Evolutionary System of Mineralogy


Minerals reveal the nature of the co-evolving geosphere and biosphere through billions of years of Earth history. “Mineral Evolution” is the study of the changing diversity and distribution of minerals, which are the consequence of a succession of physical, chemical, and ultimately biological processes through more than 4.5 billion years of Earth history. In its first presentation (Hazen et al. 2008), as well as several subsequent publications (Hazen and Ferry 2010; Hazen 2010, 2012), mineral evolution was outlined in a qualitative framing of 10 stages, from pre-terrestrial minerals preserved in meteorites to biominerals of the Phanerozoic Eon (the last 541 million years). 

Mineral evolution timeline with 10 stages, from Hazen and Ferry (2010).

An ongoing challenge has been transforming the qualitative mineral evolution narrative into a more rigorous, quantitative system of mineralogy—one that places each kind of mineral into its historical context. To advance this goal, Hazen (2019) proposed an “Evolutionary System of Mineralogy,” which is now in preparation.

  • Hazen, R.M., Papineau, D., Bleeker, W., Downs, R.T., Ferry, J.M., McCoy, T.L., Sverjensky, D.A., and Yang, H. (2008) Mineral evolution. American Mineralogist, 93, 1693-1720.
  • Hazen, R.M. and Ferry, J.M. (2010) Mineral evolution: Mineralogy in the fourth dimension. Elements, 6, #1, 9-12.
  • Hazen, R.M. (2010) The evolution of minerals. Scientific American, 303, #3, 58-65.
  • Hazen, R.M. (2012) The Story of Earth: The First 4.5 Billion Years, from Stardust to Living Planet. New York: Viking, 306 p. Softcover edition (2013).
  • Hazen, R.M. (2019) An evolutionary system of mineralogy: Proposal for a classification based on natural kind clustering. American Mineralogist, 104, 810-816.  DOI: 10.2138/am-2019-6709 


In mid-2018, I began to confront a knotty problem that had been with me for more than a decade: Is there a coherent, internally consistent way to place the qualitative narrative of “mineral evolution” into a more quantitative and rigorous framework? Since the pioneering conceptual studies of the twentieth century (Bowen 1928; Gastil 1960; Zhabin 1979), the idea of an evolving mineral realm has had intrinsic appeal. All geoscientists realize that minerals provide the most robust and information-rich testimony for billions of years of cosmic history. From the oldest presolar moissanite grains, now dated at ~7 billion years (Heck et al. 2020), to the biominerals of our teeth and bones forming in real time, the mineral kingdom holds keys to unlocking secrets of planetary evolution through deep time.

For more than 60 years, from the time I would spend hours every month as a middle-school student studying the fabulous “Dana Collection” of the American Museum of Natural History in New York City, I embraced the classification system of the International Mineralogical Association’s Commission on New Minerals, Nomenclature and Classification (IMA-CNMNC). I proudly displayed a growing collection on groaning bedroom shelves, with hand-written labels citing name, formula, crystal system, and locality. I learned early on that nothing in mineralogy is more fundamental than structure and composition; each species bears its identity by virtue of a unique combination of those two attributes. Today, the IMA-CNMNC has approved almost 6000 mineral species, each with a unique combination of idealized chemical composition and crystal structure. 

IMA protocols define each mineral species by its unique combination of idealized chemical composition and crystal structure.

But in 2018 I was faced with a dilemma. The emerging historical narrative of mineralogy, in which new kinds of minerals arise through a combination of physical, chemical, and ultimately biological processes, did not always appear to fit comfortably into the IMA-CNMNC scheme. I was torn. I debated whether I needed to abandon IMA protocols and start over, thus rejecting decades of deeply ingrained reverence for Dana and his ever-evolving system, or somehow build on the richness of the IMA Foundation.

In a sense, the answer was provided by James Dwight Dana, himself, who wrote in his transformative third edition of his System of Mineralogy (Dana 1850, p.5): “The science of Mineralogy has made rapid progress in the past six years; chemistry has opened to us a better knowledge of the nature and relations of compounds; and philosophy has thrown new light on the principles of classification. To change is always seeming fickleness. But not to change with the advance of science, is worse; it is persistence in error.”

The third edition of James Dwight Dana’s System of Mineralogy has provided inspiration for this effort. 

The result was a short paper, “An evolutionary system of mineralogy: Proposal for classification of planetary materials based on natural kind clustering” (Hazen 2019) that suggested a complementary approach to mineral classification—one that embraces the non-ideality of real minerals. I proposed a classification method that exploits the inherent evolving “messiness” of planetary materials by grouping them according to “natural kind clusters” (Boyd 1999). Natural solids can be categorized by distinct “kinds” according to their distinctive combinations of non-ideal atomic structures, complex chemical compositions, variable physical properties, and diverse modes of origin. Cluster analysis based on observed ranges of properties in natural specimens thus complements and amplifies, though not supersedes, the present mineral classification scheme of the IMA-CNMNC. By recognizing natural kinds of planetary materials, this mineral classification approach incorporates an evolutionary component in addition to chemistry and structure.

Diamond illustrates the contrast between a “mineral species,” defined by IMA-CNMNC protocols as pure carbon in the diamond crystal structure, and “natural kinds,” each of which represents a different historical process that imparts idiosyncratic attributes to diamond specimens, including trace and minor elements, isotopes, solid and fluid inclusions, size and shape, structural defects, and more.

