1: \documentclass[12pt]{iopart}
2: \usepackage{iopams}
3: \newcommand{\be}{\begin{equation*}}
4: \newcommand{\ee}{\end{equation*}}
5: \begin{document}
6:
7: \title[Novel Features of the Energy Momentum Tensor]
8: {Novel Features of the Energy Momentum Tensor of a Casimir
9: Apparatus in a Weak Gravitational Field}
10:
11: \author{Giuseppe Bimonte\ddag\P, Enrico Calloni\ddag\P,
12: Giampiero Esposito\P, Luigi Rosa\ddag\P}
13:
14: \address{\ddag Dipartimento di Scienze Fisiche, Complesso Universitario
15: di Monte S. Angelo \\
16: Via Cintia, Edificio 6, 80126 Napoli, Italy}
17: \address{\P INFN, Sezione di Napoli, Complesso Universitario di Monte S.
18: Angelo \\
19: Via Cintia, Edificio 6, 80126 Napoli, Italy}
20:
21: \abstract The influence of the gravity acceleration on the
22: regularized energy-momentum tensor of the quantized
23: electromagnetic field between two plane parallel conducting plates
24: is derived. A perturbative expansion, to first order in the
25: constant acceleration parameter, of the Green functions involved
26: and of the energy-momentum tensor is derived by means of the
27: covariant geodesic point splitting procedure. The energy-momentum
28: tensor is covariantly conserved and satisfies the expected
29: relation between gauge-breaking and ghost parts.
30: \endabstract
31:
32: \section{Introduction}
33:
34: An important property of quantum electrodynamics is that suitable
35: differences of zero-point energies of the quantized
36: electromagnetic field can be made finite and produce measurable
37: effects such as the tiny attractive force among perfectly
38: conducting parallel plates known as the Casimir effect
39: \cite{Bord01}. This is a remarkable quantum mechanical effect that
40: makes itself manifest on a macroscopic scale. For perfect
41: reflectors and metals the Casimir force can be attractive or
42: repulsive, depending on the geometry of the cavity, whereas for
43: dielectrics in the weak-reflector approximation it is always
44: attractive, independently of the geometry \cite{Bart01}. The
45: Casimir effect can be studied within the framework of boundary
46: effects in quantum field theory, combined with zeta-function
47: regularization or Green-function methods, or in more physical
48: terms, i.e. on considering van der Waals forces \cite{Kamp68} or
49: scattering problems \cite{Grah02}. Casimir energies are also
50: relevant in the attempt of building a quantum theory of gravity
51: and of the universe \cite{Isha05}.
52:
53:
54: For these reasons, in Ref. \cite{Call02} we evaluated the force
55: produced by a weak gravitational field on a rigid Casimir cavity.
56: Interestingly, the resulting force was found to have opposite
57: direction with respect to the gravitational acceleration;
58: moreover, we found that the current experimental sensitivity of
59: small force macroscopic detectors would make it possible, at least
60: in principle, to measure such an effect \cite{Call02}. In Ref.
61: \cite{Call02}, calculations were based on simple assumptions and
62: the result can be viewed as a reasonable ``{\it first order}''
63: generalization of $T_{\mu\nu}$ from Minkowski to curved
64: space-time. The present paper is devoted to a deeper understanding
65: and to more systematic calculations of the interaction of a weak
66: gravitational field with a Casimir cavity. To first order in our
67: approximation the former value of the force exerted by the field
68: on the cavity is recovered.
