Distant ionospheric photoelectron energy peak observations at Venus

The dayside of the Venus ionosphere at the top of the planet's thick atmosphere is sustained by photoionization. The consequent photoelectrons may be identified by specific peaks in the energy spectrum at 20–30 eV which are mainly due to atomic oxygen photoionization. The ASPERA-4 electron spectrometer has an energy resolution designed to identify the photoelectron production features. Photoelectrons are seen not only in their production region, the sunlit ionosphere, but also at more distant locations on the nightside of the Venus environment. Here, we present a summary of the work to date on observations of photoelectrons at Venus, and their comparison with similar processes at Titan and Mars. We expand further by presenting new examples of the distant photoelectrons measured at Venus in the dark tail and further away from Venus than seen before. The photoelectron and simultaneous ion data are then used to determine the ion escape rate from Venus for one of these intervals. We compare the observed escape rates with other rates measured at Venus, and at other planets, moons and comets. We find that the escape rates are grouped by object type when plotted against body radius. & 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Photoionization by sunlight is the principal production process in many planetary ionospheres. As neutrals from the planetary atmospheres are ionized, ions and photoelectrons result. The solar spectrum, together with the composition of the atmosphere, provides photoelectrons with particular energies. The energy of the ionizing photon beyond the ionization potential of the gas gives the emerging photoelectron kinetic energy. In particular, there are peaks in the photoelectron spectrum in the 20-30 eV region, with particular energies for the various ionospheres. A summary of the expected peak energies for Venus, Earth, Mars and Titan is given by Coates et al. (2011) (see Table 1 in reference), based on theoretical ideas including Mantas and Hanson (1979) and Nagy and Banks (1970). The ionosphere is immersed in the Venus-solar wind interaction region, and in this paper we further investigate the interaction between the ionosphere and its environment.
A substantial amount of theoretical and simulation work has been done on the Venus-solar wind interaction, its ionosphere and the morphology of the interaction. On the large scale, the magnetic field configuration at Venus is due to the solar wind interaction with this unmagnetized planet (Luhmann, 1995). A bow shock, where the solar wind slows, is heated, and is deflected around the planet. The Interplanetary Magnetic Field (IMF) penetrates the shock and 'piles up' in front of the planet. The fields interact with the ionosphere as they drape around Venus (Luhmann and Cravens, 1991;Law and Cloutier, 1995). Several simulations have successfully reproduced the large scale features of the interaction, using both MHD (e.g. Ma et al., 2013) and hybrid (e.g. Brecht and Ferrante, 1991;Kallio and Jarvinen, 2012) models. The approaches were compared by Kallio et al. (2011).
Within the Venus ionosphere, the features unique to photoelectron energy spectra include sharp peaks in the primary photoelectron spectrum in the 20-30 eV range. These are from the ionization of neutrals by intense solar HeII 30.4 nm radiation. In addition, a reduction at $ 60 eV is predicted due to a drop in the solar spectrum near 16 nm (e.g. Nagy and Banks, 1970;Mantas and Hanson, 1979;Fox, 2007). The peaks in this range are primarily produced by O rather than CO 2 , because of the lower altitude of  (Schunk and Nagy, 2000;Coates et al., 2008). The primary ionization process clearly introduces specific energy peaks in the energy spectrum (e.g. Gan et al., 1990;Cravens et al., 1980). These specific spectral features can be used to identify ionospheric photoelectrons.
A detailed comparison of a multi-stream kinetic model with ASPERA-ELS ionospheric photoelectron data was made by Cui et al. (2011). A good agreement was found between the observed and modelled photoelectron peak energies and flux decrease features, and fair agreement between the observed and modelled absolute fluxes was found when the magnetic field direction is included.
Observations of photoelectrons at Venus, Earth, Mars and Titan were summarized by Coates et al., (2011). At Venus, the first wellresolved measurements of ionospheric photoelectrons in the ionosphere were given by Coates et al., 2008 using data from Venus Express. The flux of photoelectrons stayed fairly constant throughout the observation region in the Northern ionosphere, and from this it was inferred that the photoelectrons were produced during the ionization of oxygen at a lower altitude ($ 200 km) below the observation point (250-700 km). The observed photoelectrons were produced in the denser region of the atmosphere where atomic oxygen is abundant and then transported to the observation altitude.
One particular aspect of photoelectron observations, seen at all four objects, is the observations of ionospheric photoelectrons at locations well away from their production point, as well as locally in the ionosphere near the production point. These remote locations include the Earth's magnetosphere (at up to 6.6 R E , Coates et al., 1985), in the Martian tail (at up to 3 R M , Frahm et al., 2006aFrahm et al., , 2006bCoates et al., 2011), in Titan's tail (at up to 6.8 R T , Coates et al., 2007;Wellbrock et al., 2012) where the observations were used to estimate ion loss rate , and at Venus (at up to 1.45-1.5 R V , Tsang et al., accepted for publication; Coates et al., 2011). Tsang et al. (accepted for publication) also observed that at times near and beyond the terminator, the clear two-peak signature observed on the day side broadens into one peak, perhaps the result of scattering processes between the production and observation point (Jasperse, 1977;Jasperse and Smith, 1978). At Mars, photoelectron peaks were used to estimate the total loss rate of electrons from the Martian ionosphere, which was then suggested to be equal to the ion loss rate (Frahm et al., 2010). At Venus, escape rates appear to be dependent on upstream conditions (e.g. Brace et al., 1987, Barabash et al., 2007bMcEnulty et al., 2010).
In this paper, we present new case studies of photoelectrons seen in the Venus tail at larger distances than seen previously, and we make a new estimate of the ion escape flux from Venus using these measurements. We also summarize the observed plasma escape rates at various objects throughout the solar system as derived from in-situ measurements by spacecraft, and for the first time order these observed rates by body mass.

