Dirac 场是满足 Dirac 方程的旋量场 ψ(x)\psi(x)。现在我们尝试将 Dirac 场量子化。

我们从自由 Dirac 场的拉格朗日量出发:

L=ψˉ(im)ψ(1)\mathcal{L} = \bar{\psi}(i\cancel{\partial} - m)\psi \tag{1}

正则动量为:

π=Lψ˙=ψˉiγ0=iψ(2)\pi = \frac{\partial \mathcal{L}}{\partial \dot{\psi}} = \bar{\psi}i\gamma^0 = i\psi^\dagger \tag{2}

于是哈密顿量可以写为:

H=d3xψˉ(iγ+m)ψ=d3xψ(iγ0γ+γ0m)ψ(3)\begin{aligned} H &= \int d^3x\bar{\psi}(-i\bm{\gamma}\cdot\nabla+m)\psi\\ &= \int d^3x \psi^\dagger(-i\gamma^0\bm{\gamma}\cdot\nabla+\gamma^0m)\psi \end{aligned}\tag{3}

Dirac 场的量子化

对 Dirac 场进行量子化的尝试

我们尝试仿照对 Klein-Gordon 场量子化的方法对 Dirac 场进行量子化。我们先在薛定谔绘景下讨论:对于 Dirac 场,其平面波解在上一篇中已经找到:

us(p)eipxvs(p)eipxs=1,2u^s(\bm{p})e^{i\bm{p}\cdot\bm{x}}\quad v^s(\bm{p})e^{-i\bm{p}\cdot\bm{x}}\quad s= 1,2

或者写为:

us(p)eipxvs(p)eipxs=1,2u^s(\bm{p})e^{i\bm{p}\cdot\bm{x}}\quad v^s(-\bm{p})e^{i\bm{p}\cdot\bm{x}}\quad s= 1,2

我们同样在动量表象中写出:

ψ(x)=d3p(2π)3eipxψ(p)\psi(\bm{x}) = \int \frac{d^3p}{(2\pi)^3}e^{i\bm{p}\cdot\bm{x}}\psi(\bm{p})

那么 ψ(p)\psi(\bm{p}) 应当是 us(p)u^s(\bm{p})vs(p)v^s(-\bm{p}) 的线性组合。考虑到 ψ\psi 为复值场,我们需要使用两套阶梯算子,于是得到:

ψ(p)=s=1,2(a^psus(p)+b^psvs(p))\psi(\bm{p}) = \sum_{s=1,2}(\hat{a}_{\bm{p}}^su^s(\bm{p}) + \hat{b}_{-\bm{p}}^sv^s(-\bm{p}))

那么:

ψ(x)=d3p(2π)312Epeipxs=1,2(apsus(p)+bpsvs(p))(4)\psi(\bm{x}) = \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}e^{i\bm{p}\cdot\bm{x}}\sum_{s=1,2}(a_{\bm{p}}^su^s(\bm{p}) + b_{-\bm{p}}^sv^s(-\bm{p})) \tag{4}

类似有:

ψ(x)=d3p(2π)312Epeipxγ0s=1,2(apsuˉs(p)+bpsvˉs(p))(5)\psi^\dagger(\bm{x}) = \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}e^{-i\bm{p}\cdot\bm{x}}\gamma^0\sum_{s=1,2}(a_{\bm{p}}^{s\dagger}\bar{u}^s(\bm{p}) + b_{-\bm{p}}^{s\dagger}\bar{v}^s(-\bm{p})) \tag{5}

现在我们需要确定一个对易关系进行量子化。我们尝试考虑:假设 aps,bpsa_{\bm{p}}^s,b_{\bm{p}}^s 满足如下对易关系,在 Klein-Gordon 场中我们正是这么做的:

[apr,aqs]=[bpr,bqs]=(2π)3δ(3)(pq)δrs(6)[a_{\bm{p}}^r,a_{\bm{q}}^{s\dagger}]=[b_{\bm{p}}^r,b_{\bm{q}}^{s\dagger}] = (2\pi)^3\delta^{(3)}(\bm{p}-\bm{q})\delta^{rs} \tag{6}

则能验证 ψ(x)\psi(\bm{x}) 具有如下对易关系:

[ψ(x),ψ(y)]=d3pd3q(2π)612Ep2Eqei(pxqy)×r,s=1,2([apr,aqs]ur(p)uˉs(q)+[bpr,bqs]vr(p)vˉs(q))γ0=d3p(2π)312Epeip(xy)s(us(p)uˉs(p)+vs(p)vˉs(p))γ0=d3p(2π)312Epeip(xy)[(γ0Epγp+m)+(γ0Ep+γpm)]γ0=d3p(2π)3eip(xy)=δ(3)(xy)×14×4\begin{aligned} [\psi(\bm{x}),\psi^\dagger(\bm{y})] &= \int \frac{d^3pd^3q}{(2\pi)^6}\frac{1}{\sqrt{2E_{\bm{p}}2E_{\bm{q}}}}e^{i(\bm{p}\cdot\bm{x}-\bm{q}\cdot\bm{y})}\\ &\quad \times\sum_{r,s=1,2}([a_{\bm{p}}^r,a_{\bm{q}}^{s\dagger}]u^r(\bm{p})\bar{u}^s(\bm{q}) + [b_{-\bm{p}}^r,b_{-\bm{q}}^{s\dagger}]v^r(-\bm{p})\bar{v}^s(-\bm{q}))\gamma^0\\ &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}e^{i\bm{p}\cdot(\bm{x}-\bm{y})}\sum_s (u^s(\bm{p})\bar{u}^s(\bm{p})+v^s(\bm{-p})\bar{v}^s(\bm{-p})) \gamma^0\\ &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}e^{i\bm{p}\cdot(\bm{x}-\bm{y})}[(\gamma^0E_{\bm{p}} - \bm{\gamma}\cdot\bm{p} + m)+(\gamma^0E_{\bm{p}} + \bm{\gamma}\cdot\bm{p} - m)]\gamma^0\\ &= \int \frac{d^3p}{(2\pi)^3}e^{i\bm{p}\cdot(\bm{x}-\bm{y})} = \delta^{(3)}(\bm{x}-\bm{y})\times 1_{4\times4} \end{aligned}

