Transport and coarsening of gold nanoparticles in an orogenic deposit by dissolution–reprecipitation and Ostwald ripening

The role of nanoparticle gold in the formation of gold deposits has long been debated. Evidence in support of this process is generally limited to epithermal-, Carlin- and seafloor massive-sulfide-type deposits where gold nanoparticles are associated with primary fluid related processes. At the Kenty orogenic deposit in Ontario, Canada, gold has been remobilized from early pyrite through secondary coupled dissolution-reprecipitation processes to form high-grade gold external to pyrite. Here we report gold nanoparticles and related textures that help document this gold coarsening process. A combination of focused ion beam technology and transmission electron microscopy provides a rare glimpse of gold coarsening frozen in time, which includes nanoparticles trapped in iron oxides and rutile and non-oriented attachment of gold nanoparticles to bulk gold suggesting coarsening via Ostwald ripening. The processes documented are applicable to orogenic deposits formed through Earth’s history, and may explain the formation of ultra high-grade ore zones. Gold nanoparticles identified in the Kenty orogenic gold deposit were liberated by dissolution processes, transported along fluid pathways and coarsened through a form of Ostwald ripening, according to nanoscale observations and chemical analyses.

Geological setting, the Kenty Au deposit, and Au remobilization. The Abitibi greenstone belt (AGB) is one of the most Au endowed Archean terranes in the world and accounts for a significant portion of the total Au endowment in the Superior Province of Canada 16 . The Swayze area, which is the southwestern extension of the AGB (Fig. 1), contains volcanic and plutonic rocks with a wide range of compositions (ultramafic to felsic) as well as metasedimentary rocks (both chemical and clastic). These rocks were deposited and emplaced between ca. 2750 Ma and 2670 Ma 17,18 . Metamorphosed rocks in the western AGB attained temperatures of 350-450°C and maximum pressures of~3 kbar (10 km depth) during the final construction of the Superior Craton at ca. 2670 Ma to 2640 Ma 19 . Although metamorphosed, the prefix "meta" is herein omitted from the rock names for simplicity.
The Kenty orogenic Au deposit ( Fig. 1) is located in the central part of the Swayze area. It was discovered in the 1930s and is characterized by high-grade ore zones with coarse gold. Kenty is an excellent natural laboratory to study Au remobilization processes because it has a simple mineralization history and has escaped the multitude of overprinting events that complicate the interpretation of ore-forming processes in many orogenic deposits 4 . The Au mineralization is hosted in altered pillow basalts and granitoid rocks 4 , the latter dated at 2684.7 ± 1.5 Ma (U-Pb zircon). Two generations of Au mineralization are present: (1) early Au in pyrite localized to ankerite alteration zones marginal (i.e., <1-2 m) to quartz veins; and (2) later coarse gold lining fractures that cut the quartz veins 4 . The late coarse gold commonly lines conjugate fractures that formed in the hinge of the folded quartz veins with microfolded gold-rutile intergrowths in the wallrocks of the veins 4 .
Our previous research 4 has interpreted the late coarse gold at Kenty to be a product of remobilization from earlier pyrite based on: (1) textures indicative of dissolution-reprecipitation processes 20 that increase towards the quartz veins hosting coarse gold. These textures reflect the destruction of primary chemical zonation, porosity development, and the formation of secondary inclusions (e.g., native gold, sulfides, and tellurides) consistent with trace elements (e.g., Au, Ag, Te, Cu) enriched in the primary zones; (2) SEM-EDS (energy dispersive spectroscopy) imaging and laser ablation inductively coupled mass spectrometry (LA-ICP-MS) elemental mapping of pyrite that shows progressive loss of Au and associated elements commensurate with the degree of pyrite modification and its proximity to quartz veins; (3) elements that are spatially associated with Au in pyrite (e.g., Ag, Cu, Te) are also associated with, and incorporated into, late coarse gold phases exterior to pyrite (e.g., Cu and Ag in native gold, petzite, hessite, calaverite); and (4) the Au content of primary pyrite (i.e., 100 s to 1000 s ppm) is sufficient to account for abundance of coarse gold hosted in late fractures 4 . A summary of this dissolution-reprecipitation model (Fig. 2) highlights the progressive loss of Au from pyrite with its increasing textural and chemical modification.
