Gold at pyrite margins
A critical review of growth and reequilibration evidence with implications for
Pac Lang Vietnam
Abstract
Gold associated with the outer domains of pyrite is interpreted in terms of mineral growth or subsequent
remobilization. Yet spatial association, concentration contrast, relative timing, atomic occurrence and
metal provenance are different propositions. This focused critical review compares fifteen selected studies
of pyrite chemistry, growth, replacement and deformation, together with published evidence from
Vietnam. The comparison identifies three recurring interpretive problems: equating an outer domain with
an overgrowth, inferring atomic gold occurrence from bulk or microbeam chemistry, and treating late
gold as evidence of local recycling. Recent crystallographic studies further complicate a simple growthversus-replacement distinction by introducing dislocation-mediated interfacial re-equilibration. We
propose a claim-based framework that links each interpretation to spatially registered observations,
explicit alternatives and a stopping rule when discrimination is unresolved. Analytical mixing and the
distinction between concentration and metal inventory are incorporated into the framework. Pac Lang
illustrates how the approach can define a tractable research question without presupposing gold-rich rims
or a deposit classification. The review provides an interpretive framework and a prospective workflow,
not new analytical results or a validated classifier. Its central conclusion is that pyrite margins become
informative records of mineralization only when geometry, chemistry and chronology are evaluated
separately before being integrated.
Keywords: pyrite; gold; core–rim texture; arsenic; dissolution–reprecipitation; crystallography; Pac Lang

1 Introduction
An outer domain in a pyrite grain may be conspicuous in an image, yet its geological meaning is not selfevident. Calling the domain a rim describes position. Calling it an overgrowth assigns a process and a
relative age. Calling it an auriferous overgrowth adds a compositional assertion, while calling it a product
of gold remobilization makes a claim about material transfer. These interpretations are often discussed
together, although they require different evidence. The distinction matters whenever mineral-scale
observations are used to reconstruct ore formation or to select targets for further analysis.
Studies of Au in arsenian pyrite established important relationships between composition and gold
occurrence [1,2]. Laser imaging subsequently linked elemental zoning to paragenetic interpretation [3].
More recent work has resolved nanoscale heterogeneity, chemically bound gold, defect-associated
enrichment and multiple mechanisms of mineral re-equilibration [4–9]. These developments expand what
can be learned from pyrite but also increase the number of explanations compatible with a seemingly
simple core–rim texture. A photograph of a peripheral gold particle is therefore not a substitute for a
process test.
This review addresses a restricted question: what observations justify interpretations of gold associated
with pyrite margins? It does not attempt a comprehensive account of gold transport, global deposit
classification or analytical instrumentation. Instead, it compares selected mechanistic studies with
counterexamples, separates the claims commonly made from those observations, and develops a practical

1

framework for evaluating them. The proposed contribution is interpretive: a transparent connection
between a claim, the measurement required to support it and the alternatives that remain possible.
Pac Lang, northeastern Vietnam, provides the application context. Published work describes
mineralization and overprinting relationships [10], while another study discusses nearby intrusive rocks
and their possible relationship to the ore system [11]. These publications motivate investigation of pyrite–
gold relationships, but they do not supply a new core–rim analytical dataset for this review. Accordingly,
the application is prospective. No gold-rich rim, remobilization event or genetic classification is claimed
as a new finding at Pac Lang.

