Higher Spin from Tensor Products: The Three Carriers and the Absence of a Fourth

Introduction

The biquaternion algebra carries three carriers and no fourth. The centre $\mathbb{B}_{C}$ carries spin $0$, the minimal left ideals carry spin $\tfrac12$, and the material sector carries spin $1$ on its scalar-free part. That list is complete, and this article says why it is complete and what lies beyond it.

The list is complete for a reason that has nothing to do with dimensions. The algebra is simple, and a simple algebra has one simple module. The carriers are three and the simple module is one, so the third carrier cannot be a module at all: it is the adjoint action, and the centre is the fixed line of that same action. The completeness of the list is therefore the completeness of the isotypic decomposition of the two actions the algebra has — left multiplication and conjugation — and each of the two contributes what it contributes: the left action one carrier, the adjoint action two.

Beyond spin one, the algebra stops being the right object. A field of spin $s>1$ is a field with more than one spinor index, and an index is a tensor factor. The carrier of spin $s$ is therefore a module of $\mathbb{B}^{\otimes 2s}$, not of $\mathbb{B}$, or a constrained subspace of a product of modules of $\mathbb{B}$. This article follows the construction: the tensor powers of the defining module, their Clebsch–Gordan decomposition, the symmetric powers that isolate the pure spin, and the constraints that remove the admixture.

What this article is not. It is not an argument that the algebra admits only spins $0$, $\tfrac12$ and $\tfrac{1}{2}$. That statement is false under the corpus's definition of spin: the spin is a property of the carrier, and $\mathbb{B}$ has a carrier of spin $1$. What this article proves is the neighbouring statement — that the algebra has three carriers, that a fourth does not exist, and that the integer spin it does carry is not a module but a conjugation.

Plan. §The Three Carriers of the Base Algebra names them. §What a Module Can Carry proves that a module carries only spin $\tfrac12$ and explains why that is a statement about modules and not about the algebra. §The Spin-Zero Carrier places the scalar in the centre. §Tensor Products generates everything above one. §The Explicit Cases reads spin one, $\tfrac32$ and two against the rule. §Tensor Powers of the Algebra states the same at the level of the algebra. §Why There Is No Fourth Carrier gives the reason, §The Ceiling in Context compares it with the Clifford case, and §The Two Index Families gives the bimodule form.

The Algebra and Its Defining Module

Simplicity

The algebra is isomorphic to the algebra of $2\times2$ complex matrices,

$$ \mathbb{B} = \mathbb{C}\otimes_\mathbb{R}\mathbb{H} \;\cong\; M_2(\mathbb{C}), $$

and $M_2(\mathbb{C})$ is simple: it has no two-sided ideals other than $0$ and the whole algebra, and its centre is the one-dimensional space of scalar matrices. In the framework's basis the centre is the complex scalar subspace $\mathbb{B}_{C} = \{Q_te_t\}$, and the scalar imaginary $i$ is central. Simplicity is the algebraic fact behind everything in this article: a simple algebra has a single isomorphism class of simple modules, and all of its representation theory is generated by that one module.

Simplicity is also why the three spins cannot be three modules. One simple module serves the spin-$\tfrac12$ carrier many times over; the spin-zero and spin-one carriers are not modules at all. The word spin therefore attaches to the carrier — the subspace together with the action — and not to the module.

The defining module

The algebra acts on the two-dimensional complex space of column vectors, and this is its defining module $S$. In the framework's language $S$ is the spinor module, and its two dimensions are the two spinor components; the companion article on the spinor module and its Lorentz action develops it, and the companion Dirac article uses a Peirce carrier of $\mathbb{B}$, a minimal left ideal here, as its concrete form. Any two minimal left ideals of $\mathbb{B}$ are isomorphic, and each is isomorphic to $S$; concretely, the matrices with only one nonzero column form a minimal left ideal of dimension two.

The defining module as a minimal left ideal

The concrete realisation is worth recording, because it is the form in which the framework's fields are written. Let $e$ be a rank-one idempotent of $\mathbb{B}$, that is, an element with

$$ e^2 = e, \qquad e \neq 0,1 . $$

In the matrix representation the matrix unit $E_{11}$ is such an idempotent, and the left ideal $\mathbb{B}e$ consists of the matrices with only the first column possibly nonzero. Its complex dimension is two, since the first column is an arbitrary vector of $\mathbb{C}^2$, and it is minimal: it contains no nonzero left ideal smaller than itself. Minimality is the algebraic statement that the field it carries is irreducible, and the isomorphism of any two minimal left ideals is the statement that the algebra has only one irreducible carrier of this kind. The elements of the ideal are the spinors, and left multiplication by a general element of $\mathbb{B}$ is the action that makes the ideal a module.

The idempotent also decomposes the algebra itself,

$$ \mathbb{B} = \mathbb{B}e \oplus \mathbb{B}(1-e) \;\cong\; S\oplus S, $$

the second summand being the matrices with only the second column possibly nonzero. The algebra is therefore two copies of its defining module, and this decomposition is the concrete form of the general structure theorem: the algebra's own left-module content is that of two spinors. The decomposition is a decomposition of the left-module structure. It is distinct from the trace decomposition

$$ \mathbb{B} = \mathbb{B}_{C} \oplus \mathbb{B}_{V}, $$

into the scalars and the scalar-free part, which is the decomposition under conjugation and carries the scalar and vector representations respectively. The two decompositions have different transformation characters, and keeping them apart is what keeps the module content (spin one-half) distinct from the conjugation content (spin zero and spin one). Both are carriers; only the first is a module.

Every module is a sum of copies of the defining module

The representation theory of the algebra is exhausted by the following statement.

