betaFeedback

WorksPhysics and Astronomy

113^{113}Cd ββ-decay spectrum and gAg_{\rm A} quenching using spectral moments

Abstract

We present an alternative analysis of the 113^{113}Cd ββ-decay electron energy spectrum in terms of spectral moments μnμ_n, corresponding to the averaged values of nthn^{\rm th} powers of the ββ particle energy. The zeroth moment μ0μ_0 is related to the decay rate, while higher moments μnμ_n are related to the spectrum shape. The here advocated spectral-moment method (SMM) allows for a complementary understanding of previous results, obtained using the so-called spectrum-shape method (SSM) and its revised version, in terms of two free parameters: r=gA/gVr=g_{\rm A}/g_{\rm V} (the ratio of axial-vector to vector couplings) and ss (the small vector-like relativistic nuclear matrix element, ss-NME). We present numerical results for three different nuclear models with the conserved vector current hypothesis (CVC) assumption of gV=1g_{\rm V}=1. We show that most of the spectral information can be captured by the first few moments which are simple quadratic forms (conic sections) in the (r, s)(r,\,s) plane: an ellipse for n=0n=0 and hyperbolae for n≥1n\geq 1, all being nearly degenerate as a result of cancellations among nuclear matrix elements. The intersections of these curves, as obtained by equating theoretical and experimental values of μnμ_n, identify the favored values of (r, s)(r,\,s) at a glance, without performing detailed fits. In particular, we find that values around r∼1r\sim 1 and s∼1.6s\sim 1.6 are consistently favored in each nuclear model, confirming the evidence for gAg_{\rm A} quenching in 113^{113}Cd, and shedding light on the role of the ss-NME. We briefly discuss future applications of the SMM to other forbidden ββ-decay spectra sensitive to gAg_{\rm A}.

Introduction

The search for the rare process of neutrinoless double beta decay (0νββ0\nu\beta\beta), as well the study of its implications for the fundamental nature of the neutrino field (Dirac or Majorana), represent a major topic in both particle and nuclear physics [1, 2, 3]. The predicted rate of this decay, as well as of other weak-interaction nuclear processes, depends sensitively on the effective value of the weak axial-vector coupling gAg_A that, in nuclear matter, appears to be generally different [4, 5] from the vacuum value gAvac=1.276g_A^{\mathrm{vac}} = 1.276 [6, 7]. In particular, effective quenching factors q=gA/gAvac<1q = g_A/g_A^{\mathrm{vac}} < 1 have been discussed in a variety of observed β\beta and ββ\beta\beta decays, whose Gamow-Teller (GT) nuclear matrix elements are reduced by factors of qq and q2q^2, respectively; see, e.g., [4, 5, 8, 9, 10, 11].

While the theoretical connections among different values of qq and their physical origin are still subject to research [4, 12, 13, 14], from a phenomenological viewpoint it makes sense to study observables that appear to be particularly sensitive to possible quenching effects. In this context, highly forbidden non-unique β\beta-decays offer a very interesting avenue, since both their calculated decay rates and energy-spectrum shapes are found to change very rapidly around quenched values gA∼1g_A \sim1, due to subtle cancellations among large nuclear matrix elements [15].

For the fourth-forbidden non-unique β\beta decay of 113Cd^{113}\mathrm{Cd}, detailed experimental data are available for the energy spectrum S(we)S(w_e) in terms of the β\beta energy wew_e [16, 17]. The data constrain both the overall decay rate (or, analogously, the half-life) and the energy spectrum shape (as a function of wew_e). It is highly nontrivial to match the corresponding theoretical predictions with data, since the values of gAg_A that best fit the decay half-life are not necessarily the same that best fit the decay spectral shape and may be in conflict [15, 17, 18], although both indicate large NME cancellations. These two approaches to constraining gAg_A in 113Cd^{113}\mathrm{Cd} β\beta decay, dubbed as half-life and spectrum-shape methods [18], have only recently been reconciled by varying a small parameter multiplied by the vector coupling gVg_V, namely, the so-called small relativistic nuclear matrix element ss-NME [19] (V431(0)\mathcal{V}_{431}^{(0)} in the notation of Behrens and Bühring [20]), the estimates of which, from the conserved vector current (CVC) hypothesis or based on specific nuclear models, are rather uncertain but crucial for forbidden decays [21, 22].

In particular, by treating the ss-NME as a free parameter to be determined by data in a revised version of the spectrum-shape method (SSM) [19], both the 113Cd^{113}\mathrm{Cd} half-life and spectrum data in [19, 23] have been found to match well the theoretical predictions of different models, with consistent values of the quenching factor and the ss-NME. These nontrivial results, which represent a strong indication for gAg_A quenching in 113Cd^{113}\mathrm{Cd}, deserve in our opinion further analysis, aiming at a better understanding of the comparison of theory and data, in light of recent and future investigations of forbidden β\beta-decay spectra in other nuclides [22, 24, 25]. In particular, we aim at reducing the relevant spectral information to a relatively small set of quantities or parameters to be studied.

