Ternary plot
A ternary plot, also known as a ternary diagram or triangular plot, is a barycentric coordinate system represented as an equilateral triangle that visualizes the relative proportions of three variables whose values sum to a constant, typically 100% or 1.[1] Each vertex of the triangle corresponds to 100% of one component, with interior points indicating mixtures based on their distances from the vertices, allowing for the depiction of compositional data in a two-dimensional space.[2]
The origins of ternary plots trace back to the 18th century in studies of color mixing, with early conceptual foundations laid by Isaac Newton in his 1718 work Opticks, where he explored the blending of primary colors.[3] Key developments followed, including Johann Heinrich Lambert's 1768 photometric applications and James Clerk Maxwell's 1857 experiments on color perception, which formalized triangular representations for three-part systems.[3] By the late 19th century, J. Willard Gibbs extended their use to thermodynamics in 1873, plotting phase equilibria for three-component mixtures, paving the way for broader adoption in physical sciences.[3]
Ternary plots are constructed by drawing an equilateral triangle and overlaying a grid of contour lines parallel to each side, marking percentage intervals (often in 10% increments) for each component to facilitate precise plotting and reading of data points.[2] They are widely applied in geosciences for classifying rocks, minerals, sediments, and soils based on elemental or mineralogical compositions, as well as in petrology to map phase diagrams and in environmental science for assessing water quality and geochemical trends.[2] In chemistry and materials science, these diagrams illustrate alloy compositions, polymer blends, and reaction pathways, enabling the identification of stability fields and transitions in ternary systems.[4]
Fundamentals
Definition and Purpose
A ternary plot, also known as a ternary diagram or de Finetti diagram, is a barycentric coordinate system that visualizes the proportions of three variables, typically denoted as A, B, and C, where A + B + C equals a constant value such as 1 or 100%.[1] This representation is particularly suited for compositional data in mixtures, ensuring that the variables are non-negative and their ratios sum to unity, thereby avoiding the need for negative values in traditional Cartesian plotting.[5]
The primary purpose of a ternary plot is to facilitate the analysis of relative abundances in three-component systems, enabling researchers to identify phases, detect trends, and delineate boundaries within constrained datasets.[2] In fields such as chemistry and materials science, it is used to map phase equilibria in alloys or ternary mixtures, while in geology and earth sciences, it helps classify rock compositions, soil types, or sediment proportions based on mineral or chemical end-members.[6] For instance, atmospheric gas mixtures can be plotted to assess proportional changes in components like nitrogen, oxygen, and trace gases.[7]
Key advantages of ternary plots include their compact depiction of data subject to a constant sum constraint, which simplifies the visualization of complex interdependencies that would otherwise require higher-dimensional projections.[1] They also support the overlay of contours and isolines to represent additional gradients, such as temperature or pressure, enhancing interpretive depth without expanding the graphical space.[6] The basic structure consists of three axes, each corresponding to one variable and intersecting at 120-degree angles to form an equilateral triangle, with vertices labeled for A, B, and C; points within the plot represent weighted averages of these components based on their distances from the axes.[2]
Historical Development
The ternary plot originated in the 18th century within studies of color mixing and photometry. Isaac Newton laid foundational concepts in his 1704 work Opticks, where he described triangular arrangements to represent combinations of primary colors, establishing the barycentric coordinate system implicit in such diagrams.[8] This approach was advanced by Johann Heinrich Lambert in 1770, who employed triangular representations in his Mémoire sur la partie photométrique to quantify light intensities from multiple sources, marking an early formal use of the geometry for compositional analysis.[8]
By the mid-19th century, the diagram transitioned to scientific applications beyond optics. James Clerk Maxwell's 1857 paper in the Transactions of the Royal Society of Edinburgh utilized ternary triangles to model color perception through additive mixing of red, green, and blue primaries, influencing later graphical methods in physical sciences.[8] In thermodynamics, J. Willard Gibbs applied the plot in 1873 to depict phase equilibria in ternary systems, as detailed in his Graphical Methods in the Thermodynamics of Fluids, providing a visual tool for representing compositional stability in multi-component mixtures.[8] This thermodynamic framework was extended by H.W.B. Roozeboom in 1893, whose Zeitschrift für physikalische Chemie articles introduced ternary phase diagrams for chemical systems, standardizing their use in analyzing solubility and reactions.[8]
The adoption of ternary plots in geology and petrology accelerated in the early 20th century, driven by needs in mineral and rock composition analysis. G.A. Rankin pioneered their geological application in 1915, using a ternary diagram for the CaO-Al₂O₃-SiO₂ system in his American Journal of Science study, which illustrated melting relations and phase boundaries relevant to silicate minerals.[8] Norman L. Bowen further propelled their development in the 1920s through experimental petrology, notably in his 1928 book The Evolution of the Igneous Rocks, where ternary diagrams depicted crystallization paths in systems like diopside-anorthite-forsterite, elucidating magma differentiation processes. These contributions solidified ternary plots as essential for interpreting igneous and metamorphic compositions, influencing classifications in earth sciences.