69:
70: We consider a plane-parallel Casimir cavity, made of ideal
71: metallic plates, at rest in the gravitational field of the earth,
72: with its plates lying in a horizontal plane. We evaluate the
73: influence of the gravity acceleration $g$ on the Casimir cavity
74: but neglect any variation of the gravity acceleration across the
75: cavity, and therefore we do not consider the influence of tidal
76: forces. The separation $a$ between the plates is taken to be much
77: smaller than the extension of the plates, so that edge effects can
78: be neglected. We obtain a perturbative expansion of the
79: energy-momentum tensor of the electromagnetic field inside the
80: cavity, in terms of the small parameter $\epsilon \equiv 2
81: ga/c^2$, to first order in $\epsilon$. For this purpose, we use a
82: Fermi \cite{Misn73}, \cite{Marz94} coordinates system $(t,x,y,z)$
83: rigidly connected to the cavity. The construction of these
84: coordinates involves only invariant quantities such as the
85: observer's proper time, geodesic distances from the world-line,
86: and components of tensors with respect to a tetrad \cite{Marz94}.
87: This feature makes it possible to obtain a clear identification of
88: the various terms occurring in the metric. In our analysis we
89: adopt the covariant point-splitting procedure \cite{Chri76},
90: \cite{Dewi75} to compute the perturbative expansion of the
91: relevant Green functions. Gauge invariance plays a crucial role
92: and we check it up to first order by means of the Ward identity.
93:
94: With our notation, the $z$-axis coincides with the vertical
95: upwards direction, while the $(x,y)$ coordinates span the plates,
96: whose equations are $z=0$ and $z=a$, respectively. The resulting
97: line element for a non-rotating system is therefore \cite{Misn73}
98: $$
99: ds^{2}= -c^{2} \left(1+\epsilon {z\over a} \right) dt^{2}
100: +dx^{2}+dy^{2}+dz^{2} + {\rm O}(|x|^{2}) =
101: \eta_{\mu\nu}dx^{\mu}dx^{\nu} - \epsilon \frac{z}{ a} c^{2}
102: dt^{2},
103: $$
104: where $\eta_{\mu \nu}$ is the flat Minkowski metric ${\rm
105: diag}(-1,1,1,1)$.
106:
107: \section{The Energy Momentum Tensor and the Point-Splitting
108: Procedure}
109: %{Green Functions}
110:
111: In the Point-Splitting procedure the Energy-Momentum Tensor
112: \begin{equation*} T^{\mu \nu} \equiv {2\over \sqrt{-g}}{\delta S
113: \over \delta g_{\mu \nu}} \nonumber\label{(2.7)}
114: \end{equation*}
115: is obtained by introducing an auxiliary quantity $\langle
116: T^{\mu\nu'}(x,x') \rangle $ which involves the action of a
117: differential operator on the Hadamard function \cite{Chri76},
118: \cite{Dewi75}. In the coincidence limit
119: \begin{equation*}
120: {\langle T^{\mu\nu}(x)\rangle=\lim_{x'\rightarrow x}\langle
121: T^{\mu\nu'}(x,x')\rangle},
122: \end{equation*}
123: $\langle T^{\mu\nu}(x)\rangle$
124: is worked out. For QED (we use the Lorenz gauge \cite{Lore67}
125: to obtain the standard wave operator on the potential)