We start from the basic property that any smooth spectrum S(we)S(w_e) can be characterized by (and reconstructed from) the series of its moments μn\mu_n, namely, by the spectrum-averaged values of wenw_e^n for n≥0n \ge0 [26, 27]. This approach to spectra allows for the unification of the half-life method (connected to μ0\mu_0) and the SSM (connected to μ1,2,3,…\mu_{1,2,3,\ldots}) in a single “spectral-moment method (SMM).” We show that a few moments μn\mu_n can capture with high accuracy the whole spectral information in terms of the two free parameters, r=gA/gVr = g_A/g_V and s=ss = s-NME, where gVg_V is the vector coupling (assumed to be unity as in [19]). Furthermore, since the moments μn\mu_n are simple quadratic forms in (r,s)(r, s), the information contained in the infinite family of spectra S(we∣r,s)S(w_e \mid r, s) can be eventually discretized, with significant conceptual and numerical advantages.

Concerning experimental data, herein we use the absolute 113Cd{}^{113}\mathrm{Cd} β\beta-decay spectrum of [16] including the energy calibration and systematics assessment performed in [28]. Constraints on (r,s)(r, s), as obtained by comparing theoretical and experimental moments, are interpreted in terms of intersections of iso-moment curves (an ellipse for μ0\mu_0 and hyperbolae for μ1,2,3,…\mu_{1,2,3,\ldots}). In each of three different nuclear models for the NME, such intersections are closest for rather similar values of the (r,s)(r, s) parameters. Since most of the relevant features appear to be captured by just the first few moments, the method can be pragmatically applied to future forbidden β\beta-decay measurements in different nuclei, where the available spectral data might be more limited than for 113Cd{}^{113}\mathrm{Cd}.

Our work is structured as follows. In Sec. II we discuss the spectral-moment method, the adopted notation, and the numerical values of the first few moments. In Sec. III we discuss the implications of equating the theoretical and experimental moments in terms of the (r,s)(r, s) parameters. We find evidence for r∼1r \sim1, corresponding to a multiplicative renormalization of the axial-vector coupling, as well as for ∣s∣∼1.6|s| \sim1.6 (with a preference for positive values of ss), corresponding to an additive contribution to vector-like NME, consistent with earlier findings utilizing a different approach and independent data [19]. We summarize our results and consider further applications of the SMM in Sec. IV. Technical aspects about NMEs and quadratic forms are discussed in Appendix A and B, respectively.

Spectral moments: method, notation and numerical values

In this Section we define the notation used in the SMM to describe the 113Cd{}^{113}\mathrm{Cd} β\beta-decay spectra (experimental and theoretical) in terms of a truncated set of spectral moments. Numerical values for such moments are also derived.

β\beta-decay spectrum notation

Following [15, 18, 19], we introduce a dimensionless energy parameter wew_e,

we=Weme=1+Teme,(1)w_e = \frac{W_e}{m_e} = 1 + \frac{T_e}{m_e}, \tag*{(1)}

where WeW_e and TeT_e are, respectively, the total and kinetic energies of the electron with mass mem_e.

The energy spectrum S(we)S(w_e) is defined as the fractional number of decays nen_e per single nucleus and per unit of time tt and of energy wew_e

S(we)=d2nedt dwe,(2)S(w_e) = \frac{d^2 n_e}{dt\,dw_e}, \tag*{(2)}

where we∈[1,w0]w_e \in[1, w_0], and the endpoint w0w_0 is set by the QβQ_\beta-value of the decay (w0=1+Qβ/mew_0 = 1 + Q_\beta/m_e). When needed, experimental (ee) and theoretical (tt) spectra are distinguished by superscripts,

Se=Sexpt,(3)S^e = S^{\mathrm{expt}}, \tag*{(3)}
St=Stheo.(4)S^t = S^{\mathrm{theo}}. \tag*{(4)}

In order to link our formalism with common nuclear physics notation, we remind that the total decay rate λ\lambda (or, equivalently, the half life t1/2t_{1/2}) is obtained by integrating S(we)S(w_e) over the interval [1,w0][1, w_0]:

λ=∫1w0S(we) dwe=ln⁡2t1/2.(5)\lambda= \int_{1}^{w_0} S(w_e)\,dw_e = \frac{\ln2}{t_{1/2}}. \tag*{(5)}

However, we shall not use either λ\lambda or t1/2t_{1/2} hereafter, for the following reason.

Due to increasingly high backgrounds at low energy, the experimental spectrum is typically reported above a detector-dependent kinetic energy threshold Tthr>0T_{\mathrm{thr}} > 0 [16, 17, 19, 28] that defines the wew_e threshold as wthr=1+Tthr/me>1w_{\mathrm{thr}} = 1 + T_{\mathrm{thr}}/m_e > 1. Therefore, the decay half life t1/2t_{1/2} can be estimated only by extrapolating the measured SeS^e spectrum down to T→0T \to0, see, e.g. [16]. However, any adopted extrapolation function may well be different from the computed theoretical spectra in the range [1,wthr][1, w_{\mathrm{thr}}] below threshold. In order to avoid potential biases, we shall thus consistently compare the experimental and theoretical spectra only in the energy range above threshold,

we∈[wthr,w0].(6)w_e \in[w_{\mathrm{thr}}, w_0] . \tag*{(6)}

Spectral moments μn\mu_n

It is well known from statistics that a smooth spectrum S(we)S(w_e), defined over an interval we∈[wthr,w0]w_e \in[w_{\mathrm{thr}}, w_0], can be described by a series of moments {μn}n≥0\{\mu_n\}_{n\geq0} [26, 29]. The zeroth moment, defined as