 Hazen (2019) presents this case for an evolutionary system of mineralogy, but it does so with an undertone of tension – a clear reflection of my struggle at the time. At each instance where my thought regarding “natural kinds” diverged from IMA species, I felt the need to justify, to defend, even to (gently, I hope) criticize the IMA approach. If you read between the lines of Hazen (2019), you can sense that struggle. The paper also inspired a critique by Hatert and colleagues (2021), who felt that I misrepresented aspects of the IMA-CNMNC procedures for defining new minerals. (To the extent that was true, I’m very sorry—it wasn’t my intent).

Over the past years, that tension has been largely released, as we have gained a clearer understanding of “mineral natural kinds,” notably their identities in the context of “historical natural kinds” (Cleland et al. 2020). We argue that the advancement of science depends upon developing classification protocols that systematize natural objects and phenomena into “natural kinds”—categorizations that are conjectured to represent genuine divisions in nature in virtue of playing central roles in the articulation of successful scientific theories. In the physical sciences, theoretically powerful classification systems, such as the Periodic Table, are typically time-independent. Similarly, the standard classification of mineral species by the IMA-CNMNC relies on idealized chemical composition and crystal structure, which are time-independent attributes selected on the basis of theoretical considerations from chemical theory and solid-state physics. However, when considering mineral kinds in the historical context of planetary evolution a different, time-dependent classification scheme is warranted. The evolutionary system of mineral classification that we propose is based on recognition of the role played by minerals in the origin and development of planetary systems. We lack a comprehensive theory of chemical evolution capable of explaining the time-dependent pattern of chemical complexification exhibited by our universe. Therefore, we recommend a bootstrapping approach to mineral classification based on observations of geological field studies, astronomical observations, laboratory experiments, and analyses of natural samples and their environments. This approach holds the potential to elucidate underlying universal principles of cosmic chemical complexification. 

In the following sections, I review the emerging “Evolutionary System of Mineralogy,” now represented by more than a dozen papers published, in press, in review, and in preparation, and outline the complementary aspects of “mineral natural kinds” and IMA mineral species.

*In part adapted from Hazen (2021) and Cleland et al. (2020). 

  • Bowen, N. L. (1928) The Evolution of the Igneous Rocks. Princeton University Press, Princeton, New Jersey.
  • Boyd, R. (1999) Homeostasis, species, and higher taxa. In R. Wilson [Editor], Species: New Interdisciplinary Essays. Cambridge, Massachusetts: Cambridge University Press, pp. 141-186.
  • Cleland, C.E., Hazen, R.M., and Morrison, S.M. (2020) Historical natural kinds and mineralogy: Systematizing contingency in the context of necessity. Proceedings of the National Academy of Sciences, 118, e2015370118 (8 p).
  • Dana, J.D. (1850) A System of Mineralogy, Comprising the Most Recent Discoveries, Including Full Descriptions of Species and their Localities, Chemical Analyses and Formulas, Tables for the Determination of Minerals, and a Treatise on Mathematical Crystallography and the Drafting of Figures of Crystals. Third Edition, Rewritten, Rearranged, and Enlarged. New York and London: George P. Putnam.
  • Gastil, G. (1960) The distribution of mineral dates in space and time. Am. J. Sci., 258, 1-35.
  • Hatert, F., et al. (2021) A comment on “An evolutionary system of mineralogy: Proposal for a classification of planetary materials based on natural kind clustering.” Am. Mineral., 106, 150-153.
  • Hazen, R.M. (2019) An evolutionary system of mineralogy: Proposal for a classification based on natural kind clustering. Am. Mineral., 104, 810-816.  DOI: 10.2138/am-2019-6709
  • Hazen, R.M. (2021) Reply to “Comment on ‘An evolutionary system of mineralogy: Proposal for a classification based on natural kind clustering’.” Am. Mineral., 106, 154-156.  
  • Heck, P.R., Greer, J., Kööp, L., et al. (2020) Lifetimes of interstellar dust from cosmic ray exposure ages of presolar grains. Proc. Natl Acad. Sci., Jan. 13, 2020, DOI: 10.1073/pnas.1904573117
  • Zhabin, A.G. (1979) Is there evolution of mineral speciation on Earth? Doklady Akademii Nauk, 247, 199-202 [in Russian]. English translation (1981) Doklady Earth Science Sections, 247, 142-144. 


The evolutionary system of mineralogy is a collaborative effort of Robert Hazen, Shaunna Morrison, and colleagues. It is being published sequentially in American Mineralogist. The following parts are published or in press:

  • PART I: Hazen, R.M., and Morrison, S.M. (2020) An evolutionary system of mineralogy, part I: stellar mineralogy (>13 to 4.6 Ga). American Mineralogist, 105, 627-651. 