2 Review design and terminology
2.1 Selection and comparison of evidence
Fifteen publications were selected purposively to cover complementary evidence: Au–As relations, direct
characterization of gold occurrence, elemental zoning, growth, replacement, dislocation-related processes
and Vietnamese geological context. Searches conducted during preparation on 30 October 2026 combined
“pyrite overgrowth”, “core–rim pyrite”, “gold zoning”, “invisible gold”, “dissolution–reprecipitation” and
“dislocation-mediated re-equilibration”, together with Me Xi and Pac Lang. Publisher pages and
institutional repositories were used to verify the selected records. This was a narrative critical review, not
a systematic search of an indexed bibliographic database; no completeness or publication-frequency claim
is made.
Each selected study was compared in terms of its geological object, reported analytical approach,
principal interpretation and relevance to the present question. The comparison retained the distinction
between an author's interpretation and the transferable lesson drawn here. Full text was accessible for
some records, whereas other comparisons were limited to publisher or institutional abstracts and
accessible extracts. Appendix A records that access limitation. Supplementary datasets were not
uniformly reprocessed, and the review therefore does not independently validate reported concentrations,
analytical calibration or thermodynamic calculations.
No pooled effect size, universal concentration threshold or percentage of deposits exhibiting a particular
mechanism was calculated. Multiple papers may also concern related geological settings; they are not
treated as statistically independent samples of global ore formation. The framework developed below is a
synthesis to guide testing, not a model trained or validated against a representative corpus.

2.2 Operational definitions
An “outer domain” is a mapped region toward a grain margin, without an assigned origin. “Rim” is used
descriptively unless growth has been supported independently. An “overgrowth” denotes material
interpreted to have grown on a pre-existing substrate. “Re-equilibration” is used as a broader description
of compositional or structural modification when its specific mechanism remains unresolved. “Late gold”
means gold-bearing material demonstrably younger than a particular feature; it does not identify the gold
source.
“Gold-bearing” indicates detection under stated analytical conditions. “Gold-enriched” requires an
explicit comparator and a justified concentration comparison. “Invisible gold” is retained as a scaledependent descriptive term rather than a single atomic species. In this review, spatial association,
chemical enrichment, relative timing, atomic occurrence and provenance are five separate claims. They
can be combined only when the supporting observations are compatible in scale and location.
2

3 What the selected studies establish
3.1 Au–As relations provide context rather than a complete process diagnosis
Reich et al. [1] combined microanalytical and high-resolution transmission-electron observations to
examine Au solubility and occurrence in arsenian pyrite. Deditius et al. [2] subsequently compared new
and published Au–As data across a broad range of hydrothermal settings. These studies provide a
compositional foundation for evaluating auriferous pyrite. Their use here is not to assign an atomic gold
species to every new measurement, but to establish why composition must be considered alongside
mineral structure and overprinting.
Pokrovski et al. [5] used high-resolution X-ray absorption spectroscopy on natural and synthetic materials
to investigate Au coordination and redox state in pyrite and iron sulfarsenides. Their structural
interpretation distinguishes As-rich and As-poor environments. This illustrates the difference between
measuring how much Au is present and determining its chemical environment. A compositional
association may motivate a structural hypothesis, while spectroscopic evidence addresses a different
aspect of that hypothesis.
The practical consequence proposed here is to keep compositional plots and direct species identification
in separate evidentiary categories. A point on an Au–As diagram can be reported without assigning an
oxidation state or asserting the absence of nanoparticles. Such assignments should be reserved for a
measurement and interpretive model that can address them.

3.2 Zoning may encode more than changing fluid composition
Large et al. [3] mapped trace elements in pyrite from four sediment-hosted gold deposits and
distinguished early gold enrichment from later hydrothermal occurrences, including fracture-hosted gold
and narrow Au–As rims. The study demonstrates the importance of registering elemental distribution with
mineral texture rather than treating a grain as chemically uniform.
Wu et al. [4] investigated pyrite from Daqiao using NanoSIMS and atom probe tomography. Their
interpretation involves fluid composition, crystal-surface structure and growth kinetics acting at different
scales. This is a counterexample to the assumption that every chemical band must record a separate
externally supplied fluid pulse. It also makes the scale of a proposed explanation part of the question
being tested.
For a new study, these results motivate mapping both the geometry and continuity of domains before
assigning generations. A sector-dependent pattern, a concentric pattern and a crosscutting feature should
not be merged into one category merely because each produces an Au contrast. The present framework
therefore begins with a description that remains independent of the preferred genetic explanation.

