H3+ is the simplest polyatomic molecule and plays a pivotal role in interstellar chemistry. It is produced abundantly via cosmic ray ionization of H2 and acts as a universal proton donor, initiating chains of ion-molecule reactions that lead to the formation of complex molecules. Observations of H3+ can be used to measure ionization rates and cloud dimensions because its production rate is linearly proportional to gas density rather than quadratic. Its infrared spectrum was first detected in 1980 and subsequent observations have revealed its ubiquity in both dense and diffuse interstellar clouds, with surprising amounts in diffuse clouds.

1. SPECIAL FEA TURE: PERSPECTIVE
؉
Interstellar H3
Takeshi Oka†
Departments of Chemistry and Astronomy and Astrophysics, Enrico Fermi Institute, University of Chicago, Chicago, IL 60637
Edited by William Klemperer, Harvard University, Cambridge, MA, and approved June 13, 2006 (received for review February 4, 2006)
؉
Protonated molecular hydrogen, H3 , is the simplest polyatomic molecule. It is the most abundantly produced interstellar molecule,
؉
next only to H2, although its steady state concentration is low because of its extremely high chemical reactivity. H3 is a strong acid
(proton donor) and initiates chains of ion-molecule reactions in interstellar space thus leading to formation of complex molecules.
Here, I summarize the understandings on this fundamental species in interstellar space obtained from our infrared observations since
؉
its discovery in 1996 and discuss the recent observations and analyses of H3 in the Central Molecular Zone near the Galatic center
that led to a revelation of a vast amount of warm and diffuse gas existing in the region.
10Ϫ17 sϪ1 by many orders of magnitude,
A
lthough it took many years to the most abundantly produced molecular
detect the first signal of inter- ion in interstellar space via cosmic ray the production rate of Hϩ per unit vol-
3
stellar Hϩ, subsequent observa-
3 ionization but also plays a crucial role in ume is given by ␨n(H2) both for dense
tions have revealed the ubiquity interstellar chemistry. Because the proton and diffuse clouds.
of this molecular ion that plays the pivotal affinity of H2 (4.4 eV) is lower than that The destruction mechanism for Hϩ, 3
role in interstellar chemistry. It has now of almost all atoms and molecules (with however, differs in dense and diffuse
been well established that Hϩ exists abun-
3 the notable exception of He, N, and O2), clouds. In dense clouds where carbon at-
dantly not only in dense molecular clouds Hϩ acts as a universal proton donor
3 oms are mostly in CO, Hϩ is mainly de-
3
as predicted by chemical model calcula- (acid) through the proton hop reaction, stroyed by reaction in Eq. 2 with X ϭ CO
tions but also in diffuse clouds, and that (we ignore O for simplicity), whereas in
the total amount of Hϩ in a diffuse cloud
3
Hϩ ϩ X 3 HXϩ ϩ H2,
3 [2] diffuse clouds where C is mostly atomic,
is orders of magnitude higher than in a Hϩ is destroyed by recombination with
3
dense cloud. The surprise continued, cul- which also has a large Langevin rate con-
electrons released from the photoioniza-
minating in the recent discovery of Hϩ in stant kX. Once protonated, HXϩ is far
3 tion of C. Equating the production rate
the (J, K) ϭ (3, 3) metastable rotational more active than the neutral X and leads
to a chain of ion-neutral reactions. For ␨n(H2) and the destruction rate
level that is 361 K above the ground level
example, a protonated oxygen atom, kCOn(Hϩ)n(CO) and ken(Hϩ)n(eϪ) for
3 3
and the resultant revelation of a vast dense and diffuse clouds, respectively, we
amount of gas with high temperature and HOϩ, produces a hydronium ion after a ϩ
series of hydrogen abstraction reactions obtain the H3 number densities
low density in the Central Molecular
Zone (CMZ) of the Galactic center. with H2, HOϩ 3 H2Oϩ 3 H3Oϩ, which n͑H3͒ dense Ϸ ͑ ␨ ͞k CO͓͒n͑H2͒͞n(CO)͔ [3]
Here, I trace this inspiring development recombines with an electron to form in-
from a unified perspective of molecular terstellar water. Without the protonation and
astrophysics and project into the future. of atoms and molecules by Hϩ, the forma-
3
tion of interstellar molecules would pro- n͑H3͒ diffuse Ϸ ͑ ␨ ͞k e͓͒n͑H2͒͞n͑eϪ͔͒. [4]
؉ ceed much more slowly. Because timely
H3 , the Initiator of Interstellar Chemistry
production of a large amount of molecules If we assume the same ␨ for dense and
Within a decade and a half of the discov-
is essential for cooling gravitationally con- diffuse clouds, and [n(H2)͞n(CO)] ϳ
ery of Hϩ by J. J. Thomson in 1911 (1), it
3
densing gas, Hϩ plays a crucial role in star [n(H2)͞n(eϪ)], n(Hϩ)diffuse is calculated to
was clear that Hϩ was the most abundant 3 3
3
formation. be 100 times lower than n(Hϩ)dense be-
3
ion in ordinary hydrogen discharges. Hog-
cause kCO (due to the rϪ4 Langevin po-
ness and Lunn (2) inferred that Hϩ was
3 Simple Chemistry and Unique
produced by a secondary process, the ion- tential) is 100 times smaller than ke (due
ization of H2 into Hϩ by electron impact
Astrophysical Probe to the long-range rϪ1 Coulomb potential).
2
followed by the ion-neutral reaction, Because 99.9% of atoms in the Universe A remarkable aspect of Eqs. 3 and 4 is
are hydrogen (92.1%) and helium (7.8%), that N(Hϩ) does not scale with cloud den-
3
Hϩ ϩ H2 3 Hϩ ϩ H.
2 3 [1] interstellar chemistry is a hydrogen-domi- sity [ϳn(H2)] but is proportional to the
nated chemistry. Initially, interstellar gas ratios of n(H2)͞n(CO) and n(H2)͞
It was later shown that this reaction is was thought to be atomic, but it has be- n(eϪ) ϳ n(H2)͞n(Cϩ), which are approxi-
extremely efficient with a high exother- come increasingly clear that molecular mately constant for typical dense and dif-
micity of 1.7 eV and a large Langevin hydrogen is the most abundant species in fuse clouds, respectively. This situation is
rate constant of kL ϭ 2 ϫ 10Ϫ9 cm3⅐sϪ1. the interstellar medium. This is true not shown schematically in Fig. 1. This special
By 1961, the efficiency of this reaction only in dense molecular clouds with densi- characteristic of Hϩ, stemming from the
3
and the dominance of Hϩ in hydrogen
3 ties of 104 cmϪ3 or higher that are gravita- fact that its production rate is linear to
plasmas were so well established that tionally bound and on their way to star concentration rather than quadratic,
McDaniel and colleagues (3) advocated formation, but also in diffuse clouds with makes Hϩ a unique astrophysical probe.