μ0=∫wthrw0S(we) dwe ,(7)\mu_0 = \int_{w_{\mathrm{thr}}}^{w_0} S(w_e)\,dw_e \, , \tag*{(7)}

encodes the overall spectrum normalization, while the first and higher moments, defined as

μn=∫wthrw0S(we)wen dwe∫wthrw0S(we) dwe(n≥1) ,(8)\mu_n = \frac{\int_{w_{\mathrm{thr}}}^{w_0} S(w_e)w_e^n\,dw_e}{\int_{w_{\mathrm{thr}}}^{w_0} S(w_e)\,dw_e} \quad(n \geq1) \, , \tag*{(8)}

encode the spectrum shape information, via the averaged values of increasingly high powers of the main variable.1

Note that, in our case, μ0\mu_0 has the dimension of an inverse time, being defined as the decay rate in the interval [wthr,w0][w_{\mathrm{thr}}, w_0] above threshold (see also Eq. (5) for the total rate in [1,w0][1,w_0]). All the other moments are instead dimensionless. When needed, moments of theoretical and experimental spectra will be distinguished by superscripts:

S=St → μn=μnt ,(9)S = S^t \ \to\ \mu_n = \mu_n^t \, , \tag*{(9)}
S=Se → μn=μne .(10)S = S^e \ \to\ \mu_n = \mu_n^e \, . \tag*{(10)}

There is a vast and interdisciplinary literature on the inverse moment problem, namely, on possible methods to reconstruct the original function S(we)S(w_e) from a finite number of moments {μn}n=0,…,N\{\mu_n\}_{n=0,\ldots,N} with some approximations [27, 30, 31]. While all methods tend to improve their accuracy for increasing NN, some may converge faster or better than others, depending on specific features of the function(s) S(we)S(w_e).

We have checked that a simple reconstruction algorithm based on an expansion in Legendre polynomials, as described in [32], is sufficient enough to allow for the reconstruction of the 113Cd{}^{113}\mathrm{Cd} spectra at subpercent level (more accurately than is needed for our purposes) in the entire parameter space relevant for this work, with just N=6N=6 moments. Representative examples of theoretical spectra reconstructed from a finite set of moments are shown below in § 2 D. Such results are consistent with (but more general than) the findings of Ref. [16], where the experimental spectrum was well approximated in terms of a 6th6^{\mathrm{th}} order polynomial function.

We shall thus limit ourselves to N=6N=6 and consider the truncated set of moments

{μn}=μ0, μ1,…,μ6 .(11)\{\mu_n\} = \mu_0,\ \mu_1,\ldots,\mu_6 \, . \tag*{(11)}

Actually, as we shall see in several ways, interesting results can be obtained by considering just the first two or three moments out of the above set.

Experimental spectrum Se(we)S^e(w_e)

In this work we consider the experimental spectrum Se(we)S_e(w_e) of 113Cd{}^{113}\mathrm{Cd} as measured in [16], after a recalibration of the energy scale and the deconvolution of resolution effects as described in [28] (see Fig. 29 therein). The experimental threshold Tthr=26 keVT_{\mathrm{thr}}=26\ \mathrm{keV} [28] and the endpoint Qβ=323.83 keVQ_\beta=323.83\ \mathrm{keV} [33] define the analysis range:

we∈[wthr,w0]=[1.051,1.634] .(12)w_e \in[w_{\mathrm{thr}},w_0] = [1.051,1.634] \, . \tag*{(12)}

In this range, the experiment observed Ne=2.222×106N_e=2.222\times10^6 events for Nd=8.858×1022N_d=8.858\times10^{22} decaying nuclei over a data-taking time t=9.929×106 st=9.929\times10^6\ \mathrm{s} [16, 28]. The corresponding decay rate provides the zeroth moment μ0e=Ne/(Ndt)\mu_0^e=N_e/(N_dt):

μ0e=2.526×10−24 s−1 .(13)\mu_0^e = 2.526\times10^{-24}\ \mathrm{s}^{-1} \, . \tag*{(13)}

Figure 1 shows the experimental spectrum in the range of Eq. (12), as taken from [28] with the above normalization. The spectrum is reported in bins of width Δwe=2 keV/me\Delta w_e=2\ \mathrm{keV}/m_e, together with total (statistical and systematic) uncertainty in each bin.

From the above Se(we)S^{e}(w_e) we derive the following values for the experimental moments (up to N=6N = 6):

μ1e=1.291 ,(14)\mu^{e}_{1} = 1.291 \, , \tag*{(14)}
μ2e=1.686 ,(15)\mu^{e}_{2} = 1.686 \, , \tag*{(15)}
μ3e=2.226 ,(16)\mu^{e}_{3} = 2.226 \, , \tag*{(16)}
μ4e=2.970 ,(17)\mu^{e}_{4} = 2.970 \, , \tag*{(17)}
μ5e=4.004 ,(18)\mu^{e}_{5} = 4.004 \, , \tag*{(18)}
μ6e=5.452 .(19)\mu^{e}_{6} = 5.452 \, . \tag*{(19)}

We postpone the discussion of related uncertainties to Sec. III.