3.3 Epitaxial growth and dissolution–reprecipitation are distinct hypotheses
Xiao et al. [12] combined textural, chemical and crystallographic observations on core–rim pyrite from
Shuiyindong and Lannigou. Similar core and rim orientations, considered with the other observations,
supported epitaxial growth of ore-stage pyrite on earlier cores. This study provides a specific example of
substrate-guided precipitation in Carlin-type deposits; it does not establish that every crystallographically
continuous outer domain elsewhere formed in the same way.
Yang et al. [7] described an As-rich, Au-poor core with arsenopyrite inclusions and an As–Au-rich
oscillatory rim at Chang'an. Their interpretation links Au release and re-enrichment to coupled
3

dissolution–reprecipitation, using micro- to nanoscale observations and geochemical reasoning. This case
shows why an apparent outward increase in Au need not be interpreted solely as addition of a simple
growth layer.
For comparison between these cases, the relevant question is not which mechanism is universally correct.
It is which combination of observations discriminates mechanisms in a particular grain. Geometric
resemblance alone cannot transfer a mechanism from one deposit to another, and the two processes need
not be mutually exclusive over an entire paragenetic history.

3.4 Defects complicate the growth–replacement dichotomy
Fougerouse et al. [6] identified enrichment of Au and other trace elements at dislocations within a
deformation-related low-angle boundary. This adds a defect-associated occurrence that may not be
resolved by conventional grain-scale chemistry. The finding is used here to distinguish a mapped
chemical domain from the smaller structures that may host some of its metal.
Fougerouse et al. [8] proposed dislocation-mediated interfacial re-equilibration as an alternative to
interface-coupled dissolution–reprecipitation. Au–As-rich bands were truncated by poorer pyrite, while
interfacial dislocations and crystallographic observations supported a process not attributed to
deformation-related boundaries. This is an important counterexample: truncation and compositional
change need not uniquely identify fluid-mediated dissolution–reprecipitation. A lack of obvious porosity
also does not restore a simple growth interpretation.
These studies address different roles of dislocations and should not be collapsed into a single
“deformation” mechanism. In the proposed workflow, EBSD-scale information is a guide to further
questions, while nanoscale evidence may be required for the process actually being claimed. Failure to
resolve a mechanism should be reported as unresolved re-equilibration rather than repaired by a more
confident label.

3.5 Counterexamples to a simple outward enrichment model
Sugiono et al. [9] reported a chemically zoned pyrite at Kanowna Belle in which an Au–As-rich banded
mantle lies between relatively depleted core and outer rim domains. Several gold occurrences coexist in
the mantle, including lattice-associated gold, inclusions and late fracture fillings. This case directly
challenges a binary core-versus-rim description when an intermediate domain carries the most
informative signal. The term “rim” must therefore be tied to a mapped region rather than assumed to
denote the principal ore stage.
Cook et al. [13] investigated textural controls on gold distribution in As-free pyrite from Dongping,
Huangtuliang and Hougou. Their work provides a counterweight to selecting only As-rich targets. Ehrig
et al. [14] examined Olympic Dam pyrite at micro- to nanoscale and documented gold-bearing
nanoparticles associated with overprinting and nanoscale structures. Together, these cases support
retaining textural context and other potential host phases when evaluating an Au–As association. They do
not imply that arsenic is unimportant; they show why its presence or absence cannot be used as a
complete standalone decision rule.

3.6 Vietnamese context
Le et al. [15] described trace-element-rich rims in Me Xi pyrite while reporting that Au in the selected
pyrites was mostly below 1 ppm. This is a useful local comparison because a prominent multielement rim
need not contain high Au concentrations. Khuong et al. [10] documented brecciation and subsequent
4

quartz–pyrite cementation at Pac Lang, as well as oxidation textures. Nguyen et al. [11] reported
approximately 253 Ma zircon ages for nearby intrusive rocks and discussed a possible magmatic
relationship. None of these statements determines the Au distribution of an unanalysed Pac Lang pyrite
rim.
Table 1. Selected contrasts that constrain interpretation. Lessons are the synthesis proposed here,
not new observations from the cited deposits.
Contrast

Evidence represented

Interpretive consequence

Composition versus
species [1,2,5]