这个结果也符合我们的预期。

上述推导中用到了 spin sum completeness relations:

us(p)uˉs(p)=γpmvs(p)vˉs(p)=γp+m\begin{aligned} \sum u^s(p)\bar{u}^s(p) = \gamma\cdot p - m\\ \sum v^s(p)\bar{v}^s(p) = \gamma\cdot p + m\\ \end{aligned}

这样,我们继续将哈密顿量对角化。可以得到:

H=d3xψ(iγ0γ+γ0m)ψ=d3xd3pd3q(2π)6ei(qp)x4EpEqrs(apruˉr(p)+bprvˉr(p))γ0(γ0γq+γ0m)(a^qsus(q)+b^qsvs(q))=d3xd3pd3q(2π)6ei(qp)x4EpEqrs(apruˉr(p)+bprvˉr(p))(γq+m)(a^qsus(q)+b^qsvs(q))=d3p(2π)312Eprs(apruˉr(p)+bprvˉr(p))(γp+m)(a^psus(p)+b^psvs(p))\begin{aligned} H &= \int d^3x \psi^\dagger (-i\gamma^0\bm{\gamma}\cdot\nabla + \gamma^0 m)\psi\\ &= \int d^3x\int \frac{d^3pd^3q}{(2\pi)^6}\frac{e^{i(\bm{q}-\bm{p})\cdot\bm{x}}}{\sqrt{4E_{\bm{p}}E_{\bm{q}}}}\sum_{rs}(a_{\bm{p}}^{r\dagger}\bar{u}^r(\bm{p}) + b_{-\bm{p}}^{r\dagger}\bar{v}^r(-\bm{p}))\gamma^0\\ &\qquad (\gamma^0\bm{\gamma}\cdot\bm{q} + \gamma^0 m)(\hat{a}_{\bm{q}}^su^s(\bm{q}) + \hat{b}_{-\bm{q}}^sv^s(-\bm{q}))\\ &= \int d^3x\int \frac{d^3pd^3q}{(2\pi)^6}\frac{e^{i(\bm{q}-\bm{p})\cdot\bm{x}}}{\sqrt{4E_{\bm{p}}E_{\bm{q}}}}\sum_{rs}(a_{\bm{p}}^{r\dagger}\bar{u}^r(\bm{p}) + b_{-\bm{p}}^{r\dagger}\bar{v}^r(-\bm{p}))\\ &\qquad (\bm{\gamma}\cdot\bm{q} + m)(\hat{a}_{\bm{q}}^su^s(\bm{q}) + \hat{b}_{-\bm{q}}^sv^s(-\bm{q}))\\ &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}\sum_{rs}(a_{\bm{p}}^{r\dagger}\bar{u}^r(\bm{p}) + b_{-\bm{p}}^{r\dagger}\bar{v}^r(-\bm{p}))(\bm{\gamma}\cdot\bm{p} + m)(\hat{a}_{\bm{p}}^su^s(\bm{p}) + \hat{b}_{-\bm{p}}^sv^s(-\bm{p}))\\ \end{aligned}

利用:

uˉs(p)(γp)us(p)=(ξspσˉξspσ)(0σpσp0)(pσξspσˉξs)=ξspσσppσξs+ξspσˉσppσˉξs=(Epσ)pσ+(E+pσ)pσ=2p2\begin{aligned} \bar{u}^s(\bm{p})(\bm{\gamma}\cdot\bm{p})u^s(\bm{p}) &= \begin{pmatrix} \xi^{s\dagger}\sqrt{p\cdot\bar{\sigma}} & \xi^{s\dagger}\sqrt{p\cdot\sigma} \end{pmatrix}\begin{pmatrix} 0 & \bm{\sigma}\cdot\bm{p}\\ -\bm{\sigma}\cdot\bm{p} & 0\\ \end{pmatrix}\begin{pmatrix} \sqrt{p\cdot\sigma}\xi^{s}\\ \sqrt{p\cdot\bar{\sigma}}\xi^{s}\\ \end{pmatrix}\\ &= -\xi^{s\dagger}\sqrt{p\cdot\sigma}\bm{\sigma}\cdot\bm{p} \sqrt{p\cdot\sigma} \xi^s + \xi^{s\dagger}\sqrt{p\cdot\bar{\sigma}}\bm{\sigma}\cdot\bm{p} \sqrt{p\cdot\bar{\sigma}} \xi^s\\ &= -(E-\bm{p}\cdot\bm{\sigma})\bm{p}\cdot\bm{\sigma} + (E+\bm{p}\cdot\bm{\sigma})\bm{p}\cdot\bm{\sigma}\\ &= 2\bm{p}^2\\ \end{aligned}

