Storey Sochi 2012
Storey Sochi 2012
Storey Sochi 2012
(2002)
Electron Temperatures and Free-Electron Energy Distributions of Nebulae from C ii Dielectronic Recombination Lines
Peter J. Storey1 and Taha Sochi1 1
Accepted XXX. Received XXX; in original form XXX University College London, Department of Physics and Astronomy, Gower Street, London, WC1E 6BT
ABSTRACT
A recently generated theoretical line list of C ii dielectronic recombination lines together with observational data gathered from the literature is used to investigate the electron temperature in a range of astronomical objects, mainly planetary nebulae. The electron temperature is obtained by a least-squares optimisation using all the reliable observed lines in each object. In addition, the subset of lines arising directly from autoionising states is used to directly determine the free-electron energy distribution which is then compared with various theoretical possibilities. The method described here can potentially determine whether there are departures from Maxwell-Boltzmann distributions in some nebulae, as has been recently proposed. Using published observations of the three planetary nebulae where the relevant lines are recorded, we nd that the data are best matched by Maxwell-Boltzmann distributions but that the uncertainties are suciently large at present that -distributions or two-component nebular models are not excluded. Key words: planetary nebulae: general atomic processes methods: numerical radiation mechanisms: general ISM: abundances stars: kinematics and dynamics.
INTRODUCTION
Recombination plays an essential role in the physical processes that occur in nebulae, the principal electron-ion recombination processes being Radiative Recombination (RR) and Dielectronic Recombination (DR). Recombination of an electron and ion may take place through a background continuum known as radiative recombination, or through a resonant recombination process involving doubly-excited states known as dielectronic recombination. The latter can lead either to autoionisation, which is a radiationless transition to a lower state with the ejection of a free electron, or to stabilisation by radiative decay to a lower bound state, possibly the ground state, with a bound electron. The RR and DR processes are closely linked and the dierence between them may therefore be described as articial; quantum mechanically they are indistinguishable. In section 2 of this paper we determine an electron temperature from dielectronic recombination lines of C+ in the spectra of a number of astronomical objects, mainly planetary nebulae, using a least squares optimisation method with theoretical data obtained from the recently-computed theo-
retical line list, SS1, of Sochi & Storey (2012) and astronomical data gathered from the literature. The theoretical list was generated using the R-matrix (Berrington et al 1995), Autostructure1 (Eissner et al 1974; Nussbaumer & Storey 1978) and Emissivity (Sochi 2010) codes with an intermediate coupling scheme where the lines are produced by DR processes originating from low-lying autoionising states with subsequent cascade decays. The method of formation of DR lines is particularly simple, often only requiring the radiative probability of the transition in question and experimentally known properties, the energy and statistical weight of the autoionising upper state. Even in more complex cases, only autoionisation probabilities are required in addition. This can be compared with the complex recombination and collisional-radiative processes involved in obtaining eective recombination coecients for transitions between low-lying ionic states (e.g. Davey et al (2000)). Determining electron temperatures from DR lines can therefore provide valuable evidence about the temperature structure of photoionised nebulae. There is a long-standing puzzle in the physics of planetary nebulae involving the discrepancy between electron
1
E-mail: [email protected] (PJS). E-mail: [email protected] (TMS). Corresponding author. c 2002 RAS
temperatures and ionic abundances derived from optical recombination lines (ORLs) and collisionally excited forbidden lines (CELs). Although the forbidden lines are much stronger than the recombination lines, they are highly dependent on temperature, making abundance determinations potentially uncertain. On the other hand, the recombination lines are weak and prone to blending and can be easily contaminated by radiation from other excitation processes such as uorescence. Despite all these dierences, there is a common feature between the results obtained from these lines; that is for all the atomic species investigated so far (mainly from the second row of the periodic table such as C, N, O and Ne) the forbidden lines in planetary nebulae normally produce lower ionic abundances than the corresponding values obtained from the recombination lines. The ratio of the ORL to the CEL abundances, the so-called abundance discrepancy factor or ADF, is case dependent and can be a factor of 30 or even more. This has cast some doubt on the validity of the previously accepted CELs analysis results, although the stability of CEL abundances and the wide variations in ORL abundances between objects suggest that we should seek the solution to the problem in the physics or origin of the ORLs. The abundance problem appears to be correlated to the dierences between the temperatures obtained from the Balmer jump of H i and that from the collisionally-excited forbidden lines where the latter is systematically higher than the former (Kholtygin 1998; Liu 2002; Tsamis et al 2007). In fact, obtaining higher electron temperatures from forbidden lines than those deduced from recombination lines is a general trend in nebular studies. Several explanations have been proposed to justify these discrepancies individually or collectively, though no one seems to be satisfactory or universally accepted. One explanation is the sensitivity of the collisionally-excited lines to temperature and structure uctuations where these uctuations within the nebular structure result in systematic underestimation of the heavy element abundances deduced from the forbidden lines. The existence of knots depleted of hydrogen with high heavy element contents within the nebular gas has been proposed as a possible reason for these uctuations and subsequent implications. The temperature inside these knots of high metallicity, and hence high opacity to stellar ultraviolet emissions and large cooling rates, is expected to be too low for ecient production of forbidden lines though it is still suciently high for the emission of recombination lines. Consequently, the recombination and collisional lines originate in dierent regions of the nebular gas with dierent elemental abundances and dierent temperatures. However, the presence of such knots in most or all planetary nebulae, as would be required to explain the systematic nature of the observations, has not been conrmed observationally (Liu et al 1995; Garnett & Dinerstein 2001; Tsamis et al 2003; Liu et al 2004). In a recent paper by Nicholls et al (2012) it is suggested that this long-standing problem in planetary nebulae and H ii regions could arise from the departure of the electron energy distribution from the presumed Maxwell-Boltzmann equilibrium condition, and that it can be resolved by assuming a -distribution for the electron energy following a method used in solar data analysis. The electron energy distribution will be the subject of section 4 where we consider only lines originating from resonance states, that is free-free
2 2.1