In the latter half of the 20th century, ternary plots evolved with computational advancements, enabling dynamic and interactive representations. During the 1970s, early digital tools emerged alongside phase diagram modeling software, facilitating automated plotting in geochemical research. By the 1980s, programs like TRIGPLOT (1984) allowed interactive ternary diagram generation on basic computers, enhancing accessibility for petrological data analysis.[9] Post-1980s standardization extended their utility to environmental science, where ternary plots became routine for visualizing pollutant mixtures and ecosystem compositions, such as in soil contamination assessments.[10] The shift to digital implementations in the 21st century supported interactive and multidimensional extensions, maintaining the diagram's centrality across disciplines.
Construction
Derivation from Cartesian Coordinates
Ternary plots rely on barycentric coordinates to represent the proportions of three components, denoted as a, b, and c, which satisfy the constant sum constraint a + b + c = 1 where $0 \leq a, b, c \leq 1. These coordinates express the position of a point as a convex combination of the three vertices of the triangle, weighted by these proportions. This system allows for the visualization of compositional data in a two-dimensional plane while preserving the relative relationships among the components.[11][1]
To map these barycentric coordinates to Cartesian (x, y) coordinates, the vertices of the ternary triangle are assigned specific positions in the plane, typically forming an equilateral triangle for symmetry. Standard vertex placements are: vertex A (corresponding to 100% a) at (0.5, \sqrt{3}/2), vertex B (100% b) at (1, 0), and vertex C (100% c) at (0, 0). The Cartesian position (x, y) of a point with barycentric coordinates (a, b, c) is then given by the weighted average:
\begin{align*}
x &= a \cdot 0.5 + b \cdot 1 + c \cdot 0 = 0.5a + b, \\
y &= a \cdot (\sqrt{3}/2) + b \cdot 0 + c \cdot 0 = (\sqrt{3}/2) a.