可得

H=d3p(2π)3s(EpapsapsEpbpsbps)(7)\begin{aligned} H = \int \frac{d^3p}{(2\pi)^3}\sum_s(E_{\bm{p}}a^{s\dagger}_{\bm{p}}a^{s}_{\bm{p}}-E_{\bm{p}}b^{s\dagger}_{\bm{p}}b^{s}_{\bm{p}})\\ \end{aligned}\tag{7}

在这里我们发现:使用 bpb^\dagger_{\bm{p}} 激发更多粒子后,体系的能量反而会下降,那么并不存在一个能量下限。这说明我们这里的量子化是存在一些问题的。我们这里先不作修正,继续在这个基础上进行讨论。


Dirac 场与因果律

我们现在转换到海森堡绘景下讨论 Dirac 场的传播子。

在海森堡绘景中,有:

eiHtapseiHt=apseiEpteiHtbpseiHt=bpseiEpte^{iHt}a_{\bm{p}}^se^{-iHt} = a_{\bm{p}}^se^{-iE_{\bm{p}}t}\quad e^{iHt}b_{\bm{p}}^se^{-iHt} = b_{\bm{p}}^se^{iE_{\bm{p}}t}

根据 (4)(5)(4)(5),立即得到在海森堡绘景中 ψ,ψˉ\psi,\bar{\psi} 的表达式:

ψ(x)=d3p(2π)312Eps=1,2(apsus(p)eipx+bpsvs(p)eipx)ψˉ(x)=d3p(2π)312Eps=1,2(apsuˉs(p)eipx+bpsvˉs(p)eipx)\begin{aligned} \psi(x) &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_{s=1,2}(a_{\bm{p}}^su^s(\bm{p})e^{-ip\cdot x} + b_{\bm{p}}^sv^s(\bm{p})e^{ip\cdot x})\\ \bar{\psi}(x) &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_{s=1,2}(a_{\bm{p}}^{s\dagger}\bar{u}^s(\bm{p})e^{ip \cdot x} + b_{\bm{p}}^{s\dagger}\bar{v}^s(\bm{p})e^{-ip\cdot x})\\ \end{aligned}

为了讨论因果律,我们需要计算如下对易子:

[ψa(x),ψˉb(y)]=d3p(2π)312Eps(uas(p)uˉbs(p)eip(xy)+vas(p)vˉbs(p)eip(xy))=d3p(2π)312Ep((γp+m)abeip(xy)+(γpm)abeip(xy))=(iγx+m)abd3p(2π)312Ep(eip(xy)eip(xy))=(iγx+m)ab[ϕ(x),ϕ(y)]\begin{aligned} [\psi_a(x),\bar{\psi}_b(y)] &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}\sum_{s}(u^s_a(\bm{p})\bar{u}^s_b(\bm{p})e^{-ip\cdot(x-y)}+v^s_a(\bm{p})\bar{v}^s_b(\bm{p})e^{ip\cdot(x-y)})\\ &=\int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}((\gamma\cdot p + m)_{ab}e^{-ip\cdot(x-y)}+(\gamma\cdot p - m)_{ab}e^{ip\cdot(x-y)})\\ &= (i\gamma\cdot\partial_x + m)_{ab}\int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}(e^{-ip\cdot(x-y)}-e^{ip\cdot(x-y)})\\ &= (i\gamma\cdot\partial_x + m)_{ab} [\phi(x),\phi(y)] \end{aligned}

我们发现,最后实际上会得到 Klein-Gordon 传播子。我们之前讨论过:Klein-Gordon 传播子对于类空间隔为零,这意味着 Dirac 传播子也具有同样的性质。

回忆 Klein-Grodon Propagator 的物理意义:

[ϕ(x),ϕ(y)]=D(xy)D(yx)[\phi(x),\phi(y)] = D(x-y)-D(y-x)

分别表示:正粒子从 y\bm{y} 传播 x\bm{x} 的概率幅与反粒子从 x\bm{x} 传播 y\bm{y} 的概率幅的叠加

对于以上得到的 Dirac 传播子来说,它可以表示为:

[ψa(x),ψˉb(y)]=0[ψa(x),ψˉb(y)]0=0ψa(x)ψˉb(x)00ψˉb(x)ψa(x)0\begin{aligned} [\psi_a(x),\bar{\psi}_b(y)] &= \langle 0 | [\psi_a(x),\bar{\psi}_b(y)] |0\rangle \\ &= \langle 0| \psi_a(x)\bar{\psi}_b(x)|0\rangle - \langle 0| \bar{\psi}_b(x)\psi_a(x)|0\rangle \end{aligned}

