 2004a) Any Doppler shift fromresidual meteor motion is negligibly small andsystematic Subsequently all of the emissions ex-pected from neutral metal atoms and singly ion-ized elements were matched to the observed line

positions For the first order there is no orderoverlap below about 740 nm due to a fast dropof optical transmittance below 370 nm

We found that many strong emission lines inthis wavelength range are due to neutral calciumWith that information in hand many other lineswere also identified (Table 1) The wavelength

HYDROGEN EMISSION IN METEORS 127

FIG 3 Extracted Ha and calcium emission linesShown are average Leonid spectra from afterglow andmeteor The comparison spectra are the Geminid and Per-seid meteors discussed in the text Dashed lines depict thebackground emission from molecular nitrogen Dottedlines in the middle two spectra are the fitted line profilesfor H and Ca emission respectively which are based onthe line positions and the instrumental line profile andare matched to fit the observed peak intensity of the lines

FIG 4 Hydrogen emission in the Gem-inid of 071528 UT (December 14 2001)from a ground-based observation fromFremont Peak Observatory in CaliforniaThe horizontal line is a star spectrum anddots are first order star images

FIG 5 Comparison of the Leonid Geminid and Per-seid meteor spectraLeonid and Geminid spectra are dis-placed upward to facilitate comparison Note the strongtelluric absorption of oxygen and water in the ground-based spectrum of the Geminid Also note the absence ofmetal atom ablation in the fainter Perseid

JENNISKENS AND MANDELL128

TABLE 1 LINE IDENTIFICATIONS

^ intensity

Leonid Leonid GeminidObserved Theory vacuum Element afterglow meteor meteor

58909 5891584 Na I (D) 1036 mdash mdash58967 5897558 Na I (D) 671 mdash mdash59193 5915753 Fe I low E 5 mdash mdash59588 5958342 Fe I low E 12 mdash mdash61047 6104412 Ca I 6 3 4861251 6123912 Ca I 15 9 5861382 6138313 Fe I 2 2 4861408 6139393 Fe I 4 3 4061599 615767561584596159876 O I 7 86 1761646 6163878x3001x5462x8146 Ca I 32 16 40061712 61707506171272 Ca I 21 11 4661758 6175049 Fe I 2 2 2361909 6193271 Fe I 2 2 4262236 6223391 Fe I low E 2 2 1562325 6232450 Fe I 2 2 3562372 mdash Unidentified 4 2 2362467 6248045 Fe I 3 2 2962558 6254284 Fe I 2 2 2962831 6282353 Fe I low E 24 2 3763028 63032416304237 Fe I 21 2 4363201 6317052976524438 (19856) Fe I (Ca I) 2 5 2663386 63370806338575 Fe I 2 2 1963499 6348849 Si II 2 14 163566 6355592 Fe I low E mdash 2 3563612 6360451 Fe I low E 117 0 7863630 6363545 Ca I mdash 2 3563736 6373125 Si II 3 7 263834 mdash Unidentified 6 2 1763871 mdash Unidentified 6 2 264030 6402083 Fe I low E 18 2 1264127 64097876413419 Fe I 2 2 3264218 6423124 Fe I 2 2 2664406 6440855 Ca I 28 20 3264428 6442718 N I 2 10 mdash64525 6451592 Ca I 10 8 9564579 645776456264553 O I 5 43 6964652 6464351 Ca I 30 18 2064738 6473450 Ca I 7 4 2964857 648660064844906485545 N I 6 42 264955 6495576 Ca I 19 8 2065020 6500733 Fe I low E 17 2 9065297 mdash Unidentified 2 2 3565459 6549381 Fe I low E 2 2 1565653 6564636 H I 6 15 9065757 6574595 Ca I 760 90a 2065955 65947346595692 Fe I 2 2 2366115 6611502 Fe I low E 2 3 2067194 6719536 Ca I 11 10 1467260 67244666725309 N I 7 12 1071512 7150120 Ca I 15 15 2373277 7328163 Ca I mdash mdash 1774250 7425686 N I mdash mdash 1474433 7444348 N I mdash mdash 4074700 7470369 N I mdash mdash 36

These are vacuum wavelengths Integrated line intensities are in arbitrary units after correction for instrument response The line in boldface type is the Ha line discussed in the text

aContaminated by afterglow

Wavelength

scale is accurate to about half a pixel 6007 nmAt the altitude of the observations (37000 ft) therewas no detectable atmospheric water vapor ab-sorption Many of the weaker features seen in thespectrum are due to vibrational and rotationalstructure in the underlying First Positive band ofN2

We found that the expected position of Ha

(65646 nm) did not align with the brightest emis-sion line identified as the intercombination65746 nm line of calcium (Poole 1979) but witha fainter emission line just next to it Figure 3shows the line profiles from row averages at po-sitions where the meteor was before and afteropening of the shutter The dashed line shows theline positions of Ha and Ca I convolved with theinstrumental line profile A good match is ob-tained after scaling the line intensities to matchthe observed lines The line intensity ratio HCaequalled 017

