Thank you all for the responses. I understand leakage current is a big concern.
How about passing the ac signal (from AD9833 like IC's) through a capacitor? like the one discussed in a paper. I'm just copying and pasting it here (forgive me if I'm violating any rules here).
It was taken from Circuit Design Considerations for Current Preamplifiers for Scanning Tunneling Microscopy by Alex Kandel
"The preamplifier transimpedance gain was measured directly as a function of frequency, using a programmable sine-wave generator (AD9833, Analog Devices) at discrete frequencies between 20 Hz and 15 kHz. In order to measure noise levels, a 0.9 Hz square wave was used as a quasi-DC current source, and the frequency spectrum of the noise was obtained through the discrete-time Fourier transform. In both the cases, low-current test signals were generated by passing the output of a function generator through a capacitor. While a series resistor could also be used for this purpose, the highvalue resistor necessary to produce picoampere-to-nanoampere currents will typically have an undesirable level of parasitic capacitance and/or inductance.3,4 The 1.05 V AD9833 output was filtered, attenuated between 10 and 1000 times, and then passed through a 4 pF capacitor, producing sine-wave signals with amplitudes between 0.01 and 0.5 nA (depending on the frequency and attenuation). A 0.15 nA square wave was produced using a triangle-wave analog oscillator circuit based on a low-noise operational amplifier, also coupled through a 220 pF capacitor."
If this can be done, they suggest the use of gas-filled capacitor...I'm unlucky, not able to find suitable commercial gas-filled capacitors.
Regarda
ArM