Measurement of the CIB has proven to be extraordinarily difficult. The only observational signature of the CIB is an isotropic background arising external to the Galaxy. There is no distinctive, predictable spectral signature of the CIB, since many sources of luminosity can contribute, and their distribution in space and time, as well as the character and distribution of cosmic dust, will determine the observed spectral energy distribution. Direct observation of the CIB is impeded by the many foreground contributors to the infrared sky brightness at all wavelengths, several of them quite bright. A further challenge is the need for absolute photometry, that is, measurements relative to a well-established zero brightness level. Instrument self-emission, stray light, and electronic offset signals must be eliminated or accurately determined.
The COBE mission was designed to accomplish the best possible search for the CIB from our location in the cosmos. It was recognized that such measurements required cryogenic instruments operated above the atmosphere for a substantial period of time so that potential sources of systematic errors in the absolute brightness measurements could be identified and evaluated. The approach was to obtain maps of high photometric quality over the whole sky and over a broad wavelength range so that foreground sources could be identified by their spatial and spectral signatures, and that uncertainties associated with discriminating them from the CIB could be minimized. The COBE spacecraft carried two instruments contributing to these results (Boggess et al. 1992). The Diffuse Infrared Background Experiment (DIRBE) was designed primarily to search for the CIB from 1.25 to 240 µm. The Far Infrared Absolute Spectrophotometer (FIRAS) was designed primarily to make a definitive measurement of the spectrum of the CMB, and to extend the search for the CIB to millimeter wavelengths. All aspects of the COBE mission, from the instruments to the spacecraft, orbit, sky scan strategy, and data processing, were designed to optimize the ability to make these difficult diffuse background measurements.
The DIRBE instrument was an absolute photometer with 10 broad
photometric bands
at 1.25, 2.2, 3.5, 4.9, 12, 25, 60, 100, 140, and 240
µm. It was designed to enable detection of CIB levels as
faint as 
I ~ 1 nW
m-2 sr-1. The DIRBE instrument is described by
Silverberg et
al. (1993),
and a summary of the DIRBE investigation and its initial results is
provided by
Hauser et
al. (1998).
Additional detailed information is given in the
COBE DIRBE
Explanatory Supplement (1997).
The DIRBE was designed for extremely strong stray light rejection,
employing an off-axis Gregorian telescope, with a pupil stop, multiple
field stops, and extensive internal and external baffling. The stray
light was demonstrated to be less than 1 nW m-2
sr-1 at all wavelengths. The instrument contained an internal
cold chopper operating at 32 Hz, which allowed continual measurement of
the sky brightness relative to that of a stable, cold internal beam
stop. It had a full beam cold shutter, closed typically five times per
orbit to establish the zero point of the photometric scale by allowing
measurement of the instrumental radiative and electronic zero point
offsets. These offsets were stable, and the uncertainty in the
determination of the offsets was 1 nW m-2 sr-1 or
less from 1.25 to 100 µm, and
was 5 (2) nW m-2 sr-1 at 140 (240) µm
respectively.
The DIRBE instantaneous field of view was 0.7° × 0.7°, a
compromise between ability to discriminate stars and ability to map the
whole sky with high redundancy every six months. Redundancy was
important because it allowed monitoring of the annual variations in
apparent sky brightness in all directions due to the Earth's motion
within the interplanetary dust (IPD) cloud. This variation is a unique
signature of the IPD contribution to the signal. The sensitivity
(1
) of the instrument from
1.25 to 100 µm was better
than 1 nW m-2 sr-1 per field of view averaged over
the ten months of cryogenic operation, and was 33 (11) nW m-2
sr-1 at 140 (240) µm respectively. The
instrument gain stability was excellent, and was monitored on short time
scales using internal stimulators. Repeated observations of stable
celestial sources provided photometric closure over the sky, and assured
reproducible photometry to ~ 1% or better for the duration of the
mission. Calibration of the DIRBE flux scale was accomplished from scans
of a few isolated infrared sources of known brightness.
The FIRAS instrument was a Fourier transform spectrometer in the form of
a polarizing Michelson interferometer, providing extremely precise
spectral comparison of the sky brightness with that of a very accurate
full beam blackbody calibrator at wavelengths from 100 µm
to 1 cm. The instrument and calibration are discussed by
Mather et
al. (1993),
Fixsen et
al. (1994),
Mather et
al. (1999),
and the
COBE FIRAS
Explanatory Supplement (1997).
The FIRAS light collector was a Winston cone with a flared horn,
providing a 7° diameter field of view and low sidelobe response
over the broad FIRAS spectral range. The sensitivity
(1) of the instrument from
500 µm to 3 mm was 0.8 nW m-2 sr-1
per field of view averaged over the ten months of cryogenic
operation. The photometric calibration errors
(1) associated with the
precision blackbody calibrator were typically 0.02 MJy/sr, corresponding
to
I ~ 0.3 (0.1) nW
m-2 sr-1 at 200 (600) µm wavelength
respectively. Over the course of the COBE mission, the FIRAS obtained
superbly calibrated absolute spectral maps of almost the whole sky.
Using FIRAS data,
Fixsen et
al. (1996) and
Mather et al. (1999)
showed that the rms deviation of the CMB spectrum from that of a
(2.725 ± 0.002) K blackbody at wavelengths longer than 476
µm was less than 50 parts per million of the peak CMB
brightness. Accurate knowledge of the CMB is necessary to search for the
CIB at submillimeter wavelengths.
Fixsen et al. (1997) demonstrated that the zero point and gain calibrations of the DIRBE and FIRAS photometric scales are consistent within the quoted uncertainties of each and the systematic uncertainties of the comparison. Since the FIRAS systematic calibration uncertainties are smaller than those of the DIRBE, these results can be used to make small systematic corrections to the DIRBE 140 and 240 µm measurements. Hauser et al. (1998) discussed the effect of using the FIRAS calibration on the DIRBE 140 and 240 µm CIB detections, but DIRBE-based results in the literature, including this paper, otherwise use the DIRBE calibration.
Hence, the DIRBE and FIRAS instruments provided extensive, consistent, high quality photometric data on which to base a search for the CIB.