Claim. Every finite-dimensional left module $M$ over $\mathbb{B}$ is isomorphic to a finite direct sum of copies of $S$,

$$ \boxed{\;M \;\cong\; S^{\oplus k}, \qquad k = \dim_\mathbb{C}\mathrm{Hom}_\mathbb{B}(S,M)\;}. $$

Proof sketch. The algebra is semisimple — a matrix algebra over a field is — so by the Artin–Wedderburn theorem every finite-dimensional module is a direct sum of simple modules. Since $\mathbb{B}$ is simple, it has exactly one simple module up to isomorphism, namely $S$. Hence $M\cong S^{\oplus k}$, and the multiplicity $k$ is the dimension of the space of module maps from $S$ to $M$, by Schur's lemma applied to each summand.

Two consequences are immediate and both matter. The multiplicity space $\mathrm{Hom}_\mathbb{B}(S,M)$ is a complex vector space, so the decomposition is a Morita equivalence between the modules of $\mathbb{B}$ and complex vector spaces: a module is a complex vector space $V$ dressed as $S\otimes_\mathbb{C}V$, and nothing else. And $S$ is the only irreducible module, so it is the only source of any representation a module of the algebra can carry — which is a statement about modules, and the next two sections say what it leaves out.

The Three Carriers of the Base Algebra

carrier set action spin invariant
the centre $\mathbb{B}_{C}=\mathbb{C}e_t$ conjugation; fixed $0$ $B^2=0$
the ideals $S_\pm=\mathbb{B}\tilde\Pi_\pm$ left multiplication $\tfrac12$ $N=0$
the material sector $\mathbb{B}_{\Phi}=\mathbb{R}(ie_t)\oplus\mathfrak{k}$ conjugation $1$ on $\mathfrak{k}$ none

The catalogue. The centre is the fixed line of the adjoint action. The ideals are the minimal left ideals, two complex dimensions each, and left multiplication acts on them irreducibly. The material sector is the anti-Hermitian subspace, four real dimensions, and under conjugation it splits into the fixed line $\mathbb{R}(ie_t)$ and the three-dimensional $\mathfrak{k}$, which is the vector representation.

The invariants are what the carrier fixes. On the ideals $N=\langle\tilde Q\rangle_{\natural}=Q_t^2+B^2$ vanishes identically; on the centre $B^2$ vanishes identically; on the material sector neither is constrained. The three carriers are therefore the three faces of the partition by $(N,B^2)$ — the degenerate-in-$N$ face, the degenerate-in-$B^2$ face, and the whole — and the four sectors of the companion article on regularity are read off from them.

Why the third carrier is not a module. Left multiplication does not preserve the traceless part: the product of an arbitrary element with a traceless element is not traceless. The adjoint action does preserve it. A carrier is defined by the action, so the traceless part is a carrier of the adjoint action and not a module of the left action, and the two facts are compatible because they are about different actions. The earlier form of this article treated the absence of a module structure as an exclusion of spin one; under the definition the corpus now uses, the carrier is the subspace with its action, and the spin follows from the dimension of the representation space. Spin one is in the algebra.

What a Module Can Carry

The defining module is spin one-half

Under the rotation subgroup the defining module is the fundamental representation,

$$ S\big|_{SU(2)} = \tfrac12, \qquad \dim_\mathbb{C} S = 2, $$

with weight set $\{+\tfrac12,-\tfrac12\}$. This is not a choice of the framework; it is the statement that the algebra's defining representation is the spinor, which is the same in the standard Lorentz-group classification: the defining module of $\mathrm{SL}(2,\mathbb{C})$ is the $(\tfrac12,0)$, and its restriction to $SU(2)$ is the spin-$\tfrac12$. The companion articles on the spinor module and on the spinor representation of the Lorentz group develop the identification in full.

Direct sums carry only spin one-half

A direct sum of $k$ copies of $S$ has weight set

$$ k\times S\big|_{SU(2)}: \qquad k\;\text{copies of}\;\{+\tfrac12,-\tfrac12\}, $$

that is, $k$ copies of the spin-$\tfrac12$ representation and nothing else. A direct sum of irreducible representations contains only those representations as summands; it cannot generate a larger spin, because the weight multiplicities of a direct sum add and the largest weight present is $+\tfrac12$. Hence:

Every finite-dimensional module over the biquaternion algebra carries a finite multiple of spin one-half, and no other spin.

This is a statement about the module category, and its scope is exactly that. It is not a statement about the algebra's carriers: the centre is a carrier of spin zero and the adjoint action is a carrier of spin one, and neither is a module. The statement is sharp and it is used twice below — once to explain why the higher-spin fields are not modules, and once to show that no fourth module-like carrier exists.

The dimension obstruction, and why it is not enough

The modules have even complex dimension, $2k$, since each copy of $S$ contributes two. This excludes every representation of odd complex dimension, and among the Lorentz irreducibles $(s,s_{\mathrm R})$ those of odd dimension include $(1,0)$ and $(0,1)$, of dimension three, and more generally every $(s,0)$ and $(0,s_{\mathrm R})$ with $s$ and $s_{\mathrm R}$ integers. The representations with at least one half-integer label, whose dimension $(2s+1)(2s_{\mathrm R}+1)$ is even, escape this first filter; and the representation $(\tfrac12,\tfrac12)$ also escapes it, because its complex dimension is four. Those cases are excluded by the weight content instead. For the four-vector,

$$ (\tfrac12,\tfrac12)\big|_{SU(2)} = 1\oplus0, \qquad \{+1,0,0,-1\}, $$

which is not the weight content of any direct sum of copies of $S$. A module of dimension four would be $2S$, with weight set $\{+\tfrac12,-\tfrac12,+\tfrac12,-\tfrac12\}$, and that is a different representation of the rotation group. The dimension argument alone would leave the four-vector undecided; the weight argument decides it. This is why the statement about modules is stated in terms of the module category and not in terms of dimensions.