As indicated already, the theoretical data of the C ii dielectronic recombination transitions and subsequent cascade decay are obtained from the SS1 line list of Sochi & Storey (2012) which consists of 6187 optically-allowed transitions with their associated data such as emissivity and eective recombination coecients. The autoionising states involved in the transitions of this list consist of 64 resonances belonging to 11 symmetries (J = 1, 3, 5, 7, 9, 11 half even and J = 1, 3, 5, 7, 9 half odd) which are all the resonances above the threshold of C2+ 1s2 2s2 1 Se with a principal quantum number n < 5 for the combined electron. These include 61 theoretically-found resonances by the K-matrix method (Sochi & Storey 2012) plus 3 experimental ones which could not be found due to their very narrow width. The bound states involved in these transitions comprise 150 energy levels belonging to 11 symmetries (J = 1, 3, 5, 7, 9 half even and J = 1, 3, 5, 7, 9, 11 half odd). These include 142 theoretically found by R-matrix, which are all the bound states with eective quantum number between 0.1-13 for the outer electron and 0 l 5, plus 8 experimental top states which are the levels of the 1s2 2s2p(3 Po )3d 4 Fo and 4 Do terms. The theoretical and computational backgrounds for the atomic transition calculations including the emissivity thermodynamic model are given in Sochi (2010) and Sochi & Storey (2012). The calculations were performed using an elaborate C2+ ionic target in the intermediate coupling scheme. The list has also been validated by various tests including comparison to literature data related to autoionisation and radiative transition probabilities and eective dielectronic recombination coecients. The theoretical parameters for the bound and resonance states were also compared to the available experimental data from the National Institute of Standards and Technology2 and found to agree very well both in energy levels and in ne structure splitting. Processes other than dielectronic recombination have not been considered in the atomic scattering and transition model of the SS1 list, so the results of SS1 are incomplete for states likely to be populated by radiative recombination or collisional excitation and de-excitation. Any transition in which the upper state has an excited ion core (usually 2s2p(3 Po )) will have negligible population by radiative recombination in typical nebular conditions and is therefore well represented by only dielectronic recombination and subsequent cascade processes. This includes all free-free and free-bound transitions plus those bound-bound transitions involving excited ion core states. However, for the determination of temperature we also
2
least squares optimisation procedure which is outlined in the following section. It should be remarked that the observed ux used in the least squares procedure is the de-reddened ux obtained by correcting for extinction and other sources of error as stated by the data source and not the raw ux data. Therefore, there should be no ambiguity when we use observed ux in the following sections.
2.3
We carried out a search for C ii recombination line data in the literature in which over 140 data sets related mainly to planetary nebulae were catalogued. All data sets that comprise only bound-bound transitions with no doubly-excited core upper state were removed. The remaining data sets were subjected to a renement process in which the ux of all the observational lines in each data set were normalised to the ux of a reference observational line in the set, which is usually chosen as the brightest and most reliable, while the emissivity of all the theoretical lines in the set were normalised to the emissivity of the corresponding theoretical line. The ratio of the normalised observed ux to the normalised theoretical emissivity of each line were then plotted on common graphs as a function of temperature on log-linear scales. A sample of these graphs is presented in Figure 4 for the planetary nebula NGC 5315. All lines that did not approach the ratio of the reference line within an arbitrarilychosen factor of 3 were eliminated. The arbitrary factor of 3 was chosen as an appropriate limit considering practical factors that contribute to errors in the collection of observational data. The renement process also involved the utilisation of graphs in which the ratio of theoretical emissivity to observational ux of all lines in a certain data set was plotted on a single graph as a function of electron temperature. A sample of these graphs is shown in Figure 1. Some lines were also eliminated for various reasons related mainly to an established or suspected misidentication of the line or its intensity. For example, the wavelength of the alleged C ii line may not match with any known theoretical transition. Also, the absence of a strong line in the observational data associated with the presence of a much weaker line with no obvious reason casts doubt on the identication of the weaker line. The line may also be eliminated because its intensity ratio relative to another well-established line does not comply with the ratio obtained from theory. A very few lines were also out of the wavelength range of our line list and hence were eliminated due to lack of theoretical emissivity data. The selected rened data sets were then subjected to a
c 2002 RAS, MNRAS 000, 1??
In our least squares calculations we use a single tting parameter, which is the electron temperature, while the observations are the ux data of the C ii recombination lines that we obtained from the literature. All blended C ii lines in the observational list are combined by considering them as a single line with a single ux, while C ii lines blended with non-C ii lines are eliminated. To compare the theoretical emissivity to the observational ux, the theoretical emissivity of each line is normalised to the total theoretical emissivity of all the lines involved in the least squares procedure, while the observational ux of that line is normalised to the total observational ux of these lines. The normalised theoretical emissivities corresponding to a particular observational ux are added when the observational ux is given for a whole multiplet. The 2 dened by the following equation
N
2 =
i=1
no Ii nt i 2 I no
i
(1)
is computed, where i is an index running over all the N no lines involved in the least squares procedure, Ii and nt are i the normalised observational ux and normalised theoretical emissivity of line i respectively, is the number of degrees 2 no of freedom, and I no is the variance of Ii . This variance i is computed from formulae given in Sochi (2012). For the data sets with given observational errors the reported errors were used while for the data sets with no reported error a Poisson distribution was assumed and the error on the observed ux was assumed to be proportional to the square root of the ux. In some data sets, the observational error was given for some lines only, and hence the average of the given errors was assigned to the missing ones. In some cases when the reported error was unrealistically small resulting in large 2 , the 2 curve was scaled to unity at the minimum to obtain a more realistic error estimate. The temperature of the object is then identied from its value at the minimum 2 , while the condence interval is identied from the values of the temperature corresponding to the values of 2 1 on the lower and upper sides. In min some cases, the 2 curve was too shallow on one side and hence it resulted in a broad condence interval on that side. In the following section we present results for those objects where there were sucient adequate observations to derive a temperature.