3
the possibility of detecting interstellar densities of Ϸ102 cmϪ3. In the Galaxy, We can use it as a yardstick to measure
Hϩ ‘‘because this molecular ion may be
3 those clouds together contain perhaps the dimensions of clouds, because ob-
present under some circumstances to the Ϸ80% of interstellar matter whose total
virtual exclusion of Hϩ.’’
2 mass is on the order of Ϸ3% of the total
In 1973, after the avalanche of radio mass of the stars (6). This vast amount of Conflict of interest statement: No conflicts declared.
astronomical discoveries of interstellar H2 is continuously ionized by the ubiqui- This paper was submitted directly (Track II) to the PNAS
molecules triggered by Townes and col- tous cosmic rays to produce Hϩ. Because
3
office.
leagues, Watson (4) and Herbst and Kl- the reaction rate of Eq. 1, kLn(H2), is †E-mail: t-oka@uchicago.edu.
emperer (5) showed that Hϩ is not only
3 higher than the ionization rate ␨ ϳ 3 ϫ © 2006 by The National Academy of Sciences of the USA
www.pnas.org͞cgi͞doi͞10.1073͞pnas.0601242103 PNAS ͉ August 15, 2006 ͉ vol. 103 ͉ no. 33 ͉ 12235–12242

2. tional spectrum. Modern ab initio calcula-
tions have shown that Hϩ, although well
3
bound in the ground state, does not have
bound electronically excited states except
for a weakly bound triplet state (for a re-
view, see ref. 11). Thus, the only spectrum
for astronomical observation of Hϩ is the
3
infrared transition of the doubly degener-
ate ␯2 vibration, which has its band origin
at 4 ␮m. A remarkable ab initio theoreti-
cal calculation from 1974–1976 by Carney
and Porter (12) predicted the band origin
as well as some spectroscopic parameters
with good accuracy. A project to find this
spectrum in absorption was launched in
my laboratory then at the Herzberg Insti-
tute of Astrophysics in 1975, at a time
when there had been no reported infrared
spectrum of any molecular ion. Using the
difference frequency laser spectrometer
developed by Pine (13), the spectrum was
found in 1980 (14). Luckily, this spectrum
appears in a wavelength region known to
ϩ
astronomers as the L window, which does
Fig. 1. A schematic diagram showing relations of H2, CO, and H3 number densities, n(X), versus cloud not have strong spectra of other mole-
ϩ
density, n(H). Note that n(H2) and n(CO) scale with n(H), whereas n(H3 ) is independent of n(H).
cules. This makes the infrared spectrum of
Hϩ observable from ground-based obser-
3
served Hϩ column density is simply pro- for similar clouds in which ␨ are compara- vatories, because absorptions from atmo-
3
spheric molecules do not interfere badly.
portional to the pathlength L, N(Hϩ) ϭ
3 ble. N(Hϩ) is also the most direct probe
3
n(Hϩ)L instead of integrals as for N(CO)
3 to measure ␨ in clouds whose dimensions Detection of Interstellar H3؉
and N(H2). This relation allows us to de- are known. These characteristics make Hϩ 3 in Dense Clouds
termine ␨L reliably from observed column a unique and powerful astrophysical The search for interstellar Hϩ was initi-
3
densities. For a dense cloud and a diffuse probe. ated in June 1980 immediately after the
cloud, Eqs. 3 and 4 give paper on the laboratory spectrum was
The Infrared Spectrum
␨LϭN͑Hϩ͒ densek CO͓n͑CO͒͞n͑H2͔͒ submitted. The technological level of the
3 Hϩ is the simplest polyatomic molecule.
3 astronomical infrared spectroscopy, how-
Its three protons are well bound by two ever, was far behind that of radio astron-
Ϸ 6.4 ϫ 10 Ϫ13 cm3⅐sϪ1N͑Hϩ͒ dense [5]
3
electrons in an equilateral triangle struc- omy, perhaps by more than 20 years, and
and ture as initially shown theoretically by the first search using the FTIR spectrome-
Coulson (10). Because of the symmetry, ter on the 4-m Mayall Telescope ended a
␨L ϭ N͑Hϩ͒ diffusek e͓n͑eϪ͒͞n͑H2͔͒
3 Hϩ has no permanent electric dipole mo-
3 miserable failure (15). Several negative or
ment and hence no fully allowed rota- inconclusive results were published by this
Ϸ 8.3 ϫ 10 Ϫ11 cm3⅐s Ϫ1N͑Hϩ͒ diffuse͞f,
3
[6]
respectively, where f in Eq. 6 is a fraction
of hydrogen atom in the molecular form
f ϭ 2n(H2)͞[2n(H2) ϩ n(H)]. In calculat-
ing the numerical values of Eqs. 5 and 6,
kCO ϳ 2 ϫ 10Ϫ9 cm3⅐sϪ1 (7), the C͞H
ratio after carbon depletion to dust of
Ϸ1.6 ϫ 10Ϫ4 (8) and ke ϭ 2.6 ϫ 10Ϫ7
cm3⅐sϪ1 (9) are used. The numerical factor
in Eq. 5 is accurate perhaps to within
20%, whereas the numerical factor of Eq.
6 has a higher uncertainty because (i) the
value of ke is less well established than
kCO and, unlike kCO, ke depends on the
temperature; and (ii) carbon may not be
completely ionized. These will be dis-
cussed in more detail later. Nevertheless,
by using the observed Hϩ column densi-
3
ties, we can obtain fairly reliable values of
␨L. Unfortunately, separation of ␨ and L ϩ
Fig. 2. The first detection of interstellar H3 at the UKIRT toward two young stellar objects that are deeply
is not easy. However, observations of embedded in their natal dense molecular clouds. The doublet at 3.66808 ␮m and 3.66852 ␮m are the
N(Hϩ) give good relative measures of L
3
ϩ
R(1, 1)u and R(1, 0) transitions of para- and ortho-H3 , respectively. (Reproduced from ref. 18.)