The earliest stage of mineral evolution commenced with the appearance of the first crystals in the universe at >13 Ga and continues today in the expanding, cooling atmospheres of countless evolved stars, which host the high-temperature (T > 1000 K), low-pressure (P < 10-2 atm) condensation of refractory minerals and amorphous phases. Most stardust is thought to originate in three distinct processes in carbon- and/or oxygen-rich mineral-forming stars: (1) condensation in the cooling, expanding atmospheres of asymptotic giant branch stars; (2) during the catastrophic explosions of supernovae, most commonly core-collapse (Type II) supernovae; and (3) classical novae explosions, the consequence of runaway fusion reactions at the surface of a binary white dwarf star. Each stellar environment imparts distinctive isotopic and trace element signatures to the micro-and nanoscale stardust grains that are recovered from meteorites and micrometeorites collected on Earth’s surface, by atmospheric sampling, and from asteroids and comets. Although our understanding of the diverse mineral-forming environments of stars is as yet incomplete, we present a preliminary catalog of 41 distinct natural kinds of stellar minerals, representing 22 official International Mineralogical Association (IMA) mineral species, as well as 2 as yet unapproved crystalline phases and 3 kinds of non-crystalline condensed phases not codified by the IMA.


Characteristic pressure-temperature-composition regimes of stellar minerals. A. Eleven major mineral-forming elements and six select minor elements commonly found in stardust.  B. Estimated pressure-temperature formation ranges of most stellar primary condensate minerals, which formed via relatively low-pressure, high-temperature condensation in the turbulent atmospheres of highly evolved stars.


A.    B.

Dust-forming stars. (Hubble Space Telescope images, courtesy of NASA).  A. Image of star V838 Monocerotis, a “planetary nebula” that formed from dust and gas surrounding an asymptotic giant branch (AGB) star.  B. Image of the Crab Nebula—the remnants of a supernova.

Figure 3. Electron microscope images of stellar minerals. (A) Cross-section of a 1-micron diameter “onion” AGB graphite with central khamrabaevite (TiC) inclusion; (B) 13-micron diameter “cauliflower” SN-II graphite grain—a composite of smaller crystallites; (C) 4.5-micron diameter euhedral “mainstream” AGB moissanite (SiC) crystal; (D) 1.4-micron diameter euhedral AGB corundum (Al2O3) crystal. [A,B,C: E.K. Zinner, Treatise Geochem., 1, 17-39 (2014); D: A. Takigawa et al., Astroph. J. Lett., 862, L13 (2018)]. 

  • PART II: Morrison, S.M., and Hazen, R.M. (2020) An evolutionary system of mineralogy, Part II: Interstellar and solar nebula primary condensation mineralogy (> 4.565 Ga). American Mineralogist, 105, 1508-1535. 

Part II considers the formation of primary crystalline and amorphous phases by condensation in three distinct mineral-forming environments, each of which increased mineralogical diversity and distribution prior to the accretion of planetesimals > 4.5 Ga: 

Interstellar molecular solids: Varied crystalline and amorphous molecular solids containing primarily H, C, O, and N condense in cold, dense molecular clouds in the interstellar medium (10 < T < 20 K; P < 10-13 atm). With the possible exception of some nano-scale organic condensates preserved in carbonaceous meteorites, the existence of these phases is documented primarily by telescopic observations of absorption and emission spectra of interstellar molecules in radio, microwave, or infrared wavelengths.

Nebular and circumstellar ice: Evidence from IR observations and lab experiments suggest that cubic H2O (“cubic ice”) condenses as thin crystalline mantles on oxide and silicate dust grains in cool, distant nebular and circumstellar regions where T ~100 K.  

Primary condensed phases of the inner solar nebula: The earliest phase of nebular mineralogy saw the formation of primary refractory minerals that solidified through high-temperature condensation (1100 < T < 1800 K; 10-6 < P < 10-2 atm) in the solar nebula more than 4.565 billion years ago. These earliest mineral phases originating in our solar system formed prior to the accretion of planetesimals and are preserved in calcium-aluminum-rich inclusions, ultra-refractory inclusions, and amoeboid olivine aggregates.

Stage II of mineral evolution introduces for the first time several mineral groups that played important roles in planetary evolution, including garnet, melilite, clinopyroxene, and quartz. Like several of the phases described among the stellar condensation minerals of Stage I, many of which occur in 2 or 3 different natural kinds based on very different isotopic signatures associated with different kinds of stars (Hazen and Morrison 2020), several primary nebular minerals occur as condensates in CAIs, AOAs, and URIs – different mineral-forming environments that impart distinctive combinations of compositional, morphological, and petrologic attributes.


A. B.



The temperature, pressure, and compositional characteristics of primary interstellar and solar nebular condensates result in a distinctive second phase of the evolutionary system of mineralogy. A. These minerals formed at a wide range of temperatures via low-pressure (P < 0.01 atm) condensation in interstellar and nebular environments. B. Interstellar minerals formed primarily from C, H, N, O, and probably S – five of the most abundant elements in the cosmos. C. Primary minerals in CAIs, URIs, and AOAs formed principally from 16 essential major elements, with important additional contributions from 7 minor elements.



A Hubble Space Telescope image of a portion of the Eagle Nebula (NGC 6611 and IC 4703), dubbed “The Pillars of Creation,” displays a star-forming region. The core regions of this structure are cooler areas, dense molecular clouds where molecular solids condense. (Photo courtesy of NASA)


  • PART III: Hazen, R.M., Morrison, S.M., and Prabhu, A. (2021) An evolutionary system of mineralogy, Part III: Primary chondrule mineralogy (4.566 to 4.561 Ga). American Mineralogist, 106, 325-350.