126: \begin{equation*} S[A_{\mu},\chi,\psi]=\int \left[-{1\over
127: 4}F_{\mu \nu}F^{\mu \nu} -{1\over
128: 2}(\nabla^{\mu}A_{\mu})^{2}+\chi^{; \alpha}\psi_{; \alpha}
129: \right]\sqrt{-g}~d^{4}x, \label{(2.2)}\nonumber
130: \end{equation*}
131: %
132: \bigskip
133: one gets
134: \begin{equation}
135: \langle T^{\mu \nu} \rangle = \langle T_{A}^{\mu \nu} \rangle
136: +\langle T_{B}^{\mu \nu}\rangle+\langle T_{\rm gh}^{\mu
137: \nu}\rangle, \label{(2.8)}\nonumber
138: \end{equation}
139: with
140: \begin{eqnarray} \langle F_{\rho\alpha}F_{\tau\beta}\rangle &=&
141: \lim_{x'\rightarrow x} \frac{1}{4}\left[H_{\alpha\beta';\rho\tau'}
142: + H_{\beta\alpha';\tau\rho'}-H_{\alpha\tau';\rho\beta'}-
143: H_{\tau\alpha';\beta\rho'}
144: \right. \nonumber \\
145: & & \left.- H_{\rho\beta';\alpha\tau'}-
146: H_{\beta\rho';\tau\alpha'}+H_{\rho\tau';\alpha\beta'}+
147: H_{\tau\rho';\beta\alpha'}\right], \label{(2.11)}\\
148: \langle T_{A}^{\mu\nu} \rangle &=& \lim_{x'\rightarrow x}
149: \left[-\frac{1}{4}\left(g^{\mu\rho}g^{\nu\tau}
150: -\frac{1}{4}g^{\mu\nu}g^{\tau\rho} \right) g^{\alpha\beta} \langle
151: F_{\rho\alpha}F_{\tau\beta}\rangle \right] \label{(2.12)} \\
152: \langle T_{B}^{\mu\nu} \rangle &=& \lim_{x'\rightarrow x} \Biggl[
153: -\frac{1}{4}g^{\alpha\beta}\left(
154: g^{\mu\rho}g^{\nu\tau}+g^{\mu\tau}g^{\nu\rho}-g^{\mu\nu}g^{\tau\rho}\right)
155: \left(H_{\beta\tau';\alpha\rho}+
156: H_{\tau\beta';\rho\alpha'}\right) \nonumber \\
157: & & + \frac{1}{8}g^{\alpha\beta}g^{\mu\nu}g^{\rho\tau}
158: \left(H_{\beta\tau';\alpha\rho'}+
159: H_{\tau\beta';\rho\alpha'}\right)\Biggr]\label{(2.13)} \\
160: \langle T_{\rm gh}^{\mu\nu}\rangle &=& \lim_{x'\rightarrow x}
161: \Biggl[ -\frac{1}{4}\left(
162: g^{\mu\alpha}g^{\nu\beta}+g^{\mu\beta}g^{\nu\alpha}
163: -g^{\mu\nu}g^{\alpha\beta}\right) \left(H_{;\alpha\beta'} +
164: H_{;\beta\alpha'} \right)\Biggr]
165: \end{eqnarray}
166: having defined
167: \begin{eqnarray*}
168: H_{\mu \nu}(x,x') & \equiv & \langle [A_{\mu}(x),A_{\nu}(x')]_{+}
169: \rangle\equiv H_{\mu \nu'},\\
170: H(x,x') & \equiv & \langle [\chi(x),\psi(x')]_{+} \rangle ,
171: \end{eqnarray*}
172: \be[A_{\alpha;\rho},A_{\beta;\tau} ]_{+} \equiv
173: \lim_{x'\rightarrow x} \frac{1}{2}\left\{
174: [A_{\alpha';\rho'},A_{\beta;\tau}]_{+}
175: +[A_{\alpha;\rho},A_{\beta';\tau'}]_{+} \right\}.\ee
176:
177: Since we need a recursive algorithm for the evaluation of Green
178: functions, it is more convenient to work with the Feynman Green
179: function instead of the Hadamard Green function. They are related
180: through
181: \begin{equation*}
182: H(x,x')=-2i[G(x,x')-\overline{G}(x,x')]
183: \end{equation*}
184: %
185: where $\overline{G}(x,x')=\frac{1}{2}[G^++G^-]$. The photon Green
186: function $G_{\lambda \nu'}$ in a curved spacetime with metric
187: $g_{\mu \nu}$ solves an equation of the form \cite{Bimo04} ($g_{\mu \nu'}$
188: being the parallel displacement bivector) \be \sqrt{-g}P^{\;