12236 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0601242103 Oka

3. and other groups. In the meantime the last few years, is outside the scope of this have been notoriously contradictory. A
first nonterrestrial Hϩ spectrum was ob-
3 paper, and readers are referred to refs. 24 wide range of values spanning over four
served in 1987 as strong emission from and 25. orders of magnitude had been reported by
Jupiter, and emissions from Uranus and ؉
different experimental techniques (see ref.
Saturn followed. The exciting develop- Enigma of H3 in Diffuse Clouds 30 for a review). Also, modern theoretical
ment of Hϩ spectroscopy in planetary
3 If an observation confirms a prediction, it calculations had always given orders of
ionospheres is outside the scope of this is gratifying but not really exciting. Luck- magnitude lower value than the one used
article, and readers are referred to reviews ily, it did not take too long before our above. Such a low value of ke would im-
(16, 17). In the early 1990s, the first large observations revealed something unex- mediately explain the large observed Hϩ 3
two-dimensional infrared array detectors pected. On July 11, 1997, during our sur- column density in diffuse clouds. How-
were installed in spectrometers, and this vey of dense clouds (23), we detected very ever, the experiments by Larsson and col-
greatly enhanced the sensitivity and reli- strong Hϩ lines toward two infrared
3 leagues (30) using the ion storage ring
ability of infrared spectroscopy, setting the sources GC IRS 3 and GCS 3-2 near the have consistently given a higher value, on
stage for the next development. Galactic center (26). Because the visual the order of 10Ϫ7 cm3⅐sϪ1, since 1993. Fi-
Interstellar Hϩ was detected on April
3 extinction AV ϳ 30 mag toward the Ga- nally in 2003, Kokoouline and Greene
29, 1996 at the United Kingdom Infrared lactic center is caused more by diffuse (31, 64) presented an extensive theoretical
Telescope (UKIRT) toward the infrared clouds than dense clouds, this suggested calculation including the Jahn-Teller ef-
stars AFGL 2136 and W33A, two bright the previously unforeseen possibility that fect and gave a value of ke that agrees
young stellar objects that are deeply em- Hϩ may exist abundantly also in diffuse
3 with the experiment (9, 30). For now the
bedded in the natal dense clouds. After clouds. experimental value of ke ϭ 2.6 ϫ 10Ϫ7
confirming on July 15 that the doublet Indeed, our observation later that night cm3⅐sϪ1 at 23 K (9) used above for Eq. 6
Doppler shifts by an expected amount due revealed a high column density of Hϩ to-3 seems the most reliable one. However,
to the motion of the earth, a paper was ward Cygnus OB2 No. 12, the archetypical because this rate constant is of over-
submitted. The extremely weak R(1, 1)l– sightline obscured by diffuse clouds. This whelming importance, more theoretical
R(1, 0) doublet is shown in Fig. 2 (18). faint red star discovered by Morgan and and experimental studies are desirable.
Their observed equivalent widths (inte- colleagues in 1954 is arguably the most The second suspect was the possibility
grated absorption) gave Hϩ column densi-
3 intrinsically luminous star in the Galaxy, that the ratio n(H2)͞n(eϪ) is higher than
ties of 4 ϫ 1014 cmϪ2 and 6 ϫ 1014 cmϪ2 and its visual extinction of AV ϭ 10.3 is the value used above due to an incom-
for AFGL 2136 and W33A, respectively, the highest known among optical stars plete ionization of carbon atom. This was
corresponding to ␨L ϳ 260 cm⅐sϪ1 and (27). Nevertheless, its visual extinction is negated when we detected Hϩ with a col-
3
380 cm⅐sϪ1. If we use the canonical ␨ only Ϸ1⁄10 that toward dense cloud sources umn density of 8 ϫ 1013 cmϪ2 toward the
value of Ϸ3 ϫ 10Ϫ17 sϪ1 (see, for exam- like AFGL 2136 and W33A (AV ϳ 100), prototypical optical star ␨ Per (9). This
ple, refs. 19 and 20), we obtain path- indicating, if we assume the same dust-to- bright star (V ϭ 2.86) with low visual ex-
lengths L of 3 pc and 4 pc for the clouds gas ratio, that a 10-times-smaller amount tinction (AV ϭ 0.9) is sufficiently unat-
that are reasonable orders of magnitude. of gas occupies a 10-times-longer path in tenuated in the UV to allow observation
The temperatures of the clouds were de- the diffuse cloud than in the dense clouds. of the forbidden spectrum of Cϩ at
termined from the intensity ratios of the The Hϩ number densities given in Eqs. 3
3 2,325.4 Å. The observation showed that
doublet to be Ϸ35 K, which was also rea- and 4 therefore predict that the Hϩ col-
3 indeed the carbon atoms are mostly
sonable. For over two decades since 1973, umn density in the diffuse clouds is 1⁄10 of ionized.
the presence of Hϩ had been assumed.
3 the dense clouds, which would have been Those results suggest strongly that ␨ in
Now, this species was finally observed hard to detect. Contrary to the expecta- diffuse clouds is indeed 10 times higher
with the amount predicted by model cal- tion, we observed N(Hϩ) ϭ 3.8 ϫ 1014
3 than in dense clouds. This is reasonable
culations. A big sigh of relief was heard cmϪ2 (26, 28), which is comparable to the because the soft components of the cos-
from the community of interstellar chem- value toward AFGL 2136 and W33A. mic ray are attenuated by high-density gas
istry (21, 22). From Eq. 6, we obtain ␨L ϳ 3 ϫ 104 (32). The high value of ␨ ϳ 3 ϫ 10Ϫ16 sϪ1
Although it took 15 years to detect the cm⅐sϪ1 (even for f ϭ 1), 100 times higher is beyond the limit set from observed
first signal of interstellar Hϩ, once discov-
3 than that of dense clouds. The canonical abundances of HD and OH (33, 34), but a
ered, it seemed to be observable every- value of ␨ ϭ 3 ϫ 10Ϫ17 sϪ1 gives an ex- recent paper by Liszt (35) has shown that
where as long as a dense cloud had suffi- tremely long path length of Ϸ300 pc, higher values of ␨ are favored if neutral-
ciently long pathlength (23). By now, Hϩ 3 which occupies a sizable fraction of the ization of atomic ions on dust grains is
has been observed in more than a dozen distance to the star (1.7 kpc)! Such a huge properly taken into account. Liszt warns,
dense clouds, and the measured values of cloud is unknown. Moreover, the density however, that larger values of ␨ do not
␨L are 70–330 cm⅐sϪ1, just about the right of the cloud, if distributed over such a simply increase the Hϩ number density
3
order of magnitude for canonical values long distance, would be Ϸ10 cmϪ3 from because the higher cosmic ray ionization
of the cosmic-ray ionization rate ␨ and the standard dust to gas ratio (29), hardly will introduce ionization of H in addition
pathlength L. The temperatures of clouds an environment to contain a large fraction to that of C.
