Dear Matthias Troyer and Synge Todo,
I'm sorry for the late response but I was away from my computer several days! Thank's for your help!!
I increased the number of SWEEPS and I have finally found the square root behaviour of the error bars as a function of the number of SWEEPS. I imagine (I didn't had time to check yet) the method of estimation of the error bars makes an underestimation of the variance.
Thanks again for your time
João Lopes