189: \lambda}_\mu(x) G_{\lambda\nu'}=g_{\mu\nu'}
190: \delta(x,x')=g_{\mu\nu} \delta(x,x'). \ee On expanding (this is,
191: in general, only an asymptotic expansion) \be G_{\lambda\nu'} \sim
192: G^{(0)}_{\lambda\nu'}+\epsilon \; G^{(1)}_{\lambda\nu'} + {\rm
193: O}(\epsilon^{2}), \ee we get, to first order in $\epsilon$,
194: \begin{eqnarray}
195: \Box^{0} G^{(0)}_{\mu\nu'} &=& J^{(0)}_{\mu\nu'}, \\
196: \Box^{0} G^{(1)}_{\mu\nu'} &=& J^{(1)}_{\mu\nu'},
197: \end{eqnarray}
198: where
199: \begin{eqnarray*}
200: J^{(0)}_{\mu\nu'} &\equiv& -\eta_{\mu\nu}\delta(x,x'), \\
201: \epsilon J^{(1)}_{\mu\nu'} & \equiv & {z\over a} \epsilon \left
202: ({\eta_{\mu\nu} \over
203: 2}+\delta^0_{\mu}\delta^0_{\nu}\right)\delta(x,x') +2
204: \eta^{\rho\sigma}\Gamma^\tau_{\sigma\mu} G^{(0)}_{\tau\nu',\rho} +
205: \eta^{\rho\sigma}\Gamma^\tau_{\rho\sigma} G^{(0)}_{\mu\nu',\tau}
206: -{z\over a} \epsilon G^{(0)}_{\mu\nu',00}
207: \end{eqnarray*}
208: with $\Box^0\equiv\eta^{\alpha\beta}
209: \partial_\alpha\partial_\beta=
210: -\partial_0^2+\partial_x^2+\partial_y^2+\partial_z^2$.
211:
212: To fix the boundary conditions we note that, on denoting by ${\vec
213: E}_{t}$ and ${\vec H}_{n}$ the tangential and normal components of
214: the electric and magnetic fields, respectively, a sufficient
215: condition to obtain
216: $$
217: \left . \vec{E}_{t} \right |_{S}=0,~~ \left . \vec{H}_{n} \right
218: |_{S}=0,
219: $$
220: on the boundary $S$ of the device, is to impose Dirichlet boundary
221: conditions on \cite{Espo97}
222: $$
223: A_0(\vec{x}),A_1(\vec{x}),A_2(\vec{x})
224: $$
225: at the boundary $z=0$, $z=a$. The boundary condition on $A_3$ is
226: determined by requiring that the gauge-fixing functional, here
227: chosen to be of the Lorenz type, should vanish on the boundary
228: (the boundary conditions on all components of $A_\mu$ are then all
229: preserved under gauge transformations \cite{Espo97}
230: provided the same boundary
231: condition on ghost fields is imposed, i.e. homogeneous Dirichlet).
232: This implies
233: $$
234: \left . A^\mu_{;\mu} \right |_{S}= 0\Rightarrow \left . A^3_{;3}
235: \right |_{S} = \left . (g^{33}\partial_3
236: A_3-g^{\mu\nu}\Gamma^3_{\mu\nu}A_{3} )\right |_{S}=0 ~.
237: $$
238: To first order in $\epsilon$, these conditions imply for Green
239: functions the following:
240: \begin{eqnarray*}
241: \left . G^{(0)}_{\mu\nu'} \right|_{S} &=& 0, ~~ \left . \partial_3
242: G^{(0)}_{3\nu'} \right |_{S} ~=~ 0,~~\mu=0,1,2, \;
243: \forall \nu', \\
244: \left . G^{(1)}_{\mu\nu'} \right |_{S} &=& 0, ~~ \left .
245: \partial_3 G^{(1)}_{3\nu'} \right |_{S} ~=~ -{1 \over 2a} \left .
246: G^{(0)}_{3\nu'} \right |_{S},~~\mu =0,1,2, \; \forall \nu',
247: \end{eqnarray*}
248: hence we find that the third component of the potential $A_\mu$
249: satisfies homogeneous Neumann boundary conditions to zeroth order
250: in $\epsilon$ and inhomogeneous boundary conditions to first
251: order.