24ϳ50 K are also reasonable. For now, of H2 and thus Hϩ. It is more likely from
3 Overall, the high column density of
the Hϩ chemistry in dense clouds is a
3 the observed value of AV ϳ 10.3 that L is Hϩ in the diffuse interstellar medium is
3
happy story. Clouds with very high densi- on the order of 30 pc and that ␨ is 10 a major enigma, along with the large
ties (n Ն 106 cmϪ3) and low temperatures times higher than the canonical value. abundance of CHϩ, which is yet to be
(T ϳ 10 K), however, are a different This applies to all diffuse clouds as dis- understood.
story. In such clouds, most molecules cussed later. Before jumping to this con- ؉
other than H2 and its isotopomers may be clusion, however, we needed to examine The Ubiquitous H3
frozen onto grain, and the slightly exo- the assumptions made for calculating ␨L Although not completely understood,
thermic reaction with HD may become in Eq. 6. the large column densities of Hϩ in dif-
3
the main destruction process, leading to The first suspect was the recombination fuse clouds have been well established
extremely high deuterium fractionation. rate constant ke. Through the years, the observationally (36). Cygnus OB2 No.
This interesting subject, developed in the laboratory measurements of this value 12 [V ϭ 11.4, E(B Ϫ V) ϭ 3.35] and No.
Oka PNAS ͉ August 15, 2006 ͉ vol. 103 ͉ no. 33 ͉ 12237

4. Mauna Kea but also partly by using the
4-m Mayall Telescope on Kitt Peak. The
absorption spectra were all starting from
the lowest two rotational levels, (J, K) ϭ
(1, 1) and (1, 0). The advent of 8-m-class
telescopes equipped with high-resolution
spectrometers around the turn of the cen-
tury has led us to the next major develop-
ment. Before going into this, however, I
need to discuss the rotational-level struc-
ture of Hϩ in the ground vibrational state.
3
Fig. 4 shows rotational levels of Hϩ 3
labeled by the rotational quantum number
J and its projection to the symmetry axis
K. Because the proton is a fermion with
spin 1⁄2, the totally symmetric levels (0, 0),
(2, 0), etc. are forbidden (broken lines)
and levels with K ϭ 3n are occupied by
ortho-Hϩ (I ϭ 3⁄2; red lines) and K ϭ 3nϮ
3
1 by para-Hϩ (I ϭ 1⁄2; blue lines). The Ϯ
3
signs beside levels show parity of the
state, (Ϫ1)K. Readers are referred to ref.
37 for more details.
Because Hϩ is an equilateral triangle,
3
there is no permanent dipole moment and
hence no ordinary rotational spectrum.
However, as initially considered for NH3
(38), a breakdown of the geometrical sym-
metry due to centrifugal distortion may
cause a small dipole moment and hence
spontaneous emissions. Such emissions,
that are slow and almost academic for
NH3, are much faster for Hϩ because of
3
its smaller mass (39) and are competitive
ϩ
Fig. 3. Observed H3 column densities versus color excess E(B Ϫ V) for dense clouds (Upper) and diffuse
with collisions. For example, para-Hϩ in
3
ϩ
clouds (Lower). Note that the two categories of clouds have comparable N(H3 ), although E(B Ϫ V), and the (2, 2) level decays to the (1, 1) ground
hence the amount of gas in the sightline, is 10 times lower for diffuse clouds than from dense clouds. level with a life time of 27 days, which
(Reproduced from refs. 23 and 36.) corresponds to a critical density (at which
collision rates become comparable to the
spontaneous emission rate) of Ϸ200 cmϪ3,
5 (9.2, 1.99), Wolf-Rayet stars WR 118 i.e., the fractional abundance of Hϩ in the
3 just about the number density of diffuse
(22.0, 4.13), WR 104 (13.5, 2.10), WR gas is an order of magnitude higher in clouds! On the other hand, ortho-Hϩ in3
121 (11.9, 1.68), the prototypical diffuse diffuse clouds than in dense clouds. Our the (3, 3) level cannot decay because the
interstellar bands sightline HD18 3143 attempt to find ‘‘translucent clouds’’ that ortho 3 para transition is forbidden and
(6.9, 1.28), HD 20041 (5.8, 0.70), and ␨ may have N(Hϩ) intermediate between
3 the (2, 0) level is missing. Hϩ in the (3, 3)
3
Per (2.9, 0.31) all showed clear signals dense and diffuse clouds is so far nega- level is metastable and decays to lower
of Hϩ, where E(B Ϫ V) is the color ex-
3
tive. Sightlines with relatively high AV levels only through collisions. Thus, in a
cess that is Ϸ1⁄3 of AV. Hϩ is ubiquitous;
3
such as the Wolf-Rayet stars (AV ϭ 13–5) low-density cloud of sufficiently high tem-
it exists in diffuse clouds as well as in and the Stephenson objects (AV ϭ 16–12)
dense clouds. all gave values on the straight line of dif-
The observed Hϩ column densities in
3
fuse clouds, suggesting that those so-called
dense and diffuse clouds versus their color ‘‘translucent sightlines’’ are simply pileups
excess are shown in Fig. 3. It is seen that of diffuse clouds and are not due to
N(Hϩ) is approximately proportional to
3
‘‘translucent clouds.’’