A final comment is in order. In [16], by extrapolating the observed spectrum below threshold (i.e., for we∈[1,1.051]w_e \in[1, 1.051]), the total number of decays was estimated to be Ne′=2.40×106N_e' = 2.40 \times10^{6} for the exposure NdtN_d t, corresponding to a decay rate λ=Ne′/(Ndt)=2.73×10−24 s−1\lambda= N_e'/(N_d t) = 2.73 \times10^{-24}\ \mathrm{s}^{-1} and to the quoted half-life t1/2=ln⁡2/λ=8.04×1015 yt_{1/2} = \ln2/\lambda= 8.04 \times10^{15}\ \mathrm{y}. As previously mentioned, we do not use extrapolated quantities such as λ\lambda or t1/2t_{1/2} in this work.

Theoretical spectrum St(we)S^{t}(w_e) and its free parameters (r,s)(r,s)

As discussed in [18] and references therein, the theoretical β\beta-decay spectrum St(we)S^{t}(w_e) can be generally expressed as

St(we)=ln⁡2κC(we) p we (w0−we)2F0(Z,we) .(20)S^{t}(w_e) = \frac{\ln2}{\kappa} C(w_e)\,p\,w_e\,(w_0-w_e)^2 F_0(Z,w_e)\, . \tag*{(20)}

where C(we)C(w_e) is the so-called shape factor, pp is the electron momentum in units of mem_e, and F0(Z,we)F_0(Z,w_e) is the Fermi function with the final nuclear state having Z=49Z = 49. In contexts where extreme precision is required, small correction terms accounting for, e.g., radiative effects and atomic screening become important. For the purposes of this work these ≈1%\approx1\% corrections are insignificant but have been included as described in [18]. The conversion constant κ\kappa reads:

κ=2π3ln⁡2me5(GFcos⁡θC)2=6289 s ,(21)\kappa= \frac{2\pi^{3}\ln2}{m_e^{5}(G_F\cos\theta_C)^2} = 6289\ \mathrm{s}\, , \tag*{(21)}

where GFG_F is the Fermi constant and θC\theta_C is the Cabibbo angle. For non-unique decays the shape factor has a complicated expression including universal kinematic factors and nuclear matrix elements (NME). The latter capture all the nuclear-structure dependent information regarding the decay. In the formalism of Behrens and Bühring [20] the NME arise from a multipole expansion of the nuclear current. The NME are then expanded as a power series resulting in an expression including vector NME VMKLs(m){}^{V}\mathcal{M}^{(m)}_{KLs} and axial-vector NME AMKLs(m){}^{A}\mathcal{M}^{(m)}_{KLs}. The matrix elements with the smallest angular momenta KK and LL allowing for the decay dominate, with the first term in the power series, m=0m = 0, being by far the most important. For fourth-forbidden unique decays there are four leading-order NME, with the dominant matrix elements being VM440(0){}^{V}\mathcal{M}^{(0)}_{440} and AM441(0){}^{A}\mathcal{M}^{(0)}_{441}, and with significantly smaller contributions coming from VM431(0){}^{V}\mathcal{M}^{(0)}_{431} and AM541(0){}^{A}\mathcal{M}^{(0)}_{541}. The expansion can be further carried out to next-to-leading order, resulting in a total of 45 NME [18] that depend on wew_e and on powers of the nuclear radius R=1.2A1/3 fm=5.8 fmR = 1.2 A^{1/3}\ \mathrm{fm} = 5.8\ \mathrm{fm}. The NME need to be numerically computed with specific nuclear models. Following [18, 19] we consider: the microscopic interacting boson-fermion model (IBFM-2), the microscopic quasiparticle-phonon model (MQPM), and the interacting shell model (ISM). First we discuss general aspects of the spectrum structure, and then report model-dependent numerical results in terms of spectral moments.

In general, C(we)C(w_e) is a sum over squares and products of linear combinations of NME, each being multiplied by either the vector coupling gVg_V or the axial-vector coupling gAg_A. The couplings arise in the formalism of beta decays as a means to normalize the hadron current when moving from the quark level to the level of nucleons, and each axial-vector matrix element is always preceded by gAg_A and each vector matrix element by gVg_V. By defining the ratio

r=gAgV(22)r = \frac{g_A}{g_V} \tag*{(22)}

one can formally write St(we)S^t(w_e) as a quadratic form in rr [15],

St=gV2(SVt+rSVAt+r2SAt),(23)S^t = g_V^2\left(S^t_V+rS^t_{VA}+r^2S^t_A\right), \tag*{(23)}

where SAtS^t_A includes only axial-vector NME, SVtS^t_V only vector NME, and SVAtS^t_{VA} is a mix of vector and axial-vector NME. Hereafter we shall assume as in [19], in accordance with the conserved vector current (CVC), that

gV=1 ,(24)g_V = 1 \,, \tag*{(24)}

while rr will be treated as a free theoretical parameter to be fixed by the data. We shall comment on deviations from Eq. (24) in Sec. III.