DERIVED TEMPERATURES
In this section we present the astronomical objects that have been investigated. The objects are mainly planetary nebulae and the physical parameter of interest is the electron
5 4.5 4 3.5 3
4.5 4 3.5 3
2.5 2 1.5 1 0.5 0 1000 2000 3000 4000 5000 6000 Temperature (K) 7000 8000 9000 10000
0 0
0.5
1 Temperature (K)
1.5 x 10
2
4
Figure 1. The ratio of theoretical emissivity to observational ux as a function of temperature for the selected C ii lines from the NGC 7009 spectra of Fang & Liu (2011).
Figure 2. Temperature dependence of 2 for NGC 7009 with the inclusion of line 4267, where T = 5800 K at 2 min with a condence interval between 4961 6318 K.
temperature of the line emitting regions. We also include three objects which are not planetary nebulae for comparison, where similar techniques have been used in the past. The theoretical and observational data for the transitions used in this investigation are given in Tables 1 and 2.
2
3.1
NGC 7009
NGC 7009 is a bright planetary nebula which has a doubleringed complex spatial structure with a rich recombination spectrum and a relatively large ADF. The observational data for this object were obtained from Fang & Liu (2011) where 9 lines, listed in Table 2, were chosen following the selection process. In Figure 1 the ratio of theoretical emissivity to observational ux is plotted against electron temperature on a linear-linear graph for these lines. If there were no errors in the observational uxes or theoretical emissivities and the nebula was at a single uniform temperature, these curves would all intersect at the same temperature. The 2 graphs, with and without 4267, are given in Figures 2 and 3. As seen, The rst indicates a temperature of about 5800 K while the second a temperature of about 5500 K, which are in good agreement. We note also that in Figure 1 the curves for 4267 and the DR doublet 8794, 8800 show very good agreement for T > 8000K. A This indicates both that the theoretical emissivities of these lines are entirely consistent and also that the observational data for this object for these three lines are accurate. In Table 3 we list electron temperatures from the cited literature derived from a range of ions and three types of spectral features: CELs, ORLs and the Balmer and Paschen discontinuities. As can be seen, the values that we obtained here are in a broad agreement with the temperature obtained from several recombination lines. The lower electron temperature from optical recombination lines compared to the values obtained from collisionally-excited lines is consistent with the trend of the discrepancy between the abun-
1000
2000
3000
7000
8000
9000 10000
Figure 3. Temperature dependence of 2 for NGC 7009 with the exclusion of line 4267, where T = 5500 K at 2 min with a condence interval between 4697 6156 K.
dance and temperature results of ORL and CEL (higher ORL abundance and lower electron temperature as compared to the CEL abundance and temperature). If this discrepancy is caused by departures from a Maxwell-Boltzmann distribution by the free electrons, we might expect objects with the largest ADF to show the largest departures, so we also list ADF values from the literature in Table 3. NGC 7009 has the largest ADF of the objects in our sample of planetary nebulae. We discuss its electron energy distribution in Sec 4.
3.2
NGC 5315
NGC 5315 is a young dense planetary nebula in the southern constellation Circinus located at a distance of about 2.6 kpc
c 2002 RAS, MNRAS 000, 1??
3.5
a = 4267.15 = 5145.16 a = 6791.47 a = 6800.68 a BB BB BB BB 3d2D 16 < 4f2F* 20 3s4P* 6 < 3p4P 6 3s4P* 4 < 3p4D 4 3s4P* 6 < 3p4D 6
NGC 2867
10
10 NFER
10
10
10
0.5
1 Temperature(K)
1.5 x 10
2
4
NGC 2867 is a compact planetary nebula with comparatively small size and fairly strong surface brightness located at a distance of about 0.5 kpc in the southern constellation Carina. The observational data of this object were obtained from Garc a-Rojas et al (2009) where two knots have been studied: one labelled NGC 2867-1 and the other NGC 28672. Two lines, given in Table 2, were selected for the optimisation process. The NGC 2867-1 result indicates T 14300 K while NGC 2867-2 result indicates T 16000 K. The dierence in temperature value may be caused by the dierence in the physical conditions of the two knots. Table 3 presents electron temperatures derived in previous works from transitions of dierent species. As seen, our values are signicantly higher than most of the values reported in the literature. However, this may be explained by the complex structure of this nebula and the possibility of dierent lines being originating from dierent regions with very dierent physical conditions.
Figure 4. Ratio of normalised observed ux to normalised theoretical emissivity (NFER) versus temperature on log-linear scales for the planetary nebula NGC 5315.
3.6
DQ Herculis 1934
with an interesting complex ower shape appearance. The observational data, which consist of 4 selected lines indicated in Table 2, were obtained from Peimbert et al (2004). Figure 4 displays the ratio of the normalised observed ux to the normalised theoretical emissivity versus electron temperature for these lines. Our 2 calculations, with and without line 4267, indicate T 7400 K and T 6500 K respectively. Table 3 presents some values of the electron temperature of NGC 5315 as reported in the cited literature. Again, the temperature is lower than that derived from the collisionally-excited lines.