The four-vector escapes the module category and does not escape the algebra. Its $SU(2)$ content $1\oplus0$ is exactly the adjoint content of $\mathbb{B}$, and the four-vector is the two-sided action on the material sector. A representation that is not a module of $\mathbb{B}$ can still be a carrier of $\mathbb{B}$; that is the whole difference between the two words, and it is the difference the corpus's definition of spin turns on.

The Spin-Zero Carrier

The spin-zero part of the algebra is not a gap in the module content. It is a carrier, the centre, and it stands beside the other two in the catalogue.

The centre. $\mathbb{B}_{C} = \{Q_te_t\}$ is the space of scalars, and under conjugation it is the trivial representation of the Lorentz group on a one-dimensional complex space. The centre is not a left module over $\mathbb{B}$ — left multiplication sends $\lambda e_t$ to $\lambda Q$, which leaves the centre as soon as $Q$ is not central — but it is invariant under conjugation, and that is the action that makes it a carrier. The scalar field of the companion Klein–Gordon article lives here: it is an element of the centre, and its spin is zero because scalars are Lorentz invariant.

The invariants of products. The tensor product of the defining module with its conjugate contains a one-dimensional invariant subspace, the trace part, and the antisymmetric square

$$ \boxed{\;\Lambda^2 S \cong \mathbb{C}, \qquad \dim_\mathbb{C}\Lambda^2 S = 1\;} $$

is the trivial representation. Spin zero therefore also appears as the invariant part of a tensor product, and this is the form in which it enters the constraint structure of higher-spin fields.

The endomorphism space. By Schur's lemma, $\mathrm{End}_\mathbb{B}(S) = \mathbb{C}$ is one-dimensional and carries the trivial representation of the Lorentz group, so the space of module maps from the defining module to itself is a spin-zero object. This is the module-theoretic restatement of the centre: the scalar that multiplies a module map is the same scalar that lives in the centre, and the line of the identity endomorphism is where the two descriptions meet.

Why the carrier and the two other objects are one thing. The centre, the trace part of $S\otimes\bar S$, and the endomorphism line are three descriptions of the same trivial representation. The first is the carrier, the second is the construction that produces it inside a tensor product, and the third is its algebraic identity. The spin-zero carrier is therefore native to the algebra in a way that no higher-spin carrier is: it needs no tensor product and no constraint, because the invariant line is already there.

The External Claim of One Equation for All Spins

There is a claim in the literature that appears to contradict the completeness of the catalogue, and it is worth stating why it does not. Gsponer and Hurni, in a four-page contribution to the Cornelius Lanczos centenary conference (1994; arXiv:hep-ph/0112317), report that the quaternionic equation of Lanczos, in the generalized form whose Lagrangian they attribute to Einstein and Mayer (1933), describes particles of spin $0$, $\tfrac12$, $1$ and $\tfrac32$ in one framework. Read as a statement about modules over $\mathbb{B}$, that claim is false: the algebra's module category has one simple object, every module is a sum of copies of the two-complex-dimensional defining module, and the spin content of any module is therefore a whole number of copies of spin one-half. Read as a statement about carriers, it is half true: the algebra does carry $0$, $\tfrac12$ and $1$, and the $\tfrac32$ needs one tensor factor.

The claim is not about modules over $\mathbb{B}$, and it survives the module statement because of what it is about. The Einstein–Mayer construction promotes the mass to a Clifford coefficient and treats wave equations of different spins as semivectors of one larger system — the carrier is a tensor object, graded by the spin it carries, exactly the object the previous section describes as a module of $\mathbb{B}^{\otimes 2s}$ rather than of $\mathbb{B}$. The corpus's own higher-spin fields have the same status: the Proca field is a biquaternion-valued four-vector, the Rarita–Schwinger field is a biquaternion-valued four-vector further tensored with the defining module, and neither is a module over the base algebra. A single generalized equation over such a carrier can therefore contain several spins at once, and the claim's "unified description" is a statement about the carrier, not a counterexample to the classification.

Three qualifications keep the comparison honest, and all three are properties of the source rather than of the algebra. It is four pages, it exhibits no derivation, and its title is a question. The spin content is asserted, not exhibited; in particular no constraint structure is given that would select the pure spin-$s$ parts from the carriers, which is the work this article's spin-one and spin-three-halves sections do explicitly. And the equality of the two statements — "one equation for all spins" and "the algebra carries the spins $0$, $\tfrac12$ and $1$" — is the corpus's interpretation of the former, not an equivalence asserted by the source. What the source does establish, if its construction is sound, is a point of history: that the generalized-mass programme of the 1930s was already a higher-spin unification.

One part of the source's spin content is exhibited, and it points the same way as the catalogue. The paper lists the four rest-frame solutions of the Dirac–Lanczos equation as $D_0 = 1$, $\vec e_x$, $i\vec e_z$ and $i\vec e_y$ — spin up and down, particle and antiparticle — and then obtains the spin-$\tfrac32$ solutions by applying the operator $I = (\,)i\vec e_x$: "another quartet of solutions … together with the first one make the spin $\tfrac32$ solutions", while "pure spin $\tfrac12$ states are thus eigenstates of $I$". The higher-spin carrier is therefore built from the spin-$\tfrac12$ quartet by one further operator, which is a doubling of the lower carrier and not a counterexample to the classification above. It also supplies part of the selection statement that the qualifications above ask for: the eigenstates of $I$ are exactly the pure spin-$\tfrac12$ states, so $I$ is a constraint operator of the kind the spin-one and spin-three-halves sections of this article construct explicitly. What remains absent is the constraint structure for the spin-$\tfrac32$ part itself — the $\gamma$-tracelessness of Rarita–Schwinger — which the source does not give.

