How do I turn a symbolic function into a matrix?

I'm trying to multiply multiple symbolic equations together to create an equation (NFKernel) and then turn that equation into a 100 by 100 matrix.

The reason is that the math is much easier when not in matrix form, plus I need the matrix entries later for eigenvalue work. I know that if I could write out the equation for NFKernel and then use ngrid I can easily turn the equation into a matrix. The problem is that creating this large equation (NFKernel), I'm not sure how to make Matlab recall the parts of the equation so that I can turn it into a matrix. What I want is something where I run the script using symbolic variables up until the end of the command, where there would be some sort of "recall that NFKernel = this bugger" and then use ngrid (or something similar) to make my matrix.

Does anyone have any idea how to do this? I've tried:

    syms f(x,y); 
    f(x,y) = NFKernel
    [x,y] = ndgrid(linspace(1/10,500), linspace(1/10, 500))
    subs (f(x,y))
    NFKernel = double(ans)

but my computer, or the program, I'm not sure which, goes so slowly it can never finish.

Here's the exact problem:

syms x y 
NFSurv = .0723*log(x)+.6829
NFGrowrate = 0.646598*x+4.7886;
NFGrowstdv = 4.747;
NFGrow = (1/(NFGrowstdv*((2*pi).^(0.5))))*(2.7182818).^(-(0.5*(((y-NFGrowrate)/NFGrowstdv).^(2))))
NFP= NFGrow*NFSurv
NFbz = ((1-0.0338)*(0.06238*y-0.51856)).^(-0.33); 
NFPbz = 0.881*log(x)+0.374;
newseed = 2; 
NFC0 = (1/(2*pi*2.21).^(0.5))*(2.7182818).^(-0.5*(((newseed-2.480)/2.21).^(2)));
NFFec= NFPbz*NFbz*NFC0
NFKernel = NFP+NFFec
I need NFKernel as a 100 by 100 matrix.