Because there was no answer, I chose to implement the function myself. The following notebook contains a litteral implementation of the algorithm found in
Rousseeuw, P. J., & Van Driessen, K. (1999) A fast algorithm for the minimum covariance determinant estimator. Technometrics, 41, 212-223.
The only addition is the Consistency-to-normal-model factor; see
Pison, G., Van Aelst, S., & Willems, G. (2002) Small sample corrections for LTS and MCD. Metrika (2002) 55: 111. doi:10.1007/s001840200191
This factor makes the estimate comparable to the variance-covariance matrix of the full data set if there was not contaminant and the population was truly normal.
The notebook can be found at MCD.nb. It is distributed without no warranty. For small samples,it was compared with the R implementation of the algorithm and the results were identical. If you spot problems, let me know.