As discussed in detail in [18], the quadratic form in Eq. (23) entails delicate cancellations among large NME for r∼1r \sim1, where agreement between theory and data can be usually found in terms of either the spectrum normalization [15] or its shape [17], but not both at the same time (as far as rr is the only free parameter) [15, 18]. In particular, the main NME cancellation term turns out to be the square of a binomial, up to subleading NME terms ϵ\epsilon and ϵ′\epsilon':

St∝(V ⁣M440(0)−α A ⁣M441(0)r+ϵ)2+ϵ′(25)S^t \propto\left({}^V\!M^{(0)}_{440}-\alpha\,{}^A\!M^{(0)}_{441}r+\epsilon\right)^2+\epsilon' \tag*{(25)}

where α\alpha is a numerical coefficient of O(1)O(1) and the M\mathcal{M}’s are “large” NME, typically of O(102)O(10^2)–O(103)O(10^3) in units of fm4\mathrm{fm}^4; see [18] for details of the notation and for numerical M\mathcal{M} values in the various nuclear models. For r∼1r\sim1, it turns out that the two large M\mathcal{M}’s tend to cancel each other, leaving a residual of O(1)O(1) [18]. Therefore, a subleading term ϵ∼O(1)\epsilon\sim O(1) may still play a significant role, especially if its numerical value is rather uncertain.

It was realized in [19, 21, 22] that this role can be played by so-called small relativistic NME, dubbed as ss-NME in [19] and here just as ss for simplicity, where:

s=V ⁣M431(0) ,(26)s = {}^V\!M^{(0)}_{431}\,, \tag*{(26)}

in units of fm3\mathrm{fm}^3. On the one hand, with very simple (though unrealistic) assumptions related to the nuclear-structure aspects of the decay, the CVC hypothesis would imply (in our adopted units) the numerical relation V ⁣M431(0)=0.0678(R−1)V ⁣M440(0){}^V\!M^{(0)}_{431}=0.0678(R^{-1}){}^V\!M^{(0)}_{440} [19], leading to expected values s∼O(1–10)s\sim O(1\text{--}10). On the other hand, numerical evaluations of ss either give s=0s=0 due to model-specific limitations relating to a restricted model space (in the IBFM-2 and ISM models) or to s≃0.4s\simeq0.4 (in the MQPM model) [18]. A more detailed discussion of the uncertain estimates of s=V ⁣M431(0)s={}^V\!M^{(0)}_{431} as compared with V ⁣M440(0){}^V\!M^{(0)}_{440} and A ⁣M441(0){}^A\!M^{(0)}_{441} is presented in Appendix § 6.

Given such uncertainties, in [19] the ss-NME was simply assumed as a free parameter, presumably of O(1)O(1), to be constrained by comparison with the data (together with gAg_A). It turns out that, in this way, both the experimental spectrum shape and its normalization can be well reproduced theoretically [19]. In the same spirit, we treat (r,s)(r,s) as free parameters in our analysis.

Figure 2 shows three representative theoretical spectra St(we)S^t(w_e) calculated in the IBFM-2 model (dashed lines) for three different (r,s)(r,s) values. Their accurate reconstruction through a set of moments truncated at N=6N=6 is also shown in the left panel (dotted lines). Analogously, the right panel shows the reconstruction truncated at N=2N=2: it can be seen that the main qualitative features of the spectra are already captured by using just the first three moments μ0\mu_0, μ1\mu_1 and μ2\mu_2. Similar results hold for the spectra calculated in the MQPM and ISM models (not shown). The analysis in Sec. III will confirm that, in general, a few moments are enough to derive useful indications on the (r,s)(r,s) parameters.

Image not extracted

Figure 2. Representative examples of theoretical spectra StS^t as calculated in the IBFM-2 model (dashed lines), for three representative pairs of the (r,s)(r, s) parameters. The reconstruction of the spectra based on a truncated set of moments is also shown (dotted lines) for N=6N = 6 (left panel) and N=2N = 2 (right panel).

In particular, by using as free parameters both rr and ss, it appears from Figure 2 that one can alter both the peak position and the normalization of the theoretical spectrum, and may thus hope to match the experimental spectrum in Figure 1. It turns out that this result is achieved for typical values r∼1r\sim1 (confirming gAg_A quenching) and s∼O(1)s\sim O(1) (in the expected numerical ballpark), where the large NME cancellations mentioned in the context of Eqs. (25),(26) take place. We shall recover very similar results, not only by using an independent data set [16, 28] (as reported in Figure 1), but by adopting a different perspective, based on the following generalization of the quadratic form in Eq. (23).

Figure 1. Experimental energy spectrum for 113Cd^{113}\mathrm{Cd} β\beta decay (central values and total 1σ1\sigma errors) as taken from [28]. See the text for details.

We observe that, since ss adds to the NME terms multiplied by gVg_{\mathrm{V}}, it must appear up to second power in St(ω)S^t(\omega). Moreover, a mixed dependence ∝r⋅s\propto r \cdot s must emerge from the VA term in Eq. (23). Therefore, the spectrum St(we)S^t(w_e) must be a quadratic form in both rr and ss, as well as any integral over it involving wenw_e^n (with n≥0n \ge0),

∫wthrw0St(we)wen dwe=∑i+j≤2aijnrisj(27)\int_{w_{\mathrm{thr}}}^{w_0} S^t(w_e)w_e^n \,\mathrm{d}w_e = \sum_{i+j\le2} a_{ij}^n r^i s^j \tag*{(27)}

where the numerical coefficients aijna_{ij}^n (with nn being a superscript, not a power) are expressed in units of s−1\mathrm{s}^{-1} as the spectrum StS^t. The 0th0^{\mathrm{th}} moment μ0t(r,s)\mu_0^t(r, s) corresponds to the above quadratic form with n=0n = 0, while the nthn^{\mathrm{th}} moment μnt(r,s)\mu_n^t(r, s) for n≥1n \ge1 corresponds to a ratio of quadratic forms (with index nn at numerator and zero at denominator).

