Hi SA
Speculative answer:  I don't believe that the function F is a likely candidate for a Laplace transform.  What you have is (looking at just the first term)
F = const/s*(s+a*x^c)*(s+b*x^f)
which is actually (by mistake?)
 (const/s)*(s+a*x^c)*(s+b*x^f)
i.e. two factors involving s in the numerator.  For the inverse transforn, that leads to stuff like the derivative of a delta function.  What seems more likely is
 const/(s*(s+a*x^c)*(s+b*x^f))
 with all the s factors in the denominator.  Similarly for the second term.
Both terms have a factor of s in the denominator.  If   
 invLaplace(g(s)/s) = G(t)
then removing the s in the denominator effectively multplies by s and gives the time domain derivative,   
invLaplace(g(s)) = dG(t)/dt.
The code below does both cases.
F =  d1/(s*(s+a*x^c)*(s+b*x^f)) + d2/(s*(s+k)*(s+a*x^c)*(s+b*x^f)); 
Fs = d1/(  (s+a*x^c)*(s+b*x^f)) + d2/(  (s+k)*(s+a*x^c)*(s+b*x^f)); 
f_fun = matlabFunction(f);  
fs_fun = matlabFunction(fs);  
title('Inverse Laplace Transform Plot')
title('Inverse Laplace Transform Plot')