E(B Ϫ V) (and thus to AV), indicating The total amount of Hϩ in a diffuse
3
that the cloud dimension is approximately clouds should be 100 times more than
proportional to AV; that is, the variations in a dense cloud because the pathlength
of AV for these samples of clouds are due through a diffuse cloud is 10 times higher
to variations of pathlength rather than than that through a dense cloud. The rela-
density. From a least squares fitting of the tive number of dense and diffuse clouds is
observed values given in Fig. 3, we have not known, but it is very likely that the
total amount of Hϩ in dense clouds is
3 ϩ
Fig. 4. Rotational levels of H3 in the ground vibra-
N͑Hϩ͒ dense Ϸ 3.6 ϫ 10 12 cmϪ2 Av
3 [7] much less than that in diffuse clouds! ϩ
tional state. In the CMZ so far observed, most H3 are
؉ populated in three levels: the (1, 1) ground level
and Ideal Level Structure and Metastable H3 ϩ
(para-H-3 ), the (1, 0) ortho-ground level, and the (3,
The spectra so far discussed were mostly 3) ortho-metastable level. Blue arrows indicate spon-
N͑Hϩ͒ diffuse Ϸ 4.4 ϫ 10 13 cmϪ2 Av,
3 [8] obtained by using the 3.6-m UKIRT on taneous emissions. See ref. 41 for more detail.
12238 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0601242103 Oka

5. perature, a very nonthermal distribution is actually a chemical reaction and protons cross-dispersion grating that allows high-
results between the metastable (3, 3) level scramble in the activated complex (Hϩ)*.
5 resolution spectroscopy (R ϳ 20,000)
and the unstable (2, 2) level. Thus, unlike for neutral molecules, H2, without sacrificing wavelength coverage.
Hϩ in other higher rotational levels in
3 H2O, and NH3, collision-induced transi- The observed R(3, 3)l and R(1, 1)l lines
Fig. 4 are all unstable and decay quickly tions from ortho to para level of Hϩ are
3
are shown overlapped on the same veloc-
to lower levels; the life times of the (3, 0), allowed as shown by laboratory experi- ity scale together with spectral lines of the
(3, 1), (3, 2), and (2, 1) levels are 3.8 h, ments (44). We have used transition prob- CO 2-0 overtone band in Fig. 5. These
7.9 h, 16 h, and 20 days, respectively (39, abilities based only on the principle of spectra were observed toward GCS 3-2, a
40). Thus, these levels are not populated detailed balancing and calculated n(3, 3)͞ young star in the Quintuplet cluster (48)
n(1, 1) and n(3, 3)͞n(2, 2) as a function of and the brightest infrared source (L ϭ
unless the gas has both high temperature
temperature and density of clouds. Read- 2.7) in the CMZ.
and high density, which we have not The sharp peaks in the R(1, 1)l and CO
found so far for Hϩ. For a low-tempera-
3
ers are referred to ref. 41 for more
spectra [but not in the R(3, 3)l spectrum]
ture cloud, only the lowest two levels are details.
are absorptions by molecules in relatively
populated. The (1, 0) ortho level is 32.9 K ؉ cold and dense regions largely in the in-
Discovery of Metastable H3 in the
above the (1, 1) para level, approximately tervening spiral arms. Gas in the CMZ
Central Molecular Zone
the temperature of the clouds (23, 36). produces the R(3, 3)l spectrum and the
The ratio of spin statistical weight (2I ϩ The CMZ, a region of radius Ϸ200 pc of
‘‘trough’’ of the R(1, 1)l spectrum that are
1), 4:2, is approximately compensated by the Galactic center containing a large
nearly identical. These two featureless
the Boltzmann factor and the spectral mass (Ϸ5 ϫ 107 MJ) of dominantly
spectra with high-velocity dispersion rang-
molecular gas (45), is the treasure house ing from Ϫ160 km⅐sϪ1 to ϩ30 km⅐sϪ1
strength to make the intensities of the
of Hϩ. The Hϩ column density toward
3 3 demonstrate the energetic, turbulent na-
R(1, 1)u and R(1, 0) lines in Figs. 3
the CMZ is an order of magnitude higher ture of the gas in the CMZ. Their nearly
comparable.
than in other sightlines in the Galactic equal intensities, corresponding to a (3,
On the other hand, the (3, 3) metasta-
disk (46). Because the visual extinction 3)͞(1, 1) excitation temperature of Ϸ200
ble level is 361 K above the (1, 1) ground toward the CMZ of AV ϳ 30 mag is
level, and its population is a sure sign of K, demonstrate a high kinetic tempera-
caused more by diffuse clouds than by ture. Further analyses based on more
high temperature. The observed popula- dense clouds, a large N(Hϩ) is expected
tion ratio n(3, 3)͞n(1, 1) serves as a ther-
3 spectra taken under higher resolution at
from Eq. 8, but the observed Hϩ column
3 the 8-m Gemini South Telescope on
mometer. Meanwhile, the observed ratio density toward the Galactic center is Cerro Pachon, Chile, and UKIRT and a
n(3, 3)͞n(2, 2) serves as a densitometer higher by another factor of Ϸ5! detailed model calculation (41) have led
because the population of the (2, 2) un- Interstellar Hϩ in the metastable (3, 3)
3 to a revelation of a new category of
stable level is a sensitive function of cloud level was discovered in this environment clouds.
density. The four rotational levels, the (1, by using the 8.2-m Subaru Telescope (47).
1) ground level, the (1, 0) ortho ground The discovery was facilitated by the wide Warm and Diffuse Clouds in the CMZ
level, the (2, 2) unstable level, and the (3, wavelength coverage of its IRCS spec- It had been known mainly from radio-
3) metastable level, form an ideal level trometer equipped with an echelle and a astronomical observations that the CMZ
system for observing both cold and warm
clouds. I cannot think of a better arrange-
ment than this provided by nature!
Thermalization of Hϩ and the resultant
3
steady-state rotational distribution de-
pends on a subtle interplay between radia-
tive and collisional interactions. Although
selection rules for spontaneous emission
are rigorous and the aforementioned life-
times are very accurate (perhaps to within
5%) (40), our knowledge of state-to-state
collision-induced transitions is very lim-
ited. The simple level structure, however,
allows us to calculate thermalization with
some accuracy based on rough assump-
tions. We have used a simplifying assump-
tion that collision-induced transitions oc-
cur completely randomly without selection
rules (41). Such an assumption, used for
strong collisions by van Vleck and Weiss-
kopf in the early days of microwave spec-
troscopy (42), ignores the subsequently
established nearly rigorous ortho 4͞3
para rule and the approximate selection
rules for other quantum numbers govern-
ing collision-induced transitions (43). This
is justified to a large extent for collisions
between Hϩ and H2 because, unlike for
3
neutral molecules, ϩ
Fig. 5. Spectra of H3 (Upper) and CO (Lower) observed by the Subaru Telescope toward GCS 3-2, the
brightest infrared source near the Galactic center. The R(3, 3)l line (in red) shows the presence of a vast
Hϩ ϩ H2 3 ͑Hϩ͒* 3 Hϩ ϩ H2
3 5 3 [9] ϩ
amount of metastable H3 , which signifies high temperature. (Reproduced from ref. 46.)