In practice, to evaluate the aijna_{ij}^n coefficients for a given nuclear model at fixed nn, one just needs to calculate the energy spectrum St(we∣r,s)S^t(w_e \mid r, s) at six arbitrary points (r,s)(r, s), evaluate the integral on the l.h.s. of Eq. (27), and solve in the six unknowns {aijn}\{a_{ij}^n\}.2 Table I reports the numerical {aijn}\{a_{ij}^n\} values (in units of 10−24 s−110^{-24}\,\mathrm{s}^{-1}) up to N=6N = 6, for the three nuclear models discussed in this work.

Modelnna20na_{20}^na02na_{02}^na11na_{11}^na10na_{10}^na01na_{01}^na00na_{00}^n
IBFM-20+2.8998+0.7914+2.8163−5.8662−2.9952+3.1087
IBFM-21+3.6046+1.0225+3.5722−7.2731−3.7984+3.8552
IBFM-22+4.5331+1.3358+4.5835−9.1214−4.8722+4.8357
IBFM-23+5.7690+1.7640+5.9492−11.574−6.3214+6.1368
IBFM-24+7.4305+2.3542+7.8102−14.862−8.2949+7.8803
IBFM-25+9.6855+3.1738+10.368−19.312−11.005+10.239
IBFM-26+12.774+4.3204+13.912−25.390−14.758+13.460
MQPM0+20.086+0.7914+7.5565−41.059−7.8113+21.146
MQPM1+24.911+1.0225+9.5846−50.905−9.9058+26.223
MQPM2+31.253+1.3358+12.298−63.839−12.706+32.892
MQPM3+39.676+1.7640+15.962−81.005−16.485+41.740
MQPM4+50.976+2.3542+20.955−104.01−21.632+53.598
MQPM5+66.281+3.1738+27.817−135.15−28.700+69.640
MQPM6+87.205+4.3204+37.325−177.68−38.485+91.542
ISM0+17.509+0.7914+7.0682−33.572−6.8313+16.166
ISM1+21.713+1.0225+8.9655−41.627−8.6632+20.050
ISM2+27.238+1.3358+11.504−52.208−11.113+25.151
ISM3+34.576+1.7640+14.932−66.252−14.418+31.919
ISM4+44.418+2.3542+19.603−85.078−18.920+40.990
ISM5+57.748+3.1738+26.024−110.56−25.103+53.262
ISM6+75.969+4.3204+34.920−145.36−33.663+70.018

Table 1. Coefficients aijna_{ij}^n of the quadratic forms parametrizing the theoretical moments up to N=6N = 6, in each of the three nuclear models considered in this work. The aijna_{ij}^n are expressed in units of 10−24 s−110^{-24}\,\mathrm{s}^{-1}.

Summarizing, we have discretized the continuous information contained in the infinite family of spectra St(we∣r,s)S^{t}(w_e \mid r, s) into a small number of moments μnt\mu_n^t, each depending on simple quadratic forms in the free parameters rr and ss (involving six coefficients aijna_{ij}^n at any nn). This approach greatly simplifies the numerical calculation of the theoretical moments, as well as their comparison with the experimental moments μne\mu_n^e, as discussed below.

Spectral moment method: Comparison of theory and data

In this Section we explore the implications of equating a finite set of theoretical and experimental moments:

μnt(r,s)=μne(n=0,1,…N).(28)\mu_n^t(r, s) = \mu_n^e \quad(n = 0, 1, \ldots N). \tag*{(28)}

Each of the above equations sets a quadratic form in (r,s)(r, s) equal to a constant, and thus leads to a conic section in the corresponding coordinates. It turns out that, for n=0n = 0, the conic section is a slanted and elongated ellipse, while for n≥1n \ge1 the conics form a bundle of hyperbolae. In the ideal case (perfect match between theory and data), all these curves would intersect in a single (r,s)(r, s) point; in real cases, the various crossing points will cluster with some dispersion around a preferred (r,s)(r, s) region. The smaller the dispersion, the better the agreement between the experimental and theoretical moments and spectra. In Appendix B we discuss general features of the conic sections and of their crossings, that allow to visualize the effects of large NME cancellations, as well as to interpret previous fit results obtained in [19] through the revised spectrum shape method. Below we show our results in the (r,s)(r, s) plane and discuss the preferred parameter values in the three nuclear models considered.

Figure 3 shows the loci of points in the (r,s)(r, s) plane fulfilling Eq. (28) up to N=6N = 6, for the model IBMF-2 (left), MQPM (middle), and ISM (right). In each panel, one can see (part of) the slanted ellipse determined by the zeroth moment, and the bundle of hyperbolae determined by the first and higher moments. The two regions where the ellipse and the bundle cross each other correspond to positive and negative values of ss, and are enlarged in the lower set of panels. As discussed in Appendix B, in principle there are two other regions of crossing, close to the extremal sides of the ellipse and thus beyond scale (not shown), that would correspond to unphysical values of rr (much smaller or much larger than unity).