Tensor Products: the Generation of Higher Spin

The decomposition rule

Everything beyond spin one is obtained by tensoring the defining module with itself. The tensor powers decompose by the standard Clebsch–Gordan series for products of spin-$\tfrac12$: the highest spin in $S^{\otimes n}$ is $\tfrac n2$, the spins fall in integer steps, and the multiplicity of spin $s$ is

$$ n_s = \binom{n}{\frac n2 - s} - \binom{n}{\frac n2 - s - 1}, $$

with the binomial coefficient taken to vanish for a negative lower index. The small cases, computed from the weights, are

$$ S^{\otimes2} = 1\oplus0, \qquad S^{\otimes3} = \tfrac32\oplus2\times\tfrac12, \qquad S^{\otimes4} = 2\oplus3\times1\oplus2\times0, $$

of dimensions $4, 8, 16$, in agreement with the multiplicities $n_1=n_0=1$ for $n=2$, $n_{3/2}=1$, $n_{1/2}=2$ for $n=3$, and $n_2=1$, $n_1=3$, $n_0=2$ for $n=4$. The pattern is that spin $s$ appears for the first time in $S^{\otimes 2s}$: spin one in the square, spin three-halves in the cube, spin two in the fourth power, and in general the $2s$-fold product.

Spin one is already in $S^{\otimes2}$, and it is already in $\mathbb{B}$. The square contains $1\oplus0$, and the spin-one summand is the vector representation, which the material sector of the base algebra carries under conjugation. So the first entry of the series is not a new object: it is the adjoint carrier, seen from the tensor side. The higher entries are new, and they need the powers.

Symmetric powers and the pure spin carriers

The product $S^{\otimes 2s}$ contains spin $s$ together with lower spins, and the pure spin-$s$ carrier is the totally symmetric part. Its weight set is $\{+s,+s-1,\dots,-s\}$, one of each, so

$$ \boxed{\;\mathrm{sym}^{2s}(S)\big|_{SU(2)} = s, \qquad \dim_\mathbb{C}\mathrm{sym}^{2s}(S) = 2s+1\;}, $$

and its complex dimension is $2s+1$, odd: the carrier is not a direct sum of copies of $S$, which is the module-theoretic reason that a pure integer-spin carrier above one is not a module over $\mathbb{B}$. The antisymmetric parts carry the lower spins: $\Lambda^2 S$ is the trivial representation, of spin zero, as recorded above. In the Lorentz-group language the symmetric powers give the two conjugate three-dimensional families

$$ \mathrm{sym}^{2s}(S) = (s,0), \qquad \mathrm{sym}^{2s}(\bar S) = (0,s), $$

which for $s=1$ are the self-dual and anti-self-dual halves of the companion spin-one article. The general $(s,s_{\mathrm R})$ representation is reached by mixing the two families,

$$ (s,s_{\mathrm R}) \;=\; \mathrm{sym}^{2s}(S)\otimes_\mathbb{C}\mathrm{sym}^{2s_{\mathrm R}}(\bar S), $$

so the whole finite-dimensional representation theory of the Lorentz group is generated by the defining module and its conjugate, at the level of tensor products.

The Explicit Cases

The two preceding articles of the subcategory are the concrete instances of the general rule, and it is worth reading them against it.

Spin one

Spin one is spin $s=1$, and its pure carriers are the symmetric squares

$$ \mathrm{sym}^2(S) = (1,0), \qquad \mathrm{sym}^2(\bar S) = (0,1), $$

of complex dimension three each. These are the self-dual and anti-self-dual halves of the field strength, and the companion spin-one article shows that they are carried by the complex three-dimensional vector part of the algebra, on which the Hodge dual is minus left multiplication by $i$. The four-vector representation $(\tfrac12,\tfrac12)$ from which the two-form is built is $S\otimes\bar S$, of dimension four; its weight content is $1\oplus0$, not the weight content of any direct sum of copies of $S$. The Proca field of the companion article is an element of the material sector, which is a real form of $S\otimes\bar S$, and its equation reduces it to the three physical states of a massive spin-one particle. The construction uses the tensor product twice: once to make the four-vector from $S$ and $\bar S$, and once to make the field strength's two halves as symmetric squares.

The base case of the catalogue. Spin one needs no higher tensor power, because the algebra already holds the adjoint carrier. What the tensor product adds is the two-form, which is the antisymmetric part of the square rather than the square itself. So the relation between the algebra and its first tensor power is a relation of two subspaces of one object: the traceless part of $\mathbb{B}$ under conjugation, and the alternating square of the module.

Spin three-halves

Spin three-halves is spin $s=\tfrac32$, and its carrier is obtained by tensoring the vector with the spinor,

$$ \left(\tfrac12,\tfrac12\right)\otimes\left[\left(\tfrac12,0\right)\oplus\left(0,\tfrac12\right)\right] = \left(1,\tfrac12\right)\oplus\left(0,\tfrac12\right)\oplus\left(\tfrac12,1\right)\oplus\left(\tfrac12,0\right), $$

of dimension sixteen, whose $SU(2)$ content is $2\times\tfrac32\oplus4\times\tfrac12$. The Rarita–Schwinger field of the companion article is exactly this object, and its constraints — the algebraic trace and the subsidiary conditions — remove the four spin-$\tfrac12$ multiplets, leaving the pure spin-$\tfrac32$ quartet. The construction uses the tensor product threefold, and the price of the higher spin is visible in the constraint: the product contains far more lower spin than high spin, and the constraints are what select the wanted piece.