Au–As chemistry and direct structural
investigation

Concentration and atomic occurrence require
different claims

Fluid evolution versus
growth kinetics [3,4]

Spatially resolved chemical zoning

Band count alone does not establish fluidpulse count

Growth versus reequilibration [7,8,12]

Chemical interfaces and crystallographic
relationships

Core–rim appearance is not a unique process
indicator

Outermost rim versus
internal mantle [9]

A metal-rich intermediate domain

A binary core–rim classification can omit the
critical domain

As-rich targeting versus
wider host search
[13,14,15]

As-poor pyrite, nanoparticles and low-Au
rims

Target selection must allow counterexamples

4 A framework linking claims to observations
4.1 Association and enrichment
The first claim is spatial association: Au occurs within or adjacent to a mapped region. It requires
coordinates that connect the compositional observation to the image and geological sample. An isolated
high-Au analysis has limited value for a rim hypothesis if the analytical footprint overlaps an inclusion or
if its position cannot be recovered. Conversely, a small particle identified compositionally can establish a
local occurrence without establishing the mean Au content of the surrounding pyrite.
The second claim is enrichment relative to a defined comparator. A study should specify whether it
compares outer and inner domains of the same grain, different grains within a sample, or different sample
groups. These comparisons answer different questions. The proposed default for a core–rim hypothesis is
paired within-grain comparison followed by testing across independent samples. Analytical spots are
nested observations, not independent geological replicates.
Where the spatial resolution is inadequate, the appropriate stopping point is an unresolved mixed-domain
measurement. A visually impressive elemental map should not be converted into a concentration ratio
unless the map is quantitative and the regions of interest are analytically defensible. Non-detection should
retain its stated detection limit; it should not silently become zero.

4.2 Timing and process
A crosscutting gold-bearing filling can support a relative age younger than the domains it cuts. That
observation does not identify where the gold came from. Likewise, apparent enclosure in a twodimensional section establishes a spatial relationship on that surface, but a connection outside the section
5

may remain possible. Strong timing interpretations should explain which alternative geometries have been
tested and which remain unresolved.
Process interpretation follows the timing test rather than replacing it. The proposed alternatives include
precipitation on a pre-existing substrate, fluid-related replacement, defect-mediated re-equilibration and
later filling of fractures or interfaces. Multiple alternatives may survive. The framework treats this as an
informative outcome: it defines the additional observation that would discriminate them, or establishes
that the present paper should stop at a more limited conclusion.

4.3 Atomic occurrence and provenance
Atomic occurrence concerns the form in which Au is hosted. Provenance concerns its source. Neither can
be deduced solely from relative age or a concentration contrast. For example, an Au-poor core adjacent to
an Au-rich exterior is compatible with external metal supply; it does not demonstrate transfer from that
core. A proposed recycling mechanism should identify a plausible donor, an available inventory and
evidence linking the donor to the recipient.
The burden of evidence increases when a claim moves from one grain to an orebody or a regional event.
A grain-scale relationship can be robust without being representative of the deposit. In the proposed
reporting structure, each conclusion carries its spatial scope: spot, domain, grain, sample, ore type or
deposit. This prevents repeated images of one exceptional grain from functioning as apparent depositscale replication.
Table 2. Proposed claim-to-evidence matrix. No single row is a universal diagnostic criterion.
Claim

Evidence needed for the stated claim

Alternative or limitation to retain

Au is associated with an
outer domain

Registered image and compositional
detection at that location

Adjacent inclusion, fracture connection or
mixed analytical volume

Outer pyrite is Auenriched

Valid quantitative comparison with a
defined inner domain

Detection-limit effects, selective targeting and
mixed phases

Gold-bearing material is
younger

Direct crosscutting or other defensible
relative-age relationship

A spatial association without demonstrated
sequence

The outer domain is an
overgrowth

Compatible interfaces, chemistry and,
when relevant, crystallography

Replacement or re-equilibration preserving
some continuity

Gold was locally
remobilized

Donor–recipient relationship and a
credible inventory argument

External supply or mixed local and external
sources

Gold has a specified
atomic occurrence

A method and model addressing the
relevant structural scale

Unresolved nanoparticles, defects or
averaging

6

Figure 1. Conceptual alternatives for gold associated with pyrite margins. The diagrams are schematic hypotheses,
not microscopy or observations from Pac Lang. The same peripheral position may belong to growth, re-equilibration
or a crosscutting structure; an internal mantle can also be the enriched domain. Colours distinguish domains and
candidate Au locations rather than measured concentrations.

