Cosmology has entered the era of precision cosmic microwave background (CMB) measurements. Since the original detection of temperature perturbations on large angular scales by the COBE satellite (Smoot et al. 1992), there has been a myriad of new detections, resulting in a data set spanning roughly two orders of magnitude in angular scale (Lineweaver et al. 1997; White et al. 1994). The extraction of cosmological information requires careful control and understanding of all possible sources of signal contamination. The current quest for high precision determination of cosmological parameters (Jungman et al. 1996; Knox 1995) demands a correspondingly greater understanding of all foregrounds. In particular, the Galaxy, via synchrotron, dust and free-free emission (Bremsstrahlung), represents a source of foreground brightness fluctuations which all experiments must reckon with. These three contaminating emissions define a "valley" in the brightness-frequency plane centered around 90 GHz, representing the point of smallest Galactic contamination (Kogut et al. 1996a). Although, clearly, CMB efforts are concentrated in this "valley", Galactic signals must nonetheless be carefully removed to extract the purely cosmological fluctuations and to achieve the desired precision on cosmological parameters.
The removal of these foregrounds is usually done in one of two ways. With sufficient frequency coverage and a high signal-to-noise ratio, a spectral analysis of the CMB data alone can in principle distinguish the Galactic foregrounds from the CMB signal. The other approach is to use sky maps made at other frequencies as templates and to extrapolate a given foreground emission into the CMB bands according to its spectral dependence. Even when the quality of the CMB data permits the former technique, the second approach provides an important, external check on the removal procedure. For synchrotron emission, one usually uses the 408 MHz Haslam map (Haslam et al. 1981) and the 1420 MHz survey (Reich & Reich 1986) as a template (uncertain spatial variations of the synchrotron frequency index renders the procedure slightly less straightforward than one would hope). The IRAS all sky survey serves as a useful template for dust emission on angular scales under , and it is usually augmented with DIRBE maps on larger angular scales (as with synchrotron emission, uncertainty in the exact slope of the dust emission law introduces an unfortunate complication).
In this paper, we address the question of Galactic free-free emission in relation to CMB anisotropy measurements (two recent reviews are given by Smoot 1998 and Bartlett & Amram 1998). Among the three Galactic sources of troublesome microwave emission, free-free emission is the most difficult to control. This is because the only frequency range in which it dominates over dust and synchrotron emission is in the CMB valley; in other words, one cannot extrapolate maps made at much lower or higher frequencies into the CMB valley to remove free-free contamination. What is needed is a tracer of the warm ionized interstellar medium (WIM) responsible for free-free emission. Given that at high Galactic latitudes there is minimal extinction from dust, one expects Hydrogen line emission in the excited gas to be a good possibility for such a tracer.
The line emission is measured in Rayleighs (1R photons cm-2 s-1 ster-1 erg cm-2 s-1 ster-1 at Å) and may be expressed in terms of the temperature and emission measure, EM, of the WIM for Case B recombination:
where K; this expression is valid for temperatures (e.g. Reynolds 1990), more accurate formulae are given by Valls-Gabaud (1998). Free-free emission depends on the same quantities (given here for pure Hydrogen and in the limit as ):
where is the brightness temperature, the observation frequency is Hz, and is the thermally averaged gaunt factor, which to 20% for few is
(e.g. Smoot 1998). Thus, the free-free brightness associated with a given intensity is approximately
Valls-Gabaud (1998) discusses more accurate expressions. There does not, as of yet, exist a complete survey of the sky in , and the distribution of the warm ionized medium (WIM) of our galaxy remains somewhat of a mystery. Local sources pose the most serious difficulties for efforts to measure the Galactic emission. The Earth's geocorona emits in with an intensity of R, depending on the season, the solar activity and the solar depression angle. This is an order of magnitude larger than the typical signal we expect at high Galactic latitude. In addition, there is an OH line from the atmosphere at Å. Fortunately, the Earth's motion through the Galaxy displaces the Galactic signal relative to the local emission, and thus the cleanest way to extract a Galactic signal is by use of a high-resolution spectrometer. Reynolds has developed this approach with a double Fabry-Perot system (Reynolds 1990) to study the Galactic emission on degree angular scales with pointed observations and a small-area survey below the Galactic Plane (Reynolds 1992; Reynolds 1980). This has culminated in the construction of WHAM (Wisconsin Mapper), which is currently surveying the entire northern sky at 1 degree resolution (see http://www.astro.wisc.edu/wham/).
Other groups have recently surveyed areas in the north using narrow band filters (Gaustad et al. 1996; Simonetti et al. 1996). This technique has the advantage of much greater simplicity and lower cost; the inconveniences are that one must remove the stellar contribution by extrapolation of off-band filters and that the Geocoronal emission cannot be subtracted correctly. Nevertheless, if the geocoronal emission is stable and uniform across the field-of-view (survey area) during the observations, then useful upper limits on the anisotropy of the Galactic signal can be obtained. Both Gaustad et al. (1996) and Simonetti et al. (1996) have placed limits on the possible contamination of CMB observations at the North Celestial Pole and concluded that the Saskatoon (Wollack et al. 1997; Netterfield et al. 1997) experiment is unaffected by free-free contamination.
The situation is actually rather more complicated. Leitch et al. (1997) have recently reported the detection of a foreground signal around the NCP in data taken from the Owens Valley Radio Observatory. The signal has a spectral index favoring free-free emission, and it is well correlated with IRAS maps of the area. Such a correlation between free-free emission and dust emission has also been remarked by the COBE team in the DMR data at high Galactic latitudes (Kogut et al. 1996a; 1996b). If the foreground seen around the NCP is indeed due to Bremsstrahlung, then the intensity is 60 times larger than the limits implied by the narrow band observations in ! As discussed by Leitch et al. (1997), this could be explained by a gas at K, instead of K. It is interesting to note that K is the virial temperature of our Galactic halo. Although difficult to understand how, another possibility is that the narrow band observations are missing something. A further possibility is that this signal is due to the rotational emission of very small spinning dust grains (Draine & Lazarian 1998). In any case, present data are not sufficient to yield a complete understanding of the importance of free-free contamination for CMB observations (Smoot 1998; Bartlett & Amram 1998).
In this paper, we present some of our observations at high galactic latitude in the Southern Hemisphere. The telescope and detector system were optimized for a survey of the Galactic Plane in at a resolution of 9 arcsecs (Le Coarer et al. 1992), and so it is not the most appropriate instrument with which to constrain the distribution of the WIM on CMB angular scales ( degree). Nevertheless, by summing over pixel elements in the roughly field-of-view, we have been able to reach a sensitivity of R on a scale comparable to CMB measurements. Our goal was to check for free-free emission in the region of sky where Schuster et al. (1993, SP91) and Gunderson et al. (1995, SP94) detected microwave fluctuations.
© European Southern Observatory (ESO) 1998
Online publication: September 8, 1998