考虑到任意正/反粒子产生算子 a^ps,b^ps\hat{a}_{\bm{p}}^s,\hat{b}_{\bm{p}}^s 作用于真空态 0|0\rangle 应当有:

a^ps0=b^ps0=0\hat{a}_{\bm{p}}^s|0\rangle = \hat{b}_{\bm{p}}^s|0\rangle = 0

如此 Dirac 传播子的第二项为零。Dirac 传播子将只剩下正粒子从 x\bm{x} 传播到 y\bm{y} 的项。这是我们所不希望看到的,现在我们希望 Dirac传播子中也包含有关反粒子的传播项。结合之前有关体系能量下限的讨论:我们这里对 Dirac 场的量子化不正确的。这意味着:量子化条件 (6)(6) 是不合适的。现在我们以 Dirac 传播子应当包含正/反粒子的传播项作为一个基本的出发点,对 Dirac 场进行量子化。


Dirac 场的量子化

首先考虑 0ψ(x)ψˉ(y)0\langle 0 |\psi(x)\bar{\psi}(y)|0\rangle 应当对应正粒子从 yyxx 的传播。所以:ψˉ(y)0\bar{\psi}(y)|0\rangle0ψ(x)\langle 0|\psi(x) 应当只包含正粒子项。这意味着 ψˉ(y)0\bar{\psi}(y)|0\rangle 应当只包含 apsa^{s\dagger}_{\bm{p}};同理,0ψ(x)\langle 0|\psi(x) 只包含 apsa_{\bm{p}}^s。因此,我们得到 0ψ(x)ψˉ(y)0\langle 0|\psi(x)\bar{\psi}(y)|0\rangle 的形式应当为如下:

0ψ(x)ψˉ(y)0=0d3p(2π)312Epraprur(p)eipx×d3p(2π)312Eqsaqsus(q)eiqy0\begin{aligned} \langle 0|\psi(x)\bar{\psi}(y)|0\rangle &= \langle 0|\int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}} \sum_r a_{\bm{p}}^r u^r(\bm{p})e^{-ipx}\\ &\quad \times \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{q}}}} \sum_s a_{\bm{q}}^{s\dagger} u^s(\bm{q})e^{iqy}|0\rangle\\ \end{aligned}

矩阵元 0apraqs0\langle 0|a_{\bm{p}}^ra_{\bm{q}}^{s\dagger}|0\rangle 的形式应当如下:

0apraqs0=(2π)3δ(3)(pq)δrsA(p)\langle 0|a_{\bm{p}}^ra_{\bm{q}}^{s\dagger}|0\rangle = (2\pi)^3 \delta^{(3)}(\bm{p}-\bm{q})\delta^{rs}\cdot A(\bm{p})

如此可得:

0ψa(x)ψˉb(y)0=(iγx+m)abd3p(2π)312Epeip(xy)A(p)(8)\langle 0|\psi_a(x)\bar{\psi}_b(y)|0\rangle = (i\gamma\cdot\partial_x+m)_{ab}\int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}e^{-ip(x-y)}\cdot A(\bm{p})\tag{8}

现在我们来考虑 A(p)A(\bm{p}) 的性质:

  1. 0apsaps0=aps02>0A>0\langle 0| a_{\bm{p}}^s a_{\bm{p}}^{s\dagger} |0\rangle = |a^{s\dagger}_{\bm{p}}|0\rangle|^2 > 0 \Rightarrow A > 0

0ψ(x)ψˉ(y)0Λ0Λ12ψ(x)ψˉ(y)Λ1210=0ψ(x)ψˉ(y)0\begin{aligned} \langle 0|\psi(x)\bar{\psi}(y)|0\rangle & \overset{\Lambda}{\rightarrow} \langle 0|\Lambda_{\frac{1}{2}}\psi(x)\bar{\psi}(y)\Lambda^{-1}_{\frac{1}{2}}|0\rangle\\ &= \langle 0|\psi(x)\bar{\psi}(y)|0\rangle \end{aligned}

(8)(8) 式为一个洛伦兹不变量,这要求 A(p)A(\bm{p}) 为一个洛伦兹标量。可得:

A(p)=A(p2)=A(m2)A(\bm{p}) = A(p^2) = A(m^2)

AA 为一个正常数。

同理有:

0ψˉb(y)ψa(x)0=(iγx+m)abd3p(2π)312Epeip(xy)B(9)\langle 0|\bar{\psi}_b(y)\psi_a(x)|0\rangle = -(i\gamma\cdot\partial_x+m)_{ab}\int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}e^{ip(x-y)}\cdot B \tag{9}

对于类空间隔,虽然我们不能得到 0[ψ(x),ψˉ(y)]0=0\langle 0|[\psi(x),\bar{\psi}(y)]|0\rangle = 0,但是通过设定 A=B=1A=B=1,将会得到(光锥外):

0ψa(x)ψˉb(y)0=0ψˉb(y)ψa(x)0\langle 0|\psi_a(x)\bar{\psi}_b(y)|0\rangle = -\langle 0|\bar{\psi}_b(y)\psi_a(x)|0\rangle

这仍然能保证因果律。这是因为所有的可观测量都是由偶数个旋量场的乘积项组成。对于任意的可观测量 O1,O2\mathcal{O}_1,\mathcal{O}_2,对于类空间隔均有:

[O1,O2]=0[\mathcal{O}_1,\mathcal{O}_2] = 0

举例说明:考虑 A=ψaψa,B=ψbψb\mathcal{A} = \psi^\dagger_a\psi_a,\mathcal{B} = \psi^\dagger_b\psi_b
那么有:

[A,B]=[ψaψa,ψbψb]=ψaψaψbψbψbψbψaψa=ψaψaψbψb+ψbψaψbψa=ψaψaψbψb+ψaψbψaψb=ψaψaψbψbψaψaψbψb=0\begin{aligned} [\mathcal{A},\mathcal{B}]&= [\psi^\dagger_a\psi_a,\psi^\dagger_b\psi_b]\\ &= \psi^\dagger_a\psi_a\psi^\dagger_b\psi_b-\psi^\dagger_b\psi_b\psi^\dagger_a\psi_a\\ &= \psi^\dagger_a\psi_a\psi^\dagger_b\psi_b+\psi^\dagger_b\psi^\dagger_a\psi_b\psi_a\\ &= \psi^\dagger_a\psi_a\psi^\dagger_b\psi_b+\psi^\dagger_a\psi^\dagger_b\psi_a\psi_b\\ &= \psi^\dagger_a\psi_a\psi^\dagger_b\psi_b-\psi^\dagger_a\psi_a\psi^\dagger_b\psi_b = 0\\ \end{aligned}

因此,关于 Dirac 传播子的讨论启发我们可以使用如下 反对易关系 作为量子化条件:

{ψa(x),ψb(y)}=δ(3)(xy)δab{ψa(x),ψb(y)}={ψa(x),ψb(y)}=0\begin{aligned} \{\psi_a(\bm{x}),\psi_b^\dagger(\bm{y})\} &= \delta^{(3)}(\bm{x}-\bm{y})\delta_{ab}\\ \{\psi_a(\bm{x}),\psi_b(\bm{y})\} &= \{\psi^\dagger_a(\bm{x}),\psi^\dagger_b(\bm{y})\} = 0\\ \end{aligned}

这等价于阶梯算子的如下反对易关系:

{apr,aqs}={bpr,bqs}=(2π)3δ(3)(pq)δrs\{a_{\bm{p}}^r,a_{\bm{q}}^{s\dagger}\}=\{b_{\bm{p}}^r,b_{\bm{q}}^{s\dagger}\} = (2\pi)^3\delta^{(3)}(\bm{p}-\bm{q})\delta^{rs}

在薛定谔绘景中,将 ψ,ψˉ\psi,\bar{\psi} 用阶梯算子表示为:

ψ(x)=d3p(2π)312Eps(apsus(p)eipx+bpsvs(p)eipx)=d3p(2π)312Epseipx(apsus(p)+bpsvs(p))ψˉ(x)=d3p(2π)312Eps(bpsvˉs(p)eipx+apsuˉs(p)eipx)=d3p(2π)312Epseipx(bpsvˉs(p)+apsuˉs(p))\begin{aligned} \psi(\bm{x}) &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_s (a_{\bm{p}}^su^s(\bm{p})e^{i\bm{p}\cdot \bm{x}} + b_{\bm{p}}^{s\dagger}v^s(\bm{p})e^{-i\bm{p}\cdot \bm{x}})\\ &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_s e^{i\bm{p}\cdot \bm{x}}(a_{\bm{p}}^su^s(\bm{p}) + b_{-\bm{p}}^{s\dagger}v^s(-\bm{p}))\\ \bar{\psi}(\bm{x}) &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_s (b_{\bm{p}}^s\bar{v}^s(\bm{p})e^{i\bm{p}\cdot \bm{x}} + a_{\bm{p}}^{s\dagger}\bar{u}^s(\bm{p})e^{-i\bm{p}\cdot \bm{x}})\\ &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_s e^{i\bm{p}\cdot \bm{x}}(b_{\bm{p}}^s\bar{v}^s(\bm{p}) + a_{-\bm{p}}^{s\dagger}\bar{u}^s(-\bm{p}))\\ \end{aligned}

现在将哈密顿量对角化:

H=d3xψˉ(iγ+m)ψ=d3xd3pd3q(2π)6ei(p+q)x4EpEqrs(bpsvˉs(p)+apsuˉs(p))(γq+m)(aqsus(q)+bqsvs(q))=d3p(2π)312Eprs(bpsvˉs(p)+apsuˉs(p))(γp+m)(apsus(p)+bpsvs(p))=d3p(2π)3Eps(apsapsbpsbps)=d3p(2π)3Eps(apsaps+bpsbps)\begin{aligned} H &= \int d^3x \bar{\psi} (-i\bm{\gamma}\cdot\nabla + m)\psi\\ &= \int d^3x\int \frac{d^3pd^3q}{(2\pi)^6}\frac{e^{i(\bm{p}+\bm{q})\cdot\bm{x}}}{\sqrt{4E_{\bm{p}}E_{\bm{q}}}}\sum_{rs}(b_{\bm{p}}^s\bar{v}^s(\bm{p}) + a_{-\bm{p}}^{s\dagger}\bar{u}^s(-\bm{p}))\\ &\qquad (\bm{\gamma}\cdot\bm{q} + m)(a_{\bm{q}}^su^s(\bm{q}) + b_{-\bm{q}}^{s\dagger}v^s(-\bm{q}))\\ &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}\sum_{rs}(b_{\bm{p}}^s\bar{v}^s(\bm{p}) + a_{-\bm{p}}^{s\dagger}\bar{u}^s(-\bm{p}))(-\bm{\gamma}\cdot\bm{p} + m)(a_{-\bm{p}}^su^s(-\bm{p}) + b_{\bm{p}}^{s\dagger}v^s(\bm{p}))\\ &= \int \frac{d^3p}{(2\pi)^3}E_{\bm{p}}\sum_s(a^{s\dagger}_{\bm{p}}a^{s}_{\bm{p}}-b^{s}_{\bm{p}}b^{s\dagger}_{\bm{p}})\\ &= \int \frac{d^3p}{(2\pi)^3}E_{\bm{p}}\sum_s(a^{s\dagger}_{\bm{p}}a^{s}_{\bm{p}}+b^{s\dagger}_{\bm{p}}b^{s}_{\bm{p}})\\ \end{aligned}