Instrumentation
We use data from the Electron Spectrometer (ELS) and the Ion Mass Analyzer (IMA), part of the Analyzer of Space Plasmas and Energetic Atoms (ASPERA-4) experiment on ESA's Venus Express (VEx) spacecraft (Barabash et al., 2007a). The ELS has an energy resolution of 8%, designed to provide well resolved measurements of photoelectrons at Venus (Barabash et al., 2007a;Coates et al., 2008).

Ionospheric photoelectrons in the Venus tail region
The first measurements of photoelectrons in the Venus tail region were given by Tsang et al., accepted for publication, which studied Venus Express data on 3, 4, and 30 June 2006, where photoelectrons were observed at 1.45 R V in the tail region. An additional example with photoelectrons seen at 1.5 R V on 20 April 2008 by Venus Express was shown by Coates et al. (2011). During this pass, photoelectrons were observed in the local ionosphere (interval A) as well as in the tail of Venus (interval B), and averaged spectra were shown from both intervals (Coates et al., 2011, Fig. 3). These tail photoelectrons were also associated with an increased flux of low energy ions. This led to the suggestion that polar windtype escape along the draped magnetic field around Venus may be occurring, associated with an ambipolar electric field set up by relatively energetic photoelectrons (Hartle and Grebowsky, 1995;Coates et al., 2008Coates et al., , 2011. Hybrid simulations of the Venus environment  showed that draped fields with the suggested configuration were possible for suitable upstream conditions, at least early during interval B. Similar studies were done at Mars (Liemohn et al., 2006) and at Titan . These simulations support the hypothesis that such field morphology is consistent with the observations. We now present two new examples where ionospheric photoelectrons are seen in the tail: 8 May 2013 and 15 September 2009. These additional events are not the result of a full survey of the data, which is beyond the scope of the current paper. However, the two new events are measured (1) in the dark tail and (2) at larger distance along the tail, providing significant additional case studies.