Oka PNAS ͉ August 15, 2006 ͉ vol. 103 ͉ no. 33 ͉ 12239

6. Second, note that the CO spectrum is
composed largely only of sharp lines,
whereas the R(1, 1)l spectrum shows a
deep and wide trough from Ϫ150 to ϩ30
km⅐sϪ1. We interpret the latter to be due
to Hϩ in the CMZ. The broad relatively
3
featureless absorption reflects the high-
velocity dispersion of the gas.
Third, note that the R(3, 3)l spectrum
(second from top) of the metastable Hϩ 3
is broad and mimics the trough of the
R(1, 1)l spectrum The strength of the R(3,
3)l line is definitive evidence for the high
temperature of the gas in the CMZ.
Fourth, note that the R(2, 2)l spectrum
(third from top) shows no clear absorp-
tion, indicating that the (2, 2) unstable
level is not much populated although it is
much lower than the (3, 3) level. This
strongly inverted population between (2,
2) and (3, 3) due to the (2, 2) 3 (1, 1)
spontaneous emission is definitive evi-
dence of the low density of the clouds.
More detailed analyses give more quan-
titative information on the gas in the
CMZ, for which readers are referred to
ref. 57. Here, we summarize some main
conclusions. Of the total Hϩ column den-
3
sity of (4.3 Ϯ 0.3) ϫ 1015 cmϪ2 toward
GCS 3-2, approximately one-quarter
(1.2 ϫ 1015 cmϪ2) is in cold clouds mostly
in the intervening spiral arms, and the
rest is in hot clouds in the CMZ. Approxi-
mately half of the latter (1.6 ϫ 1015 cmϪ2)
ϩ
is in clouds with velocities near Ϫ100 km
Fig. 6. Observed H3 (top three traces) R(1, 1)l, R(3, 3)l, and R(2, 2)l lines and CO 2-0 overtone R(1) line toward
sϪ1; their average temperature is (270 Ϯ
GCS 3-2 observed by the Gemini South Telescope and the UKIRT. The vertical scaling of the R(3, 3)l and R(2, 2)l
spectra is multiplied by a factor of 2 and that of the CO spectrum is divided by 2 for clarity. (Reproduced from
70) K, and the density is lower than 50
ref. 57.) cmϪ3, although uncertainty of the latter
may be high because of uncertainty of the
collision cross section (a factor of 2). The
contains many dense molecular clouds as J ϭ 7 3 6 (50, 51) and J ϭ 16 3 15 large negative velocity suggests strongly
with a high volume-filling factor (45, 49). (52), but they are limited to the two spe- that these clouds are associated with the
This area, which is perhaps Ͻ10Ϫ5 of the cial regions of high activity, Sgr A and Sgr large-scale structure moving away from
Galaxy in volume, contains a few percent B (51). A more extensive presence of hot the Galactic center at high speed initially
of the total interstellar medium in the clouds had been claimed using NH3 in proposed as the Expanding Molecular
Galaxy. Radio-emission lines with critical high metastable levels up to (7, 7) (53, 54) Ring (EMR) (58, 59). It exists at Ϸ180 pc
densities higher than 104 cmϪ3, such as and H2 up to J ϭ 7 (55), but the reported from the center, at about the border of
CO J ϭ 2 3 1, 3 3 2, CS 1 3 0, HCN 1 densities of 104 and 103.5 to 104 cmϪ3 (56) the CMZ. The remaining Hϩ with lower
3
3 0, etc., provide definitive evidence of may well be overestimates. velocities are associated with somewhat
high density, whereas emission lines with Our recent observation and analysis of lower-temperature and higher-density
critical densities lower than 103 cmϪ3 such Hϩ have revealed a vast amount of gas
3 clouds but are all in the CMZ.
with high temperature (Ϸ250 K) and low The (3, 3) metastable Hϩ has been de-
3
as CO 1 3 0, NH3 inversion lines (all J,
density (n ϳ 100 cmϪ3) in the CMZ (57). tected also toward many other young stars
K), and H2 rotational lines and absorption
The gas has a huge volume and must fill within Ϸ30 pc of the center. The location
lines such as OH, H2CO, etc. do not. Nev-
the space between the aforementioned of the eight stars so far observed that all
ertheless high densities had often been
dense clouds. Observed spectra of Hϩ and
3 showed the metastable spectra are given
claimed from the latter spectra based on CO that provide evidence for this conclu- in Fig. 7. Their spectra show that the hot
circumstantial evidences, and perhaps this sion are shown in Fig. 6. and diffuse gas is distributed widely in the
contributed to the overestimates of both First, note the similarity between the CMZ, although its velocity depends on
the volume-filling factor and the fraction sharp absorption lines of the R(1, 1)l spec- the individual sightlines. Readers are re-
of interstellar medium in the CMZ by an trum (top trace) and the CO spectrum ferred to ref. 57 for more details of this
order of magnitude (45, 49). (bottom trace) with three main peaks at and many other interesting observations.
It had also been claimed that a signifi- Ϫ52, Ϫ32, and Ϫ5 km⅐sϪ1. They are due Our attempt at detecting metastable
cant portion of the dense clouds had high to molecules in the cold and relatively Hϩ in sources in the Galactic disk where
3
temperature of 150ϳ300 K from radio dense environment mostly in the interven- high column densities of Hϩ had been
3
and far-infrared spectroscopy. The most ing spiral arms, the 3-kpc arm, the 4.5-kpc observed has so far been unsuccessful.
direct evidence for this has been the CO arm (Scutum), and the local arm (Sagit- The warm and diffuse environment seems
emission from high rotational levels such tarius), respectively. unique to the CMZ.
