4 Model fits to the ultraviolet resonance lines of HD77581/Vela X-1 & HD153919/4U1700-37

 

 

Table 3: SEI model parameters for fits to the observed profiles and variability of UV resonance lines in the spectra of HD 77581/Vela X-1 and HD 153919/4U1700-37.

Line	T	$\alpha _1$	$\alpha_{2}$	$\sigma$	$v_\infty$	q
HD 77581/Vela X-1:
N  V	3	1	-1	0.45	600	2.9
Si  IV	300	-1	0	0.45	600	2.9
C  IV	200	0	0	0.45	600	2.9
Al  III	20	-1	0	0.45	600	2.9
HD 153919/4U1700-37:
N  V	30	0	0	0.15	1700	$>\!4$
Si  IV	$2\times10^4$	-2	0	0.15	1700	$>\!4$
C  IV	5000	-2	0	0.15	1700	$>\!4$

The modified SEI code is used to fit the line profiles at two orbital phases ($\phi =0$, X-ray eclipse, and $\phi =0.5$) of N  V, Si  IV, C  IV and Al  III resonance lines in HD 77581/Vela X-1 and N  V, Si  IV and C  IV in HD 153919/4U1700-37. The models are constrained by the following conditions: (i) the values for $\gamma$, $v_\infty$, $\sigma$ and q are the same for all lines in a given HMXB; (ii) the value for $x_{\rm X}$ follows from the orbital parameters (Table 1); (iii) the dominant ionization stage is estimated from the spectral type (Lamers et al. 1999). In HD 77581 N²⁺, Si³⁺ and C²⁺ are expected to dominate, while in HD 153919 probably N³⁺, Si⁴⁺ and C⁴⁺ dominate. Al²⁺ is assumed to dominate in HD 77581, as the observed Al  III line is very strong. The ionization is uniform throughout the undisturbed wind, i.e. $\alpha_2=0$ (except N⁴⁺ in HD 77581). Note that the N  V line suffers from strong absorption by the wing of Ly-$\alpha$. A sudden jump in intensity near the maximum of the Si  IV and C  IV emission in HD 153919/4U1700-37 is due to numerical difficulties. These lines were calculated by sampling the intrinsic line profile over $\pm5\sigma$with $I_{\rm g,max}=3333$ (instead of $\pm3\sigma$ with $I_{\rm g,max}=1111$), suppressing the spikey feature by half. The resulting line profiles are shown in Figs. 9, 10 and 14 and the model parameters are listed in Table 3.

Figure 9: SEI models that approximately fit the observed line profiles and variability of Si  IV, C  IV and Al  III in the UV spectrum of HD 77581/Vela X-1. Parameters are summarised in Table 3, and v=1 corresponds to 600 km s-1.

Figure 10: Same as Fig. 9, but for the N  V line in the UV spectrum of HD 77581/Vela X-1 (see also Table 3).

The orbital modulation of the strong wind lines in HD 77581 (with the exception of N  V) is mainly due to the HM-effect. Some detailed changes are probably related to the accretion flow in the system (Sadakane et al. 1985; Kaper et al. 1994); since these structures are not included in the model, they cannot be reproduced. The modified SEI model can naturally explain the (near) absence of orbital modulation in the resonance lines of HD 153919, which is one of the main motivations for this study.

4.1 HD 77581/Vela X-1

4.1.1 Strong wind lines

Adopting the ionization structure of the stellar wind as expected for single stars of the same effective temperature as HD 77581, and adopting $v_\infty=600$ kms^-1, the shape and variability of the Si  IV, C  IV and Al  III (²S-²P⁰ resonance doublet: restwavelengths 1854.716 and 1862.790, separation 1302 kms^-1) line profiles are reproduced rather well (Fig. 9, compare with Fig. 11).