Corroborating evidence for the presence ofboth Ha and the Ca I line comes from the spec-trum of an intrinsically slower Geminid (363 606 kms rather than 716 6 04 kms) shown inFig 4 A comparison of the details of the raw datanear the Ha line in Figs 3 and 4 reveal many sim-ilarities Though the meteor was not observed ina wide-field imager it must have been locatednear the horizon judging from the presence of thestrong telluric oxygen O2 B-band (686ndash688 nm)absorption in this spectrum (Fig 5) The inte-grated intensity over the final part of the Gemi-nid spectrum is about 11 magnitude fainter thanthat for the Leonid fireball The line intensity ra-tio HCa equalled 041 (Fig 3)

The Geminid spectrum is rich in metal atomemission lines Air plasma emissions were alsodetected The exposure was started before the me-teorrsquos peak brightness At positions just before theposition of the meteor at the onset of exposure(upper part of Fig 3) afterglow emission was ob-served Geminids are not known to have flareslike the Leonids but some more-rapid-than-ex-pected variations of the light curve can be recog-nized shortly after the beginning of the exposureThe Geminid produced a repetitive series of darkbands that are less well separated than in theldquoHardquo meteor indicative of rotation of the mete-oroid at a rate of about 40 cycless

No such calcium or hydrogen emission was ob-served in a fainter 23 magnitude Perseid meteorobserved at 103526 UT (August 13 1999) asshown in Fig 5 (Jenniskens et al 2000a) The Per-

seid meteor was detected too by a second cam-era and appeared high in the sky As expected ithas less telluric O2 B-band absorption The spec-trum has a lower spectral resolution due to a lessfavorable orientation of the meteor trajectory rel-ative to the dispersion direction of the gratingBecause of the much weaker metal atom ablationlines this spectrum illustrates the thermal emis-sions from oxygen and nitrogen atoms Note thatthis spectrum too has an Ha line as strong as thatin the Leonid and Geminid spectra as well as afaint calcium intercombination line with inten-sity ratio HCa equal to 069 (Fig 3)

Temporal variation

The temporal variation of various representa-tive emission lines in the ldquoHardquo Leonid is shownin Fig 6 In the afterglow phase neutral metalatoms of sodium iron and calcium which derivefrom widely varying volatile minerals but havesimilar excitation energies follow initially thesame exponential increase in intensity Sodiumremains in the same proportion to iron and cal-cium during the afterglow implying that therewas no differential ablation in this part of the tra-jectory Minerals containing sodium did not ab-late earlier than the less volatile minerals con-taining calcium The light curve reflects theexponential increase in brightness of the meteorleading up to a flare that occurred just beforeopening of the shutter

The ldquoHardquo meteorrsquos spectrum is dominated byair plasma emissions from nitrogen moleculesand atoms but metal atom ablation lines are alsoobserved (Table 1) Low excitation lines of iron(Ei 4 eV) are comparatively weak but calciumand sodium lines are stronger than in the after-glow earlier in the trajectory This suggests thata significant part of the calcium emission is fromthe meteor phase For T 5 4400 K we calculatedan abundance ratio of [Ca][Fe] equal to 013 Ingood agreement with this finding is the result ofRietmeijer and Nuth (2000) who calculated thatcomet 1PHalleyrsquos dust has a [Ca][Fe] ratio of012 Hence in this particular meteor the calciumwas ablated as efficiently as iron The decrease oflight from the strong sodium and calcium lines isexponential and follows the same slope as the airplasma emissions reflecting a weakening of themeteor after the earlier flare The ionized siliconemission has an interesting flare at row 15 (Fig7) that was not seen in other lines including the

HYDROGEN EMISSION IN METEORS 129

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

JENNISKENS AND MANDELL128

TABLE 1 LINE IDENTIFICATIONS

^ intensity

Leonid Leonid GeminidObserved Theory vacuum Element afterglow meteor meteor

58909 5891584 Na I (D) 1036 mdash mdash58967 5897558 Na I (D) 671 mdash mdash59193 5915753 Fe I low E 5 mdash mdash59588 5958342 Fe I low E 12 mdash mdash61047 6104412 Ca I 6 3 4861251 6123912 Ca I 15 9 5861382 6138313 Fe I 2 2 4861408 6139393 Fe I 4 3 4061599 615767561584596159876 O I 7 86 1761646 6163878x3001x5462x8146 Ca I 32 16 40061712 61707506171272 Ca I 21 11 4661758 6175049 Fe I 2 2 2361909 6193271 Fe I 2 2 4262236 6223391 Fe I low E 2 2 1562325 6232450 Fe I 2 2 3562372 mdash Unidentified 4 2 2362467 6248045 Fe I 3 2 2962558 6254284 Fe I 2 2 2962831 6282353 Fe I low E 24 2 3763028 63032416304237 Fe I 21 2 4363201 6317052976524438 (19856) Fe I (Ca I) 2 5 2663386 63370806338575 Fe I 2 2 1963499 6348849 Si II 2 14 163566 6355592 Fe I low E mdash 2 3563612 6360451 Fe I low E 117 0 7863630 6363545 Ca I mdash 2 3563736 6373125 Si II 3 7 263834 mdash Unidentified 6 2 1763871 mdash Unidentified 6 2 264030 6402083 Fe I low E 18 2 1264127 64097876413419 Fe I 2 2 3264218 6423124 Fe I 2 2 2664406 6440855 Ca I 28 20 3264428 6442718 N I 2 10 mdash64525 6451592 Ca I 10 8 9564579 645776456264553 O I 5 43 6964652 6464351 Ca I 30 18 2064738 6473450 Ca I 7 4 2964857 648660064844906485545 N I 6 42 264955 6495576 Ca I 19 8 2065020 6500733 Fe I low E 17 2 9065297 mdash Unidentified 2 2 3565459 6549381 Fe I low E 2 2 1565653 6564636 H I 6 15 9065757 6574595 Ca I 760 90a 2065955 65947346595692 Fe I 2 2 2366115 6611502 Fe I low E 2 3 2067194 6719536 Ca I 11 10 1467260 67244666725309 N I 7 12 1071512 7150120 Ca I 15 15 2373277 7328163 Ca I mdash mdash 1774250 7425686 N I mdash mdash 1474433 7444348 N I mdash mdash 4074700 7470369 N I mdash mdash 36

