IL NUOVO CIMENTO
Vol. ?, N. ?
?
The GSI Time Anomaly: Facts and Fiction
C. GlUNTip}
(f3) /JVFJV, Sezione di Torino, Via P. Giuna 1, 1-10125 Torino, Italy
Summary. — The claims that the GSI time anomaly is due to the mixing of
neutrinos in the final state of the observed electron-capture decays of hydrogen-like
heavy ions are refuted with the help of an analogy with a double-slit experiment.
It is a consequence of causality. It is shown that the GSI time anomaly may be
caused by quantum beats due to the existence of two coherent energy levels of the
decaying ion with an extremely small energy splitting (about 6 x 10 _16 eV) and
relative probabilities having a ratio of about 1/99.
La Thuile 2009, Les Rencontres de Physique de La Vallee d'Aoste
1-7 March 2009, La Thuile, Aosta Valley, Italy
A GSI experiment [1] observed an anomalous oscillatory time modulation of the
electron-capture decays
140 Ce 58 + + v. ,
(1)
(2)
140p r 58+
2 Pm e
142-D m 60+ ^142 ^60+ + y
The hydrogen-like ions 140 p r 58+ and 142 p m 60 + were produced by fragmentation of a
beam of 152 Sm with 500-600 MeV energy per nucleon on a 9 Be target and stored in the
ESR cooler-storage ring where they circulated with a frequency of about 2 MHz and were
monitored by Schottky Mass Spectrometry. The electron capture data are fitted by an
oscillatory decay rate with a period T ~ 7 s and an amplitude A ~ 0.2 [1].
It has been proposed [1-4] that the GSI anomaly is due to the interference of the
massive neutrinos which compose the final electron neutrino state,
(3)
\v e ) = COS^V-i) + SXDf&l^)
where i? is the solar mixing angle (see Refs. [5-10]).
In order to assess the viability of this explanation of the GSI anomaly, it is necessary to
understand that interference is the result of the superposition of two or more waves [11]. If
the waves come from the same source, interference can occur if the waves evolve different
phases by propagating through different paths. Therefore, interference occurs after wave
propagation, not at the wave source. In the case of the GSI experiment, there cannot be
© Societa Italiana di Fisica
1
2
C. GIUNTI
I
NO INTERFERENCE
Fig. 1. - Analogy between the electron-capture decay process Q and a double-slit interference
experiment.
any interference effect of v\ and v-i in the electron-capture decays |T]) and ([2]), which are
the sources of v\ and v% .
Let us illustrate these remarks through an analogy with the well-known double-slit
interference experiment with classical or quantum waves depicted in Fig. [TJ In a double
slit experiment an incoming plane wave packet hits a barrier with two tiny holes, generat-
ing two outgoing spherical wave packets which propagate on the other side of the barrier.
The two outgoing waves are coherent, since they are created with the same initial phase
in the two holes. Hence, the intensity after the barrier, which is proportional to the
squared modulus of the sum of the two outgoing waves, exhibits interference effects. The
interference depends on the different path lengths of the two outgoing spherical waves
after the barrier.
For the analogy with the double-slit experiment, let us write schematically an electron-
capture decay process of the type in Eqs. (Q} and ([2]) as
(4) I -» F + u e .
Taking into account the neutrino mixing in Eq. ([3]), we have two different decay channels:
(5) I -f F + v x , I -» F + V2 ■
The initial state in the two decay channels is the same. In our analogy with the double-
slit experiment, the initial state I is analogous to the incoming wave packet. The two
final states F + vi and F + V2 are analogous to the two outgoing wave packets. Different
weights of v\ and production due to i? ^ ir/4 correspond to different sizes of the two
holes in the barrier.
In the analogy, the decay rate of I corresponds to the fraction of intensity of the
incoming wave which crosses the barrier, which depends only on the sizes of the holes. It
does not depend on the interference effect which occurs after the wave has passed through
the barrier. In a similar way, the decay rate of I cannot depend on the interference of V\
and vi which occurs after the decay has happened.