252:
253: Now we are in a position to evaluate, at least formally (see
254: below), the solutions to zeroth and first order, and we get
255: \begin{equation}
256: G^{(0)}_{\lambda\nu'} = \eta_{\lambda\nu'} \int{ {d\omega d^2k
257: \over (2\pi)^3} e^{-i\omega(t-t')+ i {\vec k}_{\perp}\cdot({\vec
258: x}_{\perp}-{\vec x}_{\perp}')} }
259: \Bigr[(1-\delta_{\lambda 3})g_{D}(z,z')+\delta_{\lambda 3}g_{N}(z,z')\Bigr],
260: \end{equation}
261: having defined
262: \begin{eqnarray*}
263: g_{D}(z,z';\kappa) &\equiv& {\sin{\kappa( z_<)}\sin{\kappa(a-z_>)}
264: \over \kappa\sin{\kappa a} },~~~~~~~0<z,z'<a, \\
265: g_{N}(z,z';\kappa) &\equiv& - {\cos{\kappa(
266: z_<)}\cos{\kappa(a-z_>)} \over \kappa\sin{\kappa a}
267: },~~~~0<z,z'<a,
268: \end{eqnarray*}
269: where $D,N$ stand for homogeneous Dirichlet or Neumann boundary
270: conditions, respectively, $z_>~(z_<)$ are the larger (smaller)
271: between $z$ and $z'$, while ${\vec k}_{\perp}$ has components
272: $(k_x,k_y)$, ${\vec x}_{\perp}$ has components $(x,y)$, $\kappa
273: \equiv \sqrt{\omega^2-k^2}$, and
274: \begin{equation} G^{(1)}_{\mu
275: \nu'}=\int{ {d\omega d^2k \over (2\pi)^3} e^{-i\omega(t-t')+i{\vec
276: k}_{\perp}\cdot({\vec x}_{\perp} -{\vec x}_{\perp}')} \Phi_{\mu
277: \nu'} },
278: \end{equation}
279: where the $\Phi$ components different from zero are written in
280: Ref. \cite{Bimo06}. The ghost field satisfies the same equations
281: as the $22$ component of the gauge field, hence we do not write it
282: explicitly. In the following we will write simply $G_{\mu\nu'}$
283: and $G$ for the Green function of the gauge and ghost field,
284: respectively.
285:
286: We should stress at this stage that, in general, the integrals
287: defining the Green functions are divergent. They are well defined
288: as long as $x\neq x'$, hence we will perform all our calculations
289: maintaining the points separated and only in the very end shall we
290: take the coincidence limit as $x' \rightarrow x$ \cite{Endo84}. We
291: have decided to write the divergent terms explicitly so as to bear
292: them in mind and remove them only in the final calculations by
293: hand, instead of making the subtraction at an earlier stage.
294:
295: Our Green functions are found to satisfy the Ward identity \be
296: G^\mu_{\; \nu';\mu}+G_{;\nu'}=0,~~~ G^{\mu \; \; ;\nu'}_{\;
297: \nu'}+G^{;\mu}=0, \ee to first order in $\epsilon$ so that, to
298: this order, gauge invariance is explicitly preserved. Ward
299: identities imply $ \langle T_{B}^{\mu \nu}\rangle+\langle T_{\rm
300: gh}^{\mu \nu}\rangle=0 $ to first order in $\epsilon$, thus in the
301: following we do not consider them. {\em Nonetheless we explicitly
302: computed them and verified that they cancel each other}.
