In this paper we serendipitously identify X-ray cluster candidates using XMM-Newton archival observations complemented by 5-band optical photometric follow-up observations (r~23 mag) as part of the X-ray Identification (XID) programme. Our sample covers an area of ~2.1 sq. deg (15 XMM-Newton fields) and comprises a total of 21 (19 serendipitous + 2 target) extended X-ray sources to the limit f(0.5-2keV) ~ 6x10^{-15} cgs with a high probability (> 99.9%) of being extended on the XMM-Newton images. Of the 21 cluster candidates 7 are spectroscopically confirmed in the literature. Exploiting the optical data available for these fields we discover that 68% of the X-ray cluster candidates are associated with optical galaxy overdensities. We also attempt to constrain the redshifts of our cluster candidates using photometric methods. We thus construct the photometric redshift distribution of galaxies in the vicinity of each X-ray selected cluster candidate and search for statistically significant redshift peaks against that of the background distribution of field galaxies. Comparison of the photometric with spectroscopic redshift estimates for the confirmed clusters suggest that our simple method is robust out to z~0.5. For clusters at higher z, deeper optical data are required to estimate reliable photometric redshifts. Finally, using the sample of the 19 serendipitous X-ray selected cluster candidates we estimate their surface density down to f(0.5-2keV) ~ 6x10^{-15} cgs and find it to be in fair agreement with previous and recent studies.