3.3
NGC 7027
NGC 7027 is a compact, bright, young, high excitation planetary nebula with one of the hottest central stars known for a PN. Two observational data sets related to NGC 7027 were investigated: one obtained from Baluteau et al (1995), and the other obtained from Zhang et al (2005). Eight lines, listed in Table 2, were selected from Baluteau et al. The least squares optimisation of these lines indicate T 11100 K, in good agreement with values obtained by other researchers; a sample of which is presented in Table 3. In the case of Zhang et al, 20 lines, given in Table 2, were chosen. The results of least squares, with and without line 4267, indicate T 12500 K and T 12000 K respectively.
This is a peculiar old classical galactic nova originating from an accreting cataclysmic variable binary system which apparently consists of a white and a red dwarf. Our observational data of this object were obtained from Ferland et al (1984) where 2 lines, given in Table 2, were chosen for least squares. The result indicates T 1600 K, in very good agreement with some values reported in the literature (refer to Table 2) notably those of Smits (1991) and Davey (1995) which are also derived from C ii recombination lines. The higher temperatures in the literature may belong to the hot inner disk region, rather than the cool outer shell, where much higher temperatures have been derived. It should be remarked that the use of 1335 line A is a second exception (the rst is 4267 ) to our rule A of using the BB transitions only if the upper state has a doubly-excited core. The upper state of the 1335 transition is 1s2 2s2p2 which is connected to the C2+ 1s2 2s2 continuum and 1s2 2s2 nl Rydberg states by two-electron radiative processes which are usually very weak. Hence excitation of 1335 by direct radiative recombination can be neglected. A
3.7
CPD - 56 8032
3.4
IC 418
IC 418 is a bright, young, carbon-enhanced, low-excitation, highly symmetric planetary nebula with apparent ring structure located at a distance of about 0.6 kpc in the constellation Lepus. Our observational data on IC 418 come from Sharpee et al (2003) where 22 lines, presented in Table 2, were selected for least squares minimisation. The 2 plots, with and without line 4267, minimise at T 8700 K and T 7700 K respectively.
c 2002 RAS, MNRAS 000, 1??
CPD - 56 8032 is a cool late-type Wolf-Rayet star that is usually classied as WC10 or WC11. The star, which is located at about 1.3-1.5 kpc, is surrounded by a young planetary nebula with complex visible structure. Here, we try to infer the electron temperature of the stellar nebular wind surrounding CPD - 56 8032. The observational data of this object were obtained from De Marco et al (1997) where 13 lines, given in Table 2, were extracted following a selection process. The 2 plot indicates T 17300 K. This agrees, within the reported error bars, with the temperature of De Marco et al (De Marco 1996; De Marco et al 1996) who deduced a value of 185001500 K for this object using a similar least squares approach. Our value also agrees reasonably with some of the values reported in the literature; a sample of which is given in Table 3.
6
3.8
He 2-113 is another late-type WC10 Wolf-Rayet star surrounded by a planetary nebula with an apparent ring structure. There are many physical similarities between CPD 56 8032 and He 2-113 such as age, ux and distance. These similarities are reected in the strong resemblance of their observed spectra and hence they are normally investigated jointly. The observational data of this object were obtained from De Marco et al (1997) where 13 lines, given in Table 2, were extracted following a selection process. The 2 plot indicates T 16200 K which agrees very well with some previously-deduced values notably those of De Marco et al (De Marco et al 1997; De Marco & Barlow 2001).
r () br f ( r )
(7)
where a0 is the Bohr radius, R is the Rydberg constant and where r () is the radiative transition probability corresponding to the line of wavelength . The emissivity of a DR line is given by j() = 1 hc N e N + DR () 4 (8)
It has been suggested (Nicholls et al 2012) that the discrepancy between the results of ORLs and those of CELs is based on the assumption of a Maxwell-Boltzmann (M-B) for the electron distribution in the nebulae and that by assuming a dierent type of distribution, e.g. a -distribution, the ORLs and CELs might yield very similar results for the abundance and electron temperature. One way for testing this proposal is to use DR lines to directly sample the distribution and compare to the M-B and other distributions. The cross-section for dielectronic recombination can be expressed (e.g. Davies & Seaton (1969)) DR = 2 h mv
2
where N e and N + are the number density of electrons and ions respectively, h is the Plancks constant and c is the speed of light. Hence we may write for the ux of the transition with wavelength DR () where C is a proportionality factor and therefore I() = C f ( r ) = DI() + r
r 1/2
(9)
r br
(10)
r P( ) +
(2)
where P ( ) is the dielectronic recombination probability, the probability of capture of a free electron of energy and velocity v by an ion of statistical weight + with the emission of a photon via a resonance state of statistical weight r ; h and m being the reduced Plancks constant and mass of electron respectively. The recombination coecient for dielectronic recombination via the resonance r is given by DR =
res
where D is another proportionality factor. The observed intensities are taken from Table 2 and the necessary atomic parameters are given in Table 1. Thus for each DR line arising directly from an autoionising state (FF and FB transitions), we can obtain the fraction of free electrons at the energy of that state. The resulting values of f ( ) can then be compared to various theoretical electron energy distributions. We derive values of f ( ) for all the data sets that contain more than one FF or FB transition. In Figures 5-9 the results 1 are presented. In these plots of 2 f ( ) against , the M-B distribution, fMB (T, ) = 2 (kT )3/2 e kT (11)
DR v f ( ) d
(3)
vr f ( r )
res
P( ) d
(4)
appears as a straight line, where k is the Boltzmann constant, and T is the electron temperature shown at the optimum temperature obtained previously. We also show in these gures non-Maxwellian distributions for comparison, dened by (Bryans 2005) f,
where we have assumed that the resonance is narrow compared to changes in v and P ( ) so that vr and r are the electron velocity and energy at the resonance position. Several workers have shown (e.g. Bell & Seaton (1985)) that for an isolated narrow resonance P( ) d =
res
2 ( ) = 3/2
( + 1) 1 2
(+1)
1+
(12)
2 r a h r + a
(5)
where r and a are probabilities of radiative decay and autoionisation in units of inverse time. We dene a departure coecient br by br = r a + a (6)
which tends to unity for a r . We have assumed here that only dielectronic capture, autoionisation and radiative
where is a parameter characterising the distributions, while is a characteristic energy. The -distributions in these gures were calculated using the best t temperatures from section 3. We show curves for = 5 and = 15, the former to illustrate clearly how the shape of a distribution deviates from M-B in this representation, and the latter as representative of the values proposed by Nicholls et al (2012) to resolve the CEL/ORL abundance and temperature problem. In the -distribution there are more electrons than M-B at low energies and high energies and fewer at intermediate energies. The deviation of the -distribution from M-B increases as decreases. In typical nebular conditions the -distribution is greater than M-B for energies
c 2002 RAS, MNRAS 000, 1??