These are vacuum wavelengths Integrated line intensities are in arbitrary units after correction for instrument response The line in boldface type is the Ha line discussed in the text

aContaminated by afterglow

Wavelength

scale is accurate to about half a pixel 6007 nmAt the altitude of the observations (37000 ft) therewas no detectable atmospheric water vapor ab-sorption Many of the weaker features seen in thespectrum are due to vibrational and rotationalstructure in the underlying First Positive band ofN2

We found that the expected position of Ha

(65646 nm) did not align with the brightest emis-sion line identified as the intercombination65746 nm line of calcium (Poole 1979) but witha fainter emission line just next to it Figure 3shows the line profiles from row averages at po-sitions where the meteor was before and afteropening of the shutter The dashed line shows theline positions of Ha and Ca I convolved with theinstrumental line profile A good match is ob-tained after scaling the line intensities to matchthe observed lines The line intensity ratio HCaequalled 017

Corroborating evidence for the presence ofboth Ha and the Ca I line comes from the spec-trum of an intrinsically slower Geminid (363 606 kms rather than 716 6 04 kms) shown inFig 4 A comparison of the details of the raw datanear the Ha line in Figs 3 and 4 reveal many sim-ilarities Though the meteor was not observed ina wide-field imager it must have been locatednear the horizon judging from the presence of thestrong telluric oxygen O2 B-band (686ndash688 nm)absorption in this spectrum (Fig 5) The inte-grated intensity over the final part of the Gemi-nid spectrum is about 11 magnitude fainter thanthat for the Leonid fireball The line intensity ra-tio HCa equalled 041 (Fig 3)

The Geminid spectrum is rich in metal atomemission lines Air plasma emissions were alsodetected The exposure was started before the me-teorrsquos peak brightness At positions just before theposition of the meteor at the onset of exposure(upper part of Fig 3) afterglow emission was ob-served Geminids are not known to have flareslike the Leonids but some more-rapid-than-ex-pected variations of the light curve can be recog-nized shortly after the beginning of the exposureThe Geminid produced a repetitive series of darkbands that are less well separated than in theldquoHardquo meteor indicative of rotation of the mete-oroid at a rate of about 40 cycless

No such calcium or hydrogen emission was ob-served in a fainter 23 magnitude Perseid meteorobserved at 103526 UT (August 13 1999) asshown in Fig 5 (Jenniskens et al 2000a) The Per-

seid meteor was detected too by a second cam-era and appeared high in the sky As expected ithas less telluric O2 B-band absorption The spec-trum has a lower spectral resolution due to a lessfavorable orientation of the meteor trajectory rel-ative to the dispersion direction of the gratingBecause of the much weaker metal atom ablationlines this spectrum illustrates the thermal emis-sions from oxygen and nitrogen atoms Note thatthis spectrum too has an Ha line as strong as thatin the Leonid and Geminid spectra as well as afaint calcium intercombination line with inten-sity ratio HCa equal to 069 (Fig 3)

Temporal variation

The temporal variation of various representa-tive emission lines in the ldquoHardquo Leonid is shownin Fig 6 In the afterglow phase neutral metalatoms of sodium iron and calcium which derivefrom widely varying volatile minerals but havesimilar excitation energies follow initially thesame exponential increase in intensity Sodiumremains in the same proportion to iron and cal-cium during the afterglow implying that therewas no differential ablation in this part of the tra-jectory Minerals containing sodium did not ab-late earlier than the less volatile minerals con-taining calcium The light curve reflects theexponential increase in brightness of the meteorleading up to a flare that occurred just beforeopening of the shutter

The ldquoHardquo meteorrsquos spectrum is dominated byair plasma emissions from nitrogen moleculesand atoms but metal atom ablation lines are alsoobserved (Table 1) Low excitation lines of iron(Ei 4 eV) are comparatively weak but calciumand sodium lines are stronger than in the after-glow earlier in the trajectory This suggests thata significant part of the calcium emission is fromthe meteor phase For T 5 4400 K we calculatedan abundance ratio of [Ca][Fe] equal to 013 Ingood agreement with this finding is the result ofRietmeijer and Nuth (2000) who calculated thatcomet 1PHalleyrsquos dust has a [Ca][Fe] ratio of012 Hence in this particular meteor the calciumwas ablated as efficiently as iron The decrease oflight from the strong sodium and calcium lines isexponential and follows the same slope as the airplasma emissions reflecting a weakening of themeteor after the earlier flare The ionized siliconemission has an interesting flare at row 15 (Fig7) that was not seen in other lines including the