5 Analytical and sampling consequences
5.1 Mixing must be tested before a domain contrast is interpreted
A useful conceptual observation model is Cobs = Σ(fi × Ci) + error, where Cobs is a reported
concentration, fi is the effective contribution of domain or phase i to the sampled material, and Ci is its
concentration. This expression is a bookkeeping approximation, not a calibration model: unequal ablation,
matrix effects and instrument response can make a geometric area fraction different from an analytical
contribution. It nevertheless makes the central problem explicit. A measurement overlapping a thin rim,
core and small gold-bearing inclusion cannot be interpreted as pure rim chemistry without further
justification.
For a new study, the required records include beam or spot dimensions, mapped domain widths,
positioning uncertainty and the treatment of boundaries and inclusions. A pilot set should determine
whether the intended comparisons are physically resolvable. When the crucial rim is narrower than the
defensible analytical footprint, expanding the number of similar spots does not solve the identification
problem. Either the method, the target or the claim must change.
7

This review does not prescribe a single instrument configuration. Reflected-light microscopy and BSE
can locate candidate structures; compositional methods must then address the required elements and
spatial scale. If the conclusion concerns epitaxy, crystallographic evidence becomes relevant. If it
concerns nanoparticles or atomic coordination, a higher concentration measurement alone is not the
necessary next step. Method selection should be driven by the unresolved claim.

5.2 Concentration is not metal inventory
A second bookkeeping relation is M_Au = Σ(Ci × mi), where mi is the mass of each relevant domain or
phase and concentration units are converted consistently. A thin, strongly enriched rim need not contain
most of the gold in a grain, and a gold-bearing particle count does not determine the fraction of the
sample's Au inventory. Inferring a mass fraction requires domain or phase abundance as well as
composition.
For a remobilization argument, this distinction prevents an enrichment contrast from being mistaken for a
mass balance. A donor domain must have contained enough transferable gold to explain the proposed
recipient, within the uncertainty of the reconstructed system. Missing precursor volume, material removed
from the section and open-system additions may prevent closure. In that situation, a local redistribution
hypothesis can be discussed without asserting that it accounts for the entire metal inventory.

5.3 Sampling should test the preferred interpretation
The proposed design separates discovery sampling from validation sampling. Discovery targets structures
likely to reveal a relationship. Validation includes comparable grains or samples not selected solely for
conspicuous rims or visible gold. Reporting both stages makes clear whether a pattern was found through
targeted search or estimated within a defined population. This is a proposed design principle; no sampling
experiment is reported here.
The record should distinguish geological samples, polished sections, grains and analytical spots. Grainlevel contrasts can be summarized within samples before assessing consistency among samples. A large
number of spots from a single grain should not dominate an apparent deposit-wide trend. The reporting
should also include failed detections and ambiguous boundaries, with reasons for exclusion established
before selecting publication figures.
The resulting minimum reporting package is not a fixed number of analyses. It is a recoverable chain
from sample context to grain image, mapped domain, analytical location and interpretation. Its adequacy
depends on whether the decisive relationship recurs independently and whether the observations
discriminate the competing explanations actually discussed.

8

Figure 2. Proposed evidence workflow. Chemical detection, quantitative contrast and chronology are separate
branches. Mechanistic interpretation is attempted only after their compatibility is evaluated. The unresolved
outcome is intentional and indicates either a narrower conclusion or a specific additional measurement.