Various authors (e.g. Liu et al (2000)) have suggested twocomponent models to explain the ORL/CEL abundance discrepancy, in which a cold metal rich component is embedded in a hotter medium. Suppose a nebula consists of two distinct but individually homogeneous components (1 and 2) in which the electron energy distributions are assumed to be Maxwell-Boltzmann (fMB (T, )). The components have e e dierent electron number densities (N1 and N2 ) and electron temperatures (T1 and T2 ) and is the fraction of the total volume (V ) occupied by component 1. Then the power radiated in a line at wavelength will be given by
e + P1 () = V N1 N1 DR (, T1 )
hc hc
(13) (14)
e + P2 () = V (1 )N2 N2 DR (, T2 )
+ + where N1 and N2 are the number densities of C2+ in components 1 and 2 respectively. Then adding P1 and P2 to get the total power, which is proportional to the observed ux, I, and using the expression for DR in terms of an electron energy distribution, we get
I r br r ()
1 2
where is now the energy of the resonance which is the upper state of the transition of wavelength . To plot the quantity on the RHS we need to relate the number densities in the two component model to those in a single component model for which 1 I e + 2 N N f (16) MB ( , T ) r br r () We require that if T1 = T2 = T the RHS are identical so
e + e + N e N + = N1 N1 + (1 ) N2 N2
(17)
or
e e (N e )2 a = (N1 )2 a1 + (1 ) (N2 )2 a2 2+
(18)
where the as are the abundances of C relative to N e in the various components. The normalised two-component distribution is then
e e (N1 )2 a1 (N2 )2 a2 fMB ( , T1 ) + (1 ) f ( , T2 ) (N e )2 a (N e )2 a MB
(19)
(i) The minimum value of 2 occurs for 1/ = 0 in all cases, so that the best value of T = T . This is a quantitative reection of the qualitative observation that the data points suggest negative curvature of f ( )/ with respect to whereas the M-B distribution has zero curvature and -distributions have positive curvature. (ii) For values of T close to the minimum and 1/ = 0, 2 diers little from its minimum value. Indeed, if we take a change of reduced 2 of unity to estimate the condence interval on , then 10 is possible in all cases. This is simply a reection of the magnitude of the uncertainties, both in the observations with the current data sets extracted from the literature and in the calculated atomic parameters.
c 2002 RAS, MNRAS 000, 1??
The constants preceding the MB distribution functions are essentially relative fractional emission measures for the two components. We construct an illustrative two component model for NGC 7009 based on the model IH2 of Liu et al (2000) for NGC 6153, with T1 = 10000 K, T2 = 500 K and T = 5700 K e e from our work. We also assume N1 = 5000 cm3 , N2 = 3 1000 cm , a1 /a = 0.6, a2 /a = 70 and = 0.7. The value e of a2 = N (C2+ )/N2 is taken from the N (O2+ )/N (H) ratio of Liu et al (2000) in the metal-rich component of NGC 6153. With these parameters, the normalised two component distribution is 0.333 fMB ( , T1 ) + 0.667 fMB ( , T2 ) (20)
The resulting distribution is shown in Figure 5. Note that NGC 7009 has a signicantly lower ADF (3-9, see Table 3) than NGC 6153 (approximately 10, Liu et al (2000)), so
8
10
10
10
10
10
10
10
0.05
0.1
0.3
0.35
10
0.05
0.1
0.3
0.35
Figure 5. Electron distribution plot for the NGC 7009 data of Fang & Liu (2011), showing a Maxwell-Boltzmann and two -distributions ( = 5.0 and 15.0) for T = 5700 K. A twocomponent Maxwell-Boltzmann distribution with T1 = 10000 K and T2 = 500 K is also plotted. The y-axis has an arbitrary scaling. The factor used to scale the observational data points to the theoretical curves was obtained by minimising the weighted least squares dierence between the observational data points and their counterparts on a M-B distribution.
Figure 6. Electron distribution plot for NGC 7027 data of Zhang et al (2005), showing a Maxwell-Boltzmann and two distributions ( = 5.0 and 15.0) for T = 12300 K. The y-axis scaling is as in Figure 5.