HYDROGEN EMISSION IN METEORS 129

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

scale is accurate to about half a pixel 6007 nmAt the altitude of the observations (37000 ft) therewas no detectable atmospheric water vapor ab-sorption Many of the weaker features seen in thespectrum are due to vibrational and rotationalstructure in the underlying First Positive band ofN2

We found that the expected position of Ha

(65646 nm) did not align with the brightest emis-sion line identified as the intercombination65746 nm line of calcium (Poole 1979) but witha fainter emission line just next to it Figure 3shows the line profiles from row averages at po-sitions where the meteor was before and afteropening of the shutter The dashed line shows theline positions of Ha and Ca I convolved with theinstrumental line profile A good match is ob-tained after scaling the line intensities to matchthe observed lines The line intensity ratio HCaequalled 017

Corroborating evidence for the presence ofboth Ha and the Ca I line comes from the spec-trum of an intrinsically slower Geminid (363 606 kms rather than 716 6 04 kms) shown inFig 4 A comparison of the details of the raw datanear the Ha line in Figs 3 and 4 reveal many sim-ilarities Though the meteor was not observed ina wide-field imager it must have been locatednear the horizon judging from the presence of thestrong telluric oxygen O2 B-band (686ndash688 nm)absorption in this spectrum (Fig 5) The inte-grated intensity over the final part of the Gemi-nid spectrum is about 11 magnitude fainter thanthat for the Leonid fireball The line intensity ra-tio HCa equalled 041 (Fig 3)

The Geminid spectrum is rich in metal atomemission lines Air plasma emissions were alsodetected The exposure was started before the me-teorrsquos peak brightness At positions just before theposition of the meteor at the onset of exposure(upper part of Fig 3) afterglow emission was ob-served Geminids are not known to have flareslike the Leonids but some more-rapid-than-ex-pected variations of the light curve can be recog-nized shortly after the beginning of the exposureThe Geminid produced a repetitive series of darkbands that are less well separated than in theldquoHardquo meteor indicative of rotation of the mete-oroid at a rate of about 40 cycless

No such calcium or hydrogen emission was ob-served in a fainter 23 magnitude Perseid meteorobserved at 103526 UT (August 13 1999) asshown in Fig 5 (Jenniskens et al 2000a) The Per-

seid meteor was detected too by a second cam-era and appeared high in the sky As expected ithas less telluric O2 B-band absorption The spec-trum has a lower spectral resolution due to a lessfavorable orientation of the meteor trajectory rel-ative to the dispersion direction of the gratingBecause of the much weaker metal atom ablationlines this spectrum illustrates the thermal emis-sions from oxygen and nitrogen atoms Note thatthis spectrum too has an Ha line as strong as thatin the Leonid and Geminid spectra as well as afaint calcium intercombination line with inten-sity ratio HCa equal to 069 (Fig 3)

Temporal variation

The temporal variation of various representa-tive emission lines in the ldquoHardquo Leonid is shownin Fig 6 In the afterglow phase neutral metalatoms of sodium iron and calcium which derivefrom widely varying volatile minerals but havesimilar excitation energies follow initially thesame exponential increase in intensity Sodiumremains in the same proportion to iron and cal-cium during the afterglow implying that therewas no differential ablation in this part of the tra-jectory Minerals containing sodium did not ab-late earlier than the less volatile minerals con-taining calcium The light curve reflects theexponential increase in brightness of the meteorleading up to a flare that occurred just beforeopening of the shutter

The ldquoHardquo meteorrsquos spectrum is dominated byair plasma emissions from nitrogen moleculesand atoms but metal atom ablation lines are alsoobserved (Table 1) Low excitation lines of iron(Ei 4 eV) are comparatively weak but calciumand sodium lines are stronger than in the after-glow earlier in the trajectory This suggests thata significant part of the calcium emission is fromthe meteor phase For T 5 4400 K we calculatedan abundance ratio of [Ca][Fe] equal to 013 Ingood agreement with this finding is the result ofRietmeijer and Nuth (2000) who calculated thatcomet 1PHalleyrsquos dust has a [Ca][Fe] ratio of012 Hence in this particular meteor the calciumwas ablated as efficiently as iron The decrease oflight from the strong sodium and calcium lines isexponential and follows the same slope as the airplasma emissions reflecting a weakening of themeteor after the earlier flare The ionized siliconemission has an interesting flare at row 15 (Fig7) that was not seen in other lines including the

HYDROGEN EMISSION IN METEORS 129

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

nitrogen and oxygen lines except (more weakly)in the hydrogen line

Compared with sodium and calcium the emis-sion intensity of atmospheric N2 decreased morerapidly behind the meteor in the same rapidmanner as hydrogen Both emissions have rela-tively high excitation energies and the rapid de-crease is on account of that In previous work wecalculated the temperature decline behind themeteoroid from this rate of decay (BorovicIuml ka andJenniskens 2000 Jenniskens and Stenbaek-Nielsen 2004)

Judging from the increased ratio of N IN2emission (Jenniskens et al 2004b) the plasmawas slightly warmer in the meteor than in the af-terglow as expected (BorovicIuml ka and Jenniskens2000) At this time there was a detectable con-tinuum emission possibly because of dust or N2

