It's difficult to find a representative sample for this sort of thing. One can see brain size differences in the PING dataset. This was the dataset that the authors accidentally leaked with that "poverty shrinks the brain" paper. This sample is not very representative since they basically just sampled children around universities.

The PING Data Resource is the product of a multi-site project involving developmental researchers across the United States including UC San Diego; the University of Hawaii; UC Los Angeles; Children’s Hospital of Los Angeles of the University of Southern California; UC Davis: Kennedy Krieger Institute of Johns Hopkins University: Sackler Institute of Cornell University; University of Massachusetts; Massachusetts General Hospital at Harvard University; and Yale University.
The Data Resource includes neurodevelopmental histories, information about developing mental and emotional functions, multimodal brain imaging data, and genotypes for well over 1000 children and adolescents between the ages of 3 and 20.

http://pingstudy.ucsd.edu/welcome.html

Still, there is a variable for cortex total area ("cort_area.ctx.total") which has a d=.58 (W/B). The familiar gender pattern is also there.

I attach a plot.

The Black children were somewhat older, so this makes the difference smaller. Mean ages W/B 11.32 vs. 12.16. Age has non-linear effects so correcting for it requires a flexible model. I used LOESS, and after correction total cortical area d=.61. For comparison, the gender d is 1.12.

N's 203/973 for B/W.

--

Did you look at the Connectome dataset? Looks like they will oversample siblings and release genomic data too. So good for many analyses. I don't have the dataset or have the time to pursue it. Send it to me if you can get a hold of it.

http://www.humanconnectome.org/about/project/recruitment.html

PING dataset has total volume too. The differences are the same: .62 and .1.10 d for race/gender. Cortex x total volume is .89, hence the similar findings.