10
departures from a single temperature model would be expected be less extreme than modelled here. In addition, the emission measures and temperatures of the two components would be expected to dier between the two nebulae. Choosing a higher temperature (T1 ) and larger emission measure for the low temperature component, for example, would improve the agreement with the low energy data point. Hence plots such as Figure 5 should place signicant and useful constraints on the parameters of any two-component model.
10
10
CONCLUSIONS
10
0.05
0.1
We have determined an electron temperature from dielectronic recombination lines in a number of astronomical objects, mainly planetary nebulae, by analysing C ii transitions which originate from low-lying autoionising states and the subsequent decays. In the analysis we used a least squares minimisation method to nd the electron temperature which best ts all the reliable observational data. For the planetary nebulae, our results generally fall below those derived from CELs such as the [O iii] forbidden lines and above those from ORLs like the O ii permitted lines. There are exceptions however. In NGC 7027, for example, our temperatures broadly conrm the forbidden line results while the O ii ORLs yield a much lower temperature. In those objects, which are not planetary nebulae (DQ Her, CPD - 56 8032, and He 2-113), where a similar approach has been used before, we nd good agreement with the earlier results. We also nd that the theoretical line emissivities that we predict are entirely consistent with those previously published by Davey et al (2000) for 4267. Given the very dierent but relatively simple mechanism of formation of the DR lines we
0.3
0.35
Figure 7. Electron distribution plot for IC 418 data of Sharpee et al (2003), showing a Maxwell-Boltzmann and two -distributions ( = 5.0 and 15.0) for T = 8200 K. The y-axis scaling is as in Figure 5.
can conclude that the theoretical emissivities for 4267 and other similar C ii lines are reliable. We have also proposed and demonstrated a method to test directly whether the free-electron energy distribution in planetary nebulae departs from Maxwell-Boltzmann. We showed that the uxes of DR lines originating directly from autoionising states can be used to sample the free-electron energy distribution and we applied this method to our sample of objects. We showed that, for all the objects where suitable data are available, a Maxwell-Boltzmann distribution gives the best t to the observations but that the uncertainties in the observational data and atomic parameters are such that a -distribution with values of as sugc 2002 RAS, MNRAS 000, 1??
9
50.0
0.2
30.0
25.0
20.0
15.0
15.0
0 20.
0.16
0.14
0.12 1/
40.0
0.18
9.0 10.0
10.0 9.0
8.0
10
0.1
0.08
0.06
8.0
0.04
0.02
10
0.05
0.1
0.3
0.35
0 2000
3000
4000
6000
7000
50.0
20.0
15.0
15.0
20.0
25.0
30.0
40.0
9.0 10.0
20.0
15.0
15.0
20.0
25.0
30.0
50.0
7.5
7.0
9.0 10.0
40.0
Figure 8. Electron distribution plot for CPD - 56 8032 data of De Marco et al (1997), showing a Maxwell-Boltzmann and two -distributions ( = 5.0 and 15.0) for T = 17300 K. The y-axis scaling is as in Figure 5.
Figure 10. 2 derived from the dierence between f ( ) and a -distribution as a function of 1/ and T for NGC 7009.
25. 0
7.0
8000
0.2
3.0 2.8
2.6
8.0
2.4
2.2
6.0
5.0
4.0
7.0
10
8.0
6.0
5.0
4.0
3.5
2.2
0.14
2.6
2.2
2.0
0.12
2.0
0.16
0.1
7.0
0.08
2.2
8.0
2.8
0.04
1
2.0
2.6
0.06
7.0
2.0
1.2
1.4
1.6
1.8
1.9
2.4
2.4
2.6
3.0
10
0.02
6.0
5.0
4.0 3.5
2.8 x 10
3.5
3.0
3
4
10
10
Figure 11. 2 derived from the dierence between f ( ) and a -distribution as a function of 1/ and T for CPD - 56 8032.
0.05 0.1 0.15 0.2 0.25 Electron Energy (Ryd) 0.3 0.35
Figure 9. Electron distribution plot for He 2-113 data of De Marco et al (1997), showing a Maxwell-Boltzmann and two distributions ( = 5.0 and 15.0) for T = 16200 K. The y-axis scaling is as in Figure 5.
REFERENCES Baluteau J.P., Zavagno A., Morisset C., Pquignot D., e 1995, A&A, 303, 175 Barlow M.J., Hales A.S., Storey P.J., Liu X.-W., Tsamis Y.G., Aderin M.E., 2006, International Astronomical Union (Symposium S234), 2, 367 Bell R.H., Seaton M.J., 1985, J. Phys. B, 18, 1589 Berrington K.A., Eissner W.B., Norrington P.H., 1995, Comput. Phys. Communications, 92, 290 Bryans P., 2005, PhD thesis, University of Strathclyde Davey A.R., 1995, PhD thesis, University College London Davey A.R., Storey P.J., Kisielius R., 2000, A&AS, 142, 85 Davies P.C.W, Seaton M.J., 1969, ApJ, 2, 757 De Marco O., Barlow M.J., Storey P.J., 1996, Astrophys Space Sci, 238, 91 De Marco O., Storey P.J., Barlow M.J., 1996, ASP Conference Series, 96, 145
gested by Nicholls et al (2012) is not excluded. Similarly a two-component model as described by Liu et al (2000) for NGC 6153 is also not excluded, although signicant constraints are imposed on any such model by the DR lines. We have highlighted several spectral lines which make it possible to sample the low energy part of the electron energy distribution where departures from a simple one-component Maxwell-Boltzmann distribution are expected. Higher precision is needed in the measurement of the intensities of these lines to fully realise the potential of this method.
c 2002 RAS, MNRAS 000, 1??