12240 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0601242103 Oka

7. aliphatic hydrocarbon absorption, the tell-
tale sign of diffuse clouds and hence Hϩ.3
Our next plan is to observe the highly
obscured active galactic nucleus, the
Southern Hemisphere object Superanten-
nae, IRAS 19254-7245. Information ob-
tained from the Hϩ and CO spectra may
3
provide a new perspective on the nature
of active galactic nuclei.
Because cosmic-ray ionization is ubiqui-
tous, Hϩ is produced abundantly in any
3
region with molecular hydrogen, although
its steady-state concentration is not high
due to its high chemical reactivity. As long
as the pathlength in a cloud is sufficiently
large, Hϩ will be detectable in any object.
3
We have a considerable amount of data
yet to be published on dense and diffuse
clouds. In 2002, Hϩ emission from the
3
young star HD 141569A was reported as
possibly from preplanetary gas orbiting
the star (62). Unfortunately, our attempt
to confirm this emission ended up negat-
Fig. 7. Location of observed infrared stars indicated by crosses. They all showed the R(3, 3)l spectrum,
ing the original claim (63). However, de-
signifying high temperature, and extremely weak, if any, R(2, 2)l spectrum, signifying low density,
indicating the widespread presence of hot and diffuse gas. The background is the VLA 20-cm image by
tecting Hϩ in protoplanets and extra-solar
3
Yusef-Zadeh and Morris (60). (Reproduced from ref. 57.) planets remains an intriguing possibility.
Our proposal to search for Hϩ in plane-
3
tary nebulae and protoplanetary nebulae
Outlook OH, NH3, CO, CS, HCN etc.; the hot gas was accepted this year by the Subaru
Our study of the Galactic center using Hϩ 3
with high electron densities observed by Telescope after several attempts.
as a probe is in its infancy. The data so radio-wave scattering and free-free emis- Clearly there is no shortage of interest-
far accumulated and the understanding sion and absorption measurements; and ing objects to observe. The simplicity of
that we are beginning to acquire is but a the ultra-hot gas (T ϳ 108 K) evidenced the Hϩ chemistry allows us to obtain
3
tip of the iceberg. Many more observa- by the widely observed x-ray emission fresh and unique astrophysical informa-
tions are needed to obtain a comprehen- (45). Observers of each category of the tion from observed spectra. We are
sive picture of the hot and diffuse gas in gas claim a high volume-filling factor and thrilled by the prospect of obtaining unex-
the CMZ. We are in the process of ex- yet clearly they cannot coexist in the same pected new information. The next 10
tending the observation from Ϸ30 pc region. It will be extremely interesting to years will be full of surprises.
from the center to Ϸ200 pc using other find out the relation of the new category
young stars and even some late-type stars of gas to those other gases (60). Works discussed in this paper result from
distributed over the CMZ. It will take at The strong signal toward the Galactic collaboration with T. R. Geballe, with whom
least a few more years for these observa- center suggests that external galaxies must I have been working on the project of inter-
tions. Such Hϩ spectra combined with CO also have detectable Hϩ in the center. stellar Hϩ since 1982. We were later joined
3
3 3
spectra that provide complementary infor- Indeed, we have detected Hϩ toward a
3
by B. J. McCall in 1996 and M. Goto and T.
Usuda in 2000. To all of them I am deeply
mation will help lead us to a better under- highly obscured ultraluminous galaxy
indebted. I am grateful to T. R. Geballe,
standing of the gas in the CMZ. Previous IRAS 08572 ϩ 3915 NW (61). Because of B. J. McCall, and C. Chung for their critical
to our discovery of the warm and diffuse the expansion of the Universe, the spec- reading of the manuscript. My work has been
gas, three categories of gas were known in trum is red shifted by z ϭ ⌬␭͞␭ ϭ supported by a succession of National Sci-
the CMZ: the dense and mostly cold gas 0.05821. We plan to observe other ultralu- ence Foundation grants, most recently Grant
(T ϳ 50 K) observed by radio-spectra of minous galaxies that have strong 3.4-␮m PHY 03-54200.
1. Thomson, J. J. (1911) Philos. Mag. 21, 9. McCall, B. J., Huneycutt, A. J., Saykally, R. J., 17. Miller, S., Achilleous, N., Ballestes, G. E., Ge-
225–249. Djuric, N., Dunn, G. H., Semaniak, J., Novotny, balle, T. R., Joseph, R. D., Prange, R., Rego, D.,
´
2. Hogness, T. R. & Lunn, E. G. (1925) Phys. Rev. 26, O., Al-Khalili, A., Ehlerding, A., Helbey, F., et al. Stallard, T., Tennyson, J., Trafton, L. M., et al.
44–55. (2004) Phys. Rev. A 70, 052716. (2000) Philos. Trans. R. Soc. London A 358,
3. Martin, D. W., McDaniel, E. W. & Meeks, M. L. 10. Coulson, C. A. (1935) Proc. Cambridge Philos. Soc. 2485–2502.
(1961) Astrophys. J. 134, 1012–1013. 31, 244–259. 18. Geballe, T. R. & Oka, T. (1996) Nature 384,
4. Watson, W. D. (1973) Astrophys. J. 183, 11. Oka, T. (1983) in Molecular Ions: Spectroscopy, 334–335.
L17–L20. Structure, and Chemistry, eds. Miller, T. A. & 19. van Dishoeck, E. F. & Black, J. H. (1986) Astro-
5. Herbst, E. & Klemperer, W. (1973) Astrophys. J. Bondybey, V. E. (North–Holland, Amsterdam), phys. J. Suppl. 62, 109–145.
185, 505–533. pp. 73–90. 20. Lee, H.-H., Bettens, R. P. A. & Herbst, E. (1996)
6. Thronson, H. A. & Shull, J. M., eds. (1990) The 12. Carney, G. D. & Porter, R. N. (1976) J. Chem. Astron. Astrophys. Suppl. 119, 111–114.
Interstellar Medium in Galaxies (Kluwer, Dordrecht, Phys. 65, 3547–3565. 21. Miller, S. (November 28, 1996) The Guardian.