真空态满足:

aps0=bps0=0a_{\bm{p}}^s|0\rangle=b_{\bm{p}}^s|0\rangle = 0

这样我们就解决了能量本征值不存在下限的问题。

我们来考虑一组具有反对易关系算子的性质。具体来说,我们有:

{b,b}={b,b}=0{b,b}=1\begin{aligned} &\{b,b\}=\{b^\dagger,b^\dagger\}=0\\ &\{b,b^\dagger\} = 1\\ \end{aligned}

真空态 0|0\rangle 具有性质:

b0=0b|0\rangle = 0

我们定义 b0=1b^\dagger|0\rangle = |1\rangle。那么根据阶梯算子的反对易关系。我们可以得到:

b1=0b1=0b|1\rangle = |0\rangle\quad b^\dagger|1\rangle = 0

于是,bbbb^\dagger 所作用的 Hilbert 空间实际上只有 0,1|0\rangle,|1\rangle 两个态。我们定义 0|0\rangle1|1\rangle 为真空态的这两种描述都是等价的。

我们发现,Dirac 场所描述粒子具有如下性质:

  • 每个态上至多填充一个粒子

(ap)20=0(a_{\bm{p}}^\dagger)^2|0\rangle = 0

  • 波函数具有交换反对称性

apaq0=aqap0a_{\bm{p}}^\dagger a_{\bm{q}}^\dagger|0\rangle = -a_{\bm{q}}^\dagger a_{\bm{p}}^\dagger|0\rangle

这符合 Fermi-Dirac 分布,我们说:Dirac 场描述自旋为 12\frac{1}{2} 的费米子。

时空中的 Dirac 场

经过以上讨论。我们再重新总结一下对 Dirac 场的量子化过程。现在我们在海森堡绘景中讨论。

我们将 ψ,ψˉ\psi,\bar{\psi} 用阶梯算子写为:

ψ(x)=d3p(2π)312Eps(apsus(p)eipx+bpsvs(p)eipx)ψˉ(x)=d3p(2π)312Eps(bpsvˉs(p)eipx+apsuˉs(p)eipx)(10)\begin{aligned} \psi(x) &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_s (a_{\bm{p}}^s u^s(p)e^{-ip\cdot x}+b_{\bm{p}}^{s\dagger} v^s(p)e^{ip\cdot x})\\ \bar{\psi}(x) &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_s (b_{\bm{p}}^s \bar{v}^s(p)e^{-ip\cdot x}+a_{\bm{p}}^{s\dagger} \bar{u}^s(p)e^{ip\cdot x})\\ \end{aligned}\tag{10}

其中阶梯算子满足反对易关系:

{apr,aqs}={bpr,bqs}=(2π)3δ(3)(pq)δrs(11)\{a_{\bm{p}}^r,a_{\bm{q}}^{s\dagger}\}=\{b_{\bm{p}}^r,b_{\bm{q}}^{s\dagger}\} = (2\pi)^3\delta^{(3)}(\bm{p}-\bm{q})\delta^{rs} \tag{11}

等价的,ψ,ψ\psi,\psi^\dagger 满足等时反对易关系:

{ψa(x),ψb(y)}=δ(3)(xy)δab{ψa(x),ψb(y)}={ψa(x),ψb(y)}=0\begin{aligned} &\{\psi_a(x),\psi_b^\dagger(y)\} = \delta^{(3)}(\bm{x}-\bm{y})\delta_{ab}\\ &\{\psi_a(x),\psi_b(y)\} = \{\psi^\dagger_a(x),\psi^\dagger_b(y)\} = 0\\ \end{aligned}

定义真空态 0|0\rangle 满足:

a^p00=b^p00=0\hat{a}_{p}|0\rangle|0\rangle = \hat{b}_{p}|0\rangle|0\rangle = 0

哈密顿量和动量可以表示为:

H=d3p(2π)3sEp(apsaps+bpsbps)P=d3p(2π)3sp(apsaps+bpsbps)\begin{aligned} H &= \int \frac{d^3p}{(2\pi)^3}\sum_s E_{\bm{p}}(a^{s\dagger}_{\bm{p}}a^{s}_{\bm{p}}+b_{\bm{p}}^{s\dagger}b_{\bm{p}}^s) \\ \bm{P} &= \int \frac{d^3p}{(2\pi)^3}\sum_s\bm{p}(a^{s\dagger}_{\bm{p}}a^{s}_{\bm{p}}+b_{\bm{p}}^{s\dagger}b_{\bm{p}}^s) \\ \end{aligned}