This research further demonstrated that the liberated Au from pyrite was likely remobilized at temperatures as low as 335°C via fluid-mediated polymetallic melts enriched in low melting point chalcophile elements (LMCE), such as Te 4,21-26 . Indeed, a number of recent studies provide substantive evidence to support Au concentration by LMCE-rich melts in the presence of a fluid 24,25 . Seemingly enigmatic are areas at the Kenty deposit, as with many Au deposits worldwide, where zones hosting coarse gold lack evidence for LMCE-rich phases. We previously suggested that LMCE-rich melts may operate in conjunction with nanoparticle transport to account for Au upgrading at the Kenty deposit 4 , but we only provided evidence for the former mechanism. Thus, convincing evidence for the presence of gold nanoparticles in this Archean orogenic Au setting remains to be demonstrated.

Results and discussion
Gold nanoparticles, associated mineral phases, and textures. Here, we report on the occurrence of gold nanoparticles within samples from the high-grade ore assemblage at the Kenty Au deposit (Fig. 1). To demonstrate this, we chose a mineralized sample ( Fig. 3a-d) with evidence of Au remobilization 4 . The sample contains early-stage auriferous pyrite (≈800 ppm Au) in spatial association with late-stage coarse gold. Sample sections were extracted from a thin section by focused ion beam-scanning electron microscopy and thinned to foils for subsequent transmission electron microscopy (TEM) studies. The foils were extracted from the interior of the above-described pyrite grain and across its interface with gold-hematite-albite-rutile grains (Fig. 3d).
TEM examination of the FIB section from the interior of the pyrite (Supplementary Fig. 1) indicates the presence of nano-size Au-Ag-Te-Pb domains with droplet-like morphologies (Fig. 3e, f). The spatial overlap of Au-Ag-Te in the droplet suggests a telluride phase (e.g., petzite); as Pb is concentrated toward the upper part of the droplet it may be part of another phase (altaite?). In addition to the droplets, bright sub-nanometer-size domains in the pyrite (Fig. 3e, f) and LA-ICP-MS elemental maps of pyrite grains ( Fig. 2; distal and medial) also suggest that finely dispersed elemental Au, Ag, and Te occur throughout the pyrite.
The FIB section extracted along the pyrite margin ( Fig. 3d) contains gold surrounded by grains of Fe oxide, pyrite, albite, and rutile ( Fig. 4). On the lower left side of the FIB section, a large grain of Fe oxide is in contact with pyrite ( Fig. 4a, b). This Fe oxide grain also connects to Fe-oxides on the margins of gold, albite and rutile and is also in the interior of the latter two minerals (Fig. 4b). The Fe-oxides are present prior to extraction (Fig. 3d) and thinning (Fig. 4b, c) and remain throughout the thinning process even when the gap in the area between gold and albite widens (Fig. 4d).
The larger Fe oxide grain ( Fig. 4a, b) is composed of intergrown hematite, goethite, and magnetite/maghemite ( Fig. 5a, b) and it contains gold nanoparticles ranging in size from 1 to 5 nm ( Fig. 5c-e). A portion of the Fe oxide on the margin of gold that was previously attached to the larger Fe oxide grain (Fig. 4) is composed of hematite ( Fig. 5f, g). This hematite also contains gold nanoparticles that range in size from 1 to 5 nm ( Fig. 5h-j).
Observing the Fe oxide between the gold and albite (Fig. 4), a thin layer (5-10 nm thick) of the Fe oxide remained after FIB thinning on the gold surface. This thin layer contains gold nanoparticles whose lattice orientation is different from that of the coarse gold ( Fig. 6a-c).
The Fe oxide filling that occurs between the gold and rutile grain boundary (Fig. 4a, b) also contains gold nanoparticles. At the far end of this grain boundary, gold nanoparticles up to 10 nm in size occur within rutile in close proximity to the gold surface ( Fig. 6d-f).