Part III of the evolutionary system considers the formation of 43 different primary crystalline and amorphous phases in chondrules, which are diverse igneous droplets that formed in environments with high dust/gas ratios during an interval of planetesimal accretion and differentiation between 4566 and 4561 Ma. Chondrule mineralogy is complex, with several generations of initial droplet formation via a variety of proposed heating mechanisms, followed in many instances by multiple episodes of reheating and partial melting. Primary chondrule mineralogy thus reflects a dynamic stage of mineral evolution, when the diversity and distribution of natural condensed solids expanded significantly.

The catalog of 43 phases is misleading in at least two ways regarding the diversity and distribution of primary chondrule minerals. First, the primary origins of 8 of these phases (awaruite, graphite, haxonite, pentlandite, magnetite, merrillite, nepheline, and roedderite) have been questioned. For each of these minerals, some experts assert that all chondrule occurrences formed by secondary processes. Of the remaining 35 minerals, an additional 15 (sinoite, perryite, alabandite, caswellsilverite, daubréelite, niningerite, oldhamite, sphalerite, wassonite, rutile, armacolite, ferropseudobrookite, perovskite, zirconolite, and sapphirine) are extremely rare and of restricted occurrence, while 7 more (tetrataenite, cohenite, schreibersite, magnetite, ilmenite, cristobalite, and silica glass) are more widespread but volumetrically minor. Thus, only a dozen phases probably account for more than 99 vol. % of primary chondrule mineralogy. This distribution reflects the mineralogical parsimony of high-temperature assemblages of the nebula’s major rock-forming elements – in essence, a nebular manifestation of J. Willard Gibbs’ “phase rule.” We will discover a similar restricted mineral diversity among the primary phases that arise from planetesimal differentiation into mantle and core (Part IV). However, a dramatic rise in mineral diversity occurred as a consequence of pervasive alteration of these equilibrium phases – reworking by impact processes, as well as aqueous, hydrothermal, and metamorphic alteration that resulted in hundreds of new minerals before the assembly of today’s planets and moons (Part V).


Primary chondrule minerals form principally from 15 structurally essential elements, with important additional contributions from 3 additional minor elements.

  • PART IV: Morrison, S.M. and Hazen, R.M. (2021) An evolutionary system of mineralogy, Part IV: Planetesimal differentiation and impact mineralization (4.566 to 4.560 Ga). American Mineralogist, 106, in press.

The fourth installment of the evolutionary system of mineralogy considers two stages of planetesimal mineralogy that occurred early in the history of the solar nebula, commencing by 4.566 Ga and lasting for at least 5 million years: (1) primary igneous minerals derived from planetesimal melting and differentiation into core, mantle, and basaltic components; and (2) impact mineralization resulting in shock-induced deformation, brecciation, melting, and high-pressure phase transformations. 

We tabulate 90 igneous differentiated asteroidal minerals, including the earliest known occurrences of minerals with Ba, Cl, Cu, F, and V as essential elements, as well as the first appearances of numerous phosphates, quartz, zircon, and amphibole group minerals. We also record 40 minerals formed through high-pressure impact alteration, commencing with the period of asteroid accretion and differentiation. These stages of mineral evolution thus mark the first time that high pressures, both static and dynamic, played a significant role in mineral paragenesis. 

A B  


Non-chondritic meteorites. (A) A polished and etched slab (1151 gm; 10 cm maximum diameter) from the Staunton iron meteorite with taenite exsolved from kamacite in a Widmanstätten pattern; (B) A polished slab of the Esquel pallasite (1091 gm; 8 cm maximum diameter), a stony-iron meteorite, with kamacite (silver in reflected light) and forsteritic olivine (yellow in transmitted light); (C) The Cumberland Falls aubrite (1227 gm; 11 cm maximum dimension), an achondrite meteorite observed to fall in Kentucky in 1919. This polymict breccia contains a rich variety of silicate-rich clasts of different lithologies derived primarily from ordinary chondrites. [Photos courtesy of the National Museum of Natural History, Smithsonian Institution]

Essential mineral-forming elements in 90 primary asteroidal minerals.


Essential mineral-forming elements in 40 shocked meteorite minerals. All of these 14 elements are major mineral-forming elements in primary asteroidal minerals.


The ranges of pressure-temperature conditions for primary asteroidal mineralization (green) and impact mineralization (red) suggest that these processes represent distinct stages of mineral evolution.


Cosmic mineral evolution played out in a succession of stages, each of which explored new regimes of temperature, pressure, and composition, while adding to the diversity of condensed solid phases. The 130 meteorite minerals reviewed above (Tables 1 and 2) complement the 41 stellar natural kinds from Part I, 67 interstellar and primary nebular condensates of Part II, and 44 primary chondrule minerals described in Part III of this series. In Part IV we encounter pressures significantly above 1 atmosphere for the first time, both in the contexts of asteroidal interiors (to P < 0.5 Gpa) and via shock events (to P > 30 GPa). The resulting inventory of meteorite minerals includes 90 kinds formed by primary asteroidal processes, as well as 40 high-pressure impact minerals. These 130 mineral natural kinds encompass 127 approved IMA mineral species, 10 of which are lumped with other species and thus do not appear as separate natural kinds in Tables 1 and 2. In addition, we include 7 crystalline phases, either not recognized as valid IMA species (e.g., fassaite; magnesiowüstite; martensite) or awaiting possible approval (e.g., unnamed CuCrS2 and Mg-Fe silicates), as well as 6 amorphous phases. 

