We observed NGC 3516 on two occasions during the Astro-2 mission, once
for 1518 s beginning at 5:35:34 UT on 1995 March 11, and again for 2200 s
beginning at 1:32:03 on 1995 March 13. Both observations were through a
20'' aperture during orbital night when airglow is at a minimum.
The basic design of HUT is described by Davidsen et al.
(1992)Davidsen92.
Briefly, a 0.9-m mirror collects light for a prime-focus, Rowland-circle
spectrograph. A photon-counting detector sensitive from 820--1840 Å
samples the dispersed spectrum at a resolution of 2--4 Å with
0.52 Å per pixel.
Improvements to HUT, its performance during
the Astro-2 mission, and our basic data reduction procedures
are described by Kruk et al. (1995)Kruk95.
The raw data were reduced by subtracting
dark counts, correcting for scattered geocoronal Ly
emission and subtracting airglow. We then flux-calibrated the spectrum
using the time-dependent inverse sensitivity curves developed from
on-orbit observations and model atmospheres of white dwarfs.
Statistical errors for each pixel are calculated from the raw count spectra
assuming a Poisson distribution and are propagated through the data reduction
process.
As there was no evidence for variability in the UV data,
the two separate observations are weighted by their exposure times and combined
to form the mean flux-calibrated HUT spectrum of NGC 3516
shown in Figure 1.
Figure 1:
The flux-calibrated spectrum of NGC 3516 obtained with the Hopkins
Ultraviolet Telescope during the Astro-2 mission is shown.
Significant emission and absorption lines are marked.
The indicated Lyman limit is at a redshift of 0.0075, and it is
intrinsic to NGC 3516 as are
the absorption lines of O VI, N V, Si IV, and C IV.
The remaining absorption lines are likely galactic in origin.
The earth symbol indicates the residual feature produced by
subtraction of geocoronal Ly
emission.
To model the spectrum of NGC 3516 and measure properties of the continuum,
emission lines and the absorption lines, we use the IRAF
task specfit ([Kriss 1994a]).
We fit the continuum with a power law in 
.
The brightest broad emission lines (O VI 
, Ly
, and
C IV 
) are well described by power law profiles
while single Gaussian components are adequate for the weaker broad lines.
The power law profile has a functional form
,
where 
(c.f. NGC 4151, [Kriss et al. 1992]).
Additional narrow Gaussian
cores are required for Ly
and He II 
.
Single Gaussian profiles are used for all absorption lines
other than the Lyman series.
We allow extinction to vary freely following a Cardelli, Clayton, & Mathis
(1989)CCM89 curve with 
.
To model hydrogen absorption in the Galaxy and in NGC 3516,
we compute grids of transmission functions including transitions up to
. Using Voigt profiles of varying column density and
Doppler parameter, we then convolve the transmission with the
instrument resolution.
Galactic neutral hydrogen is fixed at zero redshift at a column density
of 
([Stark et al. 1992]) with a Doppler
parameter of 
.
A sharp, redshifted Lyman edge is readily apparent in the NGC 3516 spectrum.
Neutral hydrogen intrinsic to NGC 3516 is permitted to vary freely in
column density, redshift, and Doppler parameter.
Our best fit yields 
for 1671 data points between 916 Å and 1800 Å\
(we omit a region from 1207-1222 Å surrounding geocoronal Ly
).
The fitted continuum has
with 
.
In frequency space this corresponds to a
spectral index 
for 
.
This is rather steep, but it is not that unusual for a low-luminosity AGN.
For comparison, the Astro-1 spectrum of NGC 4151 showed a spectral index of
1.50 ([Kriss et al. 1992]), the FOS spectrum of NGC 1566 has 
([Kriss et al. 1991]), and the UV continuum of M 81 has 
([Ho, Filippenko, & Sargent 1996]).
Our fitted extinction is higher than the predicted Galactic reddening of
--0.03 in the maps of Burstein & Heiles (1982)BH82,
but compatible with that expected using
and a gas-to-dust ratio of
([Shull & Van Steenberg 1985]).
