The source counts, corrected for cluster contamination and lensing effects, in both the LW2 and LW3 bands, are presented in Fig. 2. At 15 µJy we have used only 5 sources (i.e. 27 sources above 60 µJy before lensing amplification correction). According to Hogg & Turner (1998), this is sufficient to avoid the positive flux-estimate Eddington bias which occurs for faint source counts.
At 7 µm, we find a source density greater than that of Taniguchi et al. (1997) by a factor of 2. We suggest that their observations may be incomplete below 50 µJy, whereas our 7 µm map is 80% complete down to 30 µJy in the central with lensing amplification correction.
At 15 µm the number counts are compatible with the results of Aussel et al. (1998) in the HDF. However, with the help of gravitational lensing we are able to extend the counts in both bands down to 30µJy. Putting all background sources at only slightly decreases the faintest counts, at the faint end, because a few suspectedly high-z faint sources would be less amplified. This shows the small dependence of our derived counts on the redshift distribution of the sample. The higher lensing-uncorrected counts at the bright end are due to small number statistical bias, where more sources than average appear in the high gain region of the A2390 cluster potential.
We find a total number density N7( 30µJy) = 3.5 arcmin-2 at 7 µm , and N15( 30µJy) = 13 arcmin-2 at 15 µm , with a slope .
The 15 µm counts show a steadily increasing excess (by more than a factor of 10) with respect to the prediction of a no-evolution model (dashed line, Franceschini et al. 1997). This confirms the steeper count slope below 1 mJy found on the Lockman Hole (Elbaz et al. 1998) and is in good agreement with the ISO HDF counts (Aussel et al. 1998). The counts are a factor 2 to 3 higher than the boundaries of the counts coming from an early analysis of the background fluctuations in the ISOCAM 15 µm map of the HDF (Oliver et al. 1997). The slope stays close to -1.5 down to 30 µJy. This source density at faint levels favours evolution models, needed to fit the counts at brighter fluxes (Elbaz et al. 1998), or single-population models (Blain et al. 1998).
Integrating the number counts over our A2390 flux range (30 µJy to 200µJy) and over other ISOCAM deep surveys (Elbaz et al. 1998) up to 50 mJy we find the resolved background to be and W m-2 sr-1 emitted respectively at 15 µm and 7 µm. If we restrict to the very reliable counts, i.e. down to 50 µJy only at 15 µm, we find a conservative background value of W m-2 sr-1. The 15µm lower limit is close to the current upper limits set by the gamma-CMBR photon-photon pair production (Stanev and Franceschini 1997). Note that our 7µm source counts power law slightly underestimates counts at higher fluxes; a slope of -1.3 would be necessary to be compatible with Flores et al. (1999a) counts of 0.45 per arcmin2 above 150 mJy).
© European Southern Observatory (ESO) 1999
Online publication: March 1, 1999