Discussion on gold nanoparticles and their chemical behavior.
Documenting the nature and origin of gold nanoparticles in an Archean orogenic deposit (i.e., Kenty) is an important step towards understanding how high-grade ore zones form in such Au deposits in general given that most previous research on this topic is restricted to epithermal, Carlin, and seafloor massivesulfide deposits [8][9][10]12,13 . Although mechanisms have been suggested for nanoparticle aggregation previously 8,10 , evidence for nanoparticle processes related to secondary remobilized gold has been lacking until now. The following discussion aims to resolve some of these aspects.
Caution must be exercised when preparing samples by FIB milling since redeposition of material is possible 27 and some TEM observations may be equivocal (e.g., Supplementary Fig. 2). Although this is a possibility, the gold nanoparticles presented in Figs. 3, 5, 6 are interpreted to be natural based on the following evidence: (1) separate and distinct gold nanoparticles are found in pyrite (Fig. 3e, f), Fe-oxides (Fig. 5) and rutile (Fig. 6d-f) with threedimensional features that confirm that the gold nanoparticles are not surface redeposition; (2) pyrite, Fe oxide, and rutile also have an association with gold at the micro-scale (Fig. 3c, d) 4 ; (3) the Fe oxide was present at the margin of gold and adjacent to pyrite (Fig. 4a-c) from the very beginning of the FIB sample thinning; (4) the gap in the FIB section widened during the FIB milling process, hence material was being removed and not added during this process (Fig. 4); (5) the Fe oxide between gold and albite narrowed during the FIB milling process (Fig. 4b-d), reinforcing the first two points; and (6) the Fe-oxides adjacent to pyrite and on the margin of gold include goethite, hematite, and magnetite/maghemite, as confirmed by selected area electron diffraction (SAED; Fig. 5a, b). A combination of Fe 2+ , Fe 3+ , and hydrous Fe-oxides within the same assemblage would simply not form at the same time as redeposition during FIB milling, which precludes this being an artifact of the process.
Model on the formation, transport, and coarsening of gold nanoparticles. An oxidizing fluid was responsible for the dissolutionreprecipitation of pyrite at the Kenty deposit (Figs. 2, 3c, d) 4 . Once Au supersaturates in a fluid 6,8-10 , gold nanoparticles would remain suspended in solution as long as their surface charges create larger net repulsive forces than net attractive forces. This phenomenon is best explained by Derjaguin-Landau-Verwey-Overbeak theory 28 , which is commonly used to predict nanoparticle behavior, although quantum effects for particles <20 nm have yet to be resolved 29 .
The surface charge of any nanoparticle is mainly affected by pH and counter-ion activity, although particle size and compositional variations (e.g., fluctuating Ag content of gold) can also affect the surface charge 28,29 . Gold nanoparticles generally have a negative charge in near neutral to basic solutions and a positive one in highly acidic solutions 10,30-32 with gold aggregation initiated where the pH favors a point of zero charge (PZC). The PZC for gold at 25°C occurs at a pH of ≈3; however, no data currently constrains the PZC for gold at conditions relevant to ore deposit formation.
Two alternative models can explain the charge-dependent transport and coarsening of gold nanoparticles at the Kenty deposit based on the evidence presented above and the aforementioned summary of current research on nanoparticle behavior. Dissolution of the early auriferous pyrite by an oxidizing fluid liberated Au (Fig. 2) and lowered the pH of the transporting fluid such that: (1) A fluid with a pH below the PZC of gold would promote the transport of positively charged gold nanoparticles. The dissolution of another phase (e.g., plagioclase) would be required to raise the pH of the fluid close to the PZC of gold to promote aggregation. This represents an intriguing mechanism, as both transport and aggregation of the gold nanoparticles are promoted by the dissolution of minerals. Albitization is a common process in Au deposits 1,3,33 and Kenty is no exception 4 , as indicated by likely dissolution of original plagioclase (~An 30 ) based on the presence of endmember albite with associated alteration phases now seen (i.e., pyrite-muscovite(sericite)-carbonate) 4 . Recent studies on epithermal deposits have provided evidence for diffusion-limited gold aggregation resulting in "fractal" dendrites 8 . These textures form through oriented attachment of gold nanoparticles, but data from the Kenty deposit only shows non-oriented attachment (Fig. 6a). Thus, gold coarsening most likely occurred through a different mechanism.