THE GSI TIME ANOMALY: FACTS AND FICTION
3
Of course, flavor neutrino oscillations caused by the interference of v\ and v-i can occur
after the decay of I, in analogy with the occurrence of interference of the outgoing waves
in the double-slit experiment, regardless of the fact that the decay rate is the incoherent
sum of the rates of production of v\ and and the fraction of intensity of the incoming
wave which crosses the barrier is the incoherent sum of the fractions of intensity of the
incoming wave which pass trough the two holes.
The above argument is a simple consequence of causality: the interference of v\ and
v-i occurring after the decay cannot affect the decay rate.
Causality is explicitly violated in Ref. [2], where the decaying ion is described by a
wave packet, but it is claimed that there is a selection of the momenta of the ion caused
by a final neutrino momentum splitting due to the mass difference of v\ and i^- This
selection violates causality. In the double-slit analogy, the properties of the outgoing
wave packets are determined by the properties of the incoming wave packet, not vice
versa. In a correct treatment, all the momentum distribution of the wave packet of the
ion contributes to the decay, generating appropriate neutrino wave packets.
The authors of Refs. [3,4] use a different approach: they calculate the decay rate with
the final neutrino state
(6) H = ]T> fe )-
fe=i
This state is different from the standard electron neutrino state, which is given by
3
(7) \"e)=J2 U eM,
fe=l
where U is the mixing matrix (in the two-neutrino mixing approximation of Eq. ([3]),
U e i = cost?, U e 2 = sintf, and J7 e 3 = 0). It is not even properly normalized to describe
one particle {(v\v) = 3). Moreover, it leads to a decay rate which is different from
the standard decay rate, given by the incoherent sum of the rates of decay into the
different massive neutrinos final states weighted by the corresponding element of the
mixing matrix [12-16]. The decay rate is given by the integral over the phase space of
the decay probability
(8) Pi-^w+u = |<^F|S|I)| 2 =
where S is the S-matrix operator. Considering the S-matrix operator at first order in
perturbation theory,
3
fc=i
FISI
(9)
S = 1 — i / d 4 i%(i) ,
4
C. GIUNTI
with the effective four-fcrmion interaction Hamiltonian
G
U w {x) = -j= cos9 c T^(x)-/ p (l - 7 5 )e(x)n(x)7 P (l - g A j 5 )p(x)
(10) = ^ cos 0q J2u: k m^W-7 5 )e(x)n(x)Y0--9A^)p(x),
v 2 fc=i
where 0q is the Cabibbo angle, one can write the matrix elements in Eq. (JSj) as
(11) (v k ,F\S\I) =U: k M k ,
with
(12) M k = ^| cos0cK,F|^)7p(l-7 5 )e(^«(^7 P (l-.9A7 5 M^|I)-
Therefore, the decay probability is given by
(13)
fc=i
This decay probability is different from the standard one [12-16], which is obtained by
summing incoherently over the probabilities of decay into the different massive neutrinos
final states weighted by the corresponding element of the mixing matrix:
(14)
P = Y^\U ek \ 2 \M h
fe=i
The analogy with the double-slit experiment and the causality argument discussed
above support the correctness of the standard decay probability P. Another argument
against the decay probability P\^ +l/ is that in the limit of massless neutrinos it does
not reduce to the decay probability in the Standard Model,
(15)
with
Psm = \M
SM
G F
id c (F, ^fVf M (a;)7 p (l - 7 5 ) e (x) n{x)Y{l - g A l b ) P {«
(16) M S M = — j= C«>s
where vf M is the SM massless electron neutrino. Indeed, for the matrix elements A4 k we
have
(17)
Mi
M
SM ■
THE GSI TIME ANOMALY: FACTS AND FICTION
5
leading to
(18)
mk— >0
\M
SM
fc=i
This is different from the SM decay probability in Eq. (fT5|) . Notice that the contribution
of the elements of the mixing matrix should disappear automatically in the limit m k — > 0.
In fact, even in the SM one can define the three massless flavors neutrinos u e , v^, v r as
arbitrary unitary linear combinations of three massless neutrinos v\, v%, v?,. However, all
physical quantities are independent of such an arbitrary transformation.
We conclude that the state in Eq. ([6]) does not describe the neutrino emitted in an
electron-capture decay process of the type in Eq. ((4]) and Refs. [3,4] are flawed.