1

emission (Jenniskens et al 2004c) Fitting the ob-served spectrum with a model air plasma spec-trum as in Jenniskens et al (2004bc) provided ameans to separate the continuum and N2 bandThe band intensity at the center of the image isuncertain because of what appears to be a reflec-tion of light in the window with a Gaussian dis-tribution peaking at the center of the CCD

The ratio of N IN2 is lower in the slower Gem-inid than in the Leonid corresponding to an

equivalent local thermodynamic equilibrium airplasma temperature of T 5 4355 K versus T 54470 K for the Leonid

The lower plasma temperature in the after-glow would correspond to a relatively higher N2 abundance if chemical equilibrium is main-tained Indeed the ratio of HaN2 line intensi-ties was higher in the meteor (Fig 6) The Ha linein the meteor spectrum also falls off less steeplythan the Ca I 657 nm emission with a light curvemore similar to the N2 air plasma emission Inthe same way the Ha line in the Geminid spec-trum follows the broadband emission from mol-ecular nitrogen rather than the meteoric calciumintercombination line which has a broad maxi-mum that peaks slightly later than the air plasmaemission (Fig 7)

From these results we calculated the abun-dances of hydrogen atoms in the Leonid meteorplasma and in the afterglow following themethod by BorovicIuml ka (1993) In the meteorplasma where nitrogen and oxygen atom emis-sions are best defined we calculate that [H][N] 511 6 03 3 1022 and [O][N] 5 30 6 03 (for theGeminid meteor) assuming that all emissionoriginates from a T 5 4400 K plasma The lattervalue differs from the expected value of 025 ifsimilar amounts of oxygen and nitrogen mole-cules were dissociated in the meteor plasma Thissuggests that less molecular nitrogen is dissoci-ated than expected If all oxygen molecules aredissociated then that [H][N] 5 92 6 17 3 1024In another Leonid meteor the fraction of dissoci-ated molecular oxygen was measured to be as lowas 15 (Jenniskens et al 2000b) or [H][N] 514 6 03 3 1024 Here N is the total amount of

JENNISKENS AND MANDELL130

FIG 7 As Fig 6 for the Geminid spectrum of Fig 4Note how the Ca I 657 nm line peaks later than other emis-sions

FIG 6 Light curves for representative emission linesand bands The bold line is the light curve of neutral hy-drogen atomic emission (H I 65653 nm) Metal atomemissions are neutral sodium (Na I 595 + 589 nm) neu-tral calcium (Ca I 657 nm) neutral iron (Fe I 636 nm)and ionized silicon (Si1 635 1 637 nm) The lines of ni-trogen atoms (N 649 nm) and nitrogen molecules (N2)trace the air plasma components ldquocontrdquo refers to a con-tinuum component Row numbers less than zero trace theafterglow emissions behind the meteor Row number 0corresponds to the start of the exposure The shift in theonset of meteor sodium emission found at the edge ofthe spectrum is due to the gradual opening of the cam-era shutter

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

nitrogen atoms in the plasma in atomic and mol-ecular form Because the observed transitions ofhydrogen oxygen and nitrogen atoms haveabout the same upper energy level these resultsare insensitive to the adopted plasma tempera-ture

DISCUSSION

Atmospheric sources of hydrogen

Let us first consider the possibility that the hy-drogen is due to water vapor in the ambient at-mosphere At the meteor plasma temperature of4400 K water is effectively dissociated into Hand OH and given sufficient time OH is disso-ciated into O and H (Fig 8 in Jenniskens et al2004c) From the [H][N] 5 14 6 03 3 1024 re-sult in the previous section and after taking intoaccount that 798 of all ambient atmosphericcompounds are in the form of molecular nitrogen(with 196 oxygen molecules and 06 oxy-gen atoms) we calculated that if all hydrogen wasdue to ambient water (with two hydrogen atomsper molecule) the abundance of water in allforms in the mesopause at the altitude of the me-teor (80ndash90 km) would be in excess of 165 ppm

However the water mixing ratio above 80 kmis thought to be typically a few ppm and rarelymore than 20 ppm (Conway et al 1999 Stevenset al 2001) Above 80 km water molecules are ef-fectively photolysed by ultraviolet photons re-sulting in a drop off of water abundance with al-titude which contributes to a sharp decrease inOH and HO2 abundances Above 85 km mostHO2 reacts with oxygen atoms to form OH andOH molecules react with oxygen atoms to formhydrogen atoms H2 is produced from H 1 HO2and is destroyed by reactions with oxygen atoms(Harris 2001) As a result most hydrogen is pre-sent in the form of hydrogen molecules andatoms The NCAR 1D TIME model (Roble 1995)predicts H H2 and H2O concentration of 28 65and 14 ppm at 80 km respectively and 13 54and 01 ppm at 90 km respectively OH and HO2

abundances are negligible sources of hydrogenWith two hydrogen atoms from each H2 and H2Omolecule this adds up to a hydrogen atom abun-dance of 19 and 12 ppm at 80 and 90 km altituderespectively These values are a factor of 8 and 14lower than observed After subtracting the con-tribution from H and H2 we calculated that if all

hydrogen was due to ambient water the abun-dance of water in all forms in the mesopause atthe altitude of the meteor (80ndash90 km) would bein excess of 150 6 20 ppm

