A common reaction when I explain that we are working with colloidal gold is that people think that such samples are very expensive because of the gold. This is not the case. Our measurement techniques are very sensitive, and thanks to the very strong optical response of colloidal gold particles, only very small amounts of gold are needed to "see something".
To illustrate this, "industry standard" colloidal gold has a concentration of gold particles such that its optical density is 1 ("OD 1", meaning that a 1 cm thick sample absorbs 90% of the light at a particular wavelength). If we take 20 nm particles, this gives a concentration of 1.1 nM (650 billion particles in a ml). The amount of gold in a typical 1 ml sample is 54 µg.
A typical commercially available 25 ml quantity of colloidal gold suspension contains the equivalent of 1.3 mg of gold. The value of the gold is then about 5 eurocents!
When we do resonant light scattering spectroscopy (see for example this paper), we may work with concentrations on the order of 0.1 pM (about 60 million particles in a ml) for 80 nm gold particles. The active detection volume of our spectrometer can be estimated to be about 1 µl (therefore containing 60000 of particles), which means that we are routinely measuring 0.3 nanograms of gold.
In our microfluidic work (such as the electrokinetics stuff and this work), the observed volume is typically 1000 times smaller still (1 picoliter), and when we look carefully through the microscope we can indeed see the particles individually. We may try counting them, even though their incessant Brownian motion complicates this task. A single 100 nm gold nanosphere has a mass of 10 femtogrammes.