The correct normalized state ((y e \v e ) = 1) which describes the electron neutrino
emitted in an electron-capture decay processes of the type in Eq. (J4j) is [9, 17]
1/2
3
£> fc )(i/ fc> F|S|]I)
fc=l
(19) = |Ei^i 2 i^i 2 ] izu; k M k \ Vk ).
In experiments which are not sensitive to the differences of the neutrino masses, as
neutrino oscillation experiments, we can approximate Mk ~ M and the state (I19p
reduces to the standard electron neutrino state in Eq. (0 (apart for an irrelevant phase
M/\M\).
With the electron neutrino state in Eq. (|19[) . the decay probability is given by
3 3
(20) JW+, e = l<^,F|S|I)| 2 = J]|(z/ fc ,F|S|I)| 2 = Y, \ u ek\ 2 \M k \ 2 .
fe=i fc=i
This is the correct standard result in Eq. (fl4)l : the decay probability is given by the
incoherent sum over the probabilities of decay into different massive neutrinos weighted
by the corresponding element of the mixing matrix.
Using Eq. (| 17[) and the unitarity of the mixing matrix, one can also easily check that
Pi^F+^ e reduces to Psm in Eq. (p~5|) in the massless neutrino limit.
Although the GSI time anomaly cannot be due to effects of neutrino mixing in the
final state of the electron-capture process, it can be due to interference effects in the
initial state. In fact, there could be an interference between two coherent energy states
of the decaying ion which produces quantum beats (see, for example, Ref. [18]). Also
in this case we can draw an analogy with a double-slit experiment. However, we must
change the setup, considering the double-slit experiment with two coherent sources of
waves depicted in Fig. [5J The two coherent sources are produced by an incoming plane
wave packet hitting a first barrier with two tiny holes. The two coherent outgoing waves
interfere in the space between the first and the second barrier. The interference at the
holes in the second barrier induces a modulation of the intensity which crosses the barrier.
6
C. GIUNTI
Fig. 2. - Analogy between quantum beats in the electron-capture decay process ((4]) and a
double-slit interference experiment with two coherent sources.
The role of causality is clear: the interference effect is due to the different phases of the
two coherent incoming waves at the holes in the second barrier, which have developed
during the propagation of the two waves along different path lengths between the two
barriers. Analogously, quantum beats in the GSI experiment can be due to interference
of two coherent energy states of the decaying ion which develop different phases before
the decay. The two coherent energy states could be produced in the creation process
of the ion, which in GSI occurs through fragmentation of a beam of heavier ions on a
target [1], as illustrated in Fig. [2j Alternatively, the two coherent energy states could be
due to interactions of the decaying ion in the storage ring.
The quantum mechanical description of quantum beats is rather simple. If the two
energy states of the decaying ion Ii and I2 are produced at the time t — with amplitudes
Ai and A 2 (with |Al| 2 + |*4 2 | 2 = 1), we have
(21) |I(t = 0))=^i|Ii)+^2|I 3 ).
Assuming, for simplicity, that the two states with energies E\ and E 2 have the same
decay rate T, at the time t we have
(22) \l{t)) = (Ai e~ iElt [I a > + A 2 e- tE2t |I 2 )) e~ rt l 2 .
The probability of electron capture at the time t is given by
(23)
P EC (i) = |<^,F|S|I(i))| 2
= [1 + A cos(AM + <p)] Pec e"
THE GSI TIME ANOMALY: FACTS AND FICTION
7
where where S is the S-matrix operator, A = 2\Ai\\A2\, AE = E 2 — E\,
(24) Pec = |K,F|S|Ii)| 2 = |<^ e ,F|S|I 2 )| 2 ,
and if is a constant phase which takes into account possible phase differences of Ai and
A 2 and of (i/ e ,F|S|Ii> and (u e ,¥\S\I 2 ).
The fit of GSI data presented in Ref. [1] gave
(25) A£( 140 Pr 58+ ) = (5.86 ± 0.07) x 10~ 16 eV , A( 140 Pr 58+ ) = 0.18 ± 0.03 ,
(26) A£( 142 Pm 60+ ) = (5.82 ± 0.18) x 10~ 16 eV , A( 142 Pm 60+ ) = 0.23 ± 0.04 .