Asteroidal processes resulted in many new mineral phases. Of the 90 primary asteroidal natural kinds, 48 minerals appear for the first time, including the earliest known members of the apatite and amphibole groups, as well as zircon, potassic feldspar, and numerous phosphate minerals. Of the 40 impact minerals, 38 occur for the first time; only diamond and clinopyroxene also formed previously at low pressure. This total of 86 new minerals in Part IV almost doubles the total of 181 phases tabulated to this stage of mineral evolution.   

  • PART V: Hazen, R.M., and Morrison, S.M. (2021) An evolutionary system of mineralogy, Part V: Planetesimal Aqueous and thermal alteration of planetesimals (4.565 to 4.550 Ga). American Mineralogist, 106, in press.

Part V of the evolutionary system of mineralogy explores phases produced by aqueous alteration, metasomatism, and/or thermal metamorphism – relicts of ancient processes that affected virtually all asteroids and that are preserved in the secondary mineralogy of meteorites. We catalog 167 historical natural kinds of minerals that formed by alteration in the parent bodies of chondritic and non-chondritic meteorites within the first 20 million years of the solar system. Secondary processes saw a dramatic increase in the chemical and structural diversity of minerals. These phases incorporate 41 different mineral-forming elements, including the earliest known appearances of species with essential Co, Ge, As, Nb, Ag, Sn, Te, Au, Hg, Pb, and Bi. Among the varied secondary meteorite minerals are the earliest known examples of halides, arsenides, tellurides, sulfates, carbonates, hydroxides, and a wide range of phyllosilicates.


Secondary minerals from chondritic and nonchondritic meteorite parent bodies formed by aqueous alteration and thermal metamorphism primarily from 23 different essential elements that appear in 5 or more minerals, with important additional contributions from 18 minor elements that appear in fewer than 5 scarce phases. Included among these elements are the earliest known appearances of minerals with essential Co, Ge, As, Nb, Ag, Sn, Te, Au, Hg, Pb, and Bi.


Aqueous alteration and/or thermal metamorphism in planetesimals dramatically increased the chemical and structural diversity of the preterrestrial mineral kingdom. In Parts I through V, we have tabulated 448 historical natural kinds of minerals representing 263 IMA-approved mineral species plus 19 as yet unapproved crystalline phases and 16 amorphous phases. Of this total of 298 diverse minerals, 120 phases (i.e., 40%) are new to Part V, including the earliest known examples of halides, arsenides, tellurides, sulfates, carbonates, hydroxides, and a wide range of phyllosilicates. 

Secondary processes also dramatically increased the chemical diversity of minerals with 41 different essential (i.e., species-defining) elements, including the earliest known appearances of essential Co, Ge, As, Nb, Ag, Sn, Te, Au, Hg, Pb, and Bi. Nevertheless, as with earlier stages of mineral evolution, secondary meteorite minerals are dominated volumetrically by relatively few phases. We estimate that only 34 minerals of the 167 listed in Part V occur widely or ever exceed 1 vol %. Those more common secondary minerals, furthermore, incorporate only 15 different essential elements, all of which are relatively abundant: H, C, O, S, P, Cl, Na, Mg, Ca, Fe, Ni, Al, Cr, Si, and Ti. By contrast, at least 95 of 167 minerals are known as volumetrically trivial phases. Thus, as in many other mineral-rich environments, relatively few mineral species are common, whereas most are rare.

  • PART VI: An evolutionary system of mineralogy, Part VI: Primary mineralogy of the earliest Hadean Eon (4.56 to 4.50 Ga). In preparation.  
  • PART VII: An evolutionary system of mineralogy, Part VI: Secondary mineralogy of the earliest Hadean Eon (4.56 to 4.50 Ga). In preparation.


Network graphs provide a useful method to visualize relationships among varied minerals and their attributes [Morrison et al., Am. Mineral., 102, 1588-1596 (2017)]. One way to illustrate the evolutionary system of mineralogy is to link mineral species (usually a species approved by the IMA-CNMNC) to its paragenetic mode (the process by which the mineral formed). In such a network, every mineral is represented by a diamond-shaped “node” that is linked to one or more nodes representing a paragenetic mode.

The network below from Part I of the System displays a bipartite force-directed network graph of stellar minerals, in which 27 phases—22 IMA approved mineral species, two additional crystalline phases not yet recognized by IMA (MoC and Fe7C3), plus three amorphous condensed phases (C, Al2O3, and silicate)—are represented by diamond-shaped nodes. These mineral nodes are linked to three types of stars (AGB, SN-II, and CNova) represented by star-shaped nodes. Compositional information is conveyed by mineral node colors: black (C-bearing), green (not C or O), blue (contains O, but not Si), and red (contains Si + O).