303:
304: \section{Energy-Momentum Tensor}
305:
306: Using eqs. (1)-(3) we get, from the asymptotic expansion
307: $T_{\mu\nu'} \sim T^{(0)}_{\mu\nu'}+{\epsilon \over a}
308: T^{(1)}_{\mu\nu'} +{\rm O}(\epsilon^{2})$,
309: \begin{eqnarray}
310: \langle T^{(0)\mu\nu'}\rangle &=& {1 \over 16\,a^4\,{\pi }^2}
311: \left( {\zeta}_{H}\left(4, {2\,a + z - {z'} \over 2\,a}\right) +
312: {\zeta}_{H}\left(4, {z'-z \over 2\,a}\right) \right) \nonumber \\
313: & & \times{\rm diag}(-1,1,1,-3),
314: \end{eqnarray}
315: where $\zeta_{H}$ is the Hurwitz $\zeta$-function
316: $\zeta_{H}(x,\beta) \equiv \sum_{n=0}^{\infty}(n+\beta)^{-x}$. On
317: taking the limit $z'\rightarrow z^+$ we find
318: \begin{equation}
319: \lim_{z' \to z^+} \langle T^{(0)\mu\nu'} \rangle =\left({\pi^2
320: \over 720 a^4} +\lim_{z' \to z^+} {1 \over \pi^2(z-z')^4}\right)
321: {\rm diag}(-1,1,1,-3),
322: \end{equation}
323: where the divergent term as $z' \rightarrow z$ can be removed by
324: subtracting the contribution of infinite space without bounding
325: surfaces \cite{Bord01}, and in our analysis we therefore discard
326: it hereafter. The renormalization of the energy-momentum tensor in
327: curved spacetime is usually performed by subtracting the $\langle
328: T_{\mu \nu} \rangle$ constructed with an Hadamard or
329: Schwinger--DeWitt two-point function up to the fourth adiabatic
330: order \cite{Chri76}, \cite{Chri78}. In our problem, however, as we
331: work to first order in $\epsilon$, we are neglecting tidal forces
332: and therefore the geometry of spacetime in between the plates is
333: flat. Thus, we need only subtract the contribution to the energy
334: momentum tensor that is independent of $a$, which is the standard
335: subtraction in the context of the Casimir effect in flat
336: spacetime \cite{Deut79}.
337:
338: In the same way we get, to first order in $\epsilon$:
339: \begin{eqnarray}
340: \lim_{z' \to z^+} \langle
341: T^{(1)\mu\nu'}\rangle &=& {\rm
342: diag}(T^{(1)00},T^{(1)11},T^{(1)22},T^{(1)33}) \nonumber \\
343: & +& \lim_{z' \to z^+} {\rm diag}\Bigr(-z'/\pi^{2}(z-z')^{4},0,0,0
344: \Bigr),
345: \end{eqnarray}
346: where
347: \begin{eqnarray}
348: T^{(1)00} &=& -{{\pi }^2 \over 1200\,a^3} + {11 {\pi }^2\,z \over
349: 3600\,a^4} - {\pi \over 60\,a^3}\, \frac{ \cos{({\pi \,z \over
350: a})}}{{\sin^{3}{ ({\pi \,z \over a})}}}, \\
351: T^{(1)11} &=& {{\pi }^2 \over 3600\,a^3} - {{\pi}^2\,z \over
352: 1800\,a^4} -{\pi \over 120\,a^3}\, \frac{ \cos{({\pi \,z \over
353: a})}}{{\sin^{3}{ ({\pi \,z
354: \over a})}}}, \\
355: T^{(1)22} &=& T^{(1)11}, \\
356: T^{(1)33} &=& -{\left( {\pi }^2\,\left( a - 2\,z \right) \right)
357: \over 720\,a^4}.
358: \end{eqnarray}
359: Incidentally we note that the tensor is covariantly conserved:
360: $\nabla\cdot T=0$ to first order in $\epsilon$.