Where the catalogue stops. $\tfrac32$ is the first spin with no carrier in the base algebra. Its carrier is a product of the vector carrier with the spinor carrier, that is, a product of the two carriers of different character — one adjoint, one modular — and that mixed product is what the base algebra cannot supply.

The pattern

The pattern that the two cases establish is general. To carry spin $s$ one needs the tensor product $S^{\otimes 2s}$, whose dimension grows as $2^{2s}$ while the pure spin-$s$ part grows only as $2s+1$; the difference is lower-spin admixture, and a field equation is required to remove it. The lower the spin, the smaller the admixture: spin one needs only the square, and its two-form needs no constraint beyond the antisymmetry; spin three-halves needs the cube, and its equation carries the algebraic trace and the subsidiary conditions; spin two, in the fourth power, needs the constraints that make a symmetric tensor into a graviton. None of these carriers is a module over the algebra, and the reason is always the same: their $SU(2)$ content is not a multiple of the defining module's.

Spin two, for the pattern's sake

The next case shows the growth of the admixture. Spin two appears first in $S^{\otimes4}$, whose $SU(2)$ content is $2\oplus3\times1\oplus2\times0$; the pure spin-two carrier is the symmetric fourth power,

$$ \mathrm{sym}^4(S) = (2,0), \qquad \dim_\mathbb{C}\mathrm{sym}^4(S) = 5, $$

and its conjugate is $(0,2)$. The symmetric traceless rank-two tensor that carries the physical graviton is a real form of the product

$$ \left[(\tfrac12,\tfrac12)\otimes(\tfrac12,\tfrac12)\right]_{\mathrm{sym},\,\mathrm{traceless}} = (1,1), $$

of complex dimension nine, restricted by the trace and divergence conditions; the two helicities $\pm2$ of the massless graviton are what survive. The dimension of the parent product is $4\times4 = 16$, of which the pure spin-two content is a small part; the same phenomenon as before, now with a larger excess. The algebra supplies the product and the conjugation, and the constraints select the spin.

Spin two is where §Tensor Products and the Absence of a Spin-Two Carrier of the mathematics article bites. The tensor square of the adjoint module does contain a $\mathcal{V}_4$, and the multiplication map of the algebra kills it. So the spin-two carrier is available in $\mathbb{B}\otimes\mathbb{B}$ as a representation and not as a product of two algebra elements; the graviton is built from two four-vectors, not from one biquaternion.

Tensor Powers of the Algebra

The generation of higher spin can also be described at the level of the algebra rather than of its module, and the description makes the role of the tensor product explicit.

Claim. The $n$-fold tensor power of the biquaternion algebra is a full matrix algebra on the $n$-fold tensor power of the defining module,

$$ \boxed{\;\mathbb{B}^{\otimes n} \;\cong\; M_{2^n}(\mathbb{C}), \qquad \text{its defining module} \;\cong\; S^{\otimes n}, \qquad \dim_\mathbb{C}\left(S^{\otimes n}\right) = 2^n\;}. $$

Proof sketch. The algebra is $M_2(\mathbb{C}) = \mathrm{End}_\mathbb{C}(S)$, and the tensor product of endomorphism algebras is the endomorphism algebra of the tensor product of the modules, $\mathrm{End}(S)\otimes_\mathbb{C}\mathrm{End}(S) \cong \mathrm{End}(S\otimes_\mathbb{C}S)$; iterating gives the statement.

The claim says that the operator that carries spin $s$ is available in the algebra's $2s$-th tensor power, whose defining module is $S^{\otimes 2s}$ and contains $\mathrm{sym}^{2s}(S)$ as the pure spin-$s$ carrier. The base algebra, which is the first tensor power, holds the three carriers of the catalogue; the second tensor power contains the four-vector and the two-form; the fourth contains the spin-two carrier. Higher spin is thus not a carrier of the biquaternion algebra but a carrier of its tensor powers, and each power is a larger matrix algebra with a correspondingly larger category of modules — and a correspondingly larger catalogue of adjoint carriers.

This is the algebra-level form of the catalogue statement. There is no tension between "the base algebra carries the spins $0$, $\tfrac12$ and $1$" and "spin $s$ appears in $\mathbb{B}^{\otimes 2s}$": the first is a statement about the subcarriers of $\mathbb{B}$, the second about the modules of a larger algebra in which $\mathbb{B}$ sits diagonally. The fields of the corpus that carry higher spin are constructed on the larger algebra, and the framework's notation is a convenient way of writing them down, not a claim that the base algebra contains them.

Why There Is No Fourth Carrier

The catalogue has three entries, and the reason is a decomposition and not a dimension count.

The two actions and their isotypic components. The algebra has two multiplications that make a subspace invariant. Left multiplication gives the regular module $\mathbb{B}\cong2S$, whose irreducible summands are the two minimal left ideals; so the left action contributes one carrier, of spin $\tfrac12$, in two copies. Conjugation gives the adjoint module $\mathbb{B}\cong Z\oplus V$, whose isotypic components are the centre and the scalar-free part; so the adjoint action contributes two carriers, of spins $0$ and $1$. Two actions, three isotypic components, three carriers. A fourth carrier would need a third action or a fourth isotypic component, and the algebra has neither.

