Supplement to:
Bollen, M. R., Blaser, P., Bruggmann, S., Wu, S., & Jaccard, S. L. (2026). Efficient microwave-assisted digestion of geological materials with HBF4: a reduced-hazard alternative to HF. Advances in Geochemistry and Cosmochemistry, 2(2), 1100. https://doi.org/10.33063/agc.v2i2.1100
Measured trace element concentrations of the standard reference materials following microwave-assisted acid digestion using HBF4.
Sample Preparation and Digestion:
The digestion protocol was tested on six certified reference materials; BHVO-2 (Hawaiian basalt, USGS), NOD-A-1 (Atlantic ferromanganese nodule, USGS), ShBOQ-1 (Boquillas Shale, USGS), SCo-2 (Cody Shale, USGS), SSAR-1 (Animas River sediment, USGS), and MESS-4 (Beaufort Sea marine sediment, NRC Canada). First, 40 to 60 mg of each reference material was weighed into 8 mL PFA digestion vials, with six replicates for each material and three procedural blanks (39 samples in total). Samples were acidified with 6 mL of HNO3 (65 % w/w, trace metal analysis grade, VWR) and left to react for two hours to dissolve easily soluble phases such as carbonates. Subsequently, 1 mL HBF4 (38 % w/w ultra-pure grade, AnalytiChem Belgium) was added to each vial before loading them into a Milestone ultraCLAVE automated microwave digestion system with a Single Reaction Chamber. A baseload solution in the reactor consisted of 300 mL 18.2 M resistivity water, 10 mL H2O2 (30 % w/w, trace analysis grade, VWR) and 5 mL H2SO4 (95 % w/w, trace metal analysis grade, VWR), which served to absorb excess microwave energy and ensure uniform heat distribution among reaction vials. The chamber was initially pressurised with N2 gas to 4 MPa to prevent sample boiling during heating. Variable microwave power was applied up to a maximum of 1000 W to sustain a temperature ramp of 4 °C/min for 45 minutes until 180 °C, corresponding to a reaction pressure of ~10 MPa, which was maintained for 105 minutes. Total operational duration, including heating, cooling and depressurisation was approximately three hours, with capacity for the digestion of up to 40 samples in each run.
Following digestion, vials were removed from the microwave digestion system and left cooling in a fume hood for one hour prior to further processing. Immediately after opening each vial, an aliquot of 70 μL was pipetted from close to the surface of the digested solution and diluted into 6.93 mL 0.5 mol/L HNO3 for elemental analysis; these aliquots are hereafter referred to as the T1 aliquots. The remaining ~6.93 mL of sample in each digestion vial were transferred into 14 mL PFA vials, and the original digestion vials were rinsed twice with 18.2 M resistivity water into their corresponding PFA vial. The combined solution was evaporated to dryness overnight at 120 °C on a hotplate within a fume hood, then redissolved in 10 mL 0.5 mol/L HNO3, hereafter referred to as the T2 aliquots. For elemental analyses, both the T1 aliquots and T2 aliquots were further diluted in 0.5 mol/L HNO3 to a total dilution factor of 10 000 relative to the initial average sample mass. A flowchart overview of the total procedure is appended as Supplementary Figure S5.
Elemental Concentration Measurements:
Elemental concentrations were determined on an Agilent 8900 Triple Quadrupole ICP-MS/MS at the Faculty of Geosciences and the Environment, University of Lausanne. Measurements were conducted under various gas modes (He, H2, O2 and no gas) using both single and triple quadrupole configurations to minimise polyatomic and isobaric interferences for each element. An internal standard solution containing Rh and Bi was introduced via in-line addition. Calibration standards were prepared gravimetrically from multi-element reference solution (CMS-1, CMS-3, CMS-4, CMS-5; Inorganic Ventures). Details of gas modes, isotope selection, internal standard correction, and mass-shifts parameters are provided in Supplementary Table S1.
Data reduction was performed in Agilent MassHunter, applying blank subtraction and internal standard correction. Limits of detection (DL) were defined as three times the standard deviation of blank replicates, corresponding to a <1 % probability that the signal arises from instrument noise. Limits of quantification (LOQ) were set to three times the DL. Concentrations below the LOQ were excluded for further analyses. To monitor for analytical accuracy, reproducibility, two secondary reference solutions (NOD-A-1, USGS - digested in aqua regia; DRIFT, in-house multi-element solution gravimetrically diluted to 1 ppb) were each measured after every 10 samples (Supplementary Figure S1).
Element recovery was evaluated using certified and information values for each reference material (Figure 1). Where such values were incomplete, they were supplemented by community-based reference data compiled in the GeoReM database (Jochum et al., 2005). For certain elements, MESS-3 and SCO-1 values were used to supplement MESS-4 and SCO-2, respectively, ensuring a consistent comparison across the widest possible array of elements and lithologies, though there may be slight compositional differences between these editions. The complete dataset of certified/accepted/community values for each reference material are presented in Supplementary Table S2. REE profiles were normalised to Post-Archean Australian Shale (PAAS; Pourmand et al., 2012). To account for variable element recovery, each REE concentration was further normalised to the mean REE concentration of the corresponding sample. This approach preserves the shape of PAAS-normalised REE pattern while compensating for overall recovery offsets, thereby allowing the detection of any systematic fractionation trends across the lanthanide group.
Analytical uncertainty and blank assessment:
Instrumental reproducibility was assessed using two reference solutions; NOD-A-1 digested in Aqua Regia, and DRIFT, and in-house mixture of elemental concentration standards at 1 ppb (uncertainty expressed as two relative standard deviations, 2RSD; Supplementary Figure S1). The 2RSD is <10 % for S, V, Mn, Fe, Co, Ni, Cu, Zn, Sr, Y, Zr, Nb, Ba, La, Ce, Tb, Tl, Pb, and U and <20 % for all other measured elements (Na, Mg, Al, Ti, Cr, As, Sb, Pr, Nd, Sm, Eu, Gd, Dy, Ho, Er, Tm, Yb, Lu, W) except for Hf (~25 %) and Th (~45 %). The potential for contamination during sample processing was evaluated by normalising the average signal intensity of procedural blanks to the mean signal of reference material samples (Supplementary Figure S2). For most elements, the procedural blank contribution was <1 % of the total signal. Only Si and Ta exceeded 10 %, though Si concentrations were below the LOQ in both blanks and T2 aliquots. Considering the negligible impact of blanks relative to the ± 10 % ICP-MS/MS analytical uncertainty, no procedural blank correction was applied to the reported concentrations. Seven outlying measurements were removed, with poor data quality attributed to erroneous sample dilution and handling. There are thus four and three replicates of the T1 and T2 aliquots from MESS-4, respectively, and four replicates of the SSAR-1 T2 aliquots. All other materials retain six replicates for each of the processing stages.