361:
362: \section{Push}
363:
364: To compute the Casimir energy we must project the energy-momentum
365: tensor along the unit timelike vector $u$ with covariant
366: components $u_\mu=(\sqrt{-g_{00}},0,0,0)$ to obtain $\rho=\langle
367: T^{\mu\nu}\rangle u_\mu u_\nu$, so that
368: \be
369: \rho = -{\pi^2 \over
370: 720a^4}+2 {g \over c^2} \left(-{{\pi }^2 \over 1200\,a^3} + {{\pi
371: }^2\,z \over 600\,a^4} - {\pi \over 60\,a^3}\, \frac{ \cos{({\pi
372: \,z \over a})}}{{\sin^{3}{ ({\pi \,z \over a})}}} \right)
373: +{\rm O}(g^{2}),
374: \ee
375: where we have replaced
376: $\epsilon$ by its expression in terms of $g$. Thus, the energy
377: stored in the Casimir device is found to be
378: \be
379: E=\int{d^3\Sigma\sqrt{-g}\langle T^{\mu\nu}\rangle u_\mu
380: u_\nu}=-{\hbar c \pi^2 \over 720} {A \over a^{3}} \left(1+{1 \over
381: 2} {g a \over c^2} \right)\equiv E_C\left(1+{1 \over 2} {g a \over
382: c^2} \right),
383: \ee
384: where $A$ is the area of the plates, $d^3\Sigma$
385: is the three-volume element of an observer with four-velocity
386: $u_\mu$, and we have reintroduced $\hbar$ and $c$.
387:
388: In the same way, the pressure on the plates is given by
389: \be
390: P(z=0)={\pi^{2}\over 240} {\hbar c \over a^{4}} \left(1+{2\over
391: 3}{ga \over c^{2}}\right), \; P(z=a)=-{\pi^{2}\over 240 } {\hbar c
392: \over a^{4}} \left(1-{2\over 3} {ga \over c^{2}}\right).
393: \ee
394: To obtain the resulting force one has to multiply each of them by the
395: redshift $r$ of the point where they act, relative to the point
396: where they are added \cite{nordt75}:
397: \be
398: r_{\tiny P_{added}(\tiny
399: P_{act})}=\sqrt{\frac{|g_{00}(P_{act})|}{|g_{00}(P_{added})|}}
400: \approx 1+\frac{g}{c^2}(z-z_Q)
401: \ee
402: to leading order in $\frac{g
403: z}{c^2}$, so that a net force \cite{bica071}
404: \be
405: F =-\frac{\pi^2
406: \hbar c}{a^4}\left[ \frac{g}{240 c^2}(z_2-z_1)-
407: \frac{4 g}{720 c^2}(z_2-z_1)\right]
408: = {\pi^{2}\over 720}{A \hbar g\over c a^{3}}=\frac{E_C}{c^2}g,
409: \ee
410: pointing upwards along the $z$-axis is obtained, in perfect
411: agreement with the early findings of \cite{jare93} and the more
412: recent results of \cite{fumi07}
413: (for other relevant references on curved spacetime calculations,
414: see \cite{Endo84, Chri78, Dewi84}).
415:
416: \section{Concluding Remarks}
417:
418: To the best of our knowledge, the analysis presented in this paper
419: represents the first study of the energy-momentum tensor for the
420: electromagnetic field in a Casimir cavity placed in a weak
421: gravitational field. The resulting calculations are considerably
422: harder than in the case of scalar fields. By using Green-function
423: techniques, we have evaluated the influence of the gravity
424: acceleration on the regularized energy-momentum tensor of the
425: quantized electromagnetic field between two plane-parallel ideal
426: metallic plates, at rest in the gravitational field of the earth,
427: and lying in a horizontal plane. In particular, we have obtained a
428: detailed derivation of the theoretical prediction according to
429: which a Casimir device in a weak gravitational field will
430: experience a tiny push in the upwards direction \cite{Call02}.
431: This result is consistent with the picture that the {\it negative}
432: Casimir energy in a gravitational field will behave like a {\it
433: negative mass}. An outstanding open problem is now how
434: to obtain an independent evaluation of our formula for the
435: energy-momentum tensor, and what its implications are for fundamental
436: physics.
437:
438: \section*{Acknowledgments}
439: The work of G. Bimonte and G. Esposito has been partially
440: supported by PRIN {\it SINTESI}. G. Esposito is grateful to the
441: Dipartimento di Scienze Fisiche of Federico II University, Naples,
442: for hospitality and support. The work of L. Rosa has been
443: partially supported by PRIN {\it FISICA ASTROPARTICELLARE}.