8 May 2013
Three types of plots are shown in Fig. 1. These are: the VEx orbital trajectory shown at the lower right, the ion and electron spectrogram at the lower left, and selected electron spectra at the top. Utilizing both the particle spectrograms and the orbital trajectory, VEx travels from the tail region (where intervals of relatively high energy solar wind halo/strahl electrons are seen, up to 0400 UT, and then again during an intensification at $ 0405 UT), through ionospheric plasma in the tail (before the terminator at $ 0425 UT) and then traversing the dayside ionosphere (0425-0431 UT), out to the magnetosheath (0431-0446 UT) then into the solar wind after the bow shock crossing at 0446 UT.
At various locations within the ionosphere, individual electron spectra highlight the observed distributions. The spectra A and B are measured before the terminator, with spectrum C measured just after the terminator. In all the spectra A, B and C, a broad peak structure in energy is observed between 10 and 20 eV, with some further detail also seen (2 peaks in spectra A and C in particular). We interpret the peaks in spectra A, B and C as due to ionospheric photoelectrons. Spectrum A in particular represents ionospheric photoelectrons observed in the dark tail region. Spectrum B appears somewhat broadened in energy; as found by Tsang et al. (accepted for publication), this may be due to scattering processes (Jasperse, 1977;Jasperse and Smith, 1978) between the production site in the dayside ionosphere and the observation site at the spacecraft. This may differ depending on which field line is being sampled.

15 September 2009
The data from 15 September 2009 are shown in Fig. 2. Here, the ELS spectrogram shows that the spacecraft starts in the tail region, as shown by the relatively energetic electrons, at up to $ 0145 UT. Intervals of ionospheric plasma are then observed at particular locations up to the terminator at $ 0226 UT. Sheath plasma is then observed after $ 0230 UT. The spectra A and B are shown from two of the intervals containing ionospheric plasma. Spectrum A is from $2 R V along the tail while spectrum B is from the terminator. Both spectra show a two-peaked structure, highly reminiscent of the spectra in Coates et al., 2008. We interpret these observations as ionospheric photoelectrons. In the case of spectrum A, this observation is an example which is furthest away from the planet (about 2.3 R V ). It is important to note that the orbital constraints on VEx preclude observations further from the planet in the Venusian tail.

Venus ion escape rate estimate
For the 15 September 2009 case, we now estimate the escape rate along the Venusian tail, using a similar method to that discussed by Coates et al. (2012) at Titan. We estimate the flux of ions (Q) along the tail using the following equation: where k¼ 2.9 Â 10 22 (cm 3 km À 1 R c À 2 ) is a constant for units conversion, n ELS (cm À 3 ) is the number density estimated from moment calculations of the ELS data, v is ion velocity relative to Venus (km s À 1 ) estimated from the peak of the ion spectrograms, and A is the area (R c 2 ) of the tail escape channel as a fraction of R V (¼6051 km), estimated from the duration of the ionospheric plasma seen near interval A. We have used an escape channel radius of 0.5 R V in this calculation based on the time of observation of the ionospheric photoelectrons on this particular orbit. By comparison with models, where larger and more complex escape channels are observed (e.g., Kallio et al. (2011), and references therein), the error on this area, and thus the escape rate estimate, could be at least 50%.
Since the parent ion from which electrons are liberated cannot be uniquely identified, the energy of the observed ion has been used to determine the ion velocity for H þ , O þ , and O 2 þ . For each of these cases, the ion loss rates are shown in Table 1 in the columns indicated as E esc . Using IMA measurements of the ion velocity for H þ and O þ (not shown), we can make more accurate estimates of the loss rate, using these mass group-resolved data. These estimates are shown in the columns marked V esc , for light (H þ ) and heavy ions (O þ ). These loss rates were calculated using Eq. (1).
Using the latter more accurate method, Table 1 shows in the Total column an estimated total loss rate of 2.2 Â 10 23 ions s À 1 for this particular interval. Using the relative composition data this can be converted to a mass loss rate of 2.0 Â 10 24 amu s À 1 .
We note that this estimate is lower than that provided in other studies (e.g., Barabash et al., 2007b). However, as mentioned above the error in this estimate could be as much as 50% due to possible uncertainties in the escape channel radius. In addition, not all of the tail is sampled by the Venus Express orbit, and it is possible that some of the escaping plasma is missed on each particular orbit. It is also worth pointing out that there will be differences in escape rate associated with different upstream conditions as well as solar cycle variations.

