Using the NONLINEARMODEL class, you can get the standard errors of the fitted variables as follows:
fitFunc = @(b,t) b(1)*exp(-b(2)*t) + b(3)*exp(-b(4)*t);
%# leave off the semicolon for nicely formatted output
mdl = fitnlm(tVector, energies, fitFunc, initialGuessForBeta)
%# SE of coefficients
mdl.Coefficients.SE