Yes, you just zero pad the input by providing the optional NFFT argument to fft()
If you use an NFFT value that is greater than the length of the input data, the DFT is interpolated. Instead of getting DFT coefficients at Fs/N where Fs is the sampling frequency, you get them at Fs/NFFT
If you have a peak that falls between two DFT bins, kFs/N and (k+1)Fs/N, then zero padding can help you resolve that peak better. It does not improve the frequency resolution of the DFT, but you can get a better picture of a single peak.