In each of the enlarged crossing regions reported in Figure 3 (lower panels), the bundle of hyperbolae shows some dispersion, that turns out to be smaller for s>0s > 0 as compared with s<0s < 0, and minimal for the IBMF-2 model.

Figure 3. Numerical results obtained by imposing the equality of theoretical and experimental moments, μnt=μne\mu_n^t = \mu_n^e (up to N=6N = 6) in the plane charted by the free parameters r=gA/gVr = g_A/g_V and s=ss = s-NME. The left, middle and right plots correspond to the IBMF-2, MQPM and ISM models, respectively. In each of the three upper panels, note (part of) the slanted ellipse determined by the zeroth moment, and the bundle of hyperbolae determined by the first and higher moments. For each model, the two lower panels zoom in the regions at s<0s < 0 (left) and s>0s > 0 (right) where the ellipse and the bundle cross each other. See the text for details.

Modelrrss
IBFM-21.125+1.617
1.029−1.675-1.675
MQPM1.068+1.557
1.020−1.662-1.662
ISM0.994+1.635
0.960−1.729-1.729

Table 2. Intersection points of the 0th0^{\mathrm{th}} and 1st1^{\mathrm{st}} moment curves in the (r,s)(r,s) plane, for each of the three nuclear models considered.

Therefore, we expect that the experimental spectrum is best matched by the theoretical spectra at s>0s>0 (w.r.t. s<0s<0), in particular for the IBFM-2 model. For definiteness, we check these expectations by calculating the StS^t spectra at the points where the μ0\mu_0 ellipse intersects the μ1\mu_1 hyperbola, whose coordinates as reported in Table II.

Figure 4 shows the theoretical spectra corresponding to the above (r,s)(r,s) values for the three nuclear models: IBFM-2 (left), MQPM (middle) and ISM (right). In each panel, the experimental spectrum (in light blue) should be compared with the blue-dashed and red-dotted spectra, referring to positive and negative values of ss in Table II, respectively. The spectra with s>0s>0 are generally slightly broader and less peaked than the experimental spectrum, while the opposite happens for the spectra with s<0s<0. The deviations from the experimental spectrum are generally smaller for s>0s>0 than for s<0s<0, and can be as small as the experimental errors for the IBFM-2 model at s>0s>0. In this context, we recall that the model IBFM-2 predicts a priori s=0s=0 [18], and that only by fixing s∼1.6s\sim1.6 one gets the good agreement with the experimental data in Figure 4; see Appendix § 6 for details.

Theoretical spectra for the three nuclear models

Figure 4. Theoretical spectra calculated at the (r,s)(r,s) points reported in Table II, for each of the three nuclear models IBFM-2, MQPM and ISM (left, middle and right panels), as compared with the experimental spectrum of Fig. 1. In each panel, the dashed blue curve refers to the case s>0s>0, and the dotted red curve to the case s<0s<0.

We thus find that all models point towards r∼1r\sim1, corresponding to a quenching factor q=gA/gAvac∼0.8q=g_A/g_A^{\mathrm{vac}}\sim0.8 assuming gV=1g_V=1; and towards ∣s∣∼1.6\lvert s\rvert\sim1.6, corresponding to a small vector-like relativistic NME in the expected ballpark of O(1)O(1), with a preference for s>0s>0. More precisely, by grouping the values in Table II, we find the best match between theory and data around

r≃0.99−1.13,(29)r \simeq0.99-1.13 , \tag*{(29)}
s≃1.56−1.64,(30)s \simeq1.56-1.64 , \tag*{(30)}

with a secondary (worse) match around r≃0.96r\simeq0.96–1.031.03 and s=−(1.66s=-(1.66–1.73)1.73), that cannot be excluded a priori from a phenomenological viewpoint (see also Appendix § 6). Remarkably, the above ranges correspond to relatively small uncertainties on the (r,s)(r,s) parameters.

It would be tempting to refine the indications in favor of r∼1r\sim1 and ∣s∣∼1.6\lvert s\rvert\sim1.6 by attaching more accurate error estimates to these parameters, as derived by detailed data fits including both experimental and theoretical uncertainties. However, in our case the theoretical shape errors are likely to be larger than the data errors (in contrast with the data analysis in [19]), as suggested by the fact that, in Figure 4, the theoretical spectra are generally outside (or at border of) the experimental error band. As a check, we have performed numerical least-squares adjustments of the theoretical

spectra by including only experimental uncertainties, either by fitting the binned spectrum with uncorrelated total errors in Figure 1 or by fitting the associated N=6N = 6 moments with their propagated covariance matrix. In both cases we obtain unreasonably high values of the minimum χ2\chi^2 per d.o.f. (except for the noted IBFM-2 case at s>0s > 0), and unreasonably tiny errors on (r,s)(r, s) at subpercent level (in all cases), that are much smaller than the realistic few-percent spread of the same parameters (see Table II). On the positive side, by using only experimental errors, the (r,s)(r, s) best fits are invariably close to the s>0s > 0 solutions reported in Eqs. (29),(30) and their associated spectra are close to those in Figure 4, within percent-level deviations (not shown). The SMM appears thus to provide reasonably correct and robust (r,s)(r, s) values, even with a limited amount of information and without the need for detailed data fits. Future improvements may include theoretical spectrum shape errors as estimated, e.g., by a detailed analysis of the inputs or approximations inherent the current next-to-leading order NME calculations in each nuclear model: a task that is beyond the scope of this work.