The dimension count agrees and is weaker. The representation spaces have dimensions $1$, $2$ and $3$, and $1+2+3=6>4=\dim_{\mathbb{C}}\mathbb{B}$, so no single module holds all three. That count is a consequence, not the reason: it is the reason no one action carries all three spins, and the reason there are three carriers is the decomposition above.

The four-vector is the same carrier, twice seen. The two-sided action on the material sector gives the four-vector $(\tfrac12,\tfrac12)$, of $SU(2)$ content $1\oplus0$. That is the adjoint content of the algebra itself, and the two carriers present the same multiset $\{+1,0,0,-1\}$ to the rotation group. They are one carrier written in two coordinates — the algebra's own conjugation, and the Lorentz two-sided action restricted to rotations — and counting them as two would be counting one object twice.

The multiplication map closes the last door. One might hope for a fourth carrier in a product: the tensor square of the adjoint module contains a spin-two summand $\mathcal{V}_4$, and a carrier of spin two inside $\mathbb{B}\otimes\mathbb{B}$ would be a fourth entry with a natural home. The multiplication map of the algebra is equivariant and kills that summand — the product of two biquaternions has no spin two — so the summand survives as a representation and not as a product of algebra elements. The mathematics article proves the vanishing; its physical reading is that a spin-two field cannot be a single biquaternion, which is the statement that the graviton needs two indices.

The adjoint action and the spin-one carrier

The companion spin-one article exhibits the spin-one structure inside $\mathbb{B}$: the inner derivations $D_\eta = \tfrac12\mathrm{ad}_{\vec e_\eta}a$, with $\mathrm{ad}_{\vec e_\eta}a(\tilde{Q})=[\vec e_\eta,\tilde{Q}]$, act on the scalar-free part of the algebra as a three-dimensional angular momentum, and the scalar-free part is four real dimensions less its scalar line, that is, exactly the carrier of spin one.

The carrier and the module are different objects, and both exist. The scalar-free part is not a submodule of $\mathbb{B}$ under left multiplication — the product of an arbitrary element with a scalar-free element is not scalar-free — and it is a subcarrier under conjugation. Nothing follows about its spin from the first fact, because the spin is read from the action that makes it a carrier. Under the definition the corpus uses, the adjoint action is the carrier, and its dimension gives $s=1$.

The bimodule reading. The conjugation action $\tilde{U}\tilde{Q}\tilde{U}^{*}$ is a bimodule action, and a bimodule over $\mathbb{B}$ is a module over $\mathbb{B}\otimes_\mathbb{R}\mathbb{B}$; correspondingly, the scalar-free part is the product $S\otimes\bar S$ with its trivial summand removed. This is a true and useful statement about where the carrier comes from. It is not a reason to deny it the name carrier: the carrier of a field is the subspace the field transforms in, and the scalar-free part is that subspace.

Why the restriction to $G$ matters for the label. The trace survives only because $\tilde{U}^{*}=\tilde{U}^{-1}$ in $G$. A general Lorentz element does not preserve it: it sends $\lambda e_t$ to $\lambda\tilde{U}\tilde{U}^{*}$, which is not a multiple of $e_t$. Under the full Lorentz group $\mathbb{B}$ is irreducible, carrying the four-vector $(\tfrac12,\tfrac12)$. The split $Z\oplus V$ is therefore a split under the rotations, not under the Lorentz group, as the companion article on the Lorentz representations records; the spin is a rotation label, and the carrier that holds it is a rotation carrier.

A spin-one module does not exist

It is instructive to try to build a module carrying spin one and to see the attempt fail, because the failure marks the boundary of the module category.

The candidate would be a three-dimensional complex space $V$ on which $\mathbb{B}$ acts linearly, with the action restricting to spin one. A three-dimensional $\mathbb{B}$-module would have dimension three, and every module has even dimension. Relaxing to four dimensions, the module would be $2S$, whose rotation content is two copies of spin one-half; no rearrangement of the two copies can produce a spin-one multiplet, because the action of $\mathbb{B}$ is fixed and is the fundamental on each copy. The only way to obtain spin one from two spin-$\tfrac12$'s is the tensor product $\tfrac12\otimes\tfrac12 = 1\oplus0$, and the algebra does realise that product: the traceless part is its spin-one summand. But the adjoint action is not a module structure, and the only simple module over $\mathbb{B}$ is two-dimensional, so no spin-one module exists.

What the failure means. It means that spin one is not accessible to left multiplication, and it says nothing against the carrier. The corpus's fields of spin one are not modules: the Proca potential is an element of $\mathbb{B}_{\Phi}$, transformed by conjugation; the field strength is an element of the vector part, transformed by the adjoint action. The transformations are conjugation and adjoint, which is the algebraic mark of the second action, and the second action is a perfectly good carrier structure.

The Ceiling in Context

The obstruction above spin one is not peculiar to the biquaternion algebra, and it is worth saying what is and is not special about this case.

It is a general feature of the spinor description of the Lorentz group that the reducible representations carrying integer spin are built from products of spinor representations: the vector is a product of a spinor and a conjugate spinor, the two-form is a product of two spinors of the same type, the graviton is a product of four, and so on. This is the standard statement that the Lorentz group's representations are labelled by pairs of half-integers and that the integer-spin representations arise when the two half-integer labels combine to an integer sum. The biquaternion algebra's contribution is to realise the spinor module concretely as a Peirce carrier, which here is a minimal left ideal, so that the tensor products become products of ideals and the symmetric parts become the symmetric powers of a two-dimensional complex space.