Bipartite force-directed network graph of stellar minerals linked to their host stars. Diamond-shaped nodes represent condensed crystalline and amorphous phases [black (C-bearing), green (not C or O), blue (contains O, but not Si), and red (contains Si + O)], whereas star-shaped nodes represent three types of host stars—asymptotic giant branch stars (AGB), Type II supernovae (SN-II), and classical novae (CNova). The sizes of nodes correspond to the number of links to other nodes. [Courtesy of Anirudh Prabhu, RPI]

The following figure displays a bipartite force-directed network graph of primary stellar, interstellar, and nebular minerals formed prior to ~4,565 Ma, in which 69 different phases, including 10 amorphous condensed phases, are represented by diamond-shaped nodes. Each of these mineral nodes is linked to one or more node representing a paragenetic mode of formation. Three different star-shaped nodes (AGB, SN-II, and CNova) represent stellar environments that impart distinctive isotopic signatures to minerals. A cloud-shaped node indicates interstellar dense molecular clouds (DMC), whereas four flattened disk icons represent different primary mineral-forming nebular environments (Circumstellar, CAI, AOA, and URI). 

Node size, shape, and color convey information. Mineral compositions are indicated by the color of diamond-shaped mineral nodes: black (C-bearing), green (lacking C or O), blue (contains O, but not C or Si), and red (contains Si + O). The sizes of mineral nodes correspond to the numbers of paragenetic modes to which they are linked. Similarly, the sizes of the star-, cloud-, and disk-shaped symbols indicate the numbers of different minerals to which they are associated.

At this early stage of mineral evolution, 8 different low-temperature interstellar and nebular condensed molecular phases (T < 100 K) form a separate network from 56 high-temperature stellar and nebular condensates (T > 1100 K). In future parts of this series, phases formed at intermediate temperatures in planetary surface environments will provide links between these two mineral-forming environments.

The bipartite force-directed network graph of primary stellar, interstellar, and nebular minerals formed prior to ~4561 Ma displays 96 different phases, including 10 amorphous condensed phases, represented by diamond-shaped nodes. Each of these mineral nodes is linked to one or more nodes representing a paragenetic mode of formation. Three different star-shaped nodes (AGB, SN-II, and CNova) represent stellar environments that impart distinctive isotopic signatures to minerals. A cloud-shaped node indicates interstellar dense molecular clouds (DMC), whereas five flattened disk icons represent different primary mineral-forming nebular environments (Circumstellar, CAI, AOA, URI, and PC). Information about mineral compositions is indicated by the color of diamond-shaped mineral nodes: black (C-bearing), green (lacking C or O), blue (contains O, but not C or Si), and red (contains Si and O). The sizes of the star-, cloud-, and disk-shaped symbols indicate the numbers of different minerals to which they are associated.

Most mineral nodes form a well-connected network, while 8 interstellar and nebular condensed molecular phases, all with T < 100 K, form a separate network from 88 high-temperature stellar and nebular condensates (T >> 300 K). In future contributions to this series, which will consider phases formed at intermediate temperatures in planetary surface environments, new links will occur between these two mineral-forming environments.

Bipartite force-directed network graph of primary stellar, interstellar, and nebular minerals linked to their modes of paragenesis. Diamond-shaped nodes represent condensed crystalline and amorphous phases [black (C-bearing), green (not C or O), blue (contains O, but not C or Si), and red (contains Si + O)]. Star-shaped nodes represent three types of host stars—asymptotic giant branch stars (AGB), Type II supernovae (SN-II), and classical novae (CNova); the cloud-shaped node represents dense molecular clouds (DMC); and five disk-shaped nodes indicate circumstellar environments, CAI, AOA, URI, and PC minerals. The sizes of paragenetic mode nodes correspond to the numbers of links to mineral nodes. Note that 8 low-temperature phases of the interstellar medium are not linked to 88 high-temperature primary phases of stellar and nebular environments.

As yet unpublished bipartite graph of 448 historical natural kinds of minerals representing 263 IMA-approved mineral species plus 19 as yet unapproved crystalline phases and 16 amorphous phases. This graph b y Anirudh Prabhu (RPI) represents Parts I to V of the evolutionary system.

Unpublished bipartite network graph of 5659 mineral species (blue nodes) linked to 57 different paragenetic modes (green nodes). An interactive online version of this graph is under development by Anirudh Prabhu. Versions of this represent all known minerals and mineral-forming mechanism. 


Two keys to identifying mineral natural kinds are (1) development of large, reliable, well-curated, and growing open-access mineral data resources that record numerous attributes of many samples; and (2) application of cluster analysis to discern discrete subsets of those samples with idiosyncratic combinations of attributes that point to distinctive formational environments. Among the project now underway:

Pyrite (FeS2):

Pyrite has formed in many different ways in meteorites, igneous rocks, hydrothermal ore deposits, through biological sulfate reduction, human-induced environmental changes, and more. Prof. Ross Large (University of Tasmania) and Prof. Daniel Gregory (University of Toronto) have been constructing a database of pyrite trace elements. Gregory et al. (2019) analyzed ~4,000 of these analyses. In work led by Prof. Shuang Zhang (now at the University of Texas; formerly at Carnegie), we are now collaborating with Gregory and colleagues to apply a variety of clustering methods to these data (Zhang et al. 2019).


Stellar Moissanite (SiC):

Moissanite is among the most common mineral phases produced in the expanding, cooling atmospheres of aged stars. Thousands of robust individual grains have been recovered from chondrite meteorites and identified by their extreme Si, C, and N isotopic anomalies. More than 17,000 stellar moissanite isotope analyses have been compiled by Stephan et al. (2020) and by Boujibar et al. (2021), who applied cluster analysis to discern different stellar origins of these grains. 


