A previously unrecognized source of the O2 Atmospheric band emission in Earth’s nightglow

See allHide authors and affiliations

Science Advances  20 Mar 2019:
Vol. 5, no. 3, eaau9255
DOI: 10.1126/sciadv.aau9255


Earth’s night sky continuously produces a faint chemiluminescence known as nightglow. Two prominent nighttime emissions around 90 km are the O2 Atmospheric and the OH Meinel band systems. Despite a plethora of studies since their identification seven decades ago, substantial gaps persist in our understanding of the mechanisms that control them. This report shows that oxygen atoms connect these two emissions: Fast, multiquantum, vibrational-to-electronic relaxation of OH(v) by O atoms activates a pathway that generates O2 Atmospheric band emission. This newly discovered source exhibits a strong altitude dependence and can contribute a majority of the observed O2 Atmospheric band emission when the peaks of the OH and O-atom layers overlap. The new findings call for a reinterpretation of Earth’s nightglow emissions and a revision of relevant atmospheric models.


Bright auroral displays have inspired awe for millennia. In contrast, the weak background radiation of Earth’s night sky has remained relatively unnoticed (1, 2). Photometric observations in the early 1900s established that starlight and scattered light combined could not account for the brightness of the sky and provided the first quantitative information on the variability of nighttime spectral features (3, 4). This knowledge planted a fertile seed for the field of upper atmospheric science known as aeronomy, and numerous studies of the nightglow ensued. Initially, emphasis was placed on emission from atomic oxygen, but by the middle of the 20th century, excited molecular species were known to be important for the nightglow (1).

Photochemical processes initiated by solar radiation in the upper atmosphere during the day yield energetic atoms and molecules, some of which emit radiation—referred to as dayglow and nightglow depending on when the emission occurs. The O2 Atmospheric and OH Meinel band emissions (Fig. 1A) are two important nightglow features (1). Resolved spectra of both emissions were first recorded from ground-based observations by astronomer Aden B. Meinel in 1948 (5). The identification of the (0,1) O2(b1g+X 3g) Atmospheric band (AB) emission near 864.5 nm was enabled by its detection in laboratory discharge afterglows a year earlier (6). In contrast, no laboratory comparison was available for the OH Meinel band emission in 1948, and its identity was confirmed 2 years later (710).

Fig. 1 The OH Meinel and O2 AB systems are important features of the nightglow.

(A) Astronomical night sky spectrum obtained using the Echelle Spectrograph and Imager on the Keck II telescope displays strong OH Meinel band emission (4 March 2000, 8:58 UT; λ/Δλ, ~7000; 50-min exposure) (14). The inset shows the weak (0,1) band of the O2 AB system near 864.5 nm. (B) Potential energy curve and vibrational levels of the OH ground state. The red arrow indicates a representative OH Meinel feature—the (8,4) band near 937 nm. (C) Observed OH vibrational population distribution in the nightglow (green) and nascent distribution from the H + O3 reaction (blue) (12, 14). (D) Low-lying electronic states of O2. The red arrow indicates the strong (0,0) O2 (b1g+X3g) emission near 762 nm that is absorbed by the atmosphere and is observable only from space.

Reliable interpretation of nightglow requires understanding how the emitting layer comes to existence. Therefore, detailed knowledge of the relevant photochemical production mechanisms is needed. The processes that consume the emitting species are just as important, and the observed intensity reflects the competition of the production and loss pathways.

The source of vibrationally excited OH in the upper atmosphere is the highly exothermic reaction of hydrogen atoms with ozone (1, 7, 8, 11, 12)H+O3ΟΗ(v=59)+O2(1)

Water photodissociation gives rise to a layer of hydrogen atoms, whereas the three-body association O + O2 + M (M = N2, O2) is responsible for the generation of ozone (1, 7, 8). The OH Meinel band emission involves transitions between rotational-vibrational levels within the OH ground state (Fig. 1B). The emission varies as a function of geographic location, time, and season. Its altitude profile exhibits a peak near 87 km with a full-width-at-half-maximum value of approximately 8 km (13). Most of the available energy from the H + O3 reaction is released into the OH product. The highest energetically allowed vibrational level, v = 9, is also the most probable nascent level (11, 12). Nevertheless, collisional relaxation efficiently redistributes the initial OH(v) excitation, and the observed emission indicates a steady-state vibrational population distribution that markedly differs from the nascent one, i.e., generated by the H + O3 reaction in the absence of collisions (Fig. 1C) (12, 14).

The maximum intensity of the O2 AB emission usually occurs near 94 km, a few kilometers below the maximum of the O-atom concentration (1518). Three-body O-atom association, O + O + M (M = N2, O2), to form molecular oxygen is considered the source of this emission at night. This process generates all the low-lying electronic states of O2 (Fig. 1D). The nascent O2 molecule is initially formed near the dissociation limit, and the shallow 5Πg electronic state at large internuclear distance is believed to play an important “gateway” role (15, 16). Collision-induced vibrational relaxation, electronic energy exchange, and intersystem crossings redistribute the nascent molecules among the low-lying O2 electronic states and eventually populate the emitting O2(b1g+) state. The O2(b1g+, v = 0) vibrational level generates the strongest emission because it is chemically inert (15, 16). The most intense emission band, (0,0) O2(b1g+X 3g) near 762 nm, is approximately 21 times stronger than the (0,1) band near 864.5 nm but can only be observed from space because it is absorbed by Earth’s atmosphere (16). Nightglow emission from higher O2(b) vibrational levels has also been observed, but it is extremely weak (16). Any subsequent mention of the O2 AB will refer to emission from the (0,0) band.

Since the 1980s (1518), the generally accepted mechanism for the generation of O2 AB emission has beenO+O+MO2*+M(M=O2,N2)(2)O2*+O2O2(b,v=0)+O2(3)O2(b,v=0)O2(X)+hν(4)where O2* is an electronically excited precursor that leads to the emitting level following collisions with ground-state O2 molecules and hν represents the emitted photon.

Despite numerous investigations of these two nightglow emissions during the past seven decades, notable gaps persist in our knowledge. Regarding the OH Meinel band emission, the applicable relaxation processes and observed rotational and vibrational population distributions are only partially understood. For the O2 AB emission, the details of its production mechanism, the relevant intermediate precursors, and their yields remain elusive. The empirical equations used in the literature to relate its intensity to the O-atom concentration generally account for the observed emission and altitude dependence (17, 18), but the underlying processes and scaling factors used have not been confirmed by laboratory experiments. Moreover, as shown in reactions 2 to 4, these processes consider only collisions of O2 with the precursor excited state. Laboratory experiments determined comparable rate constants for vibrational relaxation of O2(b, v = 1) by O and O2 and thus cast doubt on the validity of the assumption that O atoms have no role in the excitation transfer step of reaction 3 (19, 20). The aforementioned problems have stubbornly persisted for decades because these nightglow emissions depend on interactions involving atmospheric species such as atomic hydrogen and oxygen that are extremely difficult to probe in situ and in the laboratory.


Laboratory studies reveal a previously unrecognized source of nighttime O(1D) atoms and O2 AB emission

Approximately a decade ago, experiments investigating OH collisional relaxation yielded an unexpected result: The deactivation of OH(v = 9) by ground-state O(3P) atoms was found to be extremely fast, with a total removal rate constant of (4 ± 1) × 10−10 cm3 s−1 at room temperature (21). Such a large value approaches the gas-kinetic limit and, at first, could not be accommodated in atmospheric models (2224). This quandary defied a solution until recently (25, 26), when laboratory measurements demonstrated that the interaction of OH(v) with O atoms involves a fast, spin-allowed, multiquantum, vibrational-to-electronic (V-E) energy transfer pathway that generates excited O(1D)OH(v5)+O(3P)OH(0vv5)+O(1D)(5)

An example of this process for OH(v = 9), the major nascent product of H + O3, isOH(v=9)+O(3P)OH(v=3)+O(1D)(6)

Multiquantum V-E relaxation by O atoms was found to be the most efficient loss process for OH(v = 9) and provided an explanation for the unexpectedly large total removal rate constant (25).

During the day, O(1D) is generated from photodissociation of O2 and O3 by solar radiation at altitudes near 90 km. These processes do not occur at night, but the OH(high v) + O multiquantum relaxation provides a source of nighttime O(1D) atoms. This source was recently shown to result in enhanced CO2 4.3-μm emission via energy transfer from O(1D) to N2 and then to CO2 and thus explained unacceptably large discrepancies between observations and model calculations of the nighttime CO2 4.3-μm emission that persisted for several decades (27).

The O(1D) atoms generated by the OH(high v) + O multiquantum vibrational relaxation process also provide a significant source of O2 AB emission. O(1D) atoms rapidly transfer energy to ground-state O2 and generate electronically excited O2(b) molecules (28, 29)O(1D)+O2O2(b,v=0,1)+O(3P)(7)

This process is important in the dayglow (29) but has never been considered for the nighttime O2 AB emission. The availability of O(1D) from the multiquantum relaxation process of reaction 5 leads to the efficient production of O2(b, v = 0) because collisions with O2 rapidly relax O2(b, v = 1) to O2(b, v = 0) (20). From the atmospheric composition and the rate constants for O(1D) relaxation by N2 and O2, we can estimate that O2 deactivates approximately 25% of O(1D) for altitudes near the OH layer (30, 31). As a result, the OH layer provides a source of the O2 AB emission, and O atoms represent the link that connects these two nightglow emissions.

Estimates of the OH(v) + O source and observations with sounding rockets

The sounding rocket campaigns Energy Transfer in the Oxygen Nightglow (ETON; multiple rockets launched on 23 March 1982 from South Uist, Scotland) and Non-local Thermodynamic Equilibrium (NLTE; two rockets launched in March 1998 from Esrange, Sweden) measured the O2 AB emission and O atoms under geomagnetically quiet conditions (17, 32). Here, we present an analysis of these observations that demonstrates the importance of the new O2 AB emission source. Only OH(v = 9) is considered because this is the only vibrational level for which the relevant kinetics parameters have been confirmed in laboratory measurements. Additional contributions to the O2 AB emission can be expected to the extent that process 5 is active for OH(v = 5 to 8). A compelling indication that this may be the case arises from the success of recent efforts to model the OH Meinel band and CO2 4.3-μm emissions (27) and to retrieve O atoms from OH Meinel band observations (33). These calculations determined that considering OH(v = 5 to 8) multiquantum relaxation by O atoms together with OH(v = 9) + O is required to reproduce the observed OH vibrational population distribution and the magnitude of the enhancement in the CO2 4.3-μm emission. The kinetics parameters used for the OH(v = 5 to 8) + O processes were estimated on the basis of currently available knowledge (27) but have yet to be confirmed. As mentioned above, for the purpose of demonstrating the existence of the new source of the O2 AB emission in this report, we restrict the calculations to the more extensively studied OH(v = 9) + O process. Thus, the estimates for the O2 AB new source from OH(v = 9) + O reported here should be considered lower limits. Table 1, Methods, and the Supplementary Materials provide detailed information on the relevant mechanistic steps, equations, and chemical kinetics parameters used.

Table 1 Processes and relevant kinetics parameters used in the calculation of the contributions of OH(v = 9) + O multiquantum vibrational relaxation to the O2 AB emission.

When a kinetics data compilation is available, that reference is shown for the sake of brevity. Additional details are provided in Methods.

View this table:

Figure 2 demonstrates that OH(v = 9) + O multiquantum relaxation is an important source of O2 AB emission. The figure presents observations from the ETON and NLTE-2 sounding rocket campaigns together with the corresponding atomic oxygen profiles and calculations of the O2 AB emission generated from the OH(v = 9) + O source. In each case, two estimates that reflect limiting values for the removal rate constant of O2(b, v = 0) by O atoms (see detailed discussion in Methods) are shown. A key feature of the two datasets of Fig. 2 is that they are characterized by significant differences in the measured altitude profiles for O atoms (Fig. 2B). For the ETON measurement, the O-atom layer peaks near 102 km. In contrast, during the NLTE-2 flight, downward transport of the O-atom layer appears to have shifted the number density maximum to a much lower altitude, 90 km. This altitude coincides with the region where the peak of the vibrationally excited OH layer appears and provides a key test for the OH(v) + O source. The more extensive the overlap of the OH and O-atom layers, the stronger the O2 AB emission expected from OH(v) + O. Thus, despite the larger O-atom peak concentration measured during ETON, the lower altitude of the O-atom layer maximum and better overlap with the OH layer in the case of NLTE-2 explain the substantially stronger NLTE-2 O2 AB emission compared with ETON. The results of Fig. 2 demonstrate that the new pathway originating from OH(v = 9) + O relaxation is an important source of O2 AB emission. In the case of NLTE-2 measurements, where the O-atom and OH layers overlap optimally, the new pathway represents more than half of the observed emission.

Fig. 2 The OH(v) + O source of O2 AB emission depends on the altitude profile of the O-atom layer.

(A and C) Measured (0,0) O2 AB volume emission rate as a function of altitude for ETON (gray circles) and NLTE-2 (black circles) and calculated contributions of the source from OH(v = 9) + O using the lower and upper limit rate constant values of 0 (blue) and 8 × 10−14 cm3 s−1 (green) for removal of O2(b, v = 0) by O atoms. (B) Measured atomic oxygen number density as a function of altitude for ETON (gray) and NLTE-2 (black). Note that despite the larger O-atom density during ETON, the better overlap of the O-atom profile with the OH layer during NLTE-2 leads to stronger O2 AB emission originating from OH(v = 9) + O.

Figure 3 compares the ETON and NLTE-2 data with the estimates of the O2 AB emission from the OH(v) + O source and possible contributions from O + O + M recombination for upper and lower limit scenarios. Additional details on these comparisons and relevant calculations are presented in Methods and the Supplementary Materials. The rocket observations and model calculations shown in Fig. 3 show overall good agreement. This is a promising result with the caveat that all the processes of interest are not fully quantified yet. Discussion and Methods consider the limitations of the current state of knowledge. A key point to highlight is that if process 5 is confirmed to be active for OH(v = 5 to 8) with similar rate constants to OH(v = 9), then the new source of O2 AB emission will be the dominant source below approximately 95 km. Notably, the steady-state population in vibrational levels OH(v = 5 to 8) is almost one order of magnitude larger than that in OH(v = 9) (Fig. 1C) (12, 14), which implies that even smaller rate constants for OH(v = 5 to 8) + O should be expected to have non-negligible contributions.

Fig. 3 Estimated contributions from OH(v) + O and O + O + M to the O2 AB emission.

(A and C) Measured (0,0) O2 AB volume emission rate profiles for the ETON and NLTE-2 datasets (gray and black circles, respectively) and calculated contributions from OH(v = 9) + O (green) and O + O + M (violet). The sum of the two contributions is also shown (orange) (see Methods for additional information). (B and D) As in (A) and (C) for O + O + M and assuming as an upper limit that OH(v = 5 to 8) produced from the H + O3 reaction exhibits a behavior similar to OH(v = 9) + O. OH(v) + O and O + O + M are assumed to be the only active sources of the O2 AB emission.


Besides its fundamental significance for molecular energy transfer, the new OH(v) + O multiquantum vibrational relaxation pathway of reaction 5 represents an important advancement in our understanding of nightglow emissions. These processes have several implications for the nighttime upper atmosphere that relate to the determination of its composition by remote sensing, understanding the energy flow and chemical heat deposition, diurnal and seasonal variability, atmospheric circulation, wave propagation, and winds. Because the altitude region around 100 km cannot be easily studied in situ, nightglow emission signatures observed from space- and ground-based instruments are commonly used as proxies for the concentration of atmospheric species. Monitoring their intensity and temporal evolution by remote sensing provides the method of choice to study the phenomena mentioned above. These observations can be reliably interpreted only when the mechanistic details of the underlying processes are understood and quantified at the atomic and molecular level.

The H + O3 and the O + O + M reactions are two of the most important sources of chemical energy release in this region of the atmosphere, often referred to as mesosphere and lower thermosphere (MLT). Most of the energy released from the newly formed chemical bonds of the above two reactions initially appears as excited OH (vibration and rotation) for the former reaction and excited O2 (electronic, vibrational, and rotational) for the latter reaction. Collisional processes influence the pathways of the energy flow through the dense manifold of excited quantum levels and thus the heating efficiency, i.e., the fraction of the chemical energy that ends up as kinetic energy of atmospheric gases. Consequently, the rate and altitude dependence of chemical energy deposition is strongly influenced by the location, altitude profile, and transport dynamics of the atomic oxygen layer. A portion of the released energy is radiated into space, which, in turn, implies that detailed knowledge of the processes involved is required to quantify the energy balance of the nighttime MLT region and to fully understand the overall response of the atmosphere to the daily solar forcing, the solar cycle, other long-term changes, and transient disturbances such as geomagnetic storms.

O atoms represent key reactive species for the MLT and are intimately connected to nightglow emissions. An important point is that the number density of O atoms in the MLT region exhibits a steep gradient that guarantees a strong altitude dependence for the processes that O atoms influence, including the multiquantum vibrational relaxation pathways of reaction 5. For example, the volume mixing ratio of O atoms is approximately 5 × 10−4 at 85 km and increases by as much as two orders of magnitude near 100 km (17, 30). This altitude dependence also affects vertical tidal transport of O atoms. Because direct in situ measurements of the O-atom concentration are difficult, a detailed knowledge of nightglow emissions allows us to infer from them the atomic oxygen concentration profile. Recent model calculations that used observations by the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) instrument on board the Thermosphere Ionosphere Mesosphere Energetics and Dynamics (TIMED) orbiter mission and included the OH(v) + O multiquantum vibrational relaxation processes of Eq. 5 retrieved atomic oxygen concentrations 20 to 50% lower than previous analyses (33). The revised concentration of atomic oxygen reduced the mean chemical heating rate at an altitude near 90 km by 10 K/day and the mean radiative cooling by CO2 emission at 15 μm by 2 K/day, thus bringing the global annual chemical heating and radiative cooling rates near balance (33).

The OH(v) + O source of the O2 AB emission signifies that O2 electronically excited states are not generated only by the O + O + M association process, as previously thought. The emission resulting from the OH(v) + O source favors the lower altitude range in which the nighttime O2 AB emission is observed (Fig. 2, A and C). Assuming that no other source for O2 AB exists, the difference between the observed O2 AB and the OH(v) + O source can be attributed to O-atom association. Therefore, the mechanism adopted during the past four decades for O2 AB generation by O + O + M does not adequately describe the altitude dependence of the difference. Additional details and suggestions for improvements based on what we know from recent laboratory studies are provided in Methods. Looking forward, it is crucial to determine the role of the multiquantum vibrational relaxation pathways OH(v = 5 to 8) + O, so that together with OH(v = 9) + O, we can fully quantify the OH(v) + O source of the nighttime O2 AB emission. This is also a key first step for future efforts to better understand the mechanistic details of O-atom association and the cascade pathways through the quantum-state manifold of electronically excited O2.


O atoms have an even more complex and elaborate role in the upper atmosphere than previously appreciated. Their interaction with OH(v) leads to O2 AB emission, and thus, O atoms effectively connect the two emissions. The new findings call for a reinterpretation of Earth’s nightglow emissions and a revision of relevant atmospheric models. This report opens a new chapter for Earth’s upper atmospheric nightglow emissions and aims to stimulate synergistic efforts between observers, modelers, theoreticians, and experimentalists to further deepen our understanding of planetary atmospheres.


This section begins with a presentation of the formulas used to estimate the new source of O2 AB emission. Next, we provide information on the datasets and model atmosphere inputs used for the calculations. We then discuss the parameters of Table 1 and relevant uncertainties and elaborate on the results by providing additional comments and a consistency check. Last, the focus shifts to the O + O + M association reaction and the implications for this process with regard to the O2 AB emission.

Calculation of O2 AB emission from OH(v = 9) + O

The steps that control the generation of O2 AB emission from OH(v = 9) + O collisions can be summarized in reactions M1 to M9. Process M4, loss by atmospheric species, includes inelastic scattering and reaction with O atoms that leads to O2 and hydrogen atoms (31). Process M5 is the key step that generates O(1D) and effectively couples the OH Meinel and the O2 AB emissions. Loss of O(1D) by ground-state O atoms is negligible for the altitude range considered here (34)O+O2+MO3+M(M=O2,N2,O)(M1)H+O3OH(v=9)+O2(M2)O3+OO2+O2(M3)OH(v=9)+Mproducts(M=O2,N2,O)(M4)OH(v=9)+O(3P)OH(v=3)+O(1D)(M5)O(1D)+N2O(3P)+N2(M6)O(1D)+O2O2(b,v=0,1)+O(3P)(M7)O2(b,v=1)+O2O2(b,v=0)+O2(M8)O2(b,v=1)+OO2(b,v=0)+O(M9)

We applied the steady-state approximation to chemical species O(1D), OH(v = 9), and O3. The assumption of photochemical equilibrium in the mesosphere is often used in the literature (12, 3537). It is a reasonable approximation for this case that concerns rapid, local measurements of slowly varying nightglow emissions. The steady-state concentration of OH(v = 9) and O2(b, v = 0) generated from source (M5) is described by the equations[OH9]=y9 × kOO2M × kHO3 × [O] × [O2] × [M] × [H]/{LT(OH9) × LT(O3)}(M10)[O2b0]=[OH9] × yO2b × kOH9NM × kODO2 × [O] × [O2]/{LT(O1D) × LT(O2b0)}(M11)where [M] = [O2] + [N2] + [O]. The yields yi and rate constants ki are listed in Table 1. The factors LT represent the loss terms for OH(v = 9), O2(b, v = 0), O3, and O(1D), respectivelyLT(OH9)=A9+kOH9O × [O]+kOH9O2 × [O2]+kOH9N2 × [N2](M12)LT(O2b0)=AO2b+kO2bO × [O]+kO2bO2 × [O2]+kO2bN2 × [N2](M13)LT(O1D)=AOD+kODO2 × [O2]+kODN2 × [N2](M14)LT(O3)=kHO3 × [H]+kOO3 × [O](M15)

Ai are the relevant radiative decay rates for O(1D), the (0-0) band of the O2 AB emission, the total emission from O2(b, v = 0), and OH(v = 9), with values 0.009, 0.079, 0.083, and 173 s−1, respectively (31, 38, 39). The observed volume emission rate for an atmospheric species at the altitude of interest is given by the product of its radiative decay rate and number density.

Data and model atmosphere

The ETON campaign was launched on 23 March 1982 from the island of South Uist in Scotland (57.36°N, 7.38°W). Multiple rockets measured atomic and molecular nightglow emissions as well as the atomic oxygen density profile (17, 40). ETON payloads P230H and P227H, from which the observed O2 AB emission profiles were used here, were launched at 21:27:17 UT and 22:10:51 UT, respectively (17).

The NLTE campaign comprised two rockets launched from the Esrange Space Center (68°N, 21°E), Sweden in March 1998 (32, 41). The payload of the second rocket, NLTE-2, was launched on March 6 at 21:26:00 UT and included photometric measurements of the O2 AB emission with simultaneous monitoring of atomic oxygen.

The atmospheric composition, [N2], [O2], [H], and temperature inputs were taken from the NRLMSISE-00 model (30) for the time and date of each dataset. For the atomic oxygen profile, we used the observations of ETON payload P232H and NLTE-2 descent flight. In our discussion of possible sources of error, we will revisit the atomic oxygen profiles used.

Parameter inputs and relevant uncertainties

Table 1 lists information on the input parameters used to calculate the contributions of the OH(v = 9) + O multiquantum relaxation to the O2 AB emission. The table shows the values and associated uncertainty estimates at 298 K and, where available, an expression that describes their temperature dependence. The latest Evaluation on Chemical Kinetics of the Jet Propulsion Laboratory (JPL) was the reference of choice for several processes (31). Additional comments are provided next.

For the rate constant of the H + O3 reaction, experimental studies by Liu et al. (42) have corroborated the most recent JPL recommendation and provided new measurements at low temperatures. The yield of OH(v = 9) from H + O3 used in this work follows the recommendation of Adler-Golden (12), which was based on the critical evaluation by Steinfeld et al. (43). The OH(v = 9) total removal rate constant for collisions with O, O2, and N2 as well as the rate constant for the OH(v = 9) + O multiquantum relaxation adopted the values reported by Kalogerakis et al. and Sharma et al. (21, 25, 26) and assumed a temperature dependence similar to that observed by Thiebaud et al. (44). For the collisional relaxation of O2(b, v = 0) by O2, N2, and O, the values of the current JPL evaluation were used (31). Because of the uncertainty in the rate constant for O2(b, v = 0) by O, measured by Slanger and Black (45) at 298 K, and the lack of studies at low temperatures, a common practice in the literature (17, 18, 32) has been to use a lower limit of zero and a high value of 8 × 10−14 cm3 s−1, i.e., the measured rate constant at room temperature. We adopted this approach and, in Fig. 2, show results for the new source of O2 AB from OH(v = 9) + O using both limiting values of kO2bO, the rate constant for removal of O2 (b, v = 0) by O atoms.

Equations M10 to M15 used to calculate the contributions of the new source contain several kinetics parameters and variables, many of which have significant uncertainties. We will now consider the kinetics inputs before discussing specific details pertinent to the model atmospheric composition used. To estimate the uncertainty for the results presented in Fig. 2, we considered the uncertainty of all kinetic parameters in Eqs. M10 to M15 at a nominal temperature of 200 K. We used the information listed in Table 1 and an uncertainty estimate of 10% for the nascent yield y9 based on the uncertainties reported for the experimental measurements of Klenerman and Smith (46). The largest uncertainty contributions arise from parameters yO2b, LT(OH9), and kOH9NM with uncertainties of 25, 23, and 22%, respectively. The combined uncertainty of all parameters propagated through the formulas for O2 AB generated from OH(v = 9) + O amounts to an SD of 57%.

As already mentioned, the [N2], [O2], [H], and temperature inputs are the profiles of the NRLMSISE-00 model for the time of the rocket launch of interest. The atomic hydrogen layer displays a steep “ledge” near 80 km where its number density increases by almost an order of magnitude within a few kilometers. Moreover, the layer and its peak near 85 km may move during the night in a manner not accurately captured by the NRLMSISE-00 model (30). Thus, the total intensity of the new source for O2 AB could be affected, especially the slope of the emission’s onset at altitudes below 85 km.

The intensity of the OH(v) + O source of O2 AB strongly depends on atomic oxygen. Accurate determination of the O-atom profile has been a long-standing challenge for upper atmospheric studies, and simultaneous measurements of O atoms are usually missing from nightglow observations. Such information is available for the ETON and NLTE sounding rocket campaigns. Nevertheless, a discussion of substantial issues with these measurements is warranted. For ETON, Greer et al. (40) summarized the difficulties with partial detector saturation and signal interference in the measurements and how these were handled. The atomic oxygen input for our calculations is from the ETON P232H payload, the first of two payloads that performed these measurements. For the ETON O2 AB dataset, we used the average of two measurements from payloads P230H and P227H, launched 21 and 22 min before and after P232H (O-atom profile), respectively. The atomic oxygen profile from payload P232H can be considered representative of the conditions during the two O2 AB measurements that preceded and followed it. Nevertheless, the second ETON O-atom profile (P234H), recorded approximately 126 min after the first (P232H), differed significantly at most altitudes and had a peak O-atom concentration that was larger than that of P232H by approximately 35%. Thus, while the choice of the datasets aims at mitigating any inconsistencies in the measurements, we cannot exclude the possibility of systematic errors.

The NLTE data have its own set of challenges, mainly related to the calibration of the resonance fluorescence measurements. The O-atom dataset from the descent of the NLTE-2 flight was calibrated according to the work of Hedin et al. (32). These authors proposed a method that uses nightglow intensity profiles (O2 Chamberlain and AB bands) and the available empirical equations that relate the observed intensity to the O-atom concentration. From a study of several measured O-atom profiles, Hedin et al. (32) found that calibrating them in this manner resulted in an improvement of more than one order of magnitude in the spread of the retrieved O-atom concentrations. Using this approach for the NLTE-2 descent, we find an O-atom concentration peak of 4.2 × 1011 atoms/cm3 (Fig. 2). For the NLTE-2 ascent, Hedin et al. (32) obtained values from 4.4 × 1011 to 4.7 × 1011 atoms/cm3. The ascent and descent flight trajectories are separated by several kilometers and are likely subjected to different aerodynamic conditions. Given the uncertainties in the measurements and their calibration, and the fact that we now know that multiple production mechanisms are active, significant systematic errors cannot be excluded. Several authors have also considered complications in similar nightglow measurements such as the possibility of transient rocket-induced glow, nonlinearities in the measurements, or other contaminating emissions (32, 47, 48). The availability of accurate information on the atomic oxygen layer remains a challenge for the analysis and interpretation of nightglow emissions. Nevertheless, the uncertainties in the ETON and NLTE-2 atomic oxygen profiles do not put in question the conclusion that OH(v = 9) + O multiquantum relaxation is a significant source of O2 AB emission.

Consistency check for OH(v = 9) density

We can examine whether the OH(v = 9) number density implied by the calculations of the OH(v = 9) + O O2 AB emission source is realistic. Some relevant information is available because ETON payload P228H obtained the emission profile of the (8,3) OH Meinel band (49). Figure S1 presents the OH(v = 8) concentration as a function of altitude, inferred from the P228H (8,3) OH Meinel band emission measurement using an Einstein coefficient of 0.9336 s−1, reported by Xu et al. (22). The figure also presents the calculated [OH(v = 9)] using Eq. M10. For this calculation, the second O-atom measurement (P234H) was used as input because it is the closest one in time—the (8,3) OH band emission was measured ~26 min before P234H and 100 min after P232H (first O-atom measurement). The integrated OH(v = 8) population corresponding to the ETON observation is approximately 50% larger than that calculated for OH(v = 9), in very good agreement with observations of the OH(8)/OH(9) ratio for the Meinel band emission reported in the literature (12, 14, 50).

O + O + M association and the new source for O2 AB emission

Having established the important role of the source for O2 AB emission from OH(v = 9) + O multiquantum relaxation, it is informative to consider the three-body O + O + M association reaction and the implications of the new findings. The generally accepted mechanism for the formation of O2 AB emission from O-atom association was established in the 1980s (15, 17, 18). As mentioned, it involves excitation energy transfer from an initially formed electronically excited precursor to O2(b, v = 0). The exact nature of the excited precursor(s) and the detailed mechanistic steps are not well understood. This mechanism can be summarized by reactions M16 to M19O+O+MO2*+M(M=O2,N2,O)(M16)O2*+O2O2(b,v=0)+O2(M17)O2(b,v=0)+M(M=O2,N2,O)products(M18)O2(b,v=0)O2(X)+hν(M19)

Assuming steady-state conditions, the concentration of O2(b, v = 0) is given by (15)[O2b0]=yp × kOOM × [O]2 × [M] × ypO2 × kpO2 × [O2]/{LT(O2b0) × LT(O2*)}(M20)

The parameters kOOM, kpO2, yp, and ypO2 represent the rate constants and yields for processes M16 and M17, respectively. The term LT(O2*) describes the loss rate of the electronically excited precursorLT(O2*)=AO2*+kpO × [O]+kpO2 × [O2]+kpN2 × [N2](M21)

Equations M20 and M11 have a rather similar dependence on [O], [O2], and [M]. This explains, in part, why the empirical equations for the O2 AB emission available in the literature since the 1980s were found to describe reasonably well the observed altitude dependence of the emission and have been used to assess this emission and retrieve atomic oxygen profiles (17, 18, 32). Nevertheless, it is important to reiterate the fact that the nature of the O2* precursor and the relevant mechanistic pathways originating from O + O + M are not well known. Moreover, the rate constants and yields kpi and ypi have not been quantified at mesospheric temperatures. In the literature, the parameters of the relevant empirical equations were simply adjusted to account for the observed O2 AB emission intensity.

From Fig. 2, it is evident that the contribution of the source of O2 AB emission from the OH(v = 9) + O process M5 peaks at a slightly lower altitude than the observed profile. The remainder emission originates from additional contributions by OH(v = 5 to 8) + O and termolecular association O + O + M. The former can be expected to have a rather similar altitude dependence to that of OH(v = 9) + O. Therefore, given the presence of the OH(v) + O new source of O2 AB emission, the reaction mechanisms M16 to M19 provide a less than adequate description of the O + O + M contributions to the O2 AB emission, which peak at a higher altitude than those from OH(v = 9) + O. This is a telltale sign that contributions to the O2 AB emission from O + O + M beyond the previously accepted basic mechanism most likely involve additional interactions with O atoms. The mechanism summarized in reactions M16 to M19 appears to be an oversimplification. Numerous relaxation pathways can lead from the multitude of electronically excited O2 states near the dissociation limit to the O2(b, v) vibrational level manifold and eventually to O2(b, v = 0) (16). Oxygen atoms undoubtedly play a significant role in these processes. As already mentioned, laboratory experiments determined that the rate constants for vibrational relaxation of O2(b, v = 1) by O and O2 are comparable at room temperature (19, 20). The faster the relaxation of O2(b, v > 1) levels by O atoms occurs compared with the relaxation of O2(b, v = 1) by O, the larger the influence O atoms will have in the population flow through the O2(b) vibrational level manifold toward O2(b, v = 0). Although not directly relevant to O2(b, v) + O, we note that the removal of OH(v = 9) by O atoms at 298 K is faster than that of OH(v = 1) by one order of magnitude (21, 51). It follows that we could include one more process in the mechanism for O + O + M generation of O2 AB emissionO2*+OO2(b,v=0)+O(M22)

The revised Eq. M20 for the steady-state concentration of O2(b, v = 0) when process M22 is included can be written as in reference (15)[O2b0]=yp × kOOM × [O]2 × [M] × {ypO2 × kpO2 × [O2]+ypO × kpO × [O]}/{LT(O2b0) × LT(O2*)}(M23)

Figure 3 presents the ETON and NLTE-2 data together with best-fit curves using Eq. M23 and the kinetics parameters of table S1. The room temperature value of kO2bO, 8 × 10−14 cm3 s−1, was used in the calculations. The figure shows two limiting cases for the new source of the O2 AB emission. First, we considered only OH(v = 9) + O as a lower limit for the source of O2 AB. Then, for an upper limit, we assumed that OH(v = 5 to 8) produced from the H + O3 reaction behave similarly to OH(v = 9) in collisions with O atoms. The corresponding yields for processes M17 and M23, (ypO2, ypO), in Fig. 3 (A to D) have the values (0.24, 0.11), (0.02, 0.16), (0.02, 0.41), and (0, 0.30), respectively. As expected, when a maximum contribution for OH(v) + O is considered (Fig. 3, B and D), ypO2 becomes less important than ypO. The fact that the optimized values for each limiting case differ significantly between ETON and NLTE-2 (Fig. 3, A versus C, and B versus D) appears to indicate some inconsistencies in the description of the mechanism. Nevertheless, invoking process M22 improves the agreement with the observed profile and may be a first step toward a more complete description of the O2 AB emission. Better characterization is still required of the mechanistic details of O + O + M association as well as the contributions from OH(v = 5 to 8) + O.

On the basis of the observed OH(v) vibrational population distributions reported in the literature (12, 14, 50), the steady-state population of OH(v = 9) is approximately one order of magnitude less than the cumulative population in levels v = 5 to 8 (Fig. 1C). This observation suggests that most of the observed O2 AB emission below 95 km originates from OH(v) + O. We eagerly envision that future studies will fully quantify the O2 AB source from OH(v ≥ 5) + O and the elusive mechanism of O + O + M association.


Supplementary material for this article is available at

Fig. S1. Relevant information on the OH Meinel band emission from the ETON rocket P228H.

Table S1. Processes and relevant kinetics parameters relevant to the generation of O2 AB emission from O + O + M association.

Data S1. Data inputs and calculation outputs presented in the main text and Methods.

References (52, 53)

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: I thank R. N. Zare, K. Bergmann, and T. G. Slanger for continued support throughout my career; J. P. Simons and M. Brouard for introducing me to the subtleties of the hydroxyl radical; the members of SRI International’s (SRI) Molecular Physics Laboratory (1956–2014) for mentoring and remarkable legacy in upper atmospheric science; W. C. Olson for outstanding laboratory technical support; J. Hedin for providing data from the NLTE-2 sounding rocket campaign; R. D. Sharma, P. P. Wintersteiner, A. A. Kutepov, and P. A. Panka for insightful discussions; and J. Graham, K. Vamvaca, D. Matsiev, P. C. Cosby, and T. G. Slanger for comments on the manuscript. Funding: The seeds for this work were planted by a series of laboratory experimental projects at SRI over the course of two decades supported by the Aeronomy Program of the U.S. National Science Foundation (NSF) and the Geospace Science Program of the U.S. NASA. The specific data analysis and other preparations pertaining to this manuscript were supported by NSF award AGS-1441896, NASA award 80NSSC17K0638, and SRI internal research and development funds. Author contributions: The author conceived the ideas for this endeavor, performed the data analysis, and wrote the manuscript. Competing interests: The author declares that he has no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the author.

Stay Connected to Science Advances

Navigate This Article