Evaluating internal standards for the determination of gas phase mercury using silver nanoparticle assisted total reflection X-ray fluorescence

Abstract

Mercury was often used to protect cultural heritage specimens from damage by various pests. Over time, mercury is released into the air and access to archives may therefore need to be controlled due to the hazards associated with high gas phase concentrations of Hg. A procedure to analyze gas phase Hg using total reflection X-ray fluorescence (TXRF) was developed. TXRF is a small footprint instrument available in many laboratories. Silver nanoparticles (AgNPs) were prepared and used to sample Hg from air directly on the quartz carriers. The relative standard deviation (RSD) for the AgNP amount deposited on the carriers was on average about 10%. The efficiency and reproducibility of the Hg-capture on washed and non-washed AgNP-specimens was studied. Washed carriers had about 60% less Ag than non-washed specimens. Interestingly, the amount of Hg captured on washed carriers was significantly higher than on non-washed carriers. Calibration in TXRF is usually achieved using an internal standard (IS). However, acidic IS solutions like gallium (3% HNO3) led to the formation of large Ag crystals and a rather poor reproducibility was observed for Ag determination using Ga as IS i.e. 10%. Alternative standard solutions having basic to neutral pH (i.e. chromate and molybdenum) were therefore tested. The TXRF results showed good reproducibility using Mo (L-lines) as IS, yielding a RSD of 2.6% mixing the standard with the AgNP solution before carrier preparation. A lower reproducibility was obtained using Cr as IS with a RSD of 4.0%. Micro-X-ray fluorescence studies on the spatial distribution of Mo and Ag in AgNP deposits with Mo IS showed a relatively even distribution of both elements over the entire area of the residue, although neither element correlated pixel by pixel. A web-like structure of Mo was overlaid by a space-filling Ag distribution, resulting in poor correlation (r = 0.59). The spatial distribution of Ag and Cr in AgNP deposits with Cr IS showed a significant inhomogeneity on a larger length scale. Half of the deposit was usually richer in Cr than the other half, however, the correlation (r = 0.77) was higher than for AgNPs with Mo IS. The addition of Mo to an already dried and washed AgNP deposits resulted in a more inhomogeneous sample (r = 0.27).