Previous work based on IUE observations ([Kolman et al. 1993]; [Koratkar et al. 1996])
have determined 
based on the strength of the 2200 Å
absorption feature, but such a large extinction correction provides a poor
match to our data --- for 
, 
.
To show the sensitivity of the data to the extinction correction,
Figure 2 compares the observed spectrum with power-law
continua after correction for 
and 0.15.
The figure clearly shows that no extinction correction leads to a deficit in
continuum flux below 1000 Å, whereas an extinction correction of
produces an excess in flux at short wavelengths.
We suggest that the apparent strength
of the 2200 Å dip in the IUE data is not due to extinction, but rather
to the onset of broad emission from blended Fe II at 2300 Å.
Figure 2:
The figure shows the HUT spectrum of NGC 3516 on a log-log scale with
extinction corrections applied as indicated.
The spectra have been scaled up for display by the factors under each spectrum.
The solid straight lines through each spectrum are the best fit power law
continuum levels for the given extinction correction.
No correction clearly produces a deficit
in flux at short wavelengths, and 
leads to an excess in
continuum flux shortward of 1000 Å. 
provides the best fit.
Note that the broad wings on the emission lines leave clear continuum
visible only in narrow windows shortward of 980 Å and near 1130 Å,
1450 Å and 1800 Å.
For the neutral hydrogen intrinsic to NGC 3516 we find a best fit redshift
of 
, 
, and
.
This is blue-shifted by 
relative to the
systemic velocity of NGC 3516 measured using the stellar
absorption lines ([Vrtilek & Carleton 1985]).
The opaque Lyman limit requires neutral hydrogen with a minimum column
density 
(90% confidence).
Its sharpness limits b to less than 
at 90% confidence.
The effect of the assumed Doppler parameter on the shape of the intrinsic
Lyman limit is shown with an enlarged view of the Lyman-limit region in
NGC 3516 in Figure 3.
The general weakness of the Lyman absorption lines (only Ly
and
perhaps Ly
are detected)
gives an upper limit of 
at 90% confidence.
Figure 3:
This expanded view of the short-wavelength end of the HUT spectrum shows the
intrinsic Lyman limit and the O VI absorption more clearly.
The position of the Galactic Lyman limit is shown to illustrate the intrinsic
absorption. The solid, dotted, and dashed curves show the shape of the
intrinsic Lyman absorption in NGC 3516 for assumed Doppler parameters of
b = 10, 20, and 50 
, respectively. As described in the text,
at 90% confidence, 
.
The heavy and light solid lines show our fit to the O VI emission and
absorption. The light solid line shows the O VI emission profile
superposed on the powerlaw continuum. The heavy solid line shows the
contribution of the intrinsic O VI absorption to the fit.
The light solid line in the vicinity of 1037 Å shows the contribution
of intrinsic Ly
and Galactic C II 
to the fit.
Galactic absorption features in the spectrum are denoted by a ``(G)"
following the line identification.
Measured properties of the fitted emission lines are in Table 1.
The fitted absorption lines are summarized in Table 2.
All tabulated features, except for Ly
,
have a statistical significance exceeding 
.
The error matrix of the fit is used to derive the 
statistical errors
shown in the table.
Systematic errors of up to 
5% are likely to be present in the fluxes,
and our wavelength scale has a limiting accuracy of 
.
For ease in fitting, some poorly constrained parameters were linked to others
with better determined values.
For example, the velocity offset and FWHM of the C III 
and the N III 
lines were linked to share common values,
and the FWHM of He II 
was set equal to that of
He II 
.
Any entries in the tables with identical velocity offsets or FWHM's were
linked in a similar manner.
Since the intrinsic O VI absorption is difficult to see in
Figure 1, an enlarged view illustrating our fits to this
region is shown in Figure 3.
The O VI absorption doublet straddles the peak of the O VI emission
line, and the sharpness of the line peak is largely defined by the flux
escaping at wavelengths between the doublets.
Galactic C II, Galactic Ly
, and intrinsic Ly
absorption
all contribute to the deficit in flux on the blue wing
of the O VI emission.
