Would you kindly specify the procedure you used. Why are your frequencies so low? 

Sorry, I corrected the frequencies.

Actually I've used n and k of gold for different wavelengths that were on that website. Then I've used this MATLAB code:

epsrau=(nau.^2)-(kau.^2);

epsiau=diag(2*nau*kau');

epsau=epsrau+(i*epsiau);

k0=6.28*(landa.^-1);

radicalau=(diag(epsau*((1+epsau).^-1)')).^0.5;

ksppau=diag(k0*radicalau');

kfinal=real(ksppau)

omega=2*pi*3*10^8./(landa*10^(-9))

 finally I've plotted the Omega versus Re(K) diagram.

Frequencies are more like it now. I suspect that the square root operation in you code peaks up a wrong root, randomly. This happens.

I've changed the code to this:

epsrau=(nau.^2)-(kau.^2);

epsiau=diag(2*nau*kau');

epsau=epsrau+(i*epsiau);

zarib=6.28*(landa.^-1);

radicalau=sqrt(diag(epsau*((1+epsau).^-1)'));

ksppau=diag(zarib*radicalau');

k0=6.28*((landa*10^(-9)).^-1);

kfinal=real(ksppau)

W=2*pi*3*10^8./(landa*10^(-9))

but the result doesn't change.

The code is correct I've tested it for silver (Ag).

Strange

I don't know mathlab but rewring you code in C

#include

#include

#include

#include

int main()

{

FILE *f;

f = fopen("Au.txt", "r");

char s[100];

float n, k, epsr, epsi, landa;

while (!feof(f))

{

fgets(s, 100, f);

sscanf(s, "%f %f %f", &landa, &n, &k);

epsr=(n*n)-(k*k);

epsi= 2*n*k;

complex double eps=epsr+I*epsi;

double zarib=6.28/landa;

complex double radical=csqrt(eps/(1+ eps));

complex double kspp=zarib*radical;

double k0=2*M_PI/(landa*1e-9);

double kfinal=creal(kspp);

double W=2*M_PI*3e8/(landa*1e-9);

printf("%f %g \n", kfinal, W);

}

return 0;

}

produces

Mr. Goliney thank you for taking the time to rewrite this code.