Figure 11: Phase $\phi =0$ (solid) and $\phi =0.5$ (dotted) spectra for the N  V, Si  IV, C  IV and Al  III resonance lines in HD 77581/Vela X-1, with rest wavelengths indicated by vertical dashed lines and the velocity axis defined by the redmost component (labelled above the graph). The differences between the $\phi =0.25$, 0.5 and 0.75 spectra relative to the $\phi =0$ spectrum are plotted below.

Figure 12: Same as Fig. 11, but for C  II, Si  III and Fe  III.

The N  V line profile (Fig. 10, calculated with $I_{\rm g,max}=5555$), however, cannot be modelled using a normal ionization balance that would yield an extremely profound HM-effect, especially at the inner part of the absorption (Fig. 10a), which is not observed (Fig. 11). Instead, the observations seem to favour the degree of ionization to increase with distance to the primary: $\alpha_2=-1$ gives an acceptable fit, suppressing the orbital modulation of the absorption trough, while preserving some of the emission (Fig. 10b). The line profile and its modulation are very similar for $\alpha_1=0$ or -1 (Fig. 10c), only requiring a slightly larger integrated optical depth (T). If , however, much of the absorption arises from material at extreme velocities, causing the absorption profile to extend up to large positive velocities beyond the severely diminished emission. The absorption observed longward of the N  V line can be attributed to a C  III line.

The slope of the blue absorption wing (especially in Si  IV and C  IV, see Fig. 2) indicates that the level of saturation (deeper and at velocities slightly more negative than $-v_\infty$) is not reproduced well by the SEI turbulence recipe (Eq. (19)). The shocked wind structure produced by numerical simulations (Owocki 1994), however, seem to predict more material at extreme velocities (rather than gaußian turbulence). Furthermore, the turbulence may not be uniform throughout the wind, as it is assumed here.

4.1.2 Variations in addition to the HM-effect: evidence for a photo-ionization wake

The main discrepancy between the model and observations of HD 77581/Vela X-1 is the -400 to 0 kms^-1 region, where the observed intensity is often lower at $\phi =0.5$ than at $\phi =0$, contrary to what the HM-effect predicts. This may be due to the presence of a photo-ionization wake that enhances absorption at phases (Kaper et al. 1994). The photo-ionization wake, as observed in strong optical lines like the hydrogen Balmer lines and He  I lines in the spectrum of HD 77581, is situated close to the star at the trailing border of the ionization zone.

This effect is especially pronounced in N  V (Fig. 11), possibly as a result of a change in ionization degree within the photo-ionization wake. Enhanced absorption is seen between +200 and -600 kms^-1 ( $=-v_\infty$), with its maximum evolving towards more negative velocities from $\phi =0.25$ to $\phi=0.75$. Absorption at $\phi =0.25$ (or enhanced emission at $\phi =0$) extends up to +600 kms^-1.

Similar enhanced absorption is seen in Si  IV, C  IV and Al  III (Fig. 11) at -200 and -300 to -400 kms^-1 around $\phi =0.5$ and 0.75, respectively, and perhaps already around $\phi =0.25$. The enhanced absorption due to the photo-ionization wake may have reduced the amplitude of the HM-effect seen between -600 and -400 kms^-1.

Several weaker UV lines also show orbital modulation. The ²P⁰-²D resonance doublet of C  II at 1334.5323 and 1335.7077 Å (Fig. 12) is sufficiently strong to show the HM-effect between 0 and about +700 kms^-1 and between about -400 and -600 kms^-1, but slightly enhanced absorption at -300 kms^-1 around $\phi=0.75$ may be attributed to the photo-ionization wake.

Figure 13: Lightcurve for the integrated flux in the Si  III resonance lines around 1300 Å in HD 77581/Vela X-1. The dotted line indicates the flux level around $\phi =0$.

The strongly variable ³P⁰-³P multiplet UV 4 of Si  III (Fig. 12) consists of 6 components, of which 2 are blended: the pattern of variability is repeated five-fold. The absorption is diminished at -400 kms^-1 at $\phi =0.5$due to the HM-effect, but enhanced at -300 kms^-1 at $\phi=0.75$ due to the photo-ionization wake. In Fig. 13 a lightcurve is plotted for the spectral region between 1292 and 1303 Å - corresponding to -2600 and -70 kms^-1, respectively. Absorption is minimal between $\phi=0.4$ and 0.5, and strongest between $\phi=0.7$ and 0.8. The blending of the Si  III components makes it difficult to detect variability at -600 kms^-1, although there is a hint for diminished absorption at $\phi =0.5$ at this velocity in the bluemost component at 1294.543 Å. The ¹S-¹P⁰ (multiplet UV 2) resonance singlet of Si  III at 1206.51 Å could not be studied, due to the severe interstellar extinction and the strong Ly$\alpha$ geocoronal line at 1215.34 Å in the neighbouring echelle order. The subordinate singlets of ¹P⁰-¹S multiplets UV 9 and UV 10 of Si  III at 1417.24 and 1312.59 Å, respectively, do not show variability, although the lines are clearly present in the spectrum - particularly the UV 9 singlet. The ²P⁰-²S resonance doublet of Si  II at 1304.372 Å (+240 kms^-1 in Fig. 12) does not show any variability either, as it is of interstellar origin.

The spectral region around 1900 Å is dominated by numerous lines of Fe  III, of which the a⁷S-z⁷P⁰ multiplet 34 is the strongest. It consists of 3 widely separated components at 1895.456, 1914.056 and 1926.304 Å (Fig. 12). They all clearly show the diminished absorption at -400 and -600 kms^-1around $\phi =0.5$ due to the HM-effect, and enhanced absorption at -300 km s^-1 around $\phi=0.75$ due to the photo-ionization wake, exactly as observed in the Al  III resonance doublet. Some confusion arises from the presence of 2 lines of the a⁵G-z⁵H⁰ multiplet 51 of Fe  III at 1922.789 and 1915.083 Å, corresponding to -550 and -1750 kms^-1 with respect to 1926.304 Å, respectively. Multiplet 51 shows the same behaviour of the absorption near -300 and -400 kms^-1 as multiplet 34, but not the -600 kms^-1 absorption.

Figure 14: SEI models that approximately fit the observed line profiles and variability of N  V, Si  IV and C  IV in the UV spectrum of HD 153919/4U1700-37. Parameters are summarised in Table 3, and v=1corresponds to 1700 kms-1.

In conclusion, the orbital modulation of both the strong wind lines and other lines in the UV spectrum of HD 77581 indicates the presence of two absorption components that are not included in the SEI modelling. The most prominent of these can be explained by a photo-ionization wake, causing additional absorption in the line-of-sight at $v\sim-200$ kms^-1 around $\phi =0.5$and at $v\sim-300$ kms^-1 around $\phi=0.75$. In addition, at $v\sim-600$ kms^-1, i.e. the terminal velocity of the wind of HD 77581, more absorption is present outside the Strömgren zone than the SEI model reproduces. This might reflect the stellar wind structure at larger distances from the star: the velocity may be lower and the density higher after the wind has passed through the Strömgren zone compared to an undisturbed wind.

4.2 HD 153919/4U1700-37

The shape and (lack of) orbital modulation of the line profiles in HD 153919/4U1700-37 are reproduced well (Fig. 14). The exact value for qis hard to determine because of the (near) absence of the HM-effect due to a combination of strong absorption and the presence of turbulence, yet it is clear that q<4 would definitely yield a too strong HM-effect. Also, the HM-effect is predicted to appear strongest at velocities between $-v_\infty$and 0, implying that any variations in the blue absorption wing (at $\vert v\vert>v_\infty$) without accompanying HM-effect at smaller velocities must be due to some other mechanism, e.g. Raman-scattered far-UV emission lines (Kaper et al. 1990).