Micromatter makes custom standards that are widely used (http://www.micromatter.com/xrf.php).  Depending on your matrix you may have to correct for self absorption.

There are different possibilities. If you evaluate your data using simulations, you might get along with pure elemental standard(s) of lead and the other elements in your sample.

Otherwise I would recommend to generate standards that are similar to your sample, i.e. the same material (but free of lead), then you can make a set of standards with different lead concentrations.

If this is also difficult, you can also take your sample material and make a dilution series from the sample, i.e. you take 5 batches of your sample; to 4 of them you add e.g. 5ppm, 10ppm, 20ppm, 40ppm of lead in a suitable form (PbO2 or what else). Then you plot the data vs. the added amount.  If you can fit the data, you can judge whether you find the expected linear correlation and if you have a meaningful offset which would be the original concentration. 

Micromatter makes custom standards that are widely used (http://www.micromatter.com/xrf.php).  Depending on your matrix you may have to correct for self absorption.

It is very difficult and often impractical to perform XRF analyses that are both accurate and precise on powders. A case in point is the XRF determination of iron in sand when the iron contaminant is iron chromite that naturally occurs as particles. In an investigation, sand samples were ground (studies were made with -200 mesh and -400 mesh powders) and many pressed pellets were made, the accuracy and precision of the iron determination was very poor when compared to wet chemical analysis. This was due to at least two factors: (1) the differing hardness between sand and iron chromite guaranteed that the grinding process treated each constituent differently. This implied that the ground fractions might not be representative of the original sample, and (2) the lack of assurance that the distribution of iron chromite on the analytical surface of the pressed pellet was the same as in the bulk sample.

We abandoned the pressed powder approach and used instead the (lithium tetraborate flux) fusion pellet approach or wet chemical analysis.

Your need to prepare lead standards faces the same problem, i.e., assuring the homogeneity of mixed powders. Mixing of powders is very difficult. The homogeneity attained is highly scale dependent. Bulk samples of a few cubic centimeters probably will have identical compositions but the volume sample by XRF (of the order of 0.03 cubic centimeters) will likely have a different composition because of the homogeneity problem associated with mixing powders.

I would recommend that your standards and samples be prepared by the fusion technique. This will better ensure homogeneity and analytical precision. You did not mention the nature of your samples. If they have pretty much the same composition, then one set of standards should suffice. However, if they are (widely ?) different, you may need different standards to account for the matrix effect in the XRF procedure.

Is the solid phase of plant origin or a heavier matrix?  I agree with Ernst, mixing of powders is difficult and it's difficult to keep them mixed.  For plant samples we have had some success by adding a solution of the trace constituent to the matrix, snap freezing, freeze drying .and fine grinding.

This is a recurrent interest