\end{align*}
Since c = 1 - a - b, these are expressed in terms of the two independent variables a and b. This derivation follows directly from the affine combination property of barycentric coordinates, ensuring the point lies within the convex hull of the vertices.[1][11]
The inverse transformation recovers the barycentric coordinates from given Cartesian (x, y) coordinates within the triangle. Using the standard vertex positions, solve for a from the y-equation: a = (2/\sqrt{3}) y. Then, substitute into the x-equation to find b = x - 0.5a = x - (1/\sqrt{3}) y, and compute c = 1 - a - b. This process assumes the point (x, y) lies inside or on the boundary of the equilateral triangle; points outside would yield negative coordinates, which are invalid for standard ternary plots. The equations arise from solving the linear system defined by the barycentric definition.[11][1]
Key assumptions underpin this mapping: the constant sum a + b + c = 1 enforces the compositional nature of the data, preventing independent variation beyond two dimensions. At the edges of the triangle, one coordinate is zero—for instance, on the base BC (y = 0), a = 0, reducing to a binary mixture of b and c. Vertices correspond to pure components (e.g., a = 1, b = 0, c = 0). These constraints ensure all valid points represent feasible proportions without exceeding 100% for any component.[11]
Plotting Techniques
Manual plotting of ternary diagrams begins with constructing an equilateral triangle using a ruler and protractor to ensure 120-degree angles at each vertex.[2] Each vertex represents 100% of one component, while the opposite side is labeled 0%; percentages decrease linearly along each side from the vertex to the base.[2] To plot a data point, normalize the three component values to sum to 100%, then draw lines parallel to the base from the appropriate percentage marks on each axis; the intersection marks the position.[2]
Adding grids involves drawing iso-composition lines, typically in 10% or 20% increments, parallel to each side to form a network of constant-proportion contours across the triangle. These lines facilitate interpolation and visualization of composition gradients; for example, lines at 10% intervals create a fine mesh for precise point placement. Overlaying data points uses standard marking techniques, while shaded regions for fields (e.g., phase boundaries) are filled between iso-lines using hatching or color gradients to denote compositional zones.
Digital methods streamline construction through specialized software. In Python, the mpltern library extends Matplotlib to handle ternary projections; a basic scatter plot is created by specifying normalized coordinates in a ternary projection.[12] Brief pseudocode for point placement is:
import matplotlib.pyplot as plt
import mpltern
fig = plt.figure()
ax = fig.add_subplot(projection="ternary")
ax.scatter([0.3, 0.5], [0.4, 0.3], [0.3, 0.2]) # Ternary coordinates (a, b, c)
plt.show()
import matplotlib.pyplot as plt
import mpltern
fig = plt.figure()
ax = fig.add_subplot(projection="ternary")
ax.scatter([0.3, 0.5], [0.4, 0.3], [0.3, 0.2]) # Ternary coordinates (a, b, c)
plt.show()
[13] In R, the ggtern package integrates with ggplot2 for ternary diagrams; initialize with ggtern(data, aes(x, y, z)) + geom_point().[14] For Excel, add-ins like XLSTAT enable direct ternary plotting from normalized data tables, automating coordinate conversion and grid addition.[15]
Variations include the right-triangle format, which folds the equilateral triangle into a right-angled coordinate system with orthogonal X and Y axes, preserving proportions but using half the space for compactness in publications.[16] Log scales are handled via isometric log-ratio transformations to address compositional data constraints, enabling analysis of ratios rather than absolute values.[14] For non-constant sums, data is normalized by dividing each component by the total of the three, ensuring the plot represents relative proportions summing to 100%.[2]
Interpretation
Reading Values
To read values from a ternary plot, begin by identifying the position of a data point relative to the three axes, each representing one component of the mixture. The proportions are determined by drawing perpendicular lines from the point to each axis, parallel to the grid lines if present; the intersection points on the axes indicate the percentage composition for that component, and these values must sum to 100% to verify accuracy.[2][17] For example, in a plot of soil texture components (sand, silt, clay), a point might yield readings of 45% sand, 30% silt, and 25% clay after normalization.[2]