In order to explain the high 150 ppm waterabundance observed in our data it is possible thatsolid water in the form of ice particles or hydratesis evaporated and dissociated rapidly during pas-sage of the meteor At the lower boundary of themesopause heavy molecules detected in rocketexperiments are thought to be metal atom hy-drates Meteoric smoke molecules such asNaHCO3 are thought to act as nucleation sites forwater clusters [NaHCO3 (H2O)n with n 100]with sizes up to 20 nm (Plane 2000) When thetemperature drops below 150 K noctilucentclouds are observed to form at around 83 km asa result of ice particle formation Noctilucentcloud particles can grow to sizes of 50ndash800 nmJust above this altitude where the particles do notgrow as large polar mesosphere summer echoesare observed that can only be explained by thepresence of similar but smaller particles Regard-ing these smaller particles which are estimatedto be less than 50 nm their presence and proper-ties are yet to be proven experimentally

Such continuous production of water mole-cules in the plasma may in fact account for theOH ARX emission at 310 nm (Harvey 1977 Abeet al 2002 Jenniskens et al 2002) However thiswould imply that much of the hydrogen in themesopause is not in the form of molecular hy-drogen but rather in the form of hydrates Theproblem with this interpretation is the limitednumber of collisions with hot air compounds thatoccur in the brief meteor phase We know of nosimulations that study the evaporation of the hy-drates in the meteor plasma but doubt that theprocess is rapid enough to evaporate the solidgrains in the brief 002 s meteor phase

Meteoric sources of hydrogen

An alternative hypothesis is that the hydrogenemission is related to the destruction of organicsor the outgassing of water bound and trapped inminerals The hydrogen emission cannot origi-nate from the same warm T 5 4400 K plasma asthe weaker iron lines (Table 1) from which wecalculated [H][Fe] 5 6 3 106 However if bothSi II and H I in the Leonid meteor are from a hot component we calculated a reasonable [H][Si] 5 4 for the ldquoHardquo meteor After adopting a

HYDROGEN EMISSION IN METEORS 131

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

bulk comet Halley abundance ratio of [Fe][Si] 5028 (Rietmeijer and Nuth 2000) this translatesto a ratio for [H][Fe] 5 14 in close agreementwith the ldquoY2Krdquo Leonid fireball result BorovicIuml kaand Jenniskens (2000) calculated an abundanceratio of [H][Fe] 5 10 for the ldquoY2Krdquo meteor

Although the hot component is hypothesizedto be in the range of about 9000ndash15000 K(BorovicIuml ka 1994a) no reliable measurement ofthe temperature of the hot component exists Ifthe hot component is assumed to be 8000 K wecalculated [H][Fe] 5 35 and at 15000 K the ra-tio of [H][Fe] equals 9 We adopt [H][Fe] 514 6 5 if all hydrogen in the spectrum of ldquoHardquoderived from the meteoroid

The hot component is a small fraction of the to-tal plasma volume If a volume of such a hot com-ponent (VHot) is responsible for the H I emissionbut a volume of iron is collisionally excited by thewarm 4400 K plasma (VWarm) and by assumingand abundance ratio [H][Fe] 5 14 in both com-ponents we find VHotVWarm 5 6 3 1024 for thismeteor if all hydrogen is meteoric in origin Suchsmall volume makes it possible that the hot com-ponent is related to the impact excitation duringthe expansion phase at the time when moleculescollide with remnant non-thermal speeds Thosecollisions involve a larger fraction of metal abla-tion products than later thermal collisions

Comet Halley has a ratio of [H][C] 5 1 in itsrefractory organic component (Greenberg 2000)and about equal amounts of refractory organicsand silicates Assuming [H][C] 5 1 we calcu-lated that [C][Si] 5 39 6 14 This is in goodagreement with the ratio of [C][Si] 5 44 mea-sured in the spectrum from bulk comet Halleydust (Rietmeijer and Nuth 2000) It appears thatthe meteoroids of comet 55PTempel-Tuttleejected in 1767 are still as rich in organic matteras the dust of comet Halley was during the Giottospacecraft flyby in 1986 Little of the refractoryorganics were lost in interplanetary space be-tween the moment of ejection in 1767 and Earthimpact in 2001

This interpretation implies that a significantfraction of the hydrogen in the refractory organ-ics is lost during ablation possibly in a processknown as carbonization that leads to the loss offunctional groups and the formation of aromaticstructures (eg Jenniskens et al 1993)

Difficulties with the proposed interpretation ofthe origin of the Ha emission line remain TheGeminid meteor does not have detectable ionized

silicon emission which makes it remarkable thatthe hydrogen line is clearly detected The Si II linehas an upper energy state of 1007 eV lower thanthe 1209 eV of Ha If the line is merely weak thenthe corresponding [C][Si] ratio would be signif-icantly higher than that of the Leonid Of coursethe Geminid meteor also had stronger iron linesthan calcium emissions which leaves open thepossibility that not all of the silicon is evaporatedAlternatively a smaller fraction of silicon couldbe in the ionized state in a Geminid meteor notconsistent with T 5 10000 K