2.4
2.0
2.8
2.
0.18
3.5
2.0
10
0.2
3.0
2.5
3.0
0.16
0.12
2.1
1.8
0.14
3.5 3.0
2.5
3.0
3.5
4.0
0.18
2.1
2.1
3.5
1/
0.1
0.08
0.04
1.8
0.06
3.0
3.5
4.0
1.5
4.5
1.3
0 0.5
0.6
0.7
0.8
0.9
1.2
4.0
0.02
1.3
1.4
4.8
1.5 x 10
4
Figure 12. 2 derived from the dierence between f ( ) and a -distribution as a function of 1/ and T for IC 418.
0.2
8.00
5.00
3.00
0.18
1.9 0
1.4
0.16
1.
15
0.14
1.15
0.12 1/
8.00
5.00
3.00
1.90
1.4
1.1
0.08
0.06
1.15
0.04
1.0
5.00
3.00
1.90
1.05
0.02
1.40
1.15
1.4 0
1.4
0.1
1.2
1.4
1.6
1.8
2.4
2.6
2.8 x 10
1.9
3
4
Figure 13. 2 derived from the dierence between f ( ) and a -distribution as a function of 1/ and T for NGC 7027.
De Marco O., Storey P.J., Barlow M.J., 1997, MNRAS, 297, 999 De Marco O., Barlow M.J., 2001, Astrophys Space Sci, 275, 53 Eissner W., Jones M., Nussbaumer H., 1974, Comput. Phys. Communications, 8, 270 Fang X., Liu X.-W., 2011, MNRAS, 415, 181 Ferland G.J., Williams R.E., Lambert D.L., Shields G.A., Slovak M., Gondhalekar P.M., Truran J.W., 1984, ApJ, 281, 194 Garc a-Rojas J.; Esteban C., 2007, ApJ, 670, 457 Garc a-Rojas J., Pena M., Peimbert A., 2009, A&A, 496, 139 Garnett D.R., Dinerstein H.L., 2001, RevMexAA, 10, 13 Kholtygin A.F., 1998, A&A, 329, 691 Liu X.-W., Storey P.J., Barlow M.J., Clegg R.E.S., 1995, MNRAS, 272, 369
c 2002 RAS, MNRAS 000, 1??
4.5
2.5
1.8
2.5
2.1
5 1.
3.5
1.3
1.8
1.1
1.5
2.1
2.5
11
Table 1: Observed transitions used in the current study where the columns stand for: an arbitrary transition index, the wavelength in , the lower level designation and its statistical weight, the upper level designation and its statistical weight, A transition status (FF, FB or BB), the radiative transition probability in s1 , the departure coecient of the upper autoionising state in the case of FF and FB transitions and its energy in Ryd above the C2+ 1s2 2s2 1 Se ionisation threshold. The wavelengths are given as the vacuums for < 2000 and as the airs for > 2000 . The prime in the level designation indicates an A A excited core, i.e. 1s2 2s2p(3 Po ), and the 1s2 core is suppressed from all other congurations. In. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 1334.53 1335.66 1335.71 3588.91 3590.76 3590.88 3876.19 3876.39 3876.65 4267.00 4267.26 4267.26 4318.61 4323.11 4372.38 4376.58 4411.15 4618.56 4619.25 4627.50 4953.86 4958.66 4959.92 4964.74 5107.81 5113.65 5114.26 5132.95 5133.28 5143.49 5145.16 5151.08 5259.06 5259.66 5259.76 5485.91 5648.07 6098.51 6250.76 6779.94 6780.59 6783.91 6787.21 6791.47 6798.10 6800.69 6812.28 7112.48 7113.04 7115.63 8794.08 8800.28 8826.55 L 2s 2p 2 Po 2s2 2p 2 Po 2s2 2p 2 Po 3p 4 De 3p 4 De 3p 4 De 3d 4 Fo 3d 4 Fo 3d 4 Fo 2s2 3d 2 De 2s2 3d 2 De 2s2 3d 2 De 3p 4 Pe 3p 4 Pe 3d 4 Po 3d 4 Po 3d 2 Do 3d 2 Fo 3d 2 Fo 3d 2 Fo 3p 2 Pe 3p 2 Pe 3p 2 Pe 3p 2 Pe 3d 2 Po 3d 2 Po 3d 2 Po 3s 4 Po 3s 4 Po 3s 4 Po 3s 4 Po 3s 4 Po 3d 4 Fo 3d 4 Fo 3d 4 Fo 3d 4 Do 3s 4 Po 3p 2 Pe 3d 2 Do 3s 4 Po 3s 4 Po 3s 4 Po 3s 4 Po 3s 4 Po 3s 4 Po 3s 4 Po 3s 4 Po 3p 4 De 3p 4 De 3p 4 De 3p 2 De 3p 2 De 3p 2 De
2
l 2 4 4 2 4 6 10 8 6 4 6 6 2 2 4 4 4 6 8 8 2 4 2 4 4 4 2 2 4 4 6 6 8 4 6 6 4 4 6 4 2 6 2 4 4 6 6 2 4 6 6 4 6
u 4 4 6 2 2 4 12 10 8 6 8 6 4 2 4 6 6 8 10 8 2 2 4 4 4 6 4 4 6 2 6 4 6 2 4 6 4 6 4 6 4 8 2 4 2 6 4 4 6 8 8 6 6
St. BB BB BB FB FB FB FB FB FB BB BB BB FB FB FF FF FF FF FF FF FB FB FB FB FF FF FF BB BB BB BB BB FB FB FB FB BB FB FF BB BB BB BB BB BB BB BB BB BB BB FB FB FB
r () ul 2.403E+8 4.753E+7 2.869E+8 4.920E+7 4.945E+7 6.232E+7 2.629E+8 2.296E+8 2.116E+8 2.185E+8 2.340E+8 1.560E+7 3.470E+7 1.374E+7 9.895E+7 1.189E+8 1.855E+8 1.931E+8 2.288E+8 9.390E+6 2.469E+7 1.240E+7 6.017E+6 3.135E+7 2.158E+7 1.185E+8 1.060E+8 3.704E+7 2.700E+7 7.550E+7 6.354E+7 4.176E+7 2.063E+7 2.482E+7 2.010E+7 3.323E+6 1.945E+7 5.026E+7 2.733E+7 2.497E+7 1.486E+7 3.542E+7 2.946E+7 1.875E+7 5.803E+6 1.042E+7 1.708E+6 3.321E+7 3.558E+7 4.062E+7 2.034E+7 1.897E+7 1.405E+6