6 Application to Pac Lang
6.1 Separate the published context from the working hypothesis
The published brecciation–cementation relationships at Pac Lang [10] justify searching for overprinting at
grain scale. They do not establish that the cementing pyrite is consistently gold-rich, nor that an exterior
domain in every grain belongs to the same event. The proposed question is consequently neutral: does Au
distribution follow pyrite domains, crosscutting structures, or both, and which relative relationships can
be established directly?
The approximately 253 Ma intrusive ages discussed by Nguyen et al. [11] are ages of the analysed
intrusive rocks, not direct dates of pyrite rims or gold deposition. They provide geological context for a
hypothesis requiring additional tests. Similarly, an absence of intrusive rock in a particular sampled
interval would describe that interval; by itself it would not resolve the source of the ore fluid. The present
framework deliberately leaves genetic classification outside the grain-scale question.

6.2 A tractable sequence of observations
The first stage is an inventory of available specimens and their provenance. Samples should be grouped
by preserved relationships rather than assigned automatically to regional pyrite generations. Useful
comparisons include relatively preserved pyrite-bearing material, brecciated and cemented material, and
later structures crossing earlier assemblages. Oxidized specimens need separate treatment before they are

9

combined with primary sulfide observations. These are selection criteria, not a claim that such a sample
suite has already been assembled.
The next stage is spatial registration. A specimen image should connect to a section, a grain and the
location of each analysis. Candidate domains receive descriptive identifiers before process labels. A pilot
survey can then establish whether outer domains are present, whether their width permits meaningful
analysis and whether Au is detectable by the proposed method. Only after that assessment should a larger
analytical programme be fixed.
Three outcomes should remain publishable questions rather than predetermined successes. If Au follows a
mapped outer domain, the next question concerns its origin and concentration contrast. If Au follows
structures that cut several domains, the immediate conclusion concerns a later gold-bearing filling. If no
reproducible association emerges, the correct result is that the examined set does not support the proposed
rim control under the methods used. None of these outcomes alone establishes a whole-deposit gold
budget.

6.3 What would constitute a distinct empirical contribution
A follow-up paper should identify the specific relationship added to published Pac Lang descriptions.
Merely demonstrating several pyrite generations would repeat an already established complexity. A
stronger contribution would connect a repeatable Au distribution to a demonstrable sequence, quantify an
appropriately resolved contrast, or show that a plausible conventional interpretation fails in a defined
group of samples. Which contribution is available cannot be decided before the relevant observations
exist.
The analytical programme should not automatically include every advanced technique used in the
reference studies. Its purpose is to eliminate the alternatives that matter for the chosen claim. For
example, a defensible description of late fracture-hosted gold may not require atom probe tomography.
Conversely, a claim of gold incorporated in a particular structural environment cannot be secured merely
by adding more SEM images. Scope is determined by the claim, not by the length of the methods list.
Table 3. Decision points for a prospective Pac Lang study.
Pilot observation

Next decision

Conclusion not yet justified

A resolvable outer pyrite
domain occurs

Register chemistry and test the interface

Gold enrichment or overgrowth origin

Au is detected in the outer
region

Separate matrix, inclusions and fractures

Atomic occurrence or local source

A paired core–outer
contrast is reproducible

Test sample-level consistency and
analytical mixing

Most ore gold is in rims

Au-bearing filling cuts the
mapped domains

Document independent repetitions and
paragenesis

Gold was released from those domains

No association is
reproducible

Reassess the target or report the restricted
negative result

Absence of gold throughout the deposit

10

7 Discussion and limitations
7.1 The common feature is an identification problem
The selected studies do not support one mechanism for all pyrite margins. They reveal a recurring
identification problem: different processes can produce overlapping descriptions at one observational
scale. The proposed response is to make the observational target and the excluded alternatives explicit.
More precise language is not merely editorial caution; it identifies the next experiment and prevents an
attractive image from carrying several unsupported interpretations.
The framework differs from a catalogue of pyrite types by organizing evidence around claims rather than
appearances. A grain may support a strong local chronology but only a tentative process interpretation.
Another may have well-resolved chemistry but no defensibl