Therefore, the energy splitting is extremely small and the oscillation amplitude A is
significantly smaller than one.
The authors of Ref. [1] noted that the splitting of the two hyperfine Is energy levels
of the electron is many order of magnitude too large (and the contribution to the decay
of one of the two states is suppressed by angular momentum conservation). It is difficult
to find a mechanism which produces a smaller energy splitting. Furthermore, since the
amplitude A ~ 0.2 of the interference is rather small, it is necessary to find a mechanism
which generates coherently the states Ii and I2 with probabilities |.4i| 2 and \A2\ 2 having
a ratio of about 1/99!
An important question is if the coherence of Ii and I2 is preserved during the decay
time. Since the measuring apparatus monitors the ions through elastic electromagnetic
interactions with a frequency of the order of the revolution frequency in the ESR storage
ring, about 2 MHz, the coherence can be preserved only if the interaction with the mea-
suring apparatus does not distinguish between the two states. In this case the interaction
is coherent, i.e. the two states suffer the same phase shift. Since the energy splitting AE
is extremely small, I think that coherence is maintained for a long time if Ii and I2 have
the same electromagnetic properties.
In conclusion, I have shown that the GSI time anomaly cannot be due to neutrino
mixing in the final state of the observed electron-capture decays of the hydrogen-like
i40p r 58+ an( j i42p m 60+ j ons rp ne ar g umen t has been clarified through an analogy with
a double-slit experiment, emphasizing that it is a consequence of causality [11]. I have
explained the reasons why the claim in Refs. [2-4] that the GSI time anomaly is due to
the mixing of neutrinos in the final state of the electron-capture process is incorrect (see
also Refs. [19,20]). I have also shown that the GSI time anomaly may be due to quantum
beats due to the existence of two coherent energy levels of the decaying ion. However,
since the required energy splitting is extremely small (about 6 x 10~ 16 eV) and the two
energy levels must be produced with relative probabilities having a ratio of about 1/99,
it is very difficult to find an appropriate mechanism.
REFERENCES
[1] Y.A. Litvinov et al., Phys. Lett. B664 (2008) 162, larXiv:0801.2079l
[2] H.J. Lipkin, larX~iv:0805.0435l
[3] A.N. Ivanov, R. Reda and P. Kienle, larXiv:0801.2l2"n
[4] M. Faber. larXiv:0801. 32621
8
C. GIUNTI
[5] S.M. Bilenky, C. G iunti and W. Grimus, Prog. Part. Nucl. Phys. 43 (1999) 1,
|arXiv:hep-ph/9812360[
[6] M. Gonzalez-Garcia and Y. Nir, Rev. Mod. P hys. 75 ( 2003) 345 , |arXiv:hep-ph/020 2058
[7] M. Maltoni et al., New J. Phys. 6 (2004) 122, |arXiv:hep-ph/0405172|
[8] A. Strumia and F. Vissani, [arXiv:hep-ph /0606054
[9] C. Giunti and C.W. Kim, Fundamentals of Neutrino Physics and Astrophysics (Oxford
University Press, 2007).
[10] M.C. Gonzalez-Garcia and M. Maltoni Phys. Rept. 460 (2007) 1. larXiv:0704.1800l
[11] C. Giunti, Phys. Lett. B665 (2008) 92. larXiv:0805.0431l
[12] R.E. Shrock, Phys. Lett. B96 (1980) 159.
[13] B.H.J. McKellar, Phys. Lett. B97 (1980) 93.
[14] I.Y. Kobzarev et al., Sov. J. Nucl. Phys. 32 (1980) 823.
[15] R.E. Shrock, Phys. Rev. D24 (1981) 1232.
[16] R.E. Shrock, Phys. Rev. D24 (1981) 1275.
[17] C. Giunti, J. Phys. G: Nucl. Part. Phys. 34 (2007) R93, |arXiv:hep-ph/0608070"l
[18] R. Carter and J. Hu ber, Chem. Soc. Rev. 29 (2000) 305.
[19] H. Burkhardt et al.. larXiv:0804.1099l
[20] H. Kienert et al., J. Phys. Conf. Ser. 136 (2008) 022049, larXivT 0808.2389 Neutrino 2008.