Studies on nanoparticle attachment 34,35 have provided evidence for crystallization of minerals via non-oriented attachment of nanoparticles. These studies showed that when non-oriented attachment occurs near flat or negatively curved surfaces, nanoparticles dissolve and a net transfer of monomers to the larger crystal occurs (i.e., Ostwald ripening) 34,35 . Thus, the nonoriented attachment observed at Kenty (Fig. 6a) supports an interpretation whereby gold coarsened through the dissolution of gold nanoparticles and the mass transfer of Au monomers to the coarse gold lattice. The implication of this conclusion is that nonoriented attachment and Ostwald ripening may play an important role in the coarsening of gold in orogenic deposits. This also offers an alternative for gold coarsening by oriented attachment reported in epithermal environments 8 . There is only a limited understanding of gold nanoparticle behavior in natural systems owing to the changing properties of minerals at the nanoscale and different nanoparticle-hostinteractions 36 . For example, the melting temperature of gold decreases dramatically from 1064°C to <400°C for isolated nanoparticles <3 nm 13,37-39 , but becomes unpredictable for nanoparticles <2 nm where quantum effects need to be considered. In addition, the thermal stability of gold nanoparticles is also governed by their host mineral 13,36 . In isolation, gold nanoparticles are stable until they reach the size-dependent melting curve [37][38][39] , but in arsenian pyrite it has been shown that gold nanoparticles <4 nm in diameter would be unstable at temperatures relevant to orogenic deposits (i.e., 250-400°C), and would coarsen through solid-state Ostwald ripening to become stable 13 . Assuming As does not greatly affect the stability of gold nanoparticles 13 , then the minimum temperatures obtained at the Kenty deposit (335°C) 4 would indicate that gold nanoparticles within pyrite would need to be a minimum of 4 nm to be stable 13 . However, gold nanoparticles between 1 and 5 nm in diameter occur in Fe-oxides indicating that this type of matrix affects the thermal stability of gold nanoparticles differently than pyrite. Thus, our observations (Figs. 5, 6) offer rare evidence of nanoparticles trapped prior to their coarsening into larger gold domains.
Considering that LMCE and Au can form a polymetallic melt in the presence of a fluid 24,25 and that we have evidence at Kenty for polymetallic melt formation from the previous research 4 , as well as droplet-like polymetallic nano-domains within primary pyrite from this work (Fig. 3e, f), we suggest that fluid-mediated polymetallic melts and gold nanoparticle transport are both relevant to upgrading orogenic Au deposits. Where LMCE is absent from the system, then nanoparticle transport would become the dominant mechanism.
All aspects influencing the behavior of gold nanoparticles, their transport, and how they aggregate in different environments must be examined in detail if we want to better understand and resolve the processes that govern the formation of high-grade gold zones. However, many unanswered questions remain regarding nanoparticle processes in Au deposits, but discovering direct evidence for gold nanoparticles associated with Au remobilization in an Archean orogenic deposit serves to highlight that this mechanism is not restricted to a specific geologic time period or deposit setting. Future research should aim to: (1) understand the role of Ag in the behavior of gold nanoparticles because almost all natural gold contains significant Ag and many epithermal deposits are host to electrum (i.e., <80 wt.% Au, >20 wt.% Ag); (2) examine the nature of gold within precursor pyrite to understand if there is a conversion of elemental Au to nanoparticles or simply nanoparticle liberation during dissolution; (3) clarify the relationship among gold nanoparticles and other elements, nanoparticles and minerals; and (4) determine the roles temperature and particle size play in the formation and aggregation of gold nanoparticles, specifically for conditions relevant to ore deposit formation.
Implications for Au deposits and relevant disciplines. Highgrade, "bonanza-type" ore zones can form in ancient to recent Au deposit types owing to the transport and aggregation of gold nanoparticles 5,6,[8][9][10] . This mechanism can operate alone or in conjunction with fluid-mediated polymetallic melts to upgrade Au deposits. Dissolution of minerals, Au remobilization, counterion activity, and pH fluctuation are likely controls for gold nanoparticle formation, transport, and aggregation. The underlying mechanism for gold coarsening may, however, differ between orogenic (attachment, dissolution, and monomer transfer) and epithermal (diffusion-limited aggregation and orthokinetic aggregation) 8 deposits. The existence of these different gold coarsening processes has far-reaching implications for other disciplines given that gold nanoparticle research is a critical component of material science 40,41 , nanomedicine 42,43 , and nanotechnology 44,45 . A more complete understanding of natural gold nanoparticles and their interaction with other natural materials may enhance laboratory manipulation for a variety of disciplines.

Methods
Geological sampling. Rock sampling and bedrock mapping was carried out by the lead author as an employee of the Ontario Geological Survey. The Universal Transverse Mercator (UTM) co-ordinates for the location of the sample used in this study are 378743 m E, 5299846 m N using NAD83 in zone 17. The sample location can also be seen with reference to the detailed outcrop map (Fig. 1b) and is shown by a yellow star.
SEM-EDS. This work was performed with a Zeiss EVO 50 SEM at the Geoscience Laboratories in Sudbury, ON, Canada. The SEM is equipped with an Oxford X-Max 50 mm 2 EDS, and used Oxford's Aztec © software package. Operating conditions were 20 kV and ≈750 pA with variable counting times for each spot analysis (20-40 s) and used a working distance of 8.5 mm.
TEM of FIB sections. Extraction of FIB sections was conducted with an FEI Helios 650 NanoLab FIB at the 4D LABS at Simon Fraser University, Vancouver, BC, Canada. FIB sections were lifted with a platinum gas-glue and thinned with ion gas milling (Ga ions) to electron transparency. The FIB sections were analyzed with a field-emission TEM FEI Talos F200x at the Manitoba Institute for Materials, at the University of Manitoba, Winnipeg, Manitoba, Canada. Imaging in bright and darkfield mode was performed with an accelerating voltage of 200 kV and with a 16 MB ceta camera and a Fischicone high angle annular dark-field detector. Compositional analysis was performed in STEM-EDS mode with four SDD detectors. Micro-scale phases were identified in the previous research 4 and confirmed here at the nanoscale using a combination of Selected Area Electron Diffraction (SAED) and fast Fourier transform (FFT) patterns, chemical distribution maps, semi- quantitative chemical analyses, and textural features of phases previously described in the literature. TEM and STEM-EDS chemical distribution maps were processed with the TEM Imaging & Analysis (a trademark of FEI) and Esprit 1 (a trademark of Bruker Nano).
Quality control during TEM-STEM examinations. TEM and STEM imaging can cause significant damage to the sample owing to beam-sample interactions. The two main damage types that samples experience are displacement effects and breaking of chemical bonds 46 . Imaging in TEM and STEM mode can generate similar integrated doses and dose rates but in TEM mode, a defined area for the entire image is illuminated, whereas a smaller area of the sample is illuminated with a similar peak dose/rate in STEM mode than the larger area in TEM mode (during the same acquisition time) 46 . Thus, an individual nanoparticle in, for example, a FIB section is exposed to a higher peak dose/rate in STEM than TEM mode.
In this study, the gold nanoparticles were exposed to the highest peak dose/rate during the recording of chemical distribution maps with STEM-EDS (maximum 30 min). For this reason, STEM images were recorded prior to and during the recording of STEM-EDS chemical distribution maps. In addition, TEM images were taken from the same area after the recording of the STEM-EDS chemical distribution maps. STEM and TEM images taken before, during, or after the recording of the chemical maps were then compared with identify potential damage to the samples. Besides the drift of the sample, especially detectable at high magnification, no changes in the number, morphology, and texture of any mineralogical feature were observed during the recording of the STEM-EDS chemical distribution maps.

Data availability
No data sets were generated or analyzed during this study.