Toward an SOP for working with gold nanoparticles

We’ve been using gold nanoparticles (available for ~$60 for a 20mL bottle from BBI in sizes of 2, 5, 10, 15, 20, 30, 40, 50, 60, 80, 100, 150, 200, and 250 nm; for ~$80 for a 25mL bottle from Sigma Aldrich in sizes up to 400 nm) for years in the lab. This post highlights some of the previous work that has been posted to the blog and brings together some of the information in the literature and on supplier’s websites. First, some properties of gold nanoparticles. Concentration data is available from BBI’s website, and mean diameter, coefficient of variation, optical density at 520 nm nanometers (but only for sizes up to 80 nm), and gold chloride concentration are all also available for each size/batch as separate pdfs. Gold absorbs light because of surface plasmon resonance, and as particle size increases, the associated resonant frequency of light shifts to longer wavelengths of light. This means the absorbance peak for 5 nm gold is between 515-520 nm, but the peak for 40 nm gold is 530 nm, and 100 nm gold has a peak at 572 nm.

Gold nanoparticles have an inherent negative charge (which Jess found to be -58mV for 5 nm gold, and -30mV for 10 nm gold), which functions to keep them from aggregating. Because of this, Jess found KCl concentrations of 100mM will cause the gold nanoparticles to aggregate, but concentrations of 10mM KCl are fine for separations. (Note that gold in 10mM salt is not stable over long periods of time) A small concentration of salt is much better than none, because it lets us more accurately guess the Debye length (for our membrane, Jess calculated that 10mM KCl gives us a Debye length of 3 nm). Jess also did a substantial amount of work on conjugating BSA to the gold nanoparticles (her recipe) – this will keep the particles from aggregating in even stronger salts(up to 100mM KCl – look at comments), but it increases the size of the particles, is more difficult to work with, and apparently doesn’t work for separations anyways. It seems like that technique was abandoned.

When it comes to sizing the particles, we use either the Malvern or the TEM for sizing information (and sometimes these agree more than the manufacturer’s data sheet). But interestingly, we can also use the Tecan for size information – this paper (gold nanoparticle paper) gives us the formula (where is the wavelength of maximum absorbance, in nm), which is quite accurate in the range of d = 35-110 nm, and the supplemental information contains experimental data giving both size and concentration of spherical gold nanoparticles based on absorbance at set wavelengths and the position of the absorbance peak.

When we measure concentration we always use absorbance measurements – the high optical density of gold is what makes it attractive for quantifying sieving behavior in the first place. When I do separations, I begin by constructing a standard curve based on a series of dilutions of my initial concentration of gold species. I fit this data with a linear trendline in excel, and then when I collect filtrates, I back-calculate the percent of original concentration. I believe this is the same technique used in the ACS nano paper to get sieving coefficients (sieving coefficient = concentration of filtrate/concentration of feed).

Ideally we would like to have set of standard standard curves – one for each size of gold nanoparticle we have. It would also be nice to know what minimum concentration the nanoquant plate can reliably detect (takes 3 uL, has a path length of ~1 mm), and what concentration the cuvettes can detect (takes 50uL, has a path length of ~1 cm). One of the problems we’ve had in the past is that it is difficult to get consistent baseline measurements on the Tecan. The individual windows on the nanoquant plate and the two standard cuvettes we use can give water readings that are substantially different. Although acid-base (0.1M KCl and 0.1M NaOH) washes work reasonably well for the cuvettes, and ten minutes in the bath sonicator work well for the plate, blanking will still always be required, and a standard curve will need to be measured each time a sieving coefficient is desired. If we are working with larger volumes, disposable cuvette that work with volumes >70 uL are available from sigma aldrich (none of the disposable cuvettes we have downstairs work with volumes less than 500 uL). I have not been able to find a disposable cuvette that can give a measurement with only ~3 uL of sample like the nanoquant plate can.