What is special to this framework is the centre. Because the algebra is $\mathbb{C}\otimes_\mathbb{R}\mathbb{H}$, its centre is the complex scalars and it carries the scalar field natively, with the scalar imaginary $i$ as the algebra's central phase. The spin-zero carrier is therefore not an accident of an invariant subspace but a distinguished subalgebra, and the Klein–Gordon field lives there. The half-integer carrier is the module, the integer carrier is the adjoint, and everything above is a tensor power. The division of the framework's field content into a scalar part, a spinorial part, an adjoint part and a tower of tensor-product parts is exactly the division of the algebra's representations into the centre, the defining module, the traceless part, and the tensor powers — and that is the content of the catalogue statement.

The Same Catalogue in the Clifford Algebra

The biquaternion algebra is the even subalgebra of the Clifford algebra of spacetime, $\mathbb{B}\cong \mathrm{Cl}_{1,3}^{+}$, and it is worth noting that the catalogue is not an artefact of taking the even part. The full Clifford algebra is

$$ \mathrm{Cl}_{1,3} \cong M_2(\mathbb{H}), \qquad \dim_\mathbb{R}\mathrm{Cl}_{1,3} = 16, $$

which is again simple, over the real numbers; its complexification is the full matrix algebra $M_4(\mathbb{C})$, whose irreducible module is the four-dimensional Dirac spinor, of spin one-half. The same division of labour therefore holds there: the spinor is the module, the vector and the two-form are carried by the algebra under conjugation and by tensor products of the spinor module, and a module of the Clifford algebra carries no integer spin. The catalogue is a feature of the spacetime Clifford algebra as a whole, and the biquaternion formulation inherits it. What the biquaternion formulation adds is the concrete identification of the even part with the algebra of the central complex structure, which is what makes the centre available as a distinguished carrier for the scalar field.

The point has a standard physical interpretation: the objects that transform in the vector representation of the Lorentz group are not spinorial objects, and the reason a Dirac spinor is the natural home of the electron while a four-vector is the natural home of the photon is that the two are carried by different structures — the module and the algebra. The catalogue is the algebraic form of that distinction, and it explains why the framework must introduce fields beyond the algebra's modules for every spin above one-half.

The conventions and the explicit cases used above are those of five companion articles:

  • Companion article The Spinor Module in Biquaternionic Form and Its Lorentz Action, for the defining module, its conjugate, and the Lorentz action.
  • Companion article The Spinor Representation of the Lorentz Group in Biquaternionic Form, for the $(s,s_{\mathrm R})$ classification and the restriction to the rotation subgroup.
  • Companion article Maxwell's Equations in the Biquaternionic Formulation, for the four-vector and field-strength content of the spin-one case.
  • Companion article The Proca Equation: Massive Spin 1 in Biquaternionic Form, for the explicit spin-one case and its constraint structure.
  • Companion article The Rarita–Schwinger Equation: Spin 3/2 in Biquaternionic Form, for the explicit spin-three-halves case and its constraint structure.

The Two Index Families of the Bimodule

The algebra is not only an algebra over $\mathbb{C}$; it is a rank-two bimodule over $\mathbb{H}$ on each side, with two commuting actions (the mathematics article Modules over the Quaternion Algebra). Read on the spinor, the two commuting actions are two spinor index families: the left action carries one index and the right action the other, so that the four-vector module $S\otimes\bar S$ is the product of the two families, and the defining module $S$ is the carrier of one family alone.

The interpretation gives the module count of this article a second form. The module category of the algebra over $\mathbb{C}$ carries only spin one-half; over $\mathbb{H}$ the same modules carry two index families, and the failure of full $\mathbb{H}$-bilinearity — the module being of rank two and not of rank one — is read as the obstruction to treating the two families as independent. It is a labelled interpretation, and it is offered as the bimodule form of the catalogue this article establishes. The owner is the mathematics article Modules over the Quaternion Algebra.

Summary

The biquaternion algebra is $\mathbb{C}\otimes_\mathbb{R}\mathbb{H}\cong M_2(\mathbb{C})$, a simple algebra, and it carries three carriers: the centre $\mathbb{B}_{C}$ with spin $0$, the minimal left ideals $S_\pm$ with spin $\tfrac12$, and the material sector $\mathbb{B}_{\Phi}$ with spin $1$ on its scalar-free part. Two actions produce them — left multiplication, which gives the ideals, and conjugation, which gives the centre and the traceless part — and there is no third action, so there is no fourth carrier.

A module over the algebra carries only spin one-half. Every finite-dimensional module is a direct sum of copies of the two-dimensional $S$, by Artin–Wedderburn and Schur, so a module's rotation content is a multiple of spin one-half. The dimension argument alone is insufficient to exclude the four-vector, whose dimension four is even but whose weight content is $1\oplus0$; the weight argument decides it. The correct conclusion is not that the algebra has no spin one — it has, as the adjoint carrier — but that a module has none.

Spin zero is the centre, and it is native: the trivial representation of the algebra on a one-dimensional space, the same object as the invariant line $\Lambda^2 S\cong\mathbb{C}$ and as the endomorphism space $\mathrm{End}_\mathbb{B}(S)=\mathbb{C}$. The scalar field of the framework lives there, which is why the scalar carries no tensor index and no constraint.