bu
0.9595 0.9062 0.9846 0.9952 0.4063 0.6893 0.7843 0.6893 0.9929 0.9929 0.9925 0.9925 0.9985 0.9979 0.9985
0.117661 0.117441 0.228409 0.228209 0.222570 0.227110 0.227462 0.227110 0.050483 0.050483 0.050258 0.050258 0.228615 0.228411 0.228615
12
Table 2: The observed ux data for the investigated objects where the rst column is the index as given in Table 1, while the other columns give the investigated objects. For NGC 7027, the rst column belongs to the data of Baluteau et al (1995) and the second to the data of Zhang et al (2005). For NGC 2867, the rst value is related to knot 1 and the second to knot 2. The given ux is the value normalised to H = 100 value except for the CPD - 56 8032 and He 2-113 where it is given as the absolute value in units of erg.s1 .cm2 and in multiples of 1012 . The symbol xx indicates that the previous value of ux in that column is shared by the indicated lines in that multiplet. NGC NGC NGC NGC IC NGC DQ CPD He In. 7009 5315 7027 7027 418 2867 Her 56 8032 2-113 1 270 2 xx 3 xx 4 0.018 5 0.059 0.0252 6 xx xx 7 0.0069 8 xx 9 0.026 xx 10 0.8795 0.6559 0.575 0.5712 0.814(1.246) 29 11 xx xx xx xx xx xx 12 xx xx xx xx xx xx 13 0.0086 14 0.004 15 0.026 16 0.031 0.0016 17 0.019 0.0016 18 0.0021 0.009 4.1940 2.2710 19 0.0021 6.1670 3.3390 20 0.1850 0.1002 21 1.4370 0.7141 22 0.7183 0.3571 23 0.7183 0.3571 24 0.0211 3.5920 1.7850 25 0.5906 0.1973 26 3.3840 1.1300 27 2.8880 0.9646 28 0.0088 0.0044 29 0.013 xx 30 0.013 31 0.0039 0.0040 32 0.009 0.0046 33 0.009 0.0031 34 0.0032 35 xx 36 0.0004 37 0.0014 38 0.0011 39 0.0015 40 11.8 0.034 0.0109 0.045(0.079) 41 0.0070 0.0055 42 2.1 0.004 0.0022 43 0.0052 3.7 0.008 44 0.0068 4.9 0.012 0.0066 45 0.7 46 0.0028 0.009 0.0050 47 0.5 0.001 48 4.7 49 xx 0.0052 50 0.0043 51 0.0320 11.8 0.015 1.9260 0.9342 52 0.0224 1.3470 0.6533 53 0.0943 0.0457
13
Table 3: The range of electron temperature, from ORLs and a sample of CELs, in Kelvin, derived from dierent species and transitions, of the investigated astronomical objects as obtained from the literature where BD stands for Balmer Discontinuity, PD for Paschen Discontinuity, and RF for Radio Frequency. The value in the rst row of each type of transition represents the minimum and the second is the maximum. Our results, as derived in the current paper, are given as averages in the additional three rows of C ii where the optimal value is in the rst row, while the lower and upper limits of the condence interval are in the second and third rows respectively. Our rst value for NGC 7027 belongs to the data of Baluteau et al (1995) while the second belongs to the data of Zhang et al (2005). Similarly, the two values for NGC 2867 correspond to the rst and second knots respectively. In the last row of the table, the range of the abundance discrepancy factor (ADF) values of some of the investigated objects as found in the literature is given. More details about the temperature data and their references can be found in Sochi (2012). Object H i(BD) H i(PD) He i C ii This work Topt Tmin Tmax [N ii] O ii [O iii] [N ii]+[O iii] RF ADF 3.0-9.10a 1.2-3.4b 1.8c 6600 23000 1.49-1.77d NGC 7009 7200 8150 5800 5800 5040 8000 NGC 5315 8600 8600 NGC 7027 8000 12800 8000 8000 8200 10360 IC 418 >15000 NGC 2867 8950 8950 DQ Her 450 1000 CPD 56 8032 He 2-113
10000 10000
10250 10900 700 1450 14300(16000) 12120(13090) 17600(26560) 8800 11750 1600 1460 1740 2400 2500
7000 11200
a Obtained from Liu et al (2004); Tsamis et al (2004); Liu et al (2006); Barlow et al (2006); Liu (2006); Wang & Liu (2007); Tsamis et al (2008). b Obtained from Tsamis et al (2004); Garc a-Rojas & Esteban (2007). c Obtained from Liu et al (2004). d Obtained from Garc a-Rojas et al (2009).