The Netherlands). 13. Pine, A. S. (1974) J. Opt. Soc. Am. 64, 1688 – 22. Barthelemy, P. (November 30, 1996) Le Monde,
´´
7. Anicich, V. G. & Huntress, W. T., Jr. (1986) 1690. Section Aujourd’hui Sciences, p. 25.
Astrophys. J. Suppl. 62, 553–672. 14. Oka, T. (1980) Phys. Rev. Lett. 45, 531–534. 23. McCall, B. J., Geballe, T. R., Hinkle, K. H. & Oka,
8. Sofia, U. J., Lauroesch, J. T., Meyer, D. M. & 15. Oka, T. (1981) Philos. Trans. R. Soc. London A T. (1999) Astrophys. J. 522, 338–348.
Cartledge, S. I. B. (2004) Astrophys. J. 605, 272– 303, 543–549. 24. Roberts, H., Herbst, E. & Millar, T. J. (2003)
277. 16. Oka, T. (1992) Rev. Mod. Phys. 64, 1141–1149. Astrophys. J. 591, L41–L44.
Oka PNAS ͉ August 15, 2006 ͉ vol. 103 ͉ no. 33 ͉ 12241

8. 25. Roberts, H., Herbst, E. & Millar, T. J. (2004) 39. Pan, F.-S. & Oka, T. (1986) Astrophys. J. 305, Walmsley, C. M. (1986) Astron. Astrophys. 162,
Astron. Astrophys. 424, 905–917. 518–525. 199–210.
26. Geballe, T. R., McCall, B. J., Hinkle, K. H. & Oka, 40. Neale, L., Miller, S. & Tennyson, J. (1996) Astro- 54. Huttemeister, S., Wilson, T. L., Bania, T. M. &
¨
T. (1999) Astrophys. J. 510, 251–257. phys. J. 464, 516–520. Martin-Pintado, J. (1993) Astron. Astrophys. 280,
27. Massey, P. & Thompson, A. B. (1991) Astrophys. 41. Oka, T. & Epp, E. (2004) Astrophys. J. 613, 349– 255–267.
J. 101, 1408–1428. 354. 55. Rodriguez-Fernandes, N. J., Martin-Pintado, J.,
28. McCall, B. J., Geballe, T. R., Hinkle, K. H. & Oka, 42. van Vleck, J. H. & Weiskopf, V. F. (1945) Rev. Fuente, A., de Vicente, P., Wilson, T. L. &
T. (1998) Science 279, 1910–1913. Mod. Phys. 17, 227–236. Huttemeister, S. (2001) Astron. Astrophys. 365,
¨
29. Bohlin, R. C., Savage, B. D. & Drake, J. F. (1978) 43. Oka, T. (1973) Adv. At. Mol. Phys. 9, 127–206. 174 –185.
Astrophys. J. 224, 132–142. 44. Cordonnier, M., Uy, D., Dickson, R. M., Kerr, 56. Huttemeister, S., Dahmen, G., Mauersberger, R.,
¨
30. Larsson, M. (2000) Philos. Trans. R. Soc. London K. E., Zhang, Y. & Oka, T. (2000) J. Chem. Phys. Henkel, C., Wilson, T. L. & Martı ´n-Pintado, J.
A 358, 2359–2559. 113, 3181–3193. (1998) Astron. Astrophys. 334, 646–658.
31. Kokoouline, V. & Greene, C. H. (2003) Phys. Rev. 45. Morris, M. & Serabyn, E. (1996) Annu. Rev. 57. Oka, T., Geballe, T. R., Goto, M., Usuda, T. &
Lett. 90, 133201. Astron. Astrophys. 34, 645–701. McCall, B. J. (2005) Astrophys. J. 632, 882–
32. Cravens, T. E. & Dalgarno, A. (1978) Astrophys. J. 46. Geballe, T. R. (2000) Philos. Trans. R. Soc. London 893.
219, 750–752. A 358, 2503–2513. 58. Kaifu, N., Kato, T. & Iguchi, T. (1972) Nature 283,
33. Harquist, T. W., Black, J. H. & Dalgarno, A. 47. Goto, M., McCall, B. J., Geballe, T. R., Usuda, T., 105–107.
(1978) Mon. Not. R. Astron. Soc. 185, 643–646. Kobayashi, N., Terada, H. & Oka, T. (2002) Publ. 59. Scoville, N. Z. (1972) Astrophys. J. 175, L127–
34. Harquist, T. W., Doyle, H. T. & Dalgarno, A. Astron. Soc. Japan 54, 951–961. L132.
(1978) Astron. Astrophys. 68, 65–67. 48. Figer, D. F., McLean, I. S. & Morris, M. (1999) 60. Boldyrev, S. & Yusef-Zadeh, F. (2006) Astrophys.
35. Liszt, H. S. (2003) Astron. Astrophys. 398, 621– Astrophys. J. 514, 202–220. J. 637, L101–L104.
630. 49. Mezger, P. G., Duschl, W. J. & Zylka, R. (1996) 61. Geballe, T. R., Goto, M., Usuda, T., Oka, T. &
36. McCall, B. J., Hinkle, K. H., Geballe, T. R., Astron. Astrophys. Rev. 7, 289–388. McCall, B. J. (2006) Astrophys. J., in press.
Moriarty-Schieven, G. H., Evans, N. J., II, 50. Harris, A. I., Jaffe, D. T., Silber, M. & Genzel, R. 62. Brittain, S. D. & Rettig, T. W. (2002) Nature 418,
Kawaguchi, K., Takano, S., Smith, V. V. & Oka, T. (1985) Astrophys. J. 294, L93–L97. 57–59.
(2002) Astrophys. J. 567, 391–406. 51. Kim, S., Martin, C. L., Stark, A. A. & Lane, A. P. 63. Goto, M., Geballe, T. R., McCall, B. J., Usuda, T.,
37. Oka, T. & Jagod, M.-F. (1993) J. Chem. Soc. (2002) Astrophys. J. 580, 896–903. Suto, H., Terada, H., Kobayashi, N. & Oka, T.
Faraday Trans. 89, 2147–2154. 52. Genzel, R., Watson, D. M., Crawford, M. K. & (2005) Astrophys. J. 629, 865–872.
38. Oka, T., Shimizu, F. O., Shimizu, T. & Watson, Townes, C. H. (1985) Astrophys. J. 297, 766–786. 64. Kokoouline, V. & Greene, C. H. (2003) Phys. Rev.
J. K. G. (1972) Astrophys. J. 165, L15–L19. 53. Mauersberger, R., Henkel, C., Wilson, T. L. & A 68, 012703.
12242 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0601242103 Oka