The location of a point within the plot conveys the nature of the mixture. Vertices at the corners represent pure components, with 100% of one substance and 0% of the others. Points along the edges depict binary mixtures, excluding the third component entirely—for instance, a midpoint on an edge would indicate 50% of two components. Interior points signify ternary mixtures involving all three components; the centroid, or center, of an equilateral ternary plot corresponds to equal proportions of 33.3% for each.[2][18][17]
For precise extraction, rely on the plot's tick marks along the axes, which are typically spaced at intervals like 10% or 20%, allowing direct reading where lines align. Interpolation between these marks or along contour lines provides approximate values for points not exactly on a tick; for greater accuracy, trace parallel grid lines from the point to multiple axes simultaneously.[2][18] These techniques are aided by the standard contour grids used in plotting.[2]
Handling Scales and Projections
Ternary plots typically employ linear scales along the axes, where component proportions are represented as percentages summing to 100%, facilitating straightforward visualization of compositional data in fields like geology and chemistry.[2] However, for datasets with wide-ranging values, logarithmic scales are adapted through log-ratio transformations to handle non-uniform distributions and emphasize relative differences.[19] This approach, introduced in compositional data analysis, uses additive log-ratios like \ln(x_1 / x_3) and \ln(x_2 / x_3) to map proportions to unconstrained coordinates, preserving the simplex structure while accommodating skewed data.[19]
Normalization is essential when raw data do not sum to unity, achieved by dividing each component by the total and scaling to 100%, ensuring compatibility with the ternary framework regardless of absolute quantities.[1] For instance, in mineral composition studies, unnormalized weight percentages are adjusted this way to plot accurately on the diagram.[2]
Projections vary to optimize space and interpretability; the standard equilateral triangle uses barycentric coordinates, but orthogonal right-triangle projections, with perpendicular axes for two components, enhance efficiency by fitting into rectangular plots and reducing overlap in dense datasets.[16] Perspective projections, such as stereographic mappings from three-dimensional space, provide 3D-like depth for complex visualizations, though they are less common due to added complexity.[1] In probability contexts, de Finetti diagrams apply these projections to genotype frequencies or stochastic processes, treating components as probabilities summing to 1, as originally proposed for genetic equilibria.[1]
Adjustments for additional variables include isopleth lines, which contour a fourth parameter (e.g., temperature in phase diagrams) across the ternary surface, revealing trends like phase boundaries without altering the core composition.[20] Asymmetries in component importance are addressed by weighting scales or asymmetric projections, such as skewed grid lines in centered schemes, to prioritize dominant variables in unbalanced compositions.[19]
Despite these adaptations, ternary plots introduce distortions in area representation, as the projection from a 3D simplex to 2D compresses volumes unevenly, potentially misrepresenting proportional areas near boundaries.[21]
Applications
In chemistry and materials science, ternary plots, often represented as Gibbs triangles, are essential for mapping the phase behavior of three-component systems where the sum of component fractions equals unity. These diagrams facilitate the analysis of chemical mixtures by illustrating equilibrium phases, solubility limits, and transitions under controlled conditions such as temperature and pressure. For instance, in alloy design, ternary phase diagrams of systems like Fe-C-Si enable predictions of microstructures and phase stability, aiding in the optimization of steel compositions for enhanced mechanical properties.[22]
The Gibbs phase rule, F = C - P + 2 (where F is degrees of freedom, C is components, and P is phases), underpins the construction of these ternary diagrams, allowing researchers to determine invariant points and reaction boundaries in multicomponent mixtures. In materials science applications, ternary plots are used to visualize polymer blends, revealing phase separation behaviors and miscibility gaps that influence blend morphology and performance. Key interpretive tools in these diagrams include the lever rule, which quantifies phase fractions in two-phase regions by measuring relative distances along tie-lines connecting coexisting phases, and tie-lines themselves, which delineate boundaries between immiscible phases or solid solutions. Eutectic points, marking the lowest melting temperature for a ternary composition, and regions of complete solid solutions, where components form homogeneous crystalline phases, are readily identified to guide material synthesis. For example, in the design of ionic liquids, the Na-K-Cs ternary system with bis(trifluoromethylsulfonyl)amide anions exhibits a eutectic composition that lowers the melting point, enhancing fluidity and electrochemical stability for applications in energy storage.[23][24][25]
