Simply, one can opt the factorial CRD, however I would also suggest to look the Response Surface Design (if objective revolves around optimisation).

Bhim, you can run them in a factorial scheme then apply multiple regression to realize the best combination and/or the combinations among them for optimized yield. i guess it would help you. good work and luck

the real problem for plant nutrition by leaves is the control of run off and root insulation from treatment. Possibly a control with elements added to roots would test absorption. Leaf diagnostic  levels in good conditions can give the element ratios to be searched, and concentration can be applied in log progression. 

you need to utilize Qashi experimental design,then apply  factorial analysis

Best is Diagnosis Recommendation Integrated System (DRIS) which take into consideration the relative requirement of individual nutrient.

I'm still holding the idea of a proof of real need to use foliar sprays. Soil conditions like high pH and leaf diagnosis should indicate either B or Zn and Fe deficiency. Usually the Timing of treatment can match more or less plant requirement, relative to bud and flower status. B should be in sodium pentaborate form. For statistics all depends on the effect, as if a moderate one would require several replications to be detected. Date, rate and element proportion are variable in a reduced factorial using a fixed start for proportion like 1B  2Fe 2Zn and rate of .4,  4 and 40 ppm in split plot with application date. If any effect is detected you may optimize

There is not best treatment before you test it. how did you know that is the best? to make a choice of design you have to know how many treatment do you want to test? if you have five treatment use the C.R.B.D for a good precision.

Factoraial RBD with absolute control