From the abstract:
Cluster analysis of presolar silicon carbide grains, based on literature data on 12C/13C, 14N/15N, δ30Si/28Si, and δ29Si/28Si plus or minus 26Al/27Al, reveals nine clusters that agree with previously defined grain types, but also highlight new divisions. Mainstream grains reside in three clusters with different metallicities. One of these clusters has a compact core, with several grains with a specific composition, suggesting an enhanced stellar formation, possibly through a starburst as previously suggested. The addition of 26Al/27Al highlights a cluster of mainstream grains that are enriched in 15N and 26Al, which cannot be explained by current AGB models. The cluster analysis also enabled the better definition of two AB grain clusters, one with 15N and 26Al excesses and the other with 14N and 26Al excesses, in agreement with recent studies. However, their definition does not use the N solar composition as a divider, and the contour of the 26Al-rich AB cluster identified in this study, previously referred as AB1, is in better agreement with supernova models that consider non-standard mixing of H into the inner He/C zone during the pre-supernova phase. We also found a cluster with a mixture of putative nova and AB grains, which may have formed in supernova and/or nova environments. X grains make up two clusters, having either strongly correlated Si isotopic ratios, or Si isotopic ratios deviating from the 2:3 line in the Si 3-isotope plot. Finally, most Y and Z grains are grouped in a cluster, suggesting that the previous use of solar 12C/13C as a divider was arbitrary. Our results show that cluster analysis is a powerful tool to understand stellar evolution and nucleosynthesis, and also highlight the need of having more multi-element isotopic data and morphological features of presolar grains for better classification.

Hystad et al. (2021) employed the same SiC dataset in studies that used alternative clustering methods. From the abstract: We report the use of several cluster analysis techniques to evaluate the classification of presolar silicon carbide (SiC) grains. The stability of the clusters and the confidence of the individual cluster assignment of the grains are assessed using consensus clustering with resampling methods. Our analysis shows that presolar SiC grains can be divided into seven groups that are found to be highly stable with most of the grains being assigned to the same cluster for at least 90% of the times over multiple aggregated clustering. Among the seven groups, two groups are dominated by AB grains, three groups by MS grains, one group by Z grains, and one group by X grains. The further division of X grains into two groups is dependent highly on the chosen algorithm and is, therefore, uncertain. Z and Y grains are clustered jointly with MS grains with one group dominated by Z grains, pointing to their common origins from low-mass asymptotic giant branch stars. The most stable clusters have the N grains clustered together with 15N -rich AB grains. However, some methods assign N grains with X grains, but with less stable clusters. The suggested genetic relationship among 15N-rich AB, N, and X grains, is in line with the recent proposal that all three types of presolar SiC came from core collapse supernovae. We discuss the results from different clustering techniques based on our assessment of the cluster stabilities and the extent to which the cluster assignments overlap across the different methods.


Garnet Group Minerals:

As part of an ongoing collaboration with George Mason University, we have offered courses and summer undergraduate internships for students to develop mineral data resources and apply data science methods. In the summer of 2019 a group of four undergraduates, Kristen Chiama, Morgan Gabor, Isabella Lupini, and Randolph Rutledge, tackled building a database of garnet chemical analyses and other attributes (Chiama et al. 2020). We are now applying clustering methods to this database of more than 95,000 analyses. 

Other Mineral Groups:

In the summer of 2020, George Mason undergraduates assembled a database of almost 200,000 feldspar analyses. Other databases in development include chlorite, oxide spinel, and tourmaline. 

  • Boujibar, A., Howell, S., Zhang, S., Hystad, G., Prabhu, A., Liu, N., Stephan, T., Narkar, S., Eleish, A., Morrison, S.M., Hazen, R.M., and Nittler, L.R. (2020) Cluster analysis of presolar silcon carbide grains: Evaluation of their classification and astrophysical implications. Astrophysical Journal Letters,907, L39 (14 pp.).
  • Chiama, K., Rutledge, R., Gabor, M., Lupini, I., Hazen, R.M., Zhang, S., and Boujibar, A. (2020) Garnet: A comprehensive, standardized, geochemical database incorporating locations and paragenesis. Geological Society of America, Joint 69th Annual Southeastern/55th Annual Northeastern Section Meeting. DOI: 10.1130/abs/2020SE-344505
  • Gregory, D.D., Cracknell, M.J., Large, R.R., McGoldrick, P., Kuhn, S., Maslennikov, V.V., Baker, M.J., Fox, N., Belousov, I., Figueroa, M.C., Steadman, J.A., Fabris, A.J., and Lyons, T.W. (2019) Distinguishing ore deposit type and barren sedimentary pyrite using laser ablation-inductively coupled plasma-mass spectrometry trace element data and statistical analysis of large data sets. Economic Geology, 114, 771-786.
  • Hystad, G., Boujibar, A., Liu, N., Nittler, L.R., and Hazen, R.M. (2021) Evaluation of the classification of presolar silicon carbide grains using consensus clustering with resampling methods: an assessment of the confidence of grain assignments. Monthly Notices of the Royal Astronomical Society, in review.
  • Stephan, T., et al. (2020), in 51st Lunar and Planetary Science Conference (The Woodlands).
  • Zhang S, Morrison SM, Prabhu A, Ma C, Huang F, Gregory D, Large RR, Hazen R (2019) Natural clustering of pyrite with implications for its formational environment. Abstracts, American Geophysical Union Fall Meeting, EP23D-2284 


Several studies related to the evolutionary system are in review or in preparation.