Venus escape rate compared to other solar system bodies
To compare the estimated escape rate in the previous section with other estimates at solar system objects in general, including planets, comets and moons, we have assembled some of the estimates of production and escape rates made over the last few years, primarily from spacecraft observations. The results are summarized in Table 2 (adapted from Coates, 2012).
We now take these estimates and compare them with the radius of the appropriate object. In Fig. 3, we show the rates plotted as a function of body radius. Remarkably, three broad groups appear when plotting the data in this wayplanets, moons and comets. The planet and moons groups overlap to some extent as seen in Fig. 3. We should point out that for all objects the rates are in reality time dependentfor comets, for example, the rate will be dependent on heliocentric distance, and for all objects the rates will depend on  Fig. 1, except that two spectra A and B are shown.

Table 2
Neutral gas production rates for comets and other solar system objects visited by spacecraft with plasma instrumentation.

Summary and conclusions
Regarding photoelectrons at Venus, we can summarize the results presented here as follows: On 8 May 2013, photoelectrons are seen in the dark tail region where production by sunlight is certainly absent. This is indicative of transport to this observation point, probably along a draped field line connected to the ionosphere (as modelled at Mars by Liemohn et al., at Titan by Sillanpaa et al., and presented in Wellbrock et al., 2012). A preliminary study by Jarvinen et al. (2012)  On 15 September 2009, photoelectrons are seen at $ 2.3 R V , the largest distance that ionospheric photoelectrons, originating in Venus' sunlit ionosphere, have been observed along the Venus tail to date. Again, this indicates a magnetic connection as local production would be absent due to the lack of ionizing sun light.
The examples of photoelectrons in the Venus tail region presented in this paper, together with the earlier observations from Coates et al. (2008), Tsang et al. (accepted for publication), and Coates et al. (2011), show that photoelectrons at Venus may be observed at remote locations from their production point in the dayside ionosphere. As at Titan (Coates et al., 2007;Wellbrock et al., 2012), Mars (Frahm et al., 2006a(Frahm et al., , 2006b and the Earth (Coates et al., 1985), this is indicative of a magnetic connection between the ionosphere and the observation point. This is summarized in Fig. 4 for Venus.
The presence of photoelectrons in the tail, together with lowenergy ions, is possible evidence for a polar wind style escape at Fig. 3. Collection of measured production rates from solar system objects (Table 2), plotted as a function of object radius. Venus. The relatively energetic electrons set up an ambipolar electric field which draws ions out of the ionosphere. This mechanism may supplement any escape due to pickup ions (e.g. Luhmann et al., 2008, as well as from ionospheric outflow, e.g. Fedorov et al., 2008).
The result of the escape calculation presented here is a little lower than earlier estimates, but such estimates are very sensitive to the assumed area of the escape channel. While the diameter of this is determined by observations, its shape may well differ from the assumed cylindrical shape. Different shapes have been seen in simulations at unmagnetized objects (e.g. Ma and Nagy, 2007;Snowden and Winglee, 2013).
To our knowledge, Fig. 3 presents the first time that the escape rates of all of these classes of objects have been gathered and plotted in this way. The escape rates for other solar system objects show a remarkable grouping of rates for planets, moons and comets, when the escape rate is plotted against body radius. In general, the escape rates are higher for smaller sized objectsthis is visible for both the comets and for the moons when compared with planets. Although the body mass clearly plays a role in stopping thermal ion escape, and this is consistent with the observations, additional escape processes are underway (including bulk ionospheric escape and photoelectron driven escape).
Clearly, there are different escape processes at work for all objects, including thermal and non-thermal mechanisms, which contribute to all of the points. Nevertheless, some interesting trends emerge. The relatively small objects, i.e. the comets, have generally higher escape rates, likely due to their lower gravity. The moons shown are generally higher escape rates for similar size objects, in the cases of Io and Enceladus due to their intrinsic activity, and in the other cases shown because of their immersion in hot magnetospheres rather than the solar wind.
Significant modelling efforts, as well as experimental studies, are underway to understand the different escape rates at different objects. Understanding the complex escape processes are important for determining atmospheric evolution, and for determining the history of volatiles in the solar system.