There is also the more fundamental question asto whether such a hot component existsBorovicIuml ka (1994a) assigned the O I and N I linesto hot plasma because these lines originate fromhigh excitation levels In contrast the N2 FirstPositive band was assigned to the warm 4400 Kcomponent despite its high excitation energySince that time we have demonstrated that allmeteoric air plasma emissions match well to lab-oratory air plasma in local thermodynamic equi-librium at T 5 4400 K (Jenniskens et al 2000a2004b)

CONCLUSIONS

The Ha line was identified in the meteor spec-tra of three meteoroids originating from three dif-ferent comets The intensity of the high-excitationhydrogen emission followed the high-excitationN I and N2 air plasma emissions not the inten-sity of the low-excitation calcium iron andsodium lines in the meteor afterglow

Nearly independent of the adopted excitationtemperature we calculated from the HN line in-tensity ratio that if that hydrogen originated inthe ambient atmosphere then the hydrogen atomabundance at the altitude of the Leonid meteor(80ndash90 km) would have been in excess of 150 620 ppm a factor of about 11 6 4 above the ex-pected abundance of hydrogen in the form of H I H2 and H2O The evaporation and dissocia-tion of water clusters and small ice particles areprobably too slow to account for the difference

On the other hand if the hydrogen originatedfrom a high excitation temperature T 10000 Kplasma component perhaps from non-thermalcollisions in the expansion phase of the meteorand Si II emission originated from that sameprocess then it is possible that all of the observedhydrogen emission originated from hydrogen

JENNISKENS AND MANDELL132

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

atoms ablated from the meteoroid We measureda hydrogen abundance for the Leonid meteoroidof comet 55PTempel-Tuttle within a factor of 2from that measured in the dust of comet 1PHal-ley during the 1986 flyby

Unless there was 10 times more water vapor inthe upper atmosphere than expected all of theobserved hydrogen in meteor spectra can haveoriginated from the meteoroid either from ablat-ing refractory organic matter or from watertrapped and bound in minerals If the hydrogenoriginated from the meteoric organic matter thenthat organic matter lost a significant fraction of itshydrogen during ablation perhaps in a processknown as carbonization This would imply thatthe refractory organic matter was not lost in thevacuum of space since the meteoroid was ejectedby the comet in 1767

ACKNOWLEDGMENTS

This paper benefited greatly from commentsby referees Jiri BorovicIuml ka (Ondrejov ObservatoryCzech Republic) and Shinsuke Abe (ISAS Japan)NASA Ames Astrobiology Academy studentEmily Schaller operated the CCD spectrographduring the 2001 Leonid MAC mission The in-strument was developed with support of NASAAmes Research Centerrsquos Directorrsquos DiscretionaryFund We thank Mike Koop Mike Wilson andGary Palmer for technical support Mike Koopalso supported the ground-based 2001 Geminidcampaign The 2001 Leonid MAC mission wassponsored by NASArsquos Astrobiology ASTID andPlanetary Astronomy programs and executed bythe USAF 418th Flight Test Squadron GregSchmidt was the technical monitor at NASAAmes Research Center This work meets with ob-jective 31 of NASArsquos Astrobiology Roadmap(Des Marais et al 2003)

ABBREVIATIONS

CCD charge coupled device Ha hydrogenBalmer-alpha MAC Multi-Instrument AircraftCampaign UT universal time

REFERENCES

Abe S Yano H Ebizuka N Kasuga T Watanabe J-I Sugimoto M Fujino N Fuse T and Ogasawara

R (2002) First results of OH emission from meteor andafter glow Search for organics in cometary meteoroidsIn ESA SP-500 Proceedings of Asteroids Comets Meteors(ACM 2002) ESA Publications Division NoordwijkThe Netherlands pp 213ndash216

BorovicIuml ka J (1993) A fireball spectrum analysis AstronAstrophys 279 627ndash645

BorovicIuml ka J (1994a) Two components in meteor spectraPlanet Space Sci 42 145ndash150

BorovicIuml ka J (1994b) Line identification in a fireball spec-trum Astron Astrophys Suppl Ser 103 83ndash96

BorovicIuml ka J and Jenniskens P (2000) Time resolvedspectroscopy of a Leonid fireball afterglow Earth MoonPlanets 82-83 399ndash428

Brack A (1999) Life in the Solar System Adv Space Res24 417ndash433

Ceplecha Z (1971) Spectral data on terminal flare andwake of double-station meteor No 38421 (OndrejovApril 21 1963) Bull Astron Inst Czech 22 219ndash304

Chyba CF and Sagan C (1992) Endogenous productionexogenous delivery and impact shock synthesis of or-ganic molecules An inventory for the origins of lifeNature 355 125ndash132

Chyba CF and Sagan C (1997) Comets as a source ofprebiotic organic molecules for the early Earth InComets and the Origin and Evolution of Life edited by PJThomas CF Chyba and CP McKay Springer VerlagNew York pp 147ndash173

Conway RR Stevens MH Brown CM Cardon JGZasadil SE and Mount GH (1999) The Middle At-mosphere High Resolution Spectrograph InvestigationJ Geophys Res 104 16327ndash16348

Cook AF and Millman PM (1955) Photometric analy-sis of a spectrogram of a Perseid meteor Astrophys J121 250ndash270