Paragenetic Modes:

Hazen and Morrison, “On the paragenetic modes of minerals: A mineral evolution perspective.” American Mineralogist, in review.

A systematic survey of the paragenetic modes of 5659 mineral species reveals patterns in the diversity and distribution of minerals related to their evolving formational environments. The earliest minerals in stellar, nebular, asteroid, and primitive Earth contexts were dominated by relatively abundant chemical elements, notably H, C, O, Mg, Al, Si, S, Ca, Ti, Cr, and Fe. Significant mineral diversification subsequently occurred via two main processes, first through gradual selection and concentration of rarer elements by fluid-rock interactions (for example, in hydrothermal metal deposits, complex granite pegmatites, and agpaitic rocks), and then through near-surface biologically-mediated oxidation and weathering.

We find that 3349 mineral species (59.2 %) are known from only one paragenetic context, whereas another 1372 species (24.2 %) are associated with two paragenetic modes. Among the most genetically varied minerals are pyrite, albite, hornblende, corundum, magnetite, calcite, hematite, rutile, and baryte, each with 15 or more known modes of formation. 

Among the most common paragenetic modes of minerals are near-surface weathering/oxidation (1998 species), subsurface hydrothermal deposition (859 species), and condensation at volcanic fumaroles (459 species). In addition, many species are associated with compositionally extreme environments of highly differentiated igneous lithologies, including agpaitic rocks (726 species), complex granite pegmatites (564 species), and carbonatites and related carbonate-bearing magmas (291 species). Biological processes lead to at least 2707 mineral species, primarily as a consequence of oxidative weathering but also through coal-related and other taphonomic minerals (597 species), as well as anthropogenic minerals, for example as byproducts of mining (603 minerals). However, contrary to previous estimates, we find that only ~34% of mineral species form exclusively as a consequence of biological processes. By far the most significant factor in enhancing Earth’s mineral diversity has been its dynamic hydrological cycle. At least 4583 minerals – 81 % of all species – arise through water-rock interactions. 

Lumping and Splitting:

Hazen, Morrison, Krivovichev, and Downs, “Lumping and Splitting: Toward a classification of mineral natural kinds.” American Mineralogist, in review.

How does one best subdivide nature into kinds? All classification systems require rules for lumping similar objects into the same category, while splitting differing objects into separate categories. Mineralogical systems are no exception, and our work in placing mineral species within their evolutionary contexts necessitates this lumping and splitting. For mineralogical classification in the context of planetary evolution, we lump two minerals only if they: (1) are part of a continuous solid solution; (2) are isostructural or members of a homologous series; and (3) have the same paragenetic mode. A systematic survey based on these criteria suggests that 2310 (~41 %) of 5659 IMA-approved mineral species can be lumped with at least one other mineral into 667 “root mineral kinds,” of which 353 combine pairs of mineral species, while 129 lump three species. Eight mineral groups, including cancrinite, eudialyte, hornblende, jahnsite, labuntsovite, satorite, tetradymite, and tourmaline, are represented by 20 or more lumped mineral species. Lumping criteria reduce the original list of 5659 IMA-approved mineral species to 4016 root mineral kinds. 

A species may be split into two or more “mineral natural kinds” under either of two circumstances: (1) if it forms in two or more distinct paragenetic environments, or (2) if cluster analysis of the attributes of numerous specimens reveals more than one discrete combination of chemical and physical attributes. A total of 2310 IMA-approved species form by two or more paragenetic processes; however, adequate data resources are not yet in hand to perform cluster analysis on more than a handful of mineral species.

We find that 1623 IMA-approved species (~29 %) are known from only one paragenetic environment and are not lumped with another species; therefore, these IMA species are equivalent to mineral natural kinds. Greater complexity is associated with 587 species that are both lumped with one or more other species and occur in two or more paragenetic environments. In these instances, identification of mineral natural kinds may involve both lumping and splitting of IMA-approved species on the basis of multiple criteria.  

Based on the numbers of root mineral kinds, their known varied modes of formation, and predictions of minerals that occur on Earth but are as yet undiscovered and described, we estimate that Earth holds more than 10,000 mineral natural kinds.

Copper Mineral Evolution:

Morrison et al., An evolutionary system of mineralogy: Copper mineral evolution. In preparation.

This comprehensive survey of the chronological appearance of more than 750 Cu-bearing minerals will outline the full sweep of the evolutionary system with an important subset of minerals that spans the entire 4.567 billion-year history of Earth, from pre-terrestrial alteration minerals to the modern biosphere and anthropogenic phases.

Network Analysis of All Minerals:

Morrison et al. and Prabhu et al., in preparation.

Two papers in preparation will present aspects of mineral network analysis applied to all known minerals. Morrison et al. will explore embedded patterns in bipartite networks that link nodes representing minerals to nodes representing paragenetic modes. Prabhu et al. will develop an interactive, open-access platform that facilitates exploration of mineral networks with numerous filters and renderings for research and learning applications.