Electron impact contribution to infrared NO emissions in auroral conditions



[1] Infrared emissions from nitric oxide, other than nightglow, are observed in aurora, principally due to a chemiluminescent reaction between excited nitrogen atoms and oxygen molecules that produces vibrationally excited NO. The rates for this chemiluminescent reaction have recently been revised. Based on new measurements of electron impact vibrational excitation of NO, it has been suggested that electron impact may also be significant in producing auroral NO emissions. We show results of a detailed calculation which predicts the infrared spectrum observed in rocket measurements, using the revised chemiluminescent rates and including electron impact excitation. For emissions from the second vibrational level and above, the shape of the spectrum can be reproduced within the statistical errors of the analysis of the measurements, although there is an unexplained discrepancy in the absolute value of the emissions. The inclusion of electron impact improves the agreement of the shape of the predicted spectrum with the measurements by accounting for part of the previously unexplained peak in emissions from the first vibrational level.

1. Introduction

[2] Infrared auroral spectra were measured in a rocket experiment in 1983 [Espy et al., 1988]. At the time it was understood [Caledonia and Kennealy, 1982] that nighttime infrared emissions from vibrationally excited nitric oxide (NO*) were due to collisional excitation (i.e., translational-vibrational energy transfer) by O atoms (nightglow)

equation image

and the chemiluminescent reactions

equation image
equation image

where the N(2D) and N(4S) atoms are produced by dissociation of N2, dissociative recombination of NO+ and reactions of other species, all produced by the impact of auroral electrons [e.g., Barth, 1992].

[3] After new results for the cross sections of low-energy electron impact vibrational excitation of NO were obtained by swarm [Josić et al., 2001] and binary collision experiments [Jelisavcic et al., 2003], it was shown in an equilibrium calculation [Campbell et al., 2004] that such excitation could account for a significant part of the auroral NO infrared emission at a height of 120 km, assuming the NO density to be the maximum seen in observations. Here we revisit the subject for two reasons. Improvements to our computational model (e.g. time-dependent calculations, allowing modeling of solar input and molecular and eddy diffusion over a range of altitudes, and of a limited-duration aurora) [Campbell et al., 2006, 2007] allow predictions of both background and auroral-induced NO densities for the conditions of the measurements. Also, an updated rate for reaction (R2) has been determined [Duff et al., 2005].

[4] The NO density is calculated as a function of height in the atmosphere. This allows comparison with the measurement of the nightglow component (from reaction (R1)) and so provides a test of the computational model. The NO densities are used in making predictions of the infrared emissions produced by electron impact on NO, which are compared with the chemiluminescent emissions due to reaction (R2) and to the measured emissions.

2. Theory

[5] NO is produced and lost in the upper atmosphere by reactions involving nitrogen atoms or various ions which are produced by sunlight and by auroral electron impact [Barth, 1992]. Barth did not include the reaction N2(A3Σu+) + O → NO + N(2D) (R4) that has been shown [Campbell et al., 2007] to make a significant contribution to the calculated NO densities. The density of NO is determined by these production and loss processes and also molecular diffusion, reaching chemical equilibrium at higher altitudes, but not at altitudes around and below the peak density (∼107 km) due to downward transport by eddy diffusion.

[6] NO* produced by reactions (R1)(R3) and electron impact excitation either undergoes radiative decay, producing infrared radiation around 5.3 μm, or is quenched (i.e. returned to ν″ = 0) in collisions with N2, O2 and O. The distribution among the vibrational levels of NO for the excitation in the chemiluminescent reactions is given for N(2D) + O2 → NO*(ν′ = 1 − 12) + O (R2′) by laboratory measurements designated “LABCEDE” [Winick et al., 1987] and for N(4S) + O2 → NO*(v′ = 1 − 9) + O (R3′) by Duff et al. [1994]. The population density of the vibrational levels is determined by an equilibrium between the production and loss processes.

3. Computational Models

[7] A computational model has previously been shown [Campbell et al., 2007] to give good predictions of NO densities in the upper atmosphere in equatorial daytime and auroral conditions. Here it is applied for the time and place of the measurements [Espy et al., 1988], using atmospheric densities (of O, O2 and N2), neutral temperatures and solar activity (F10.7 index) given by the MSISE-90 model and electron temperatures from the International Reference Ionosphere (Space Physics Data Facility, ModelWeb, 2007, http://modelweb.gsfc.nasa.gov/models/). A moderate auroral flux of 0.5 erg.cm−2s−1 is assumed, this being half of the value used in similar calculations [Barth, 1992; Bailey et al., 2002]. A “statistical-equilibrium” calculation [Campbell et al., 2006] is run to determine the densities of N2(A3Σu+) due to both photoelectron and auroral excitation. This calculation is necessary because the large range of rates involved in excitation and de-excitation makes a time-step calculation impractical. A time-step simulation (which uses the populations of N2(A3Σu+) determined in the statistical equilibrium calculation to implement reaction (R4)) is then run at 0.5-s intervals for 71.25 hours (starting at midnight) to calculate the NO and N(4S) densities at 23:15 on the third day. This gives the "background" densities expected to be present before the onset of the observed bright aurora.

[8] The processes involved in de-excitation of NO* (radiative decay producing infrared emission and quenching in collisions with O2, N2 and O) are sufficiently fast that they cannot be correctly calculated in 0.5-s time intervals. Hence to calculate the infrared emissions in an aurora, the simulation is run again for 3 minutes in 0.001-s time steps, starting with the"background" densities, including those of NO and N(4S), determined in the 3-day calculation. The aurora produces much higher electron temperatures [Vallance Jones, 1974] than occur otherwise, so these are recalculated at 36-s intervals [Campbell et al., 2006].

[9] Reaction rates used for most processes are as given in previous publications [Cartwright et al., 2000; Campbell et al., 2006, 2007]. Exceptions are included in Table 1, which summarises the rates for reactions involved in the production, loss, excitation and de-excitation of NO. Isotropic photoelectron and auroral fluxes [Campbell et al., 2006] of secondary electrons, plus a Maxwellian flux of thermal electrons, were folded with cross sections to determine ionisation and excitation rates.

Table 1. Reaction Rates for Processes Involved in the Production, Loss, Excitation and De-excitation of NOa
No.ReactionRate and/or Description, cm3 molecule−1 s−1Reference
  • a

    Tn is the neutral gas temperature, Te is the electron temperature and hν represents an emitted photon.

(R1)NO (ν′ = 0) + O → NO*(ν′ = 1) + O6.5 × 10−11 exp(−2715/Tn)Wise et al., [1995]
(R2)N(2D) + O2 → NO + O9.7 × 10−12 exp(−185/Tn) (earlier rate)Herron [1999]
(R2)N(2D) + O2 → NO + O6.2 × 10−12 (Tn/300)Duff et al. [2005]
(R2′)N(2D) + O2 → NO*(ν′ = 1 − 12) + O“LABCEDE ” distributionWinick et al. [1987]
(R3)N(4S) + O2 → NO + O1.1 × 10−14Tn exp(–3150/Tn)Pintassilgo et al. [2005]
(R3′)N(4S) + O2 → NO*(ν′ = 0 − 9) + Odistribution between vibrational levelsDuff et al. [1994]
 N(4S) + NO → N2(ν′ ≈ 3) + O1.05 × 10−12 Tn0.5Pintassilgo et al. [2005]
 N(2D) +NO → N2 + O6 × 10−11Herron [1999]
(R4)N2(A3Σu+) + O → NO + N (2D)2.0 × 10−11Kochetov et al. [1986]
 NO (ν″ = 0) + e → NO(ν′ = 1 − 3)calculated using cross section, TeCampbell et al. [2004]
 NO(ν′) → NO(ν″) + hvvaries with ν′, νCartwright et al. [2000]
 NO(ν′) + M → NO (ν″ = 0) + M,varies with M = O, O2, N2Cartwright et al. [2000]

[10] Due to the relatively small radiative transition probabilities for the ground-state levels of N2 and O2, infrared emissions from these molecules are expected to be insignificant relative to NO emissions and therefore were not included.

4. Prediction of NO Densities

[11] The calculated NO densities for the time and place of the rocket measurements are shown in Figure 1 along with measurements by the SNOE (Student Nitric Oxide Explorer) satellite [Barth et al., 2003]. The satellite measurements are for local time 10–11 am, for periods of high geomagnetic activity around the equinoxes in 1998–2000 and are a global average for 65° geomagnetic latitude. These conditions differ from those of the rocket measurements, and hence of the current calculations, in that solar activity is estimated (using the MSISE-90 model) to be higher on average, solar input is less (as the rocket measurements were made on April 13th, closer to midsummer) and the rocket measurements were made at ∼11 pm. The estimated 18% uncertainty in the measured densities and the height resolution [Barth et al., 2003] is indicated by error bars drawn at one height.

Figure 1.

Calculated NO density at 12-hour intervals in daytime (plus, triangle, square) and nighttime (cross, open circle, solid circle) and the measurements (solid curve) with sample error bars for density and the altitude range.

[12] The calculated NO densities are shown for 11 am and 11 pm in a time-step calculation which starts with zero density on the first day. Above 130 km the NO densities oscillate between a higher daytime value and a lower nighttime value and are almost unchanged at the same time of day during the final 36 hours, showing that the production and loss processes of NO are reaching equilibrium for the time of day. The smaller densities at 11 hours are due to molecular diffusion, as more NO will move downwards when the densities at lower altitudes are (unrealistically) small. Below 130 km the NO density builds up with time but is approaching a constant value by 71 hours.

[13] The predicted NO density at 150 km in daytime is ∼50% higher than the measurements. This may be explained by the sum of the 18% uncertainty in the measurements, the uncertainties in the solar activity (∼10% [Barth et al., 2003]) and auroral flux, and the higher solar insolation at the time of year of the calculations. For altitudes 105–135 km the calculated densities agree with the measurements to within the uncertainty. The differences in the shape of the altitude profile and the height of the peak NO density may be due to an inappropriate “characteristic energy” of the auroral flux. The model used here assumes a characteristic energy of 3.1 keV [Campbell et al., 2006], but it has been shown [Bailey et al., 2002] in modeling that the height of the peak NO density is predicted to be higher for lower characteristic energies.

[14] The densities at 11 pm on the third day are replotted in Figure 2, along with those predicted by running the same calculation but with either reaction (R4) omitted or using the earlier rate for reaction (R2). There is a reduction in the NO density which increases with height when the rate of Duff et al. [2005] is replaced by the earlier rate of Herron [1999], as expected because the more recent rate increases more with increasing temperature and the temperature increases with height above 100 km. Omission of reaction (R4) reduces the NO densities only slightly at higher altitudes, as expected because the production rate (which is proportional to the populations of the reactants as well as the reaction rate) for reaction (R4) is much lower than that for reaction (R2). At lower altitudes, however, reaction (R4) has a much larger effect, producing a factor of ≥2 difference in the calculated NO density. This is because reaction (R4) produces N(2D) as well as NO, which then produces more NO by reaction (R2) so that this double path for NO production has a disproportionately large effect [Campbell et al., 2007].

Figure 2.

Calculated NO densities at 71 hours for the full calculation (solid circle) and for the same calculations without the reaction N2(A3Σu+) + O → NO + N(2D) (open circle) or using the lesser N(2D) + O2 rate (solid square).

[15] Around the peak NO density the inclusion of N2(A3Σu+) + O has a much greater effect than changing the reaction rate for the N(2D) + O2 production path. The calculated NO density profile is closer to the available measurements (in both absolute values and shape variation with altitude) when the excited N2 production is considered.

5. Prediction of NO Infrared Emissions

[16] Measurements by Espy et al. [1988] of the NO(Δν′ = 1) emissions are shown in Figure 3, as column emission rate (apparently at zenith, although this is not explicitly stated) as a function of upper level ν′. The measurements are shown by an error bar covering the range estimated by Espy et al.. Their analysis was not straight forward, as the emissions for each level are spread out in overlapping rotational structures and must all be extracted from the much larger nightglow spectrum.

Figure 3.

Column emission rates of NO(ν′ → ν′ − 1) deduced [Espy et al., 1988] from measurements (error bars), and as predicted for 5- and 40-kR aurora for nightglow (open circle), chemiluminescent reactions (square), electron impact (triangle) and the sum of these contributions (solid circle). The nightglow values are reproduced in the inset.

[17] To simulate the measurements, the calculated densities [including of NO and N(4S)] at 71.25 hours are transferred to another version of the same model, which is run at 0.001-s time steps for 3 minutes with an auroral energy flux of 3.4 erg.cm−2s−1. This corresponds to an aurora with a brightness of 557.7-nm emission of 5 kR [Campbell et al., 2006] and is intended to emulate the conditions quoted for the rocket measurements. The aurora was measured from the ground as having a brightness of ∼50 kR shortly before launch, but declined to ∼5 kR during the time of the measurements.

[18] The predictions for the NO(ν′ → ν′ − 1) emissions at the end of the application of a 3-minute 5-kR aurora are also shown in Figure 3. Values are given for the emissions from each upper vibrational level due to electron impact, chemiluminescence and the sum of these two processes.

[19] The predicted value for the nightglow for a 5-kR aurora (see inset) is almost within the error range of the measurements. This suggests that the assumed background auroral input and the calculated NO densities are reasonable for the conditions of the measurements. The uncertainty due to the magnitude of the background auroral impact is indicated by the larger predicted nightglow emissions for a 40-kR aurora.

[20] The predicted emissions for the chemiluminescent reactions are only about 12% of the measured values. As the methods of calculation, such as in calculating the column emission rates from the height profile of volume emission rates, are the same for the nightglow and chemiluminescent emissions, the reasonable value for the nightglow suggests that the discrepancy for the chemiluminescent emissions is not due to a gross mistake.

[21] As noted above the observed aurora decayed from 50 kR to 5 kR. The calculated emissions for a 40-kR aurora in Figure 3 show agreement of the predicted chemiluminescent emissions with the measurements of NO(ν′ → ν′ − 1) for ν′ = 2 − 7. This suggests the possibility that the auroral strength was still at a higher value at greater heights (where most of the chemiluminescent contribution originates) when it had declined at lower heights where most of the 557.7-nm light originates. Such differences may be possible due to the spatial variation in the auroral brightness referred to by Espy et al..

[22] Irrespective of the strength of the aurora, the shape of the spectrum of the chemiluminescent emissions is consistent with that of the measurements for ν′ = 2 − 7. However, the measurements show an enhancement for the non-nightglow component of NO(1 → 0) which is not apparent in the predictions for the chemiluminescent emission. Espy et al. suggest this enhancement could be due to the difficulty of separating the nightglow and chemiluminescent components. However, the predicted emissions due to electron impact are as large as for the chemiluminescent reactions for NO(1 → 0) in a 5-kR aurora. Adding the electron impact contribution to the chemiluminescent emissions gives a spectral shape closer to that of the measurements, although the ratio of NO(1 → 0) to NO(2 → 1) is still larger in the measurements. (The model does not include energy transfer from O2(a1Δg) to NO, but as the excitation energy of O2(a1Δg) is much larger than that of NO(ν′ = 1), it is unlikely that this would preferentially excite NO(ν′ = 1). Another possible source of excitation is collisions with fast neutral molecules, produced in charge-exchange with ions accelerated by electric fields. The modeling of such fields is beyond the scope of this work.)

[23] For the 40-kR aurora the electron impact contribution is not as large as the chemiluminescent contribution, because the increase in electron temperature is not a linear function of auroral strength [Campbell et al., 2006] while the chemiluminescent reaction scales linearly. However, the shape of the spectrum is still in better agreement with measurements when the electron impact contribution is included.

6. Conclusions

[24] Calculations of infrared emissions from vibrationally excited nitric oxide in auroral conditions have been compared with rocket measurements. Both the density of NO and the emissions from it are calculated. The role of electron impact is considered, including the importance of excited N2 in the production of NO and the contribution of direct vibrational excitation of NO. The predicted NO densities are reasonable when compared with available satellite measurements, as are the nightglow emissions implied by these densities when compared with the rocket measurements. The calculated spectrum of Δν = 1 infrared lines for nightglow, chemiluminescent reactions and direct electron impact excitation compares well in shape with the measurements, particularly in the enhancement of the auroral NO(1 → 0) emissions due to electron impact. However, the absolute values of the predicted auroral emissions are substantially less than those measured.


[25] This work was supported by the Australian Research Council.