444:
445: \section*{References}
446:
447: \begin{thebibliography}{99}
448: \bibitem{Bord01}{Bordag M, Mohideen U and Mostepanenko V M
449: 2001 {\it Phys. Rep.} {\bf 353} 1}
450:
451: \bibitem{Bart01}{Barton G 2001 {\it J. Phys.} A {\bf 34} 4083}
452:
453: \bibitem{Kamp68}{van Kampen N G, Nijboer B R and Schram K 1968
454: {\it Phys. Lett.} A {\bf 26} 307}
455:
456: \bibitem{Grah02}{Graham N, Jaffe R L, Khemani V, Quandt M, Scandurra M
457: and Weigel H 2002 {\it Nucl. Phys.} B {\bf 645} 49}
458:
459: \bibitem{Isha05}{Ishak M 2007 {\it Found. Phys.} {\bf 37} 1470;
460: Mahajan G, Sarkar S and Padmanabhan T 2006 {\it Phys. Lett.} B
461: {\bf 641} 6}
462:
463: \bibitem{Call02}{Calloni E, Di Fiore L, Esposito G, Milano L and Rosa
464: L 2002 {\it Phys. Lett.} A {\bf 297} 328}
465:
466: \bibitem{Misn73}{ Misner C, Thorne K P and Wheeler J A 1973 {\it Gravitation}
467: (Freeman, S. Francisco)}
468:
469: \bibitem{Marz94}{Marzlin K P 1994 {\it Phys. Rev.} D {\bf 50} 888}
470:
471: \bibitem{Chri76}{Christensen S M 1976 {\it Phys. Rev.} D {\bf 14} 2490}
472:
473: \bibitem{Dewi75}{DeWitt B S 1975 {\it Phys. Rep.} C {\bf 19} 295}
474:
475: \bibitem{Lore67}{Lorenz L 1867 {\it Phil. Mag.} {\bf 34} 287}
476:
477: \bibitem{Bimo04}{Bimonte G, Calloni E, Di Fiore L, Esposito G, Milano L and
478: Rosa L 2004 {\it Class. Quant. Grav.} {\bf 21} 647}
479:
480: \bibitem{Espo97}{Esposito G, Kamenshchik A Yu and
481: Pollifrone G 1997 {\it Euclidean
482: Quantum Gravity on Manifolds with Boundary} Fundamental Theories
483: of Physics {\bf 85} (Kluwer Dordrecht)}
484:
485: \bibitem{Bimo06}{Bimonte G, Calloni E, Esposito G and Rosa L 2006
486: {\it Phys. Rev.} D {\bf 74} 085011 }
487:
488: \bibitem{Endo84}{Endo R 1984 {\it Prog. Theor. Phys.} {\bf 71} 1366}
489:
490: \bibitem{Chri78}{Christensen S M 1978 {\it Phys. Rev.} D {\bf 17} 946}
491:
492: \bibitem{Deut79}{Deutsch D and Candelas P 1979 {\it Phys. Rev.} D {\bf 20} 3063}
493:
494: \bibitem{nordt75} Nordtvedt K Jr 1975 {\it Am. J. Phys.} {\bf 43} 256
495:
496: \bibitem{bica071} Bimonte G, Calloni E, Esposito G and
497: Rosa L 2007 {\it Phys. Rev.} D {\bf 76} 025008
498:
499: \bibitem{jare93} Jaekel and Reynaud 1993 {\it J. Phys. I (France)} {\bf 3} 1093
500:
501: \bibitem{fumi07} Fulling S A, Milton K A, Parashar P, Romeo A,
502: Shajesh K V and Wagner J 2007 {\it Phys. Rev.} D {\bf 76} 025004
503:
504: \bibitem{Dewi84}{DeWitt B S 1984 `The Spacetime Approach to Quantum
505: Field Theory', in
506: {\it Relativity, Groups and Topology II} eds. B.S. DeWitt and R.
507: Stora (North--Holland, Amsterdam)}
508:
509: \end{thebibliography}
510:
511: \end{document}
512: