Generalized Misorientation: Extensions and Applications of Classical Misorientation

Abstract Classical misorientation, which uses the global alignment of crystal coordinate systems as its matching condition, is a fundamental crystallographic parameter for describing grain-boundary characteristics. However, for processes dependent on local feature matching—such as slip transfer, twin transfer, and variant selection—it is difficult to directly provide a criterion for their occurrence tendency. In this work, misorientation is generalized as the minimum rotation angle required to match a prescribed geometric feature between the two grains, and a generalized misorientation is proposed, described by the unified equation Dx·G1 = Rxy·Sy·G2. The generalized misorientation takes the minimum rotation angle as its geometric core; any process for which a feature-matching operator can be defined and for which the degree of feature matching has a testable bearing on the probability or energy of that process can be incorporated. Classical misorientation is the most restrictive special case under symmetry-operation constraints, while slip, twinning, and phase-transformation misorientations are the concrete criteria for the three processes of slip transfer, twin transfer, and variant selection. The framework encompasses, in a unified language of orientation matching, the ideal orientation configurations identified by interface-structure theories (e.g., special CSL coincidence states, preferred O-lattice states), uses the minimum rotation angle to characterize the geometric deviation between the actual relative orientation and various matching targets, and establishes an energetic linkage from interface energy and process barrier to process probability. In this way, it bridges theories such as the Hall–Petch relation, variant selection, and polycrystalline deformation compatibility, providing a unified crystallographic basis for the quantitative characterization of process propensity at interfaces. Programs for computing part of the generalized misorientations are available at https://www.iamyuhua.com/newsDetail/index/id/56.

Key words Generalized misorientation; Interfacial structure; Interfacial energy; Slip; Twinning; Variant selection; Hall–Petch relation

Received: 2026-09-24

E-mail: zbzhao@imr.ac.cn, 281181183@qq.com.

Misorientation is a fundamental concept in materials science, defined as the minimum angle through which the crystal coordinate systems of two grains must be rotated so that they coincide. In the Bunge notation[1], if the orientation matrices of the two grains are G1 and G2, respectively, the misorientation matrix is

ΔG = G2·G1−1, θ = min{arccos[(tr(S·ΔG) − 1)/2]}, (1)

where S enumerates the rotational symmetry operations of the crystal, and θ is the intergranular misorientation in the conventional sense.

This quantity is concise, universal, and convenient for large-scale statistical computation, and has become a standard tool in materials science, with wide applications in grain-boundary character distribution (GBCD) analysis, the coincidence site lattice (CSL) model, grain-boundary engineering, and recrystallization texture analysis. Grain boundaries are simultaneously channels for deformation transfer and obstacles to it, and are also the sites where solid-state phase transformations nucleate preferentially. When the object of analysis shifts from the statistical characteristics of grain boundaries to these specific physical processes, whether misorientation can address process-level questions must be examined case by case.

Consider variant selection during a solid-state phase transformation as an example. In polycrystalline materials, nucleation of solid-state transformations occurs preferentially at grain boundaries, and the parent and daughter phases usually follow a specific orientation relationship. For example, the β(BCC) → α(HCP) transformation in titanium alloys obeys the Burgers orientation relationship (BOR)[2]: {110}β//{0001}α and ⟨111⟩β//⟨11-20⟩α. By crystallographic symmetry, one β grain can generate twelve equivalent α variants, whereas in practice only a few of them are actually precipitated; this is variant selection. Variant selection directly affects the transformation texture, the morphology of the microstructure, and the mechanical properties, and its prediction has long relied on empirical geometrical rules. Among these, the most widely applied is the "common {110}" (common {110}β pole) criterion: if two adjacent β grains have common or nearly common {110} planes / ⟨110⟩ directions, α variants that maintain a BOR or near-BOR orientation relationship with both adjacent β grains precipitate preferentially. Such variants usually appear as colonies that traverse both sides with a single orientation[3]. Taking the alignment of close-packed planes as its rule, the "common {110}" criterion provides a qualitative judgement of "coplanarity" and involves at most the angular deviation of a single plane normal. It is therefore difficult to compare differences in variant-selection probability between different grain boundaries; in a number of cases, grains satisfying the coplanarity condition did not undergo preferential variant precipitation.

To describe and predict the propensity for phase-transformation variant selection more precisely, the phase-transformation misorientation θPT is proposed, defined as the minimum angle through which a grain must be rotated so that it can generate a daughter phase of the same orientation as that generated from the adjacent grain[4]. Let the orientation matrices of the two adjacent parent grains be G1 and G2, and the crystal symmetry operations of the parent phase be Si. The set of parent→daughter→parent round-trip transformation matrices is denoted by {Tkl}, where Tkl = Cl−1·Ck (Ck being the parent-to-daughter transformation matrix and Cl−1 the inverse daughter-to-parent transformation matrix); {Tkl}·G1 gives all parent orientations that can generate the same daughter-phase orientation as grain 1. The candidate rotation matrix is therefore Rikl = Si·Tkl·G1·G2−1, and taking the minimum rotation angle over all (i, k, l) gives θPT. This definition generalizes the judgement of "coplanarity" into a continuous angular quantity: a small θPT indicates that the two parent grains readily precipitate daughter phases of identical orientation, that the interfacial matching between variants is good, and that the corresponding interface energy is low; a large θPT indicates the opposite.

Statistics by the author on variants at β/β grain boundaries in TC25G titanium alloy show[4] that grain boundaries with a very small θPT almost invariably undergo variant selection, that the selection probability decays rapidly as θPT increases, and that only a small number of grain boundaries exhibit selection at still larger θPT. Compared with the "common {110}" criterion, θPT still gives correct predictions for several grain boundaries where the former fails. In addition, as the cooling rate increases, the θPT threshold for pronounced variant selection shifts upward accordingly, indicating that kinetic factors act together with the geometrical criterion. The construction logic of phase-transformation misorientation is not restricted to phase transformation itself, but lies in replacing the matching target from "the crystal coordinate system as a whole" to "the specific crystallographic feature on which the process depends". This also reveals the applicable boundary of classical misorientation.

Taking concrete processes such as dislocation transfer across grain boundaries, twin propagation, and variant selection as reference points, this boundary can be summarized in three respects.

First, the criterion is "alignment of the crystal coordinate system as a whole" rather than "matching of local features". Classical misorientation describes "the minimum angle through which the crystal coordinate systems of two grains must be rotated so that they coincide completely". Actual physical processes such as slip transfer, twin transfer, and phase-transformation variant selection do not require the crystal coordinate systems of the two grains to coincide completely, but only require certain specific local features (such as slip planes and slip directions) to satisfy a matching condition. Classical misorientation corresponds to the most stringent global matching, whereas physical processes correspond to specific local matching.

Second, it applies mainly to interfaces between grains of the same crystal system. The definition of classical misorientation implicitly presupposes that the two grains belong to the same crystal system. When the two grains belong to different crystal systems (for example, an HCP grain adjacent to a BCC grain), "making the crystal coordinate systems coincide completely" becomes physically meaningless: the unit cells of different crystal systems cannot be aligned by rotation alone. Classical misorientation is therefore not directly applicable to the analysis of physical processes at phase boundaries.

Third, it yields a geometrical quantity and cannot assess or predict process probability. Classical misorientation provides only the single geometrical quantity of "rotation angle" and contains no correspondence with process probability; consequently, when judging whether a specific process such as slip transfer can occur, it serves mainly as reference information and does not yet constitute a predictive criterion in itself. All three respects point to the same conclusion: if misorientation is to serve concrete physical processes, its matching target must be changed from the coordinate system as a whole to the local features on which the process depends.

In short, this generalization amounts to a shift from global alignment to local feature matching. The matching condition is relaxed from "coincidence of the crystal coordinate systems of the two grains" to "satisfaction of matching by specific crystallographic features", and the result is the framework of generalized misorientation. This relaxation can be written in a unified mathematical form. Let the orientation matrices of grains 1 and 2 be G1 and G2, and let Sy be the rotational symmetry operations of the crystal system to which grain 2 belongs. The core equation solved in generalized misorientation is then

Dx·G1 = Rxy·Sy·G2, (2)

Here Rxy is the candidate rotation matrix and Dx is the feature-matching operator, whose definition determines "which physical process is being matched". The mathematical construction of Dx is given in Appendix A: A.1 constructs the slip-system matching operator, A.2 and A.3 deal with symmetry equivalence and the local coordinate-system convention, A.4 constructs the twinning and phase-transformation operators, A.5 describes the numerical implementation, and A.6 and A.7 treat the coordinate representations and the ordering relation, respectively. When Dx takes different forms, Eq. (2) describes the matching condition for different physical processes.

The three generalized misorientation parameters share the representational logic of the "minimum rotation angle" and the computational logic of "taking the minimum over the corresponding variants and symmetry operations", and differ only in the definition of the feature-matching operator. This commonality allows them, like classical misorientation, to be subjected to distribution statistics, forming a unified system of misorientation parameters.

Eq. (2) defines a construction rule rather than a closed set of parameters. The three parameters listed in Table 1, together with classical misorientation, are merely the four most mature cases developed so far under this construction rule. The scope of the generalized misorientation framework is delimited by the following condition: a local feature-matching condition exists that can be expressed explicitly, that is, a corresponding feature-matching operator can be written down. On this basis, if the feature can be connected to a specific physical process, and if its degree of matching has a testable influence on the probability or energy of that process, the corresponding parameter can also serve as a process criterion. In principle, any object possessing an expressible feature-matching condition allows a new generalized misorientation parameter to be defined by the same logic: examples include the orientational continuity across deformation bands and subgrain boundaries, the correlation between precipitate variants and the slip systems of the parent phase, the orientation selection between recrystallized nuclei and the deformed matrix, and even channel matching across the twinning–slip mechanisms. The framework thus provides an open construction of "definition on demand, unified computation, and process matching".

The application of phase-transformation misorientation shows[4] that this feature-matching-based way of defining misorientation can be extended to other physical processes: replacing the matching target from "the same daughter phase" to "parallel slip systems" or "parallel twin systems" gives the slip misorientation θS and the twinning misorientation θT. θS is defined as the minimum angle through which a grain must be rotated so that it has slip systems parallel to those of the adjacent grain (parallel slip planes and parallel slip directions); θT is the minimum angle for two adjacent grains to generate parallel twin systems[5]. In Ref. [5], θS is termed "slip misorientation", synonymous with its usage here. Together with classical misorientation θ, the three constitute the framework of generalized misorientation described above (Table 1).

The three misorientations have recently been placed under a unified mathematical framework, and group-theoretical simplification together with quaternion algorithms allow the three parameters to be obtained in batch from EBSD data. It should be noted that none of the three parameters is defined according to the principle of minimum energy: the minimum angle does not correspond directly to a minimum energy, although it is strongly correlated statistically with process probability (see Chapter 3), and can serve as a quantitative tool for the propensity of deformation transfer and variant selection[4,5]. For ease of application, part of the visualization and computation programs accompanying the framework is openly available at https://www.iamyuhua.com/newsDetail/index/id/56.

This paper explains how the framework of generalized misorientation provides a supplementary geometrical descriptive tool for several classical theoretical problems: first, its connection with classical misorientation and with interface-structure theory (represented by CSL and the O-lattice) is discussed; next, the energetic foundation of the framework and its relation to process probability are established; and finally, its connections with the Hall–Petch relation, the theory of phase-transformation variant selection, and the theory of polycrystalline deformation compatibility are discussed. The framework retains the rotational geometrical core of classical misorientation and extends it into a criterion that can be linked to concrete physical processes.

The nature of this paper should be made clear: it is a framework-level, theoretical discourse, and no new experiments are reported. Wherever statistical regularities and experimental observations are involved, they are cited from the literature (including previously published work by the author); their role is to provide evidence for the plausibility of the framework, not to present new experimental results.

The mathematical relation between classical and generalized misorientation follows directly from Eq. (2). When the feature-matching operator degenerates to the set of symmetry operations of grain 1 (that is, Dx = Sx), Eq. (2) becomes

Sx·G1 = Rxy·Sy·G2, (3)

This is the standard form in which classical misorientation is solved, and its minimum rotation angle is the classical intergranular misorientation. Classical misorientation is the special case of generalized misorientation in which Dx takes the set of symmetry operations, corresponding to the minimum rotation when "only lattice alignment is required and no restriction is placed on features".

Two further mathematical properties of Eq. (2) should be noted. First, the candidate rotation Rxy is defined within the local crystal coordinate system of grain 2, being sandwiched between the rotational symmetry operation Sy and the feature transformation matrix; its rotation axis is therefore expressed naturally in the crystal coordinate system of grain 2, which allows the matching geometry to be read off directly in terms of plane/direction indices and to be linked with crystallographic objects such as orientation relationships and variants. The same matching condition can also be expressed equivalently in the sample coordinate system: the two versions of the candidate rotation are related by coordinate conjugation followed by inversion, and their minimum rotation angles are strictly equal. The sample-frame expression couples directly with engineering parameters such as loading direction and stress state, and connects with performance anisotropy analysis (see Chapter 7). The interconversion of the two expressions and the proof of their equivalence are given in Appendix A.6.

Second, the computation of generalized misorientation can be expressed formally as the coupling of two factors, "feature-matching operator × intergranular relative orientation". For grains of the same crystal system, classical misorientation is precisely the minimum rotation angle obtained after the intergranular relative orientation ΔG = G2·G1−1 (Eq. (1)) is reduced by the same crystal symmetry group; the unified relation of Eq. (3) is thereby manifested, and the two kinds of misorientation share the same relative-orientation core term, the only difference being whether a symmetry operation or a feature-matching operator acts upon it. For grains of different crystal systems, ΔG can still be defined; its meaning is the relative orientation transformation between the crystal reference frames of the two crystal systems (its preferred case being the orientation relationship), and the symmetry groups on the two sides are generally different, so that enumeration over both symmetry groups can be reduced to a single side only under certain conditions. The mathematical details of the trace decomposition and the reduction conditions are given in Appendix A.6.

This unified relation also brings out a characteristic of classical misorientation: when Dx takes the set of symmetry operations, matching permits the use of all symmetry operations and makes no selection of features, so that the alignment criterion is the most stringent (requiring coincidence of the lattice as a whole), and the minimum rotation angle found under this most stringent constraint constitutes an upper bound on the generalized misorientations for grains of the same crystal system (Appendix A.7).

Processes such as slip transfer, twin transfer, and variant selection require only that some pair of features (slip systems, twin systems, or daughter-phase variants) be mutually aligned, and the resulting minimized rotation angle is generally smaller: for grains of the same crystal system, the minimum angle of generalized misorientation is never greater than that of classical misorientation. The reason is that global alignment implies the alignment of all local features: the minimizing rotation that brings the two lattices closest to coincidence must simultaneously bring some slip system (twin system or daughter-phase variant) into matching, so this rotation necessarily belongs to the candidate rotation set of generalized misorientation, and the minimum angle cannot be larger (proof in Appendix A.7). The two correspond respectively to the minimum rotation when no feature selection is made, and to the most favorable matching required by the physical process. When classical misorientation is used to analyze a physical process, it must be ascertained whether the matching corresponding to this minimum angle meets the requirements of the process.

Interfaces between different crystal systems are another common situation in materials science research, but the definition of classical misorientation presupposes that the two grains belong to the same crystal system. Consider the case of an HCP grain adjacent to a BCC grain: the translational and point-group symmetries of the two lattices are both different, and no rotation exists that makes their lattices coincide. Yet slip transfer, twin transfer, and phase-transformation variant selection still occur at such phase boundaries. Generalized misorientation replaces global alignment with feature matching: in Eq. (2), Dx takes the transformation matrix that connects the feature elements of the two crystal systems, such as a slip-system transformation matrix or a parent→daughter→parent transformation matrix. Process analysis at interfaces between different crystal systems thereby acquires a continuous angular characterization parameter.

2.1 Research tradition and limitations of interface-structure theory

The geometrical understanding of interface structure has matured into a clear theoretical system[6]. The dislocation model of low-angle grain boundaries uses the Frank–Bilby equation to describe the content and distribution of geometrically necessary dislocations in tilt and twist boundaries, while for high-angle grain boundaries, atomic-structure descriptions such as the structural unit model have been developed. For special high-angle grain boundaries, the CSL model starts from bicrystal crystallography and uses rational orientation relationships to identify the special orientations (Σ values) in which parts of the two lattices coincide, explaining the low energy of special grain boundaries in terms of their coincidence density. The O-lattice theory, proposed by Bollmann[7], unifies the above descriptions within a general framework: two interpenetrating lattices form periodic matching points (O-points), the cell walls between matching points being composed of geometrically necessary dislocations that accommodate the misorientation and the difference in lattice parameters, and the interface energy growing with dislocation density. The theories in this system all answer the same question: given the orientation relationship between two crystals and the interface geometry, what structure does the interface have, and how high is its corresponding energy? They are theories about interface structure, the interface itself being the object of understanding and control.

The research tradition of interface-structure theory determines that it does not directly address process questions such as slip transfer, twin transfer, and phase-transformation variant selection. Take the interface-science system of Sutton and Balluffi as an example: it is divided into four parts—interface structure, interface thermodynamics, interface kinetics, and interface properties; the "kinetics" part also deals with the motion of the interface itself, namely interface diffusion and the conservative and non-conservative migration of the interface as a source/sink of atomic flux[6]. The transfer of dislocations across interfaces appears only in the mechanics part, in the role of an "obstacle", while twin transfer and phase-transformation variant selection lie outside its scope. This difference stems from a difference in research focus: slip, twinning, and phase transformation are processes carried by lattice defects and crystallographic variants, and their criteria must answer "whether the process can occur and how strong is its propensity", not merely "what the structure of the interface is and what its energy state is". The entry point of the generalized misorientation framework lies precisely here: it takes process features directly as the object of matching, thereby providing a crystallographic basis for the quantitative characterization of the propensity of cross-interface processes.

2.2 Shared geometrical core: the rotation angle to the matching state

The geometrical content of interface-structure theory is not limited to the orientations of the two grains, but also includes the geometry of the interface itself; restricted to the orientation-related part, it shares the same core with generalized misorientation, namely "the rotation angle from the actual orientational state to a certain matching state", the difference lying in how the matching state is defined. The target of classical misorientation is the (ideal) state of complete lattice coincidence; the targets of slip, twinning, and phase-transformation misorientations are the feature-matching states of slip systems, twin systems, and phase-transformation variants, respectively; and the special grain boundaries described by CSL correspond to a class of local lattice coincidence states. The relation between this coincidence state and the feature-matching state can be expressed in terms of a hierarchy of constraints: when only a certain crystal plane (for example, a close-packed plane) is taken as the feature, matching requires only that the normals of the feature planes on the two sides be parallel; introducing an in-plane feature direction gives a composite "plane–direction" matching; and when the constraint is extended to the complete lattice, one enters the problem of special lattice coincidence. Taking the Σ3 CSL (60° ⟨111⟩) of FCC as an example, both its θS and its θT can be zero, that is, feature matching and lattice coincidence are achieved simultaneously here; however, feature-plane matching is generally not equivalent to a low-Σ CSL, the two being matching problems at different levels of constraint. It should be pointed out that generalized misorientation characterizes the degree of matching of the orientation-related part within the interface configuration space and cannot by itself determine the complete interface structure; in this sense the two are complementary.

2.3 CSL and the O-lattice: the matching object determined by the transformation

The construction of a CSL starts from the transformation: given a transformation between two lattices, if part of the lattice can be made to coincide strictly after the transformation, these coincident lattice points form a periodic common lattice interpenetrating both sides of the lattices, whose unit-cell volume, relative to that of the original lattice, defines the Σ value. The existence of lattice coincidence strictly restricts the orientation relationship: CSLs occur only at discrete special orientations, forming a sparse set of points in misorientation space.

The O-lattice theory generalizes the CSL concept to arbitrary orientations: given a transformation between two lattices, all zero-misfit positions (O-points) can be solved for; the O-points generally do not lie on the lattices and are separated by O-cell walls, and where the interface cuts across an O-cell wall is the potential position of a dislocation. When an O-point coincides exactly with points of both lattices, the coincident lattice points constitute a CSL; the CSL is thus the special case in which the O-lattice degenerates into lattice coincidence at special orientations. The two determine the structure of the interface by the same path, namely determining the matching object from the transformation. In transformation crystallography, the O-lattice idea has been developed into a two-stage description of interface structure[8]: in the early stage, the system tends to form a "preferred state", that is, an interface structure of near-coherent atomic matching; in the late stage, when the misfit increases and the preferred state can no longer be maintained, the interface evolves into a "singular interface" structure composed of preferred-state regions together with a dislocation array. The orientation relationships corresponding to preferred states have well-defined constraint ranges: for a primary preferred state (one-to-one matching between two lattices) the degree of matching is taken within 15% of the average of the shortest latticespacings of the two phases, while the constraint range of a secondary preferred state (many-to-one matching) is much narrower. The two-stage description gives a correspondence between geometry and energy: the closer to the preferred state, the smaller the misfit, and the lower the interface energy. This accords with the construction idea in Chapter 3 of assembling the energy distribution from the neighborhoods of characteristic points: an ideal matching configuration is a characteristic point in the configuration space, and deviation from that state corresponds to the accumulation of the corresponding defect energy.

The O-lattice also shares the same geometrical core with generalized misorientation, namely the search for crystallographic matching relations, but the direction and the matching object differ. The O-lattice determines the matching object from the transformation: given the transformation matrix of two lattices, the positional distribution of O-points is solved for, and the interface dislocation network is derived from the periodicity of the matching. Generalized misorientation determines the transformation from the feature: given the feature alignment required by the physical process (parallel slip planes and parallel slip directions, coincidence of twinning elements, identical variant orientations), the minimum rotation angle required to achieve that alignment is solved for inversely. The two constitute two directions of the same geometrical problem, the difference being concentrated in the matching object: both classical misorientation and the O-lattice take the lattice as a whole as their object, one seeking the alignment transformation and the other the zero-misfit positions; generalized misorientation replaces the matching object with process features. This replacement follows the same logic as the generalization of "global alignment → feature matching" in Chapter 1: replacing the lattice as a whole in the criterion with feature elements directly coupled to the process. The distinction of object is not a mere mathematical construct but has a physical origin. As for the geometrical compatibility condition for dislocations crossing an interface, the dislocation couples with the interface through the slip-plane normal and the Burgers vector, and has no direct relation to the misfit of the lattice as a whole (see "the physical-origin level" in Chapter 3); a process criterion must therefore take the feature elements of the defect as its matching object.

2.4 CSL deviation angle θΣ and process misorientations

The deviation from a CSL can be measured in the same angular language, namely the CSL deviation angle. The above understanding shows that generalized misorientation and CSL/O-lattice do not replace one another. For a given Σ CSL, let the set of all ideal equivalent orientations after considering crystal symmetry be CΣ, and let the actual relative orientation of the grain boundary be ΔG; the CSL deviation angle can then be defined as θΣ = dist(ΔG, CΣ). Taking Σ3 as an example, a grain boundary lying strictly at this CSL orientation has θΣ = 0, and the deviation angle of an actual boundary from this ideal orientation can be calculated directly from this expression; what the Brandon criterion provides, by contrast, is an empirical tolerance for identifying near-CSL boundaries, and the two should be distinguished. This deviation angle shares the same minimum-rotation geometrical core as Eq. (2): θS, θT, and θPT target the corresponding process feature-matching states, whereas θΣ targets the set of ideal Σ CSL equivalent orientations. Unlike θS, which requires only the matching of specific slip planes and slip directions, θΣ measures the deviation of the actual orientation from the state of periodic coincidence of the complete lattice, and can therefore generally not be reduced significantly by selecting a few local features. Both kinds of deviation follow the relation of "ideal matching state — geometrical deviation — defect accommodation": the deviation of a near-CSL boundary is accommodated by secondary grain-boundary dislocations (SGBDs) whose Burgers vectors belong to the DSC lattice, while the incomplete matching of slip systems is characterized by the residual Burgers vector after dislocation transfer. It should be pointed out that both θΣ and θS retain only the rotation angle and contain no information on the deviation axis, the boundary plane orientation, or the misfit vector; they are therefore suitable for the statistical characterization of the degree of matching, whereas the structure of a specific interface must still be analyzed in combination with the above geometrical and defect information.

2.5 Unified form θχ and the open set of characteristic points

The above understanding can be written in a unified form. Let the target orientation set of a certain matching condition χ be Mχ, and the actual relative orientation between grains be ΔG; then θχ = min drot(ΔG, Q) (Q ∈ Mχ) can be defined formally. Here Q is the target orientation satisfying the corresponding matching condition, and drot is the angular distance in rotation space. For classical misorientation, Mχ is composed of the symmetry-equivalent coincidence states of the complete lattice; for θS, θT, and θPT, Mχ is determined respectively by the matching conditions of slip systems, twin systems, and phase-transformation variants; for CSL, the special coincidence state of the complete lattice of a given Σ type can be taken as the target set; and the preferred states corresponding to the O-lattice, together with other explicitly expressible geometrical matching configurations, can likewise constitute new target sets. Each class of ideal matching state can be regarded as a class of characteristic points or characteristic subsets in the interface configuration space, and the corresponding θχ is the local coordinate describing the deviation of the actual configuration from that characteristic point (Fig. 1).

This set of characteristic points is open. In addition to the currently known slip systems, twin systems, phase-transformation variants, special CSL coincidence states, and O-lattice preferred states, other geometrical matching configurations with clear physical significance that are discovered in the future can also be incorporated into the unified equation by defining corresponding matching operators and target sets. The framework of generalized misorientation therefore amounts not merely to a summarization of existing angular parameters, but also provides a method by which the description of interface configurations can be extended continuously: identify new ideal matching states, establish their deviation angles, and analyze the structure, energy, and process behavior in the neighborhood of that characteristic point. A comparison of the various matching states with their deviation angles and accommodation defects is given in Table 2.

As the target set, Mχ is itself composed of all candidate matchings, and θS is merely the nearest distance from the actual configuration to this set. Each candidate matching can in principle serve as an independent characteristic point, its plausibility deriving from a determinate connection with a physical process (slip-system matching corresponds to slip transfer), rather than being premised on achieving the minimum angle or on coinciding with a CSL coincidence state. The process behavior of an interface is determined jointly by the set of channels closest to the actual configuration; what "determining the transformation from the feature" provides is the complete set of candidate transformations, not a single optimal transformation. This accords with the construction idea of the O-lattice, which retains all zero-misfit positions (O-elements) rather than a single optimal matching.

2.6 Interface configuration space and information boundary

For two interface configurations A and B sharing a common matching feature, their transformation matrix QAB retains complete information such as the rotation angle, the rotation axis, and the feature correspondence, and cannot be replaced by a single scalar angle; a common feature is a necessary condition for establishing a geometrical link between the two, but the interface energy cannot be determined uniquely from θχ alone. In addition to orientation, interface structure includes the boundary plane orientation as well as information such as dislocation defects and chemical state. The θχ obtained from the unified equation characterizes mainly the feature-matching part of these, depicting the geometrical distance to the specified matching set rather than the actual interface energy itself. This geometrical distance can, however, serve as a coordinate for establishing local energy relations, used only to measure the barrier that must be overcome to reach the target local feature-matching state; there is no universal one-to-one correspondence between it and the actual interface energy.

Local energy relations are established with characteristic points as their base, and their coverage depends on the distribution density of characteristic points in the configuration space. From the viewpoint of the hierarchy of constraints (Section 2.2), the classical CSL and O-lattice models determine the matching object through correspondence of the complete lattice; lattice coincidence occurs only at discrete special orientations, forming sparse characteristic points in the configuration space, and the vast configuration regions between them have no local coordinate available. The framework of generalized misorientation relaxes the matching condition to the process-related composite "plane–direction" matching, which is a lower level of constraint: the candidate matchings of processes such as slip, twinning, and phase transformation can all serve as independent characteristic points (Section 2.5), and their distribution in configuration space is far denser than that of lattice coincidence states, so that the originally blank configuration regions likewise acquire local deviation coordinates. Traditional matching states and process feature-matching states are complementary, and the distribution of characteristic points changes from sparse to dense.

The interface configuration space corresponding to the framework of generalized misorientation is thereby constructed. The ideal matching states identified by interface-structure theories such as CSL and the O-lattice, together with their locally deviating configurations, can all become characteristic points in this space, filling the blank regions between the characteristic points corresponding to existing parameters such as θS, θT, and θPT, and providing a geometrical basis for constructing the interface energy distribution from local matching neighborhoods in Chapter 3. The structural unit model assembles general interface structures from special coincidence interfaces as its base, and can be regarded as the prototype of this picture at the structural level; generalized misorientation further extends the base to process feature-matching states such as slip, twinning, and phase transformation.

Interface-structure theory provides a microscopic mechanism for the connection between classical misorientation θ and interface energy γ, and the physical validity of generalized misorientation is built precisely on this mechanism, extending this geometry–energy connection to the matching problems involved in slip, twinning, phase transformation, and special interface configurations such as CSL and the O-lattice. This chapter is developed along the following levels: at the energy level, θ serves as a representation of interface energy, while generalized misorientation serves as a representation of the process barrier; at the structure–function level, the statistical correlation of θ and generalized misorientation with physical processes (taking slip and twin transfer as examples); at the physical-origin level, the microscopic mechanism underlying the above correlation; and at the application-scope level, the extension to phase boundaries between different crystal systems. Among these, the energy level is the juncture of the two theories in Chapter 2 and also the energetic foundation of the arguments in the subsequent chapters on the Hall–Petch relation (Chapter 4), phase-transformation variant selection (Chapter 5), and deformation compatibility (Chapter 6).

3.1 Energy level: dislocation geometry and process barriers

At the energy level, under the small-angle approximation, the O-lattice dislocation spacing ℓ and θ satisfy the classical relation for low-angle grain boundaries[9]:

ℓ = |b| / (2·sin(θ/2)) ≈ |b| / θ, (4)

Here |b| is the magnitude of the Burgers vector of a dislocation. θ is inversely proportional to the dislocation spacing, and both the dislocation density and the interface energy grow with θ. For a physical process, the misfit on the two sides of the interface that is truly relevant to the process is not the misfit of the lattice as a whole, but the misfit of the process-related optimal local feature. For example, a transfer process concerns the matching of the optimal slip-system pair: a dislocation crossing an interface generally leaves a residual dislocation at the interface, whose residual Burgers vector is br; the optimal slip-system pair whose slip directions enclose an angle κ satisfies |br| = 2|b|·sin(κ/2) ≤ 2|b|·sin(θS/2) ≈ |b|·θS (κ ≤ θS). θS is the minimum rotation angle that makes the slip systems parallel (both the slip plane and the slip direction simultaneously), and the upper bound on the residual dislocation content converges accordingly; the elastic energy of residual dislocations is proportional to |br|2, so that the transfer barrier grows with θS. In other words, Eq. (4) describes the density of the interface's own dislocation network (the structural side), whereas θS describes the residual dislocation content required for a dislocation to cross the interface (the process side); the two are applications of Frank–Bilby dislocation geometry to different objects. Accordingly, θS belongs to the same family of minimum-rotation-angle parameters as the θ of Eq. (4), as do θT and θPT. They share the rotational geometrical core but differ in the matching object; for grains of the same crystal system, the candidate set of local matching contains the classical minimum rotation, so that θS, θT, and θPT never exceed θ (Appendix A.7).

For a specific physical process P, the θ in Eq. (4) can be replaced by the process-related minimum θmin; the process-related parameter set is defined as ΘP = {θi | θi is related to the features of P}. When the baseline barriers En(z) of the various θi in ΘP at their ideal matching states are approximately equal, θmin = min(θi ∈ ΘP) can characterize the lowest-barrier channel. θmin depends on the process and its matching condition; for example, when no phase transformation occurs, θPT should not be included in the ΘP of a slip or twinning process. The introduction of the unified notation θmin facilitates the description of the lowest-barrier channel in competition between multiple processes (for example, when the difference in baseline barriers between slip and twinning is negligible, min(θS, θT) can be taken for a grain boundary where both slip and twinning can occur); the composition of ΘP is determined by the feature-matching condition on which the process depends, for example, a slip process takes ΘP = {θS}.

The process-related energy corresponding to θmin can be estimated by analogy with the Read–Shockley formula, which describes the variation of low-angle grain-boundary energy with the content of geometrically necessary dislocations[9]: as θ increases, the interface dislocation content accumulates and the interface energy grows accordingly. On this basis, the energy scale corresponding to θmin can be written as

Δγ(P) = γ0θmin(A − ln θmin) (θmin > 0); Δγ(P) = 0 (θmin = 0), (5)

Here θmin is measured in radians (dimensionless); A is a process-related effective parameter, formally corresponding to the coefficient in the Read–Shockley formula, whose value must be determined separately for each process and is not carried over from grain-boundary energy parameters; the expression applies to the small-angle range in which the dislocation description is valid (θmin ≲ 15°). For grains of the same crystal system, when ΘP contains only feature-matching parameters (θS, θT, θPT), θmin ≤ θ always holds (Appendix A.7). The nature of Eq. (5) should be made clear: it is an analogical form rather than a derived result, providing only an energy scale that increases monotonically with θmin and vanishes as θmin → 0, and it is not used for the numerical fitting of interface energies. θmin thus reflects the additional barrier that must be overcome for the process to occur rather than the interface energy itself; a small θmin means that the interface readily forms the preferred-state structure for the specific process, the activation barrier is generally low, and this corresponds statistically to a higher process probability. Interface-structure theory and generalized misorientation are thereby joined at the energy level: the former provides a microscopic mechanism for the θ → γ relation, in which the density of geometrically necessary dislocations determines the elastic energy and special coincidence configurations determine the energy minima; the latter turns this mechanism into the physical basis of a process criterion, its applicable boundary likewise being delineated by structural understanding and process features. The position of this energy scale within the interface energy distribution is given in Eq. (6).

It should be noted that, because the Read–Shockley formula is established on the basis of dislocation-dominated low-angle interfaces, the description of energy by Eq. (5) remains limited. At θmin = 0, Δγ(P) vanishes; however, the equation gives only the additional energy scale arising from the characteristic rotational deviation, and not the total interface energy or the total process barrier; the latter two remain influenced by factors such as the boundary plane orientation, atomic arrangement, interface translation, chemical state, and local defect reactions, and do not generally vanish together with θmin. Therefore, to remedy the shortcoming of Eq. (5) in describing energy solely through rotational deviation, θmin must be placed in a high-dimensional configuration space containing other interface-structure variables, and the ideal matching state must be regarded as a characteristic point with a non-zero baseline energy or baseline barrier.

3.2 Interface configuration space and energy assembly

The structure of a real interface involves the orientation relationship, the boundary plane orientation, and the chemical state, among others. The interface configuration space constructed in Chapter 2 by extending interface-structure theory within the framework of generalized misorientation can accommodate the above structural factors, with the orientation relationship ΔG serving as the principal independent variable of the present framework. Ideal slip-system matching, ideal twin-system matching, ideal phase-transformation variant matching, special CSL coincidence states, and the preferred states corresponding to the O-lattice can all be regarded as known characteristic points or characteristic subsets in this space. Each characteristic point n corresponds to one kind of ideal geometrical matching, and the rotational deviation of the actual configuration from that characteristic point is described by the generalized misorientation θn.

Each characteristic point provides only a local energy baseline for its neighborhood. Let the energy at characteristic point n be En; then, within its applicable neighborhood, the local energy after deviation from this matching state can be expressed generally as En(θn, z) = En(z) + ΔEn(θn), where the deviation term satisfies ΔEn(0) = 0; when the deviation cost is approximately proportional to the baseline energy, the multiplicative form ΔEn(θn) = En(z)·fn(θn) may be adopted, that is, En(θn, z) = En(z)[1+fn(θn)]. Here the physical meaning of the local parameter function fn(θn) depends on the specific matching feature of that parameter, and it satisfies fn(θn) ≥ 0 and fn(0) = 0; z denotes the other state variables not fixed by θn, such as the boundary plane orientation, the chemical state, and the defect state. This expresses that, for a given z and within the applicable range of that local parameter, the ideal matching state is the energy baseline of the parameter, and the accommodation defects produced by deviating from it increase the energy. En is itself not a constant, but a function or distribution of the remaining state variables; when a single scalar is used to represent En, it is only a representative value after order reduction of these degrees of freedom. It must be emphasized that fn(0) = 0 means only that the deviation energy relative to that ideal matching state vanishes at θn = 0, and does not imply that En = 0. Therefore, even when parameters such as θS, θT, θPT, or θΣ are zero, the energy or process barrier of the corresponding interface is generally still non-zero. It is thus evident that ideal orientation matching does not uniquely determine the interface energy.

The root of this conclusion lies in the structure of the degrees of freedom of an interface. Ideal orientation matching fixes only the relative orientation of the two grains and imposes no constraint on the boundary plane orientation, so that the same ideal matching state corresponds to interfaces of vastly different energy at different boundary plane orientations[6]: the coherent {111} twin boundary and the incoherent tilt boundary under a strict Σ3 misorientation (60° ⟨111⟩) are typical examples, the two having the same misorientation while their interface energies differ by more than an order of magnitude[10]. In addition, even when the geometry matches perfectly, the interface still retains non-geometrical energy components such as dislocation-core–interface interactions, interface steps, and chemical state; this is the physical reason why the baseline energy En must retain its dependence on the variable z.

The local energy branches of different characteristic points can be assembled within their respective neighborhoods to give the overall energy distribution, which can be expressed formally as:

E(ΔG; z) = ∑n w̃n(θn)·[En(z) + ΔEn(θn)], ΔEn(0) = 0, (6)

Here θn ≡ drot(ΔG, Mn) is the angular distance from the actual configuration to characteristic point n, and Mn is the orientation matrix corresponding to characteristic point n (the corresponding target orientation in the target set Mχ of Section 2.5); w̃n(θn) = wn(θn)/∑mwm(θm) is the normalized local weight (wn takes larger values near characteristic point n and decays gradually beyond the corresponding neighborhood), the normalization ensuring that no systematic underestimation arises in regions where characteristic points are sparse; coincident or nearly degenerate characteristic points should be counted together. Eq. (6) is not an energy law with all parameters already specified, but an open construction framework: each ideal matching state provides one local parameter with a clear crystallographic meaning, and the various parameters jointly cover the configuration space within their respective effective neighborhoods; since the branches are separable on the configuration space, their parameters can all be determined independently from a single experiment or simulation near the corresponding characteristic point, and this is where the testability of the framework lies. As new important characteristic points and their parameters are identified (in Eq. (6) the characteristic point n can also be refined to the "channel level", see Section 2.5), the assembled result can continue to approach the true interface energy distribution.

Within this framework, Eq. (5) can be regarded as the additive branch of Eq. (6) at the identity matching point: the Read–Shockley-type deviation term (En = 0, ΔEn = γ0θn(A − lnθn)); the identity matching point corresponds to the disappearance of the interface (a single crystal) and is the absolute zero of energy, while the remaining characteristic points have En > 0 and are local energy minima. The quadratic convergence of the single-transfer barrier corresponds, in turn, to the branch with ΔEn ∝ θn2 in the neighborhood of a process feature-matching point. Both can be understood from the perspective of "accommodation defect content × single-defect energy": the former corresponds to the case where the energy of a single dislocation changes slowly and the dislocation density increases with the deviation angle; the latter corresponds to the case where the transfer channel is given and the residual Burgers vector vanishes approximately linearly with the deviation angle. The effective neighborhood of each local energy branch depends on the range of applicability of the corresponding accommodation defect and structural model; for example, the near-CSL branch can refer to the order of the Brandon tolerance, while the process branch based on a discrete dislocation description is generally limited by dislocation-core overlap beyond about 15°[6].

The relations between different types of characteristic point can be illustrated by a concrete example. Taking the FCC {111}⟨110⟩ slip systems as an example, enumeration over all slip systems shows that the θS of Σ3 (60° ⟨111⟩) is 0, whereas the θS values of Σ7, Σ9, and Σ11 are approximately 21.79°, 31.59°, and 20.05°, respectively, and the θS of Σ5 (36.87° ⟨001⟩) is the same as the classical misorientation. This shows that a CSL characteristic point can coincide with a slip-matching characteristic point only when the CSL rotational relation is compatible with the characteristics of the slip systems; in general, the two describe different types of ideal matching. Conversely, for certain special CSL interfaces, both θ and θS may be large, yet dislocations can still cross the interface relatively easily. This may originate from the special interface configuration itself providing a lower transfer barrier, while the deviation of that configuration from the ideal CSL state is small, that is, the θΣ defined in Chapter 2 is small. Accordingly, incorporating the ideal matching configurations identified by structural theories such as CSL and the O-lattice extends the range of interface configurations covered by the framework of generalized misorientation, and can remedy biases that may arise when judgements are based only on overall lattice orientation or on process feature matching.

3.3 Structure–function correlation and physical origin

At the structure–function level, interface-structure theory gives the structural characteristics of the interface (dislocation density and distribution), while generalized misorientation provides a functional criterion that can be used directly. Earlier statistical work (published separately) on equiaxed commercially pure titanium (TA1), taking grain boundaries as the unit of analysis, shows that the slip-transfer probability decreases in steps as θS increases: the transfer probability is about 94.3% for θS in the range 0°–5°, falls to about 53.4% in the range 5°–10°, and further to about 17.1% and 10.9% in the ranges 10°–15° and 15°–20°, respectively; the twin-transfer probability shows a similar decreasing trend with θT, exceeding 85% when θT ≤ 5°, falling to about 64.3% and 31.5% in the ranges 5°–10° and 10°–15°, respectively, and to less than 10% beyond 15°. The same statistics also revealed that the formation of some twins can promote the slip-transfer process between grains. This paper reports no new experiments; these results are cited only as evidence for the validity of the framework.

Data in which a purely geometrical parameter gives the process probability directly are content not addressed by interface-structure theory. At the physical-origin level, the stepped nature of process probability can be explained by a three-stage amplification chain. The first stage is the topological origin: a dislocation is a topological defect defined by its Burgers vector, and the Burgers vector is a discrete lattice vector. Slip transfer is, in essence, the preservation of the identity (defect identity) of the dislocations on the two sides of the interface, that is, when a dislocation crosses the interface, no residual dislocation is added to the interface by the transfer in the sense of Burgers vector conservation (br = 0); this identity holds only when the two sets of slip systems coincide strictly, and θS is precisely the rotation angle to that coincidence state, so that a small θS means that the interface lies in the immediate vicinity of the strict coincidence state. The second stage is the energetic quadratic convergence: at small angles, the content of geometrically necessary dislocations at the interface is proportional to the misorientation, the residual Burgers vector of the optimal slip-system pair vanishes linearly with θS, and the elastic energy of residual dislocations is proportional to the square of its magnitude, so that the transfer barrier converges quadratically with θS, that is, the angular difference is converted into a squared order of magnitude in energy. The third stage is the exponential response of probability to energy: whether by thermal activation (P ∝ exp(−ΔE/kT)) or by an athermal threshold criterion, the process probability is exponentially sensitive to the barrier height. The compounding of the three stages amplifies small changes in θS into stark differences in probability, and the high transfer probability of boundaries with small θS and the low transfer probability of boundaries with large θS in the earlier statistical work are the results of this mechanism. Here, the compounding of quadratic convergence and exponential response also gives a testable functional form: the transfer probability should decay in Gaussian fashion with θS, P ∝ exp(−cθS2) (c being a positive constant), which can presumably be compared directly with the statistically obtained transfer-probability distribution.

There are three further arguments. First, taking the minimum of θS over all variant pairs is not merely a mathematical convenience: a dislocation can cross the interface through multiple channels, and transfer occurs as long as one of them has a sufficiently low barrier, the event being dominated by the lowest-barrier channel; taking the minimum is precisely the geometrical expression of this extreme-value statistics, and θS therefore carries the principal information about process probability. Since the transfer process is mediated by defects, the dislocation couples with the interface only through the slip-plane normal and the Burgers vector, and does not depend on the misfit of the lattice as a whole; this explains why the "slip-system matching distance" θS correlates more strongly than the classical "lattice matching distance" θ. In particular, at large classical misorientations, θS can still be small, and this is precisely the main situation in which θS provides incremental information relative to θ (see Chapter 1). Second, the above arguments at the energy and probability levels all presuppose that the interface can still be described by a dislocation array, a description that fails beyond about 15° owing to dislocation-core overlap[6]. Its window of applicability and the effective window of the interface dislocation description arise from the same physical premise, and the θS threshold corresponding to the marked statistical decay of transfer probability coincides with the boundary of this window; the decay of predictive power at the high-angle end is a manifestation of the applicable boundary, not an internal contradiction of the framework. Third, a small θS is only a necessary, not a sufficient, geometrical condition for transfer to occur; the resolved shear stress must still reach its critical value, mobile dislocation sources must be available, and the boundary plane orientation must also act in concert. The probability P(transfer | θS) is therefore in fact the result of statistical averaging over hidden variables such as the local stress state and the interface normal: at the small-angle end, no combination of hidden variables can usually block transfer (P→1), while at the large-angle end transfer is mostly difficult (P→0), and the shape of the sigmoidal curve in the intermediate range is itself a measurement of the distribution of hidden variables. The misorientation spectrum of Chapter 7 (that is, a spectrum taking candidate channels or candidate variants as discrete dimensions and the corresponding misorientation angles as values) can be used to resolve this curve layer by layer.

3.4 Scope of application and connections between chapters

At the scope-of-application level, feature matching allows the analysis of generalized misorientation to extend beyond grain boundaries between grains of the same crystal system to phase boundaries between different crystal systems. An earlier analysis of slip transfer at α/β phase boundaries in TC18 titanium alloy (published separately) is a case of applying θS to interfaces between different crystal systems: the values of θS between the adjacent α1 and α2 grains within the same β grain and the β phase are 2.3° and 18.8°, respectively; the {110}⟨111⟩ slip activated in the β phase crosses the α1/β phase boundary, where θS is small, smoothly and continues to move by {10-11}⟨11-20⟩ slip within the α1 grain, whereas it is markedly obstructed at the α2/β phase boundary, where θS is large.

At this point, the energetic foundation of generalized misorientation has been set out systematically: at the level of interface configuration, the ideal matching states serve as characteristic points, and the energy branches corresponding to their local parameters jointly constitute the overall energy distribution; at the level of specific processes, the lowest-barrier channel is identified from among the candidate channels satisfying the physical constraints, and is characterized by the minimum generalized misorientation under that process constraint. Chapters 4–6 apply this foundation to three parallel problems: Chapter 4 introduces the θS distribution into the Hall–Petch relation, supplementing the empirical coefficient of that relation with crystallographic information; Chapter 5 links θPT with the interface-energy criterion of variant selection; and Chapter 6 discusses multi-mechanism deformation compatibility in combination with the three process-related misorientations. The three chapters deal with different objects and present parallel arguments, sharing the geometry–energy connection established in this chapter. The division of labor between the two levels is clear: energy assembly answers "how the interface energy distribution is constructed", while taking the minimum answers "along which channel the actual process occurs"; the minimum under restrictive conditions (for example, counting only activatable variants) is generally larger than the theoretical minimum, yet closer to the actual process.

The Hall–Petch relation is one of the classical empirical relations describing the relationship between grain size and yield strength in polycrystalline materials[11,12]:

σs=σ0+Kd−1/2, (7)

Here σs is the yield strength, d the average grain size, σ0 the resistance to deformation within the grain, and K the Hall–Petch coefficient. This relation indicates that, within a certain range of grain sizes, the yield strength usually increases as the grain size decreases. The classical Hall–Petch relation uses the average grain size to characterize the scale effect of grain-boundary strengthening, and the differences between different grain boundaries in crystallographic characteristics and dislocation-transfer ability are contained mainly in the empirical coefficient K. K is not determined by the grain size, but is an empirical parameter that varies with the material system, the microstructural state, temperature, and strain rate; d, by contrast, characterizes the scale effect of grain-boundary spacing, and the two play different roles in Eq. (7). The obstruction of dislocation motion also differs between different types of grain boundary: for example, low-angle grain boundaries generally have good slip continuity, and some high-angle grain boundaries may also have a high propensity for slip transfer because well-matched slip systems exist on the two sides. Therefore, describing the interaction between grain boundaries and dislocations using only the average grain size or the classical misorientation does not yet adequately reflect the geometrical constraint that grain boundaries impose on slip transfer.

The Hall–Petch relation describes the macroscopic yielding behavior of polycrystalline materials, whose microscopic process is the transfer of slip through the grain-boundary network; the yield strength reflects precisely the obstruction that grain boundaries offer to this transfer. The slip misorientation θS of generalized misorientation provides a process-related crystallographic parameter for this purpose. When θS is small, at least one set of geometrically well-matched slip systems exists on the two sides, and the deviations of the slip directions and slip-plane normals are small. In conjunction with the discussion of residual Burgers vectors in Chapter 3, such interfaces generally correspond to small residual Burgers vectors, so that the geometrical obstacle that dislocations must overcome in crossing the grain boundary is relatively low. Existing statistical results also show that the slip-transfer probability decreases overall as θS increases (see Sections 3.3 and 3.4). Accordingly, θS can serve as a parameter describing the geometrical conditions for slip transfer at grain boundaries, but it must not be equated with the actual grain-boundary strength.

A small θS indicates that slip channels with good geometrical continuity are relatively easy to find on the two sides of a grain boundary, which agrees with the experimental understanding that low-angle grain boundaries generally have a high capacity for slip transfer. Conversely, a large classical misorientation does not imply that θS is necessarily large; some high-angle grain boundaries may still contain well-matched slip systems. Thus, relative to classifying grain boundaries into low-angle and high-angle solely by classical misorientation, θS provides geometrical information more directly relevant to the specific slip-transfer process.

On this basis, the possible connection between the θS distribution and the Hall–Petch coefficient can be discussed. In the classical dislocation pile-up model, Hall–Petch strengthening originates from the obstruction of dislocation motion by grain boundaries, and the critical transfer resistance of different grain boundaries is related to their degree of crystallographic matching. Therefore, on the basis of this model, the macroscopic Hall–Petch coefficient can be expressed as the statistical average of the grain-boundary slip-misorientation distribution:

K2 = CG|b|∫f(θS) τ̅c(θS) dθS, (8)

Here f(θS) is the slip-misorientation distribution in the sample, obtainable by statistics from EBSD data (normalized so that ∫f(θS)dθS = 1 over the integration domain [0, θS,max], where θS,max is determined by the family of slip systems of the crystal system); τ̅c(θS) is the statistical average of the critical grain-boundary resistance under a given θS, whose intercept satisfies τ̅c(0) = τ0 > 0—even when the geometry matches perfectly, non-geometrical factors such as dislocation-core–interface interactions, interface steps, and chemical resistance still provide a baseline resistance; G is the shear modulus; and C is a proportionality coefficient containing geometrical factors, including the Schmid/Taylor orientation factor, the statistical independence of the orientation factor and the θS distribution being adopted here as a zeroth-order assumption. This equation introduces the degree of grain-boundary slip-system matching as a crystallographic factor influencing K. In strongly textured materials, the orientation-factor distribution is correlated with the grain orientations and is not necessarily statistically independent of the θS distribution; in this case the treatment can be generalized to the joint distribution of the orientation factor and θS, and the form of Eq. (8) remains unchanged. K can thereby be understood as a macroscopic statistical parameter resulting from the combined action of the material system, the interface structure, and the slip-transfer characteristics of the grain-boundary network. It should be pointed out that when θS is large (above about 15°) the dislocation-array description fails, and τ̅c(θS) converges numerically to the resistance of a random high-angle grain boundary τsat; one specific form satisfying the above limits is τ̅c(θS) = τ0 + (τsat − τ0)·[1 − exp(−θS2/2σ2)], which gives, within the small-angle window, K2 ≈ C·G·|b|·(τ0 + α⟨θS2⟩) (α = (τsat − τ0)/2σ2), which can be tested directly by EBSD statistics. In addition, Eq. (8) corresponds physically to a first-moment weighting of the grain-boundary resistance, while the percolation effect of the connectivity of the grain-boundary network (that is, the formation of continuous paths for slip transfer) is regarded as a second-order correction.

The two scales of Hall–Petch strengthening are thereby distinguished: d reflects mainly the grain-boundary spacing, that is, the spatial scale on which the mean free path of dislocations is limited; f(θS), by contrast, reflects differences in the geometrical conditions for slip transfer within the grain-boundary network. Therefore, among materials with similar average grain sizes, if the grain-boundary θS distributions differ markedly, their effective grain-boundary obstruction and the corresponding Hall–Petch coefficients may also differ.

The above connection remains a crystallographic description proposed on the basis of the grain-boundary slip-transfer mechanism, rather than a universal re-derivation of the Hall–Petch relation. θS reflects mainly the degree of geometrical matching of the slip systems on the two sides of a grain boundary, whereas actual dislocation transfer is also influenced by factors such as the boundary plane orientation, the local stress state, dislocation sources, the grain-boundary structure, and solute segregation; the specific form of τ̅c(θS) in Eq. (8) must therefore still be determined by experiment or simulation. In addition, the main supplement that generalized misorientation offers to the Hall–Petch relation lies not in changing its d−1/2 scaling, but in introducing, for the empirical coefficient K, a crystallographic description that can be obtained directly from EBSD; the core relation can be summarized as: θS distribution → grain-boundary slip-resistance distribution → K. Strictly speaking, f(θS) should be taken as the distribution of grain boundaries actually traversed by slip bands (weighted by the frequency of traversal); the distribution after screening activatable variants in combination with mechanical conditions is a direction that can be further developed in Chapter 7.

During a solid-state phase transformation, the parent phase usually forms multiple crystallographically equivalent daughter-phase variants following a specific orientation relationship. Taking the β→α transformation in titanium alloys as an example, the Burgers orientation relationship gives twelve possible α variants. The crystallographic relations among the orientation relationship, the boundary plane orientation, and growth steps in such β/α transformations have already been studied fairly systematically[13]. When only crystallographic equivalence is considered and no local constraint is introduced, all variants have the same probability of formation; in real materials, however, variant selection is widespread, that is, some variants form preferentially while others appear less frequently. Variant selection affects the transformation texture, the morphology of the microstructure, and the anisotropy of the material. Its essence lies in differences between variants in nucleation work and growth conditions, the influencing factors generally including the local stress state, the parent-phase texture, the interface energy, the transformation strain, and the crystallographic relationship between adjacent parent grains. Among these, the common constraint imposed on the daughter-phase variants by the parent phases on the two sides of the grain boundary is one of the important factors in grain-boundary variant selection, but continuous quantitative characterization of this constraint has been relatively limited.

The phase-transformation misorientation θPT in the framework of generalized misorientation provides a geometrical characterization parameter for this purpose. It should be noted that the physical mechanism of variant selection involves multiple factors such as interface energy, transformation strain energy, and local stress, and that θPT characterizes the geometrical constraint term arising from the crystallographic matching of the two parent phases, rather than the complete criterion for variant selection. Under this delimitation, the role of the parameter differs in two situations: in thermally induced transformations, the interface energy and nucleation-site constraints act mainly through the degree of crystallographic matching of the two parent phases, and θPT can be used directly as a criterion for the propensity of variant selection; in deformation-induced transformations, θPT is used in conjunction with the mechanical driving-force criterion. The definition of θPT is given by Eq. (2): when Dx takes the parent→daughter→parent round-trip transformation matrix Tkl, it represents the minimum rotation angle required for the two parent grains, after their respective variant transformations, to generate daughter phases of the same orientation. Accordingly, θPT essentially characterizes the degree of geometrical matching by which adjacent parent grains generate daughter-phase variants of identical or similar orientation. Compared with qualitative dichotomous criteria such as the presence of a "common {110}", θPT expresses this matching relation as a continuous angular quantity. Earlier statistical results for the β→α transformation in TC25G titanium alloy show a clear quantitative correlation between θPT and actual variant selection[4], so that it can serve as a crystallographic parameter for analyzing the propensity of grain-boundary variant selection.

When the θPT of adjacent parent grains is small, variants of identical orientation are readily formed preferentially during the transformation; such variants have good lattice matching with the parent phase, which weakens the interface misfit caused by the abrupt change in daughter-phase orientation. These geometrical conditions may favor grain-boundary nucleation and subsequent growth of the corresponding variants, but whether an actual interface is coherent or has the lowest interface energy cannot be judged from θPT alone. The interface structure is also influenced by factors such as lattice misfit, the boundary plane orientation, interface translation, and local atomic relaxation, consistent with the discussion in Chapters 2 and 3 of the relation between orientation matching and complete interface structure. The connection between phase-transformation misorientation and thermodynamic conditions can be discussed from the replacement of interfaces during grain-boundary nucleation. Let the parent phase be M and the daughter phase be S, and let the original parent grain-boundary energy be γMM. After the daughter phase nucleates at the parent grain boundary, part of the original M/M interface is replaced by M/S interfaces on the two sides, and the corresponding change in interface energy can be expressed as Δγ = γMS1 + γMS2 − γMM. When the M/S interfaces on the two sides are approximately equivalent, this simplifies to Δγ = 2γMS − γMM, and the condition for complete wetting is correspondingly 2γMS < γMM [6]. This relation shows that whether grain-boundary nucleation is favorable is closely related to the energy of the newly formed M/S interfaces.

When the material system and interface conditions are similar, a small θPT indicates that variant combinations well matched to the same daughter-phase orientation exist in the two parent phases, and may therefore reduce the interface-energy term associated with crystallographic orientation misfit and increase the propensity for the corresponding variant combinations to form. Here θPT is more suitably regarded as a crystallographic characterization parameter related to the interface energy and the nucleation barrier, rather than as the interface energy itself. For a specific M/S interface, its energy also depends on variables such as the interface normal, lattice misfit, interface translation, and atomic structure. After the transformation, if the daughter-phase variants formed on the two sides of the original grain boundary have a small phase-transformation misorientation, the orientation misfit of the residual S/S interface is correspondingly small. Under conditions where the small-angle grain-boundary approximation holds, the energy associated with orientation misfit can be described by the corresponding small-angle interface model. If, however, the daughter phases on the two sides form twins, low-Σ CSLs, or other special interfaces, their low energy may originate from mirror symmetry or special coincidence of the complete lattice, in which case the corresponding interface-structure analysis is still required and the interface energy cannot be judged from θPT alone.

Studies by Furuhara et al.[13] on grain-boundary α phases in β titanium alloys show that grain-boundary α often maintains a relatively strict Burgers orientation relationship with one β grain while forming a certain degree of crystallographic matching with the other β grain, thereby reducing the nucleation work. This variant-selection behavior of "maintaining a strict orientation relationship on one side and approximate matching on the other" is a concrete manifestation of the common constraint on the two parent phases characterized by θPT. Studies by Bhattacharyya et al.[14] on α/β two-phase titanium alloys likewise show that the crystallographic geometrical relationship between the α and β phases is closely associated with variant selection. In the author's study of heat treatment of Ti60 alloy in the α + β two-phase region[15], the misorientation angle was used to characterize the degree to which the actual orientation of αp and the surrounding β phase deviates from the ideal Burgers orientation relationship, and the results show that αp with smaller orientation deviations has a higher tendency to be retained. These results indicate that the orientation characteristics of the parent phase and the degree of crystallographic matching between parent and daughter phases are closely related to variant selection and phase stability.

From the definition of θPT, the crystallographic matching effects revealed by these studies can be further characterized in terms of the constraint imposed on the common daughter-phase variant by the parent phases on the two sides of the grain boundary: the parent phase on one side must satisfy the corresponding orientation relationship with the daughter phase, while the degree of deviation between the parent phase on the other side and that variant further affects the conditions for grain-boundary nucleation. Accordingly, θPT incorporates the geometrical constraints of the two parent phases on the daughter-phase variant into a single minimization, providing a continuous crystallographic parameter for grain-boundary variant selection.

Accordingly, the role of θPT in variant selection can be stated as follows: when other conditions are similar, a small θPT indicates that favorable geometrical conditions exist for adjacent parent phases to form daughter-phase variants of nearly identical orientation, and such variant combinations may have lower interface misfit and nucleation barriers, and thus a higher propensity to form statistically. Actual variant selection is still determined jointly by factors such as the interface energy, transformation strain energy, local stress, parent-phase texture, and kinetic conditions. When the influence of local stress is significant, parameters such as the Schmid factor and the external force direction can be analyzed in combination with θPT, so that the mechanical driving force and the crystallographic matching condition are considered separately. Establishing a semi-quantitative connection between the parent-phase crystallographic constraint characterized by θPT and the thermodynamics and kinetics of the transformation is a key problem in this direction.

During plastic deformation of polycrystals, the stress and strain compatibility between grains is influenced jointly by factors such as grain orientation, slip-system activation, grain-boundary structure, and local constraints. From the Taylor and Sachs models to self-consistent models and crystal-plasticity finite-element methods, the above theories describe the relationship between grain orientation and plastic deformation at different scales. For the local deformation behavior near grain boundaries, the influence of crystallographic matching conditions on processes such as slip, twinning, and phase transformation must also be considered.

Generalized misorientation provides process-related crystallographic characterization parameters for this purpose: θS, θT, and θPT describe the degree of optimal matching between slip systems, twin systems, and phase-transformation variants, respectively. These parameters do not themselves contain local stress, interface-structure, or kinetic information; they are used to judge whether candidate channels with good geometrical matching exist for the corresponding process, supplementing traditional mechanical analysis with crystallographic information rather than replacing existing deformation theories.

Taking slip transfer as an example, existing criteria belong to two levels: at the intragranular level, the Schmid factor describes the ease of activation of slip systems within a single grain under a given loading direction; at the grain-boundary level, the Luster–Morris parameter m′ describes the degree of geometrical matching between specific slip-system pairs in adjacent grains[16]; and the residual Burgers vector br characterizes the degree of misfit produced after a dislocation crosses the grain boundary, depicting the residual misfit after transfer has occurred rather than the geometrical precondition for whether transfer occurs. By comparison, θS enumerates all candidate slip systems in the grains on both sides and gives the smallest slip-system matching angle among them, and is therefore suitable for identifying well-geometrically-matched transfer channels from among large numbers of candidate slip systems. The Schmid factor is directed at a single grain, m′ at a given bicrystal slip-system pair, and θS at the overall matching of all candidate slip systems on both sides in a polycrystalline system, coming closer to a statistical description of the grain-boundary network in polycrystalline materials.

For a given slip-system pair (i, j), let the rotational deviation angle required for their alignment be θij. Since rigid-body rotation preserves the angular relations between vectors, the angle κ between slip directions and the angle ψ between slip-plane normals satisfy κ ≤ θij and ψ ≤ θij. Therefore, when θij ≤ 90°, we have mij′ = cosκcosψ ≥ cos2θij. For the slip-system pair that achieves the minimum rotation angle, cos2θS ≤ m′ ≤ 1 then holds. This relation shows that when θS is small, the corresponding slip-system pair generally has a high m′; the two are not, however, equivalent. m′ retains information on the separate matching of the slip direction and the slip-plane normal, whereas θS reflects only the minimum rotation angle required to bring the slip systems on the two sides into overall correspondence. The two therefore describe geometrical matching relations at different levels. These bounds also show that, within the small-angle range, θS and m′ can be cross-checked; and the angular representation of θS allows it to be compared statistically, on the same scale, with the other parameters of the generalized misorientation framework (see the unified form in Chapter 2).

Likewise, br provides information on the residual vector misfit after a dislocation has crossed the grain boundary, and its premise of applicability is that the dislocation itself crosses the boundary; slip transfer also involves another mechanism—the pile-up of dislocations at the interface activating new dislocation sources in the adjacent grain—in which case no single dislocation crosses the interface. Both mechanisms can be regarded as slip transfer, but their physical essence differs; θS characterizes only the geometrical matching of the slip systems on the two sides before transfer occurs and presupposes no specific mechanism, so it applies equally to both situations, whereas br applies only to the former. Therefore, in the analysis of grain-boundary slip transfer, the more reasonable approach is first to use θS to screen the possible transfer channels from among all candidate slip systems, and then to evaluate their actual transfer conditions in combination with parameters such as m′, br, the Schmid/Taylor orientation factor, and the local resolved shear stress. Chapter 7 further introduces mechanical activation conditions to extend this screening process.

In addition to slip transfer, generalized misorientation can also be used to describe the mutual influence between different deformation mechanisms. During plastic deformation, slip, twinning, and phase transformation can all cause lattice rotation, thereby changing the crystallographic conditions for subsequent deformation mechanisms. For example, after twinning occurs, the crystal orientation is reconstructed, and slip-system combinations that originally had poor geometrical matching may be transformed into orientation relationships more favorable to slip transfer. Similarly, the evolution of crystal orientation caused by slip may gradually transform originally unfavorable twinning or phase-transformation systems into favorable orientations, thereby promoting the subsequent activation of these mechanisms. A given deformation mechanism is therefore influenced not only by the initial orientation relationship, but can also change the activation conditions of subsequent processes through orientation evolution.

A direct case is as described in Section 3.3: deformation-induced twins reconstruct the local orientation, reducing the θS between the twin and the adjacent grain, and slip completes its cross-boundary transfer along this newly created small-θS channel. A similar sequential evolution of multiple mechanisms also occurs in Fe-24Ni-0.6C metastable austenitic steel[17]: as plastic deformation proceeds, the dominant deformation mechanism passes successively through dislocation slip, deformation twinning, and deformation-induced martensitic transformation, displaying marked stages and continuity. In the early stage of deformation, dislocation slip and its transfer across grain boundaries cause continuous lattice rotation and orientation evolution, bringing some grains gradually into crystallographic states more favorable to the activation of twinning (low θT); the deformation twins subsequently formed further reconstruct the local orientation relationship and build twin boundaries with a small phase-transformation misorientation θPT, providing favorable crystallographic channels for the martensitic transformation and promoting its preferential nucleation. The above results show that the different deformation mechanisms in polycrystalline materials are not independent of one another, but can form history-dependent crystallographic coupling relations through continuous orientation evolution: the deformation mechanism of the preceding stage continuously reconstructs the activation conditions of the other mechanisms in the following stage by changing the local crystal orientation. The plastic compatibility of polycrystals can therefore be realized step by step through the sequential activation of different deformation mechanisms, and generalized misorientation provides a unified crystallographic parameter for quantitatively describing this process of "orientation evolution — path reconstruction — mechanism switching".

For interfaces between different phases, generalized misorientation can likewise supplement the analysis of interfacial deformation compatibility. For example, in the ferrite/austenite interfaces of medium-Mn steel, the interface orientation relationship and the coherence state influence dislocation transfer and local plastic behavior. Ref. [18] explains the above phenomena by a "dichotomy" of interface coherency: semi-coherent interfaces correspond to those with controlled orientation relationships, and incoherent interfaces to those with random orientation relationships. For such problems, generalized misorientation can supplement this dichotomous description, which classifies interfaces by coherency, with a statistically tractable continuous angular quantity: θS describes the degree of geometrical matching between the candidate slip systems of the two phases, directly reflecting the ease or difficulty of dislocation transfer across the interface; at the same time, θT and θPT can be linked to further discuss the influence of interface-structure characteristics on deformation or transformation behavior.

From the viewpoint of multi-mechanism deformation, the classical misorientation and process-related parameters can be expressed as Φ = [θ, θS, θT, θPT], where θ describes the overall relative orientation of grains, and θS, θT, and θPT correspond to the slip, twinning, and phase-transformation processes, respectively. This representation does not mean that the different parameters can substitute for one another, but is used to distinguish the crystallographic matching conditions in different physical processes. For example, slip transfer concerns mainly θS, variant selection mainly θPT, while special grain-boundary structures must still be analyzed in combination with the CSL type and interface-geometry information.

In multiscale modeling, generalized misorientation can serve as grain-boundary crystallographic information input to crystal-plasticity models. Based on EBSD data, the distributions of θS and θT for different grain boundaries, and of θPT where applicable, can be tabulated; combined with the local stress, the boundary plane orientation, and interface-transfer models, the statistical relationship between the grain-boundary network and macroscopic plastic behavior can then be analyzed. In this workflow, generalized misorientation serves first as a crystallographic screening criterion—pre-screening, from among all candidate slip and twin systems, the channels with good geometrical matching, thereby reducing the computational cost of the crystal-plasticity model; for transformation processes that are still difficult for crystal-plasticity finite-element methods to handle, θPT can act as a crystallographic bridge in the coupling of crystal plasticity and phase-field simulation, rapidly screening out the preferred variants. It should likewise be made clear that misorientation itself is not equivalent to grain-boundary strength or transfer capability; its role is to provide a process-related crystallographic constraint within the existing mechanical framework.

In summary, the connection between generalized misorientation and the theory of polycrystalline deformation compatibility lies in the following: by extending the object of crystallographic matching from the lattice as a whole to slip systems, twin systems, and phase-transformation variants, it provides computable geometrical characterization parameters for local grain-boundary processes. The method can be combined with the Schmid factor, m′, the residual Burgers vector, and the local stress state to analyze grain-boundary compatibility behavior in multi-mechanism deformation.

7.1 Summary

The framework of generalized misorientation proposed in this paper consists essentially in generalizing misorientation from "the minimum rotation for global alignment of crystal coordinate systems" to "the angular distance from the actual relative orientation to a target set of physical feature matching". The matching target can be the feature matching of slip systems, twin systems, or phase-transformation variants, or special grain-boundary relations such as CSL; classical misorientation corresponds to the most stringent global matching and is the special case in which the constraint on the target set is strictest. Under conditions of the same crystal system, generalized misorientation is never greater than classical misorientation (Appendix A.7), and the difference between the two reflects the distinction between matching of the lattice as a whole and matching of local process features. This generalization develops misorientation from a geometrical quantity describing static crystal orientation relationships into a crystallographic parameter that can be linked to concrete physical processes, and is applicable to the analysis of interfaces between different crystal systems.

The mathematical core of the framework is the unified Eq. (2): the feature-matching operator Dx defines the matching object, and the slip misorientation θS, twinning misorientation θT, and phase-transformation misorientation θPT differ only in the physical process to which the matching operator corresponds, sharing the same minimum-rotation geometry and the computational logic of "taking the minimum over variants and symmetry operations". The construction of the three parameters, their symmetry equivalence, the invariance of the local coordinate-system choice, and the ordering relation have all been derived rigorously (Appendix A), and each parameter can be obtained in batch from EBSD data with the aid of group-theoretical simplification and quaternion algorithms. Eq. (2) also gives the construction rule for new parameters: any object possessing an explicitly expressible local feature-matching condition allows the corresponding parameter to be defined by the same logic.

All three parameters—slip, twinning, and phase-transformation misorientation—are already supported by experimental statistics: the variant-selection probability decays rapidly as θPT increases, and the slip- and twin-transfer probabilities decrease in steps as θS and θT increase, respectively. The framework shares the minimum-rotation geometrical core with interface-structure theories such as CSL and the O-lattice, and can describe, in a unified way, the deviation of the actual orientation from various matching states; on this basis, an energetic linkage from interface energy through process barrier to process probability has been established, advancing misorientation from a static geometrical description to a quantitative characterization of process propensity. Within this framework, parameters such as θS, θT, θPT, and θΣ are the deviation coordinates of the respective characteristic points, and the framework thereby has the potential to serve as a "basis" for the interface energy distribution: taking characteristic points as the base and their neighborhood branches as patches, the distributions of energy and barrier in orientation space are assembled.

The extended inferences from this energy connection are currently at different stages of maturity: the formalized connection between the θS distribution and the Hall–Petch coefficient (Eq. (8)), the generalized interface energy spectrum, and the variant-resolved misorientation spectrum are hypotheses awaiting verification, whose specific forms must be determined by experiment or simulation. It must be emphasized that generalized misorientation provides a description at the level of crystallographic matching, rather than a replacement for complete interface structure or for existing theoretical systems; the causality and range of applicability of its statistical correlations must still be tested, in combination with in-situ experiments, three-dimensional interface characterization, and crystal-plasticity simulation, in more material systems and physical processes.

7.2 Applicable boundaries and validation methodology

The framework of generalized misorientation has clear applicable boundaries. First, misorientation is not the only factor determining interface behavior. The atomic structure of the boundary plane, solute segregation, precipitates, the local stress state, and pre-existing defects can all influence slip transfer, twinning, and transformation behavior. Accordingly, small values of θS, θT, or θPT indicate only that relatively favorable crystallographic matching conditions exist, and do not imply that the corresponding process is bound to occur.

Second, the connection currently established between generalized misorientation and process behavior is manifested mainly as a statistical correlation, not as a deterministic prediction for individual grain boundaries. For example, a boundary with small θS may still fail to undergo slip transfer because of differences in boundary plane orientation, local stress state, or defect distribution. Therefore, in practical analysis, generalized misorientation should be combined with mechanical conditions and interface structure, and a single angle should not be used as an independent criterion.

Third, existing statistical studies have concentrated mainly on titanium alloy systems. Since FCC, BCC, and other HCP materials have different slip systems, twin systems, and interface structures, the connection between their feature-matching relations and process behavior may be material-dependent. The applicability of θS, θT, and θPT in different material systems therefore remains to be further tested.

In addition, the process-related misorientations defined in this paper are all determined by the minimum rotation angle among candidate variant pairs, and therefore retain mainly the information on the optimal matching channel, whereas information on the rotation axis, suboptimal variants, and multiple nearly equivalent channels is retained in the complete rotation matrix Rxy (that is, the complete set of channel-level characteristic points described in Section 2.5). For problems involving competition between multiple channels or the influence of loading direction, the misorientation spectrum, taking candidate channels or candidate variants as discrete dimensions, can be used to recover this compressed information.

Different levels of matching may also coexist within the same grain boundary. For example, a boundary with small θS may simultaneously be close to a low-Σ CSL orientation. The former describes local matching between slip systems, while the latter reflects special coincidence of the complete lattice, and the two have different physical meanings. Therefore, when validating the independent role of generalized misorientation, comparison should be stratified by CSL type, so as to distinguish the contributions of local process matching and of special grain-boundary structure.

Specifically, the relation between θS and slip-transfer probability can be analyzed on the basis of stratification by CSL/Σ type; if θS still shows statistical correlation under the same CSL conditions, this demonstrates that it provides information independent of special grain-boundary structure. At the same time, EBSD statistics reflect mainly the correlation between crystallographic conditions and process events, whereas causality must still be verified by methods such as in-situ experiments, three-dimensional interface characterization, and crystal-plasticity finite-element analysis. The framework of generalized misorientation should therefore be understood as a descriptive method of "the joint action of crystallographic matching conditions and mechanical realization", not as "misorientation determines the process".

7.3 Directions for future development

The future development of the framework of generalized misorientation can be pursued in four aspects: energy, geometry, mechanics, and application.

At the energy level, a quantitative relation between misorientation and the interface energy or process barrier can be established. At present, generalized misorientation describes process-related crystallographic misfit mainly through an angular quantity; in future, first-principles and molecular-dynamics calculations can be combined to build a database of the correspondence between different values of θS, θT, and θPT and changes in interface energy. At the same time, it should be recognized that the interface energy is not determined by a single misorientation; for special interfaces such as low-Σ CSLs and twin boundaries, factors such as lattice coincidence, the boundary plane orientation, and atomic structure must also be considered. Such relations are therefore more suitably established within a restricted material system and interface type. The database is best organized according to the scheme of "characteristic point — neighborhood — assembly": characteristic points are enumerated crystallographically and each carries a clear physical meaning, so that the continuous introduction of characteristic points continuously approaches the true overall energy distribution; in engineering practice, the important characteristic points and their neighborhoods are computed first, in order of the depth of the energy valleys (energy minima), the defect scale, and process relevance, before extending to the remaining regions.

In addition, parameters such as θS, the Schmid factor, and the boundary-plane normal can be combined into feature variables and analyzed, together with machine-learning methods, to characterize grain-boundary deformation modes, making generalized misorientation a crystallographic input in multiparameter models. The two are complementary: the characteristic points and neighborhood branches are given by the crystallographic framework, and machine learning need only fit the residuals within the branches and interpolate the blank regions between characteristic points, giving a data efficiency far higher than that of purely black-box fitting.

At the geometrical level, information on the boundary plane can be introduced to establish a generalized interface energy spectrum incorporating the slip-plane–boundary-plane relation and the projection relation of slip directions: Γ(θS, ψgb, ϕ). Here ψgb denotes the angle between the slip-plane normal and the boundary-plane normal, and ϕ the angle between the projections of the slip directions on the two sides within the boundary plane. This parameterization is not a complete description of the five degrees of freedom of a grain boundary, but a local geometrical representation selected for the slip-transfer process. Its advantage is that it can be combined directly with EBSD data analysis and linked with process probability; if the goal is to build a complete grain-boundary energy database, the traditional five-degree-of-freedom description must still be adopted.

This direction still faces problems such as insufficient experimental data, the high cost of three-dimensional grain-boundary characterization, and the material dependence of the energy–process-probability relation. In future, three-dimensional EBSD, in-situ characterization, high-throughput interface computation, and machine-learning methods can be combined to improve the efficiency of multiparameter interface characterization.

At the mechanical level, the activation conditions of slip systems can be introduced to extend the current "minimum over all variants" to "minimum over activatable variants", that is, to construct a misorientation spectrum related to mechanical conditions. At present, θS takes the minimum over all candidate slip systems, which may include slip systems that are geometrically well matched but in practice difficult to activate. By screening the activatable variants in combination with the resolved shear stress, the critical resolved shear stress, and the loading direction, and then computing their minimum misorientation, an effective θS more consistent with the actual deformation conditions can be obtained.

The misorientation spectrum can also be constructed in a channel-resolved or variant-resolved form, retaining information on the different candidate channels, and can be combined with crystal-plasticity models to analyze grain-boundary slip connectivity and local non-uniform deformation behavior. For the dwell-fatigue problem of near-α titanium alloys, the misorientation spectrum can provide parameters related to process probability for analyzing the relations among microtexture, slip transfer, and crack initiation; for example, the initiation of cleavage-plane cracks induced by basal slip in dwell fatigue is, in essence, the result of the extreme development of dislocation pile-up at interfaces with large θS, and the prediction of such localized damage can be reduced to geometrical operations on the misorientation spectrum.

In terms of application, the applicability of θS, θT, and θPT must still be tested in different material systems such as steels, magnesium alloys, and nickel-based superalloys, and microstructure and interface design based on generalized misorientation can be explored: textures can be controlled by thermomechanical treatment so that the θS or θT distributions of target grain boundaries shift in directions favorable to plasticity or strength; and θPT can be used to predict variant selection, guiding transformation heat treatments to obtain the desired texture and morphological features of the microstructure.

Generalized misorientation provides a feature-matching-based mode of connection between interface-structure theory, plastic-deformation theory, and transformation crystallography. Its principal value lies in transforming crystallographic conditions related to concrete physical processes into computable parameters, and in combining them with existing interface-structure and crystal-plasticity mechanics theories. This connection is still at a developmental stage, and its quantitative relations and range of applicability must be further improved in future through experimental, simulation, and data-driven methods.

A.1 Construction of the slip-system matching operator

A slip system i is defined by its slip-plane normal ni and its slip direction di, both expressed in the crystal coordinate system of the grain in which it lies. Taking di, ni×di, and ni as row vectors, a right-handed orthogonal matrix is constructed

Di = [diT; (ni×di)T; niT] ∈ SO(3), (A1)

Its function is to transform the crystal coordinate system into the slip local coordinate system (the first axis along the slip direction and the third axis along the slip-plane normal). The signs of the slip-plane normal and the slip direction do not change the physical slip system; in enumeration, the sign variants (±ni, ±di) are included together, so that the set of slip systems is closed under the symmetry operations of the crystal system, which is also the common premise of the equivalence in A.2 and of the ordering-relation theorem in Appendix A.7. Under the orientation-matrix convention of this paper (sample frame → crystal frame), the slip local coordinate systems of variant i in grain 1 and variant j in grain 2 in the sample coordinate system are Di·G1 and Dj·G2, respectively, and their relative rotation is

Rij = Di·G1·G2−1·Dj−1, (A2)

θS is then the minimum rotation angle of Rij taken over all variant pairs (i, j). Eq. (A2) is the concrete implementation of Eq. (2) in the slip case: the matching condition that the two slip systems are parallel (parallel slip planes and parallel slip directions) uniquely determines the candidate rotation Rij, and the set of candidate rotations is determined by the matching condition itself and is independent of the choice of local coordinate system; the symmetry-operation factor in Eq. (2) summarizes all enumeration of symmetry-equivalent orientations, and its equivalence with variant enumeration is shown in Appendix A.2.

A.2 Equivalence of symmetry operations and variant enumeration

The symmetry operations of a crystal system map slip systems onto symmetry-equivalent slip systems. Let S be a proper rotation in the rotation group of the crystal system; the matching operator of the symmetry-equivalent slip system then satisfies

DS·i = Di·S−1, (A3)

The proof requires only noting that a proper rotation preserves inner products and cross products (S(n×d) = Sn×Sd), so that the row vectors of Di·S−1 are precisely (Sdi)T, (S(ni×di))T, and (Sni)T. Eq. (A3) shows that letting the variant index range over the set of symmetry-equivalent variants, and fixing the variant while explicitly enumerating the symmetry operations, give the same set of candidate rotations. The joint enumeration of the matching operator and the symmetry operations in Eq. (2) can therefore be reduced entirely, computationally, to a double enumeration over variant pairs—and this is precisely the mathematical basis of the computational logic of "taking the minimum over the corresponding variants and symmetry operations" in the main text.

A.3 Invariance of the choice of local coordinate system

In the above construction, the choice of local coordinate system is not unique, but as long as the same convention applies to all slip systems (no separate choice for individual slip systems), any right-handed orthogonal convention satisfying "the row vectors are spanned by di and ni" differs only by a single global rotation O ∈ SO(3), that is, Di → O·Di. Eq. (A2) then becomes Rij → O·Rij·O−1, a conjugate transformation that leaves the rotation angle unchanged; the set of variant pairs traversed in the minimization is likewise unchanged. Accordingly, θS is an invariant of the local coordinate-system convention, and the construction of Di is unique in this sense.

A.4 Twinning-system and phase-transformation matching operators

The construction of twinning misorientation is entirely isomorphic to that of slip: replacing (ni, di) by the twin-plane normal K1 and the twinning shear direction η1, the twin local coordinate-system matrix Ti is constructed according to Eq. (A1). It should be noted that the twinning shear is unidirectional: η1 and −η1 correspond to opposite shear directions with different activation conditions, so the two are not merged when twin variants are enumerated, and the difference in shear direction is thereby retained in the construction of the local coordinate system. The matching operator of phase-transformation misorientation, in turn, takes Tkl = Cl−1·Ck: Ck is the lattice transformation matrix from parent to daughter phase (including the lattice correspondence and any necessary rigid-body rotation), and Cl−1 is the inverse transformation matrix from daughter to parent phase; {Tkl}·G1 gives all parent orientations that can generate the same daughter-phase orientation as grain 1. The matching condition is that the parent grains on the two sides, after their respective variant transformations, generate daughter-phase lattices of identical orientation; in Eq. (2), Dx takes Tkl and Sy takes the crystal symmetry-operation group of the parent phase. The three parameters thereby share the same mathematical framework, differing only in the physical meaning of the matching operator.

A.5 Numerical implementation

Two points should be noted regarding the computation. First, the computation of the rotation angle is best carried out using a quaternion representation: representing Rij by a unit quaternion, the rotation angle is θij = 2·arctan2(‖q‖, |q0|), where q0 is the scalar part and q the vector part; this expression is numerically stable at both ends, θ→0 and θ→π, and avoids the amplification of truncation error in arccos[(tr(R)−1)/2] at the endpoints, requiring only a constant number of quaternion multiplications per variant pair. Second, the computational complexity is of the order of the square of the number of variants: face-centered cubic crystals have 12 slip systems, and hexagonal close-packed titanium alloys have 24 slip systems in total when basal, prismatic, and pyramidal systems are included (3 basal, 3 prismatic, 6 pyramidal ⟨a⟩, and 12 pyramidal ⟨c + a⟩); when enumerating, the sign variants (±ni, ±di) are included according to A.1, so that the number of representative elements is four times the number of physical slip systems and the number of pairs is its square. The computation time for a single grain boundary is of the order of milliseconds, which can support boundary-by-boundary statistics over full-field EBSD data. The computation also outputs the variant pair (i*, j*) at which the minimum is attained, and this variant-pair information is retained in the computed rotation matrix Rxy, for use in the analysis of variant-resolved misorientation spectra in Chapter 7.

A.6 Coordinate representations of the candidate rotation and trace decomposition

The same matching condition has two equivalent representations. The crystal-coordinate-system representation is Eq. (2): the candidate rotation Rxy is sandwiched between the symmetry operation Sy and the feature transformation, and is defined in the crystal coordinate system of grain 2. The sample-frame representation instead places the relative rotation in the common sample coordinate system: denoting the transformations from the sample coordinate system to the respective local crystal coordinate systems of the grains on the two sides by Ui (containing the feature transformation and symmetry operations) and Tj (containing symmetry operations), and the parallel matching operation by P, the matching condition can be written as P·Ui·R = Tj. Since R stands at the rightmost end of the product and acts first, it is defined in the sample reference frame. The two versions of the rotation matrix are related by coordinate conjugation followed by inversion; neither conjugation nor inversion changes the rotation angle, so that the two versions give strictly equal minimum rotation angles, the difference lying only in the coordinate expression of the rotation axis: the crystal-coordinate-system representation facilitates the treatment of crystallographic matching relations between grains (variant identification, index operations, and symmetry reduction), while the sample-frame representation can be coupled directly with engineering parameters.

Furthermore, by the cyclic-permutation invariance of the matrix trace, the computation of the minimum rotation angle can be organized into the form tr[(Sx−1·Dc−1·P−1·Dd·Sy)·ΔG], where Dc and Dd are the feature local coordinate-system matrices on the two grain sides (in the slip case, the Di and Dj of A.1), P is the parallel matching operation described above, Sx and Sy are the symmetry operations on the two sides, and ΔG = G2·G1−1 (under the orientation-matrix convention from the sample coordinate system to the crystal coordinate system) is the intergranular relative orientation, independent of the sample reference frame. For grains of different crystal systems, the symmetry groups on the two sides are generally different, and enumeration over both symmetry groups can be reduced to a single side only when the feature-matching operator satisfies the conjugate closure condition K−1SαK ⊆ Sβ (where K is the transformation matrix connecting the two crystal systems, such as an orientation-relationship matrix, and Sα and Sβ are the symmetry groups on the two sides). For grains of the same crystal system, K is the identity matrix and the condition holds trivially; for grains of different crystal systems it generally holds only for the common symmetry subgroup of the two crystal systems, and enumeration over both sides is usually still required.

A.7 Ordering relation between classical and generalized misorientation

Proposition: for grains of the same crystal system, generalized misorientation is never greater than classical misorientation. Proof: suppose the classical minimizing rotation is attained by the symmetry operation S*, whose rotation angle is the classical misorientation. S* maps the feature system j of grain 2 (slip system, twin system, or daughter-phase variant) onto the symmetry-equivalent feature system S*·j of grain 1; by Eq. (A3) of A.2, the local coordinate-system matrix of this feature system is Dj·(S*)−1, and the candidate rotation corresponding to the variant pair (S*·j, j), namely Dj·(S*)−1·G1·G2−1·Dj−1, is a conjugate of (S*)−1·G1·G2−1; since neither conjugation nor inversion changes the rotation angle, its rotation angle is precisely the classical misorientation. The classical minimizing rotation therefore always belongs to the set of candidate rotations of generalized misorientation, and the minimum angle cannot be larger. Q.E.D. This conclusion presupposes that the feature-system enumeration set is closed under the symmetry operations of the crystal system (including sign variants, see A.1).

It should be pointed out that, for the CSL deviation angle θΣ discussed in Chapter 2, its target set is the special lattice coincidence state of a given Σ type, rather than the state of complete lattice identity coincidence corresponding to classical misorientation. Accordingly, in general there is no universal ordering relation between θΣ and classical misorientation θ.

[1] BUNGE H J. Texture Analysis in Materials Science[M]. London: Butterworths, 1982.

[2] LÜTJERING G, WILLIAMS J C. Titanium (Engineering Materials and Processes)[M]. 2nd ed. Berlin: Springer, 2007.

[3] ZHAO Z B, WANG Q J, HU Q M, et al. Effect of β (110) texture intensity on α-variant selection and microstructure morphology during β→α phase transformation in near α titanium alloy[J]. Acta Materialia, 2017, 126: 372-382.

[4] LIU Y H, ZHAO Z B, WANG Q J, et al. An orientation relationship between parent grains and its application to variant selection of transformed α in titanium alloys[J]. Metallurgical and Materials Transactions A, 2025, 56(3): 801-814.

[5] ZHAO Z B, WANG Q J, LIU J R, et al. Two physical quantities that can describe the propensity of grain deformation transfer: slip misorientation and twinning misorientation[J]. Journal of Liaocheng University (Natural Science Edition), 2025, 38(5): 703-706 (in Chinese).

[6] SUTTON A P, BALLUFFI R W. Interfaces in crystalline materials[M]. Oxford: Clarendon Press, 1995.

[7] BOLLMANN W. Crystal Defects and Crystalline Interfaces[M]. Berlin: Springer, 1970.

[8] ZHANG W Z. Role of interfacial structures in the development of phase transformation crystallography[C]. Beijing: Proceedings of PTM2022, 2022.

[9] MEYERS M A, CHAWLA K K. Mechanical Behavior of Materials[M]. 2nd ed. Cambridge: Cambridge University Press, 2009.

[10] MURR L E. Interfacial phenomena in metals and alloys[M]. Reading, Massachusetts: Addison-Wesley, 1975.

[11] HALL E O. The deformation and ageing of mild steel: III. Discussion of results[J]. Proceedings of the Physical Society Section B, 1951, 64(9): 747-753.

[12] PETCH N J. The cleavage strength of polycrystals[J]. Journal of the Iron and Steel Institute, 1953, 174: 25-28.

[13] FURUHARA T, TAKAGI S, WATANABE H, et al. Crystallography of grain boundary α precipitates in a β titanium alloy[J]. Metallurgical and Materials Transactions A, 1996, 27(6): 1635-1646.

[14] BHATTACHARYYA D, VISWANATHAN G B, DENKENBERGER R, et al. The role of crystallographic and geometrical relationships between α and β phases in an α/β titanium alloy[J]. Acta Materialia, 2003, 51(16): 4679-4691.

[15] ZHAO Z B, WANG Q J, LIU J R, et al. Effect of heat treatment on the crystallographic orientation evolution in a near-α titanium alloy Ti60[J]. Acta Materialia, 2017, 131: 305-314.

[16] LUSTER J, MORRIS M A. Compatibility of deformation in two-phase Ti-Al alloys: dependence on microstructure and orientation relationships[J]. Metallurgical and Materials Transactions A, 1995, 26(7): 1745-1756.

[17] LIU Y, GAO S, PARK M, et al. Super ductile steel via sequentially activating multiple strengthening mechanisms[J]. Materials Today, 2026, 100: 103497.

[18] WANG T, YANG D P, DONG M, et al. Discontinuous versus continuous yielding in medium-Mn steel: role of the coherence of ferrite/austenite interface[J]. Acta Materialia, 2026, 309: 122052.