Delsemme AH (1992) Cometary origin of carbon nitro-gen and water on the Earth Origins Life 21 279ndash298

Des Marais DJ Allamandola LJ Benner SA DeamerD Falkowski PG Farmer JD Hedges SB JakoskyBM Knoll AH Liskowsky DR Meadows VSMeyer MA Pilcher CB Nealson KH SpormannAM Trent JD Turner WW Woolf NJ and YorkeHW (2003) The NASA Astrobiology Roadmap Astro-biology 3 219ndash235

Greenberg JM (2000) From comets to meteors EarthMoon Planets 82-83 313ndash324

Harris MJ (2001) A new coupled middle atmosphereand thermosphere general circulation model Studies ofdynamic energetic and photochemical coupling in themiddle and upper atmosphere [PhD Thesis] Depart-ment of Physics and Astronomy University CollegeLondon London

Harvey GA (1977) A search for ultraviolet OH emissionfrom meteors Astrophys J 217 688ndash690

Jenniskens P and Rairden RL (2000) Buoyancy of theldquoY2Krdquo persistent train and the trajectory of the 040029UT Leonid fireball Earth Moon Planets 82-83 457ndash470

Jenniskens P and Russell RW (2003) The 2001 LeonidMulti-Instrument Aircraft Campaignmdashan early reviewISAS SP 15 3ndash15

HYDROGEN EMISSION IN METEORS 133

Jenniskens P and Stenbaek-Nielsen HC (2004) Meteorwake in high frame-rate imagesmdashimplications for thechemistry of ablated organic compounds Astrobiology4 95ndash108

Jenniskens P Baratta GA Kouchi A de Groot MSGreenberg JM and Strazzulla G (1993) Carbon dustformation on interstellar grains Astron Astrophys 273583ndash600

Jenniskens P Wilson MA Packan D Laux COBoyd ID Popova OP and Fonda M (2000a) Mete-ors A delivery mechanism of organic matter to theearly Earth Earth Moon Planets 82-83 57ndash70

Jenniskens P Lacey M Allan BJ Self DE and PlaneJMC (2000b) The dynamical evolution of a tubularLeonid persistent train Earth Moon Planets 82-83471ndash488

Jenniskens P Tedesco E Murthy J Laux CO andPrice S (2002) Spaceborne ultraviolet 251ndash384 nm spec-troscopy of a meteor during the 1997 Leonid showerMeteoritics Planet Sci 37 1071ndash1078

Jenniskens P Schaller EL Laux CO Wilson MASchmidt G and Rairden RL (2004a) Meteors do notbreak exogenous organic molecules into high yields ofdiatomics Astrobiology 4 67ndash79

Jenniskens P Laux CO Wilson MA and SchallerEL (2004b) The mass and speed dependence of meteorair plasma temperatures Astrobiology 4 81ndash94

Jenniskens P Laux CO and Schaller EL (2004c)Search for the OH (X2P) Meinel band emission in me-teors as a tracer of mineral water in comets Detectionof N2

1(A-X) Astrobiology 4 109ndash121Kondratrsquoeva ED and Reznikov EA (1985) Comet Tem-

pel-Tuttle and the Leonid meteor swarm Solar SystemRes 19 96ndash101

Lellouch E Crovisier J Lim T Bockelee-Morvan DLeech K Hanner M S Altieri B Schmitt B TrottaF and Keller HU (1998) Evidence for water ice andestimate of dust production rate in comet Hale-Bopp at29 AU from the Sun Astron Astrophys 339 L9ndashL12

Maurette M Duprat J Engrand C Gounelle M Ku-rat G Matrajt G and Toppani A (2000) Accretion ofneon organics CO2 nitrogen and water from large in-terplanetary dust particles on the early Earth PlanetSpace Sci 48 1117ndash1137

Millman P (1980) One hundred and fifteen years of me-teor spectroscopy In Solid Particles in the Solar Systemedited by I Halliday and BA McIntosh D Reidel Pub-lishing Co Dordrecht The Netherlands pp 121ndash128

Oroacute J (1961) Comets and the formation of biochemicalcompounds on the primitive Earth Nature 190 389ndash390

Plane JMC (2000) The role of sodium bicarbonate in thenucleation of noctilucent clouds Ann Geophys 18807ndash814

Poole LMG (1979) The excitation of spectral lines infaint meteor trains J Atmospheric Terr Phys 41 53ndash64

Rietmeijer FJM and Nuth JA (2000) Collected ex-traterrestrial materials Constraints on meteor andfireball compositions Earth Moon Planets 82-83325ndash350

Roble R (1995) Geophysical Monograph 87 Energetics of theMesosphere and Thermosphere Upper Mesosphere andLower Thermosphere A Review of Experiment and TheoryAmerican Geophysical Union Washington DC

Stevens MH Conway RR Englert CR SummersME Grossman KU and Gusev OA (2001) PMCsand the water frost point in the Arctic summer meso-sphere Geophys Res Lett 28 4449ndash4452

Address reprint requests toDr Peter Jenniskens

Center for the Study of Life in the UniverseSETI Institute

2035 Landings DriveMountain View CA 94043

E-mail pjenniskensmailarcnasagov

JENNISKENS AND MANDELL134