利用产生算子,单粒子态可以表示为:

p,s2Epa^p0(12)|\bm{p},s\rangle \equiv \sqrt{2E_{\bm{p}}}\hat{a}_{\bm{p}}^\dagger |0\rangle \tag{12}

如此定义得到的粒子态的内积如下:

q,rp,s=2Ep(2π)3δ(3)(pq)δrs(13)\langle \bm{q},r|\bm{p},s\rangle = 2E_{\bm{p}}(2\pi)^3\delta^{(3)}(\bm{p}-\bm{q})\delta^{rs} \tag{13}

这是洛伦兹不变的。因此,考虑对粒子态 p,s|\bm{p},s\rangle 进行洛伦兹变换 Λ\Lambda 所对应的算子 U(Λ)U(\Lambda) 应当是幺正的。考虑式 (12)(12):

U(Λ)p,s=2EpU(Λ)ap0=2EpU(Λ)apU1(Λ)U(Λ)0=2EpU(Λ)apU1(Λ)0(14)\begin{aligned} U(\Lambda)|\bm{p},s\rangle &= \sqrt{2E_{\bm{p}}}U(\Lambda)a_{\bm{p}}^\dagger |0\rangle\\ &= \sqrt{2E_{\bm{p}}}U(\Lambda)a_{\bm{p}}^\dagger U^{-1}(\Lambda) U(\Lambda) |0\rangle\\ &= \sqrt{2E_{\bm{p}}}U(\Lambda)a_{\bm{p}}^\dagger U^{-1}(\Lambda)|0\rangle\\ \end{aligned}\tag{14}

另一方面,有:

U(Λ)p,s=Λp,s=2EΛpaΛp0(15)\begin{aligned} U(\Lambda)|\bm{p},s\rangle &= |\Lambda\bm{p},s\rangle\\ &=\sqrt{2E_{\Lambda\bm{p}}}a_{\Lambda\bm{p}}^\dagger |0\rangle \tag{15} \end{aligned}

(14),(15)(14),(15) 可以得到:

U(Λ)apsU1(Λ)=EΛpEpaΛps(16)U(\Lambda)a_{\bm{p}}^{s\dagger}U^{-1}(\Lambda) = \sqrt{\frac{E_{\Lambda\bm{p}}}{E_{\bm{p}}}}a^{s\dagger}_{\Lambda\bm{p}} \tag{16}

考虑对场量的洛伦兹变换:

ψ(x)Uψ(x)U1=Ud3p(2π)312Eps(apsus(p)eipx+bpsvs(p)eipx)U1=p~=Λpd3p~(2π)312Ep~s(ap~sus(Λ1p~)eip~Λx+bp~svs(Λ1p~)eip~Λx)=d3p~(2π)312Ep~s(Λ121ap~sus(p~)eip~Λx+Λ121bp~svs(p~)eip~Λx)=Λ121ψ(Λx)\begin{aligned} \psi(x) &\rightarrow U\psi(x)U^{-1}\\ &= U\int \frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\bm{p}}}}\sum_s (a_{\bm{p}}^s u^s(p)e^{-ip\cdot x}+b_{\bm{p}}^{s\dagger} v^s(p)e^{ip\cdot x})U^{-1}\\ &\overset{\tilde{p}=\Lambda p}{=}\int \frac{d^3\tilde{p}}{(2\pi)^3}\frac{1}{\sqrt{2E_{\tilde{\bm{p}}}}}\sum_s (a_{\tilde{\bm{p}}}^s u^s(\Lambda^{-1}\tilde{p})e^{-i\tilde{p}\cdot \Lambda x}+b_{\tilde{\bm{p}}}^{s\dagger} v^s(\Lambda^{-1}\tilde{p})e^{i\tilde{p}\cdot \Lambda x})\\ &=\int \frac{d^3\tilde{p}}{(2\pi)^3}\frac{1}{\sqrt{2E_{\tilde{\bm{p}}}}}\sum_s (\Lambda^{-1}_{\frac{1}{2}}a_{\tilde{\bm{p}}}^s u^s(\tilde{p})e^{-i\tilde{p}\cdot \Lambda x}+\Lambda^{-1}_{\frac{1}{2}}b_{\tilde{\bm{p}}}^{s\dagger} v^s(\tilde{p})e^{i\tilde{p}\cdot \Lambda x})\\ &= \Lambda^{-1}_{\frac{1}{2}}\psi(\Lambda x) \end{aligned}

注意:此处和我们之前讨论的对旋量场 ψ(x)\psi(x) 的进行变换的内涵是不同的,这里 ψ(x)\psi(x) 成为了由阶梯算子构成的矩阵。

在之前,我们提到对 Dirac 场相位变换对应的守恒流为:

jμ=ψˉγμψj^{\mu} = \bar{\psi}\gamma^{\mu}\psi

对应的守恒荷为:

Q=d3xj0=d3xψˉγ0ψ=d3xd3pd3q(2π)614EpEqrs(b^prvr(p)+a^psur(p))(a^qsus(q)+b^qsvs(q))ei(p+q)x=d3p(2π)312Eprs(b^prvr(p)+a^psur(p))(a^psus(p)+b^psvs(p))=d3p(2π)3s(b^psb^ps+a^psa^ps)=d3p(2π)3s(a^psa^psb^psb^ps)\begin{aligned} Q &= \int d^3x j^0\\ &= \int d^3x \bar{\psi}\gamma^0\psi\\ &= \int d^3x \int \frac{d^3pd^3q}{(2\pi)^6}\frac{1}{\sqrt{4E_{\bm{p}}E_{\bm{q}}}}\sum_{rs}(\hat{b}_{\bm{p}}^r v^{r\dagger}(\bm{p}) + \hat{a}_{-\bm{p}}^{s\dagger}u^{r\dagger}(-\bm{p}))\\ &\qquad \cdot (\hat{a}_{\bm{q}}^s u^s(\bm{q}) + \hat{b}_{-\bm{q}}^{s\dagger} v^s(-\bm{q}))e^{i(\bm{p}+\bm{q})\cdot\bm{x}}\\ &= \int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{\bm{p}}}\sum_{rs}(\hat{b}_{\bm{p}}^r v^{r\dagger}(\bm{p}) + \hat{a}_{-\bm{p}}^{s\dagger}u^{r\dagger}(-\bm{p}))(\hat{a}_{-\bm{p}}^s u^s(-\bm{p}) + \hat{b}_{\bm{p}}^{s\dagger} v^s(\bm{p}))\\ &= \int \frac{d^3p}{(2\pi)^3}\sum_{s}(\hat{b}^s_{\bm{p}}\hat{b}^{s\dagger}_{\bm{p}}+ \hat{a}^{s\dagger}_{\bm{p}} \hat{a}^s_{\bm{p}})\\ &= \int \frac{d^3p}{(2\pi)^3}\sum_{s}(\hat{a}^{s\dagger}_{\bm{p}} \hat{a}^s_{\bm{p}} - \hat{b}^{s\dagger}_{\bm{p}}\hat{b}^s_{\bm{p}})\\ \end{aligned}

我们之前提到过,当 Dirac 场与电磁场耦合在一起时,QQ 就对应电荷。那么从上式可以得到:正粒子与反粒子携带的荷相反。


现在再来看 Dirac 传播子 Dirac Propagator。定义:

SRab(xy)θ(x0y0)0{ψa(x),ψˉb(y)}0(17)S^{ab}_R(x-y) \equiv \theta(x^0-y^0)\langle 0|\{\psi_a(x),\bar{\psi}_b(y)\}|0\rangle \tag{17}

(8)(9)(8)(9) 易得:

SR(xy)=(ix+m)DR(xy)(18)S_R(x-y) = (i\cancel{\partial}_x + m)D_R(x-y) \tag{18}

可得 SR(xy)S_R(x-y) 为 Dirac 方程的推迟格林函数:

(ixm)SR(xy)=(ixm)(ix+m)DR(xy)=(2+m2)DR(xy)=iδ(4)(xy)×14×4(19)\begin{aligned} (i\cancel{\partial}_x-m)S_R(x-y) &= (i\cancel{\partial}_x-m)(i\cancel{\partial}_x+m)D_R(x-y)\\ &= -(\partial^2+m^2)D_R(x-y)\\ &= i\delta^{(4)}(x-y)\times \bm{1}_{4\times 4} \end{aligned}\tag{19}

将:

SR(xy)=d4p(2π)4eip(xy)S~R(p)S_R(x-y) = \int \frac{d^4p}{(2\pi)^4}e^{-ip\cdot(x-y)}\tilde{S}_R(p)

代入得:

S~R=ipm=i(p+m)p2m2(20)\tilde{S}_R = \frac{i}{\cancel{p}-m} = \frac{i(\cancel{p}+m)}{p^2-m^2} \tag{20}

可得:

SR=d4p(2π)4i(p+m)p2m2eip(xy)S_R = \int \frac{d^4p}{(2\pi)^4}\frac{i(\cancel{p}+m)}{p^2-m^2}e^{-ip\cdot(x-y)}

这当然是选取了以下的 p0p^0 围线。

类似在 Klein-Gordon 传播子中的方法,我们使极点稍偏离实轴,再取积分路径为实轴。

得到:

SF(xy)=d4p(2π)4i(p+m)p2m2eip(xy)={0ψ(x)ψˉ(y)0x0>y00ψˉ(y)ψ(x)0x0<y0=0Tψ(x)ψˉ(y)0(21)\begin{aligned} S_F(x-y) &= \int \frac{d^4p}{(2\pi)^4}\frac{i(\cancel{p}+m)}{p^2-m^2}e^{-ip\cdot(x-y)}\\ &= \left\{ \begin{aligned} \langle 0|\psi(x)\bar{\psi}(y)|0\rangle \quad x^0>y^0\\ -\langle 0|\bar{\psi}(y)\psi(x)|0\rangle \quad x^0<y^0\\ \end{aligned} \right.\\ &= \langle 0|T\psi(x)\bar{\psi}(y)|0\rangle \end{aligned}\tag{21}

可以验证,SFS_F 为 Dirac 方程的格林函数。