Ternary plots offer significant advantages in visualizing chemical compatibility and stability, as they compactly represent how compositional variations affect phase equilibria across temperature gradients, enabling rapid assessment of reaction pathways and material durability without exhaustive experimentation. This graphical approach supports the prediction of behaviors in complex systems, such as identifying stable regions for alloy casting or polymer processing under thermal stress.[6]
In Geology and Earth Sciences
In petrology, ternary plots play a crucial role in classifying sandstones and inferring their tectonic provenance through diagrams like the QFL (quartz-feldspar-lithics), which visualizes the relative abundances of these framework grains based on point-counting methods such as the Gazzi-Dickinson technique. The QFL diagram divides compositions into fields representing categories like quartzarenites, arkoses, and litharenites, allowing geologists to link sandstone modal data to source terrane lithologies, such as cratonic interiors or recycled orogens. For instance, sands plotting near the Q apex indicate stable, quartz-rich provenances, while those toward the L vertex suggest volcanic or sedimentary sources. This approach, refined in seminal works, enables quantitative assessment of sediment dispersal and basin evolution.[26][27]
Ternary diagrams also facilitate the interpretation of CIPW norms, a normative mineralogy calculation derived from whole-rock chemical analyses (e.g., oxides like SiO₂, Al₂O₃, and MgO), to estimate modal mineral assemblages in igneous rocks where direct petrographic modal analysis is challenging due to alteration. By plotting norms on diagrams such as the AFM (alkalis-FeO-MgO) ternary, petrologists can approximate mineral proportions like plagioclase, pyroxene, and olivine, revealing magmatic differentiation trends and comparing them to observed modes for insights into crystallization sequences. This method is particularly valuable for altered volcanics or deep-seated intrusives, providing a standardized basis for classifying rocks like basalts or granites.[28]
In sedimentology, Folk's classification system for carbonates relies on ternary proportions of allochems (grains like skeletal fragments or ooids), micrite matrix, and sparry cement or pore space to delineate rock types, such as biosparites (grain-supported with cement) or pelmicrites (mud-supported with pellets). These proportional assessments, often visualized in ternary frameworks, help interpret depositional environments by quantifying textural maturity and energy levels, with modifiers for grain size and sorting enhancing depositional reconstructions. Similarly, the USDA soil texture triangle, a classic ternary plot, categorizes sediments based on sand (2-0.05 mm), silt (0.05-0.002 mm), and clay (<0.002 mm) percentages, identifying classes like silty clay loam that influence permeability, fertility, and erosion potential in terrestrial and fluvial systems.[29][30]
Environmental applications of ternary plots in earth sciences include assessing pollutant distributions, such as heavy metal concentrations in aquatic sediments, where diagrams plot ratios of metals like Cu, Pb, and Zn against background levels to identify anthropogenic sources versus natural enrichment in riverine or coastal settings. For gas compositions in environmental samples, ternary plots of components such as CO₂, CH₄, and N₂ reveal mixing trends and emission influences, aiding in tracing biogenic versus thermogenic origins in environmental monitoring. These visualizations support pollution indexing and flux modeling in water and air quality studies.[31][32]
Ternary plots provide field-specific insights into provenance by comparing sediment compositions to end-member lithologies (e.g., volcanic, metamorphic), as in QFL or heavy mineral ternaries that quantify source contributions and transport paths in basins like the Magdalena River. In diagenesis, they illustrate evolution stages, such as plotting pore types (intergranular, dissolution, cementation) or degree of alteration in sediments to track compaction, cementation, and secondary porosity development over burial depths. Integration with GIS enables spatial ternary maps, where software like QGIS overlays compositional data on terrain models to visualize provenance gradients or diagenetic fronts across landscapes, enhancing regional interpretations of sediment routing and alteration patterns.[33][34][35]
Examples and Case Studies
Basic Plotting Example
To illustrate the basic construction and interpretation of a ternary plot, consider a hypothetical dataset consisting of three soil samples with compositions normalized to 100% total for sand, silt, and clay content.[2] Sample A has 60% sand, 30% silt, and 10% clay; Sample B has 20% sand, 50% silt, and 30% clay; and Sample C has 10% sand, 20% silt, and 70% clay.[2] These percentages represent relative proportions determined through standard particle size analysis, such as sieving and sedimentation methods.[36]
The first step in plotting is to ensure normalization, which is already satisfied here since each sample sums to 100%.[2] To derive coordinates for a point like Sample A, construct an equilateral triangle with apices labeled for sand (bottom right), silt (bottom left), and clay (top), each at 100%.[2] Overlay contour lines parallel to each side, marked in 10% increments from 0% to 100%, decreasing away from the respective apex.[2] Locate the position by drawing a line parallel to the silt-clay side (base) at 60% from the sand apex, a line parallel to the sand-clay side at 30% from the silt apex, and a line parallel to the sand-silt side at 10% from the clay apex; their intersection marks the coordinate for Sample A.[2] Repeat for Samples B and C: Sample B intersects near the silt apex region, while Sample C lies close to the clay apex.[2]
Next, draw boundaries to delineate soil texture classes based on USDA classifications, which divide the plot into 12 categories using straight lines connecting specific percentage points.[36] For instance, the boundary between sandy loam and loam runs from 50% sand/50% silt/0% clay to 20% sand/0% silt/80% clay, adjusted for clay limits below 20%.[36] Plotting the samples places Sample A in the sandy loam region (sand >52%, clay <20%, silt 28-50%), Sample B in the silty clay loam region (sand 0–20%, silt 40–73%, clay 27–40%), and Sample C in the clay region (clay ≥40%, with sand + silt ≤60%).[36]
The resulting diagram features an equilateral triangle with labeled apices, grid lines for reading percentages, and shaded or bounded areas for texture classes, with points marked for each sample and connected to show trends.[2] For example, moving from Sample A to Sample C reveals a trend of decreasing sand and increasing clay, correlating with progressively finer soil textures that influence water retention and drainage.[36] To read proportions from a plotted point, such as Sample B, trace perpendiculars or follow grid lines to the sides: approximately 20% at the sand-silt base indicates sand, 50% along the sand-clay side indicates silt, and 30% along the silt-clay side indicates clay, confirming the input values.[2]
This exercise demonstrates the transition from raw compositional data to actionable insights, such as classifying soils for agricultural suitability—sandy loams like Sample A drain quickly but hold fewer nutrients, while clays like Sample C retain water but may compact easily—highlighting the plot's utility in visualizing multivariate relationships at a glance.[36]
Notable Ternary Diagrams
In chemistry and metallurgy, the Au-Cu-Ag ternary phase diagram, first systematically mapped in mid-20th-century studies such as those compiled in NIST evaluations, delineates solidus and liquidus boundaries critical for alloy design in jewelry and electronics, revealing peritectic reactions and solid solution fields that guide heat treatment protocols.[37] The water-ethanol-benzene ternary diagram, a classic example from azeotropic distillation literature, highlights heteroazeotrope formation and residue curves that enable industrial ethanol dehydration by breaking the binary water-ethanol azeotrope through benzene addition, influencing process engineering in biofuel production.[38]
Other notable ternary diagrams include the atmospheric O2-N2-CO2 plot, employed in climate studies to track gas ratios in ice cores and air samples as proxies for paleoclimatic ventilation and carbon cycling, with deviations from modern air composition (approximately 21% O2, 78% N2, 0.04% CO2) indicating biogenic or oxidative influences.[39] Although not strictly compositional, the RGB color space ternary diagram visualizes additive color mixing in graphics and display technologies, where vertices represent pure red, green, and blue primaries, facilitating gamut mapping and perceptual uniformity in digital imaging standards.[40]
These diagrams have profoundly impacted standardization across disciplines; for instance, the International Union of Geological Sciences (IUGS) QAPF ternary frameworks, developed from Streckeisen's 1976 modal classification, unified igneous rock nomenclature globally by defining fields for over 100 rock types based on quartz, alkali feldspar, plagioclase, and feldspathoid proportions, reducing terminological ambiguity in petrologic databases and mapping.[41]