Higher spin is generated by tensor products. The tensor power $S^{\otimes n}$ decomposes into spins $\tfrac n2,\tfrac n2-1,\dots$, so spin $s$ appears first in $S^{\otimes 2s}$, and the pure spin-$s$ carrier is the symmetric power $\mathrm{sym}^{2s}(S)=(s,0)$, of complex dimension $2s+1$; the antisymmetric parts carry the lower spins, with $\Lambda^2 S$ the scalar. Spin one is the square, and its summand is already the adjoint carrier of the base algebra; spin three-halves is the cube, realised by the Rarita–Schwinger vector-spinor, whose constraints remove the spin-$\tfrac12$ admixture; spin two is the fourth power, and the algebra's multiplication map kills the spin-two summand of $\mathbb{B}\otimes\mathbb{B}$, which is why the graviton needs two indices and not one biquaternion. The carrier of spin $s$ is the module of the $2s$-th tensor power, $\mathbb{B}^{\otimes 2s}\cong M_{2^{2s}}(\mathbb{C})$, whose defining module is $S^{\otimes 2s}$.

The completeness of the catalogue is therefore a statement about the base algebra's two actions, and the growth of spin with tensor powers is a statement about its powers. The two are consistent because they are about different algebras, and the base algebra's three carriers are exactly the three spins a field can have without a second index.

Summary of Notation

Symbol Meaning
$\mathbb{B}=\mathbb{C}\otimes_\mathbb{R}\mathbb{H}\cong M_2(\mathbb{C})$ Biquaternion algebra, simple, centre the complex scalars
$e_t=1,\vec e_x,\vec e_y,\vec e_z$ Quaternion basis, $\vec e_\eta^2=-e_t$
$i$ Central scalar imaginary, the algebra's central phase
$\mathbb{B}_{C}=\{Q_te_t\}$ Centre, the spin-zero carrier
$S_\pm=\mathbb{B}\tilde\Pi_\pm$ Minimal left ideals, the spin-$\tfrac12$ carriers
$\mathbb{B}_{\Phi}=\mathbb{R}(ie_t)\oplus\mathfrak{k}$ Material sector, the spin-one carrier on $\mathfrak{k}$
$S$ Defining (spinor) module, $\dim_\mathbb{C}S=2$, $S|_{SU(2)}=\tfrac12$
$S^{\oplus k}$ General module; every module is of this form
$\mathrm{Hom}_\mathbb{B}(S,M)$ Multiplicity space of a module, a complex vector space
$N=Q_t^2+B^2$ Quaternionic norm; $N=0$ on the ideals
$B^2$ Vector square; $B^2=0$ on the centre
$\Lambda^2 S\cong\mathbb{C}$ Invariant (spin-zero) part of the antisymmetric square
$S^{\otimes n}$ $n$-fold tensor power, spins $\tfrac n2,\tfrac n2-1,\dots$
$\mathrm{sym}^{2s}(S)=(s,0)$ Pure spin-$s$ carrier, complex dimension $2s+1$
$(\tfrac12,\tfrac12)=S\otimes\bar S$ Four-vector representation; the adjoint content $1\oplus0$
$\mathbb{B}^{\otimes n}\cong M_{2^n}(\mathbb{C})$ Tensor powers; defining module $S^{\otimes n}$
$(1,0)\oplus(0,1)$ Self-dual and anti-self-dual halves of the field strength

Further Reading

  • Rotor, Spin, Pair and Carrier: A Survey of the Closure Block, for the three carriers named and read physically, with the particle assignments.
  • From the Spin to the Carriers and the Closures, for the carriers, the quadratic element, the criterion and the proof that multiplication kills spin two.
  • Rotor, Spin, Pair and Carrier: A Survey of the Closure Block, for the partition by $N$ and $B^2$ that the three carriers cut.
  • The Self-Dual and Anti-Self-Dual Split: Spin One from the Biquaternion Material Sector, for the spin-one case.
  • The Rarita–Schwinger Equation: Spin 3/2 in Biquaternionic Form, for the spin-$\tfrac32$ case and its constraint structure.
  • Modules over the Quaternion Algebra, for the bimodule structure and the two index families.
  • Richard Brauer and Hermann Weyl, "Spinors in $n$ dimensions", American Journal of Mathematics 57 (1935) 425–449, for the spinor representations and their tensor products.
  • Michael Artin, Algebra (Prentice Hall, 1991), for the Artin–Wedderburn theorem, simple algebras, and the structure of modules over a matrix algebra.
  • Irving Kaplansky, Fields and Rings (Chicago, 1972), for the Wedderburn theorems and the classification of semisimple rings.
  • Steven Weinberg, The Quantum Theory of Fields, Vol. I: Foundations (Cambridge, 1995), for the $(s,s_{\mathrm R})$ classification of the finite-dimensional Lorentz representations and the reduction of their tensor products.
  • Roger Penrose and Wolfgang Rindler, Spinors and Space-Time (Cambridge, 1984), for the two-spinor description of the Lorentz representations and the symmetric spinor products that carry integer spin.
  • Pertti Lounesto, Clifford Algebras and Spinors (Cambridge, 2001), for minimal left ideals, the spinor module, and its relation to the even Clifford algebra.
  • Morris L. Curtis and Irving Reiner, Representation Theory of Finite Groups and Associative Algebras (Interscience, 1962), for the module theory of semisimple algebras and the Morita equivalence used here.
  • J. D. Bjorken and S. D. Drell, Relativistic Quantum Fields (McGraw-Hill, 1965), for the field-theoretic realisation of integer and half-integer spin from spinor products.
  • A. Gsponer and J.-P. Hurni, "Lanczos's Equation to Replace Dirac's Equation?", Proceedings of the Cornelius Lanczos International Centenary Conference (SIAM, 1994) 509–512 (arXiv:hep-ph/0112317), for the Einstein–Mayer generalized-mass claim of one equation describing spins $0,\tfrac12,1,\tfrac32$, recorded in the section The External Claim of One Equation for All Spins as a claim about a tensor carrier and not about modules over $\mathbb{B}$. Four pages, no derivation exhibited; the title is a question.