Similar results on (r,s)(r, s) were obtained in [19], by applying a revised spectrum shape approach to an independent set of 113Cd^{113}\mathrm{Cd} β\beta-decay data, characterized by a higher threshold (Tthr≃52T_{\mathrm{thr}} \simeq52–132132 keV in different detectors, with ⟨Tthr⟩=92\langle T_{\mathrm{thr}}\rangle= 92 keV) with respect to the data used herein [16] (having a single Tthr=26T_{\mathrm{thr}} = 26 keV). In [19], the preferred values for the free parameters were found to cluster around gA≃0.83g_A \simeq0.83–0.990.99 (somewhat lower than our gA∼0.99g_A \sim0.99–1.131.13) and around s≃1.85s \simeq1.85–2.12.1 (somewhat higher than our s≃1.56s \simeq1.56–1.641.64). We surmise that these differences may be due in part to the different 113Cd^{113}\mathrm{Cd} datasets and in part to the different approach used in the analysis. In particular, in this work the spectrum normalization is constrained through the 0th0^{\mathrm{th}} moment (i.e., the absolute decay rate μ0\mu_0 above the 16 keV threshold), while in [19] it is constrained through the decay half life (that requires a spectrum extrapolation below each of the 52–132 keV detector thresholds). As noted at the end of § 2 A, such extrapolations may lead to biases. Altering the normalization leads to anticorrelated changes between gAg_A and ss (as suggested by the μ0\mu_0 ellipse in our approach), which is what we find in comparison with [19]. Apart from these differences, a robust message emerges from [19] and from this work: quenched values of the axial-vector coupling (around gA≃0.9g_A \simeq0.9–11), accompanied by an adjustment of the small vector NME (around s≃1.6s \simeq1.6–22), are required to match the existing 113Cd^{113}\mathrm{Cd} β\beta-decay spectrum data in both normalization and shape, in each of the three nuclear models considered.

Around these (r,s)(r, s) values, large NME cancellations take place, leaving residual spectra that reasonably reproduce the experimental spectrum both in normalization (0th0^{\mathrm{th}} moment) and in shape (1st1^{\mathrm{st}} and higher moments). These phenomenological facts strongly suggest that all the models require the following adjustments: an overall multiplicative correction to the axial-vector NME, via a quenched value gA∼1g_A \sim1 (for gV=1g_V = 1); and a small additive vector-like correction, parametrized by a ss-NME with an absolute value around s∼1.6s \sim1.6. One might wonder whether gVg_V could be used as free parameters instead of ss, abandoning the CVC assumption gV=1g_V = 1. The answer is negative, since the prefactor gV2g_V^2 in Eq. (23) would only affect the zeroth moment μ0\mu_0 but would cancel out in the first and higher moments [Eq. (8)], leading to a spectrum shape depending on a single free parameter rr. It was already concluded in [15] that one cannot match both the spectrum normalization and its shape by varying just gVg_V besides gAg_A. In other words, the additive adjustment parametrized by ss cannot be traded for a multiplicative adjustment parametrized by gVg_V. At present, the CVC assumption gV=1g_V = 1 can thus be safely maintained in the context of 113Cd^{113}\mathrm{Cd} β\beta decay.

Summary and perspectives

In this work we have studied the normalization and shape of the electron energy spectrum of the fourth-forbidden β\beta decay of 113Cd^{113}\mathrm{Cd} with a novel approach, coined the spectral-moment method (SMM), based on a truncated set of spectral moments μn\mu_n. The 0th0^{\mathrm{th}} moment is related to the normalization, while the 1st1^{\mathrm{st}} and higher moments are related to the shape. As in [19], we have assumed that the spectra depend on two free parameters: an axial-vector coupling parameter r=gA/gVr = g_A/g_V (for gV=1g_V = 1) and a small relativistic vector-like NME parameter ss. We have shown that: each moment is a quadratic form in (r,s)(r, s); iso-moment curves are ellipses for μ0\mu_0 and hyperbolae for μ1,2,3,…\mu_{1,2,3,\ldots}; the intersections of a few moment curves are enough to derive interesting constraints on (r,s)(r, s), without detailed data fits (see also Appendix B).

In particular, by equating the theoretical moments with the experimental ones, as derived from the data in [28], the following results emerge: the intersection of the μ0\mu_0 and μ1\mu_1 curves provides r∼1r \sim1 and ∣s∣∼1.6|s| \sim1.6; the case s>0s > 0 results in a smaller spread of intersections with higher-moment curves and is thus preferred, as also confirmed by visual inspection of the spectra. Nuclear model considerations also suggest s≥0s \geq0, although s<0s < 0 cannot be excluded a priori (see also Appendix § 6). The spread of the (r,s)(r, s) values in Table II, at the level of a few percent at least, exceeds the purely experimental uncertainties and call for (currently unquantified) theoretical spectrum-shape uncertainties. In any case, our results are in the same ballpark of those obtained in [19] with a different methodology and independent data. In general, the derived (r,s)(r, s) values provide evidence for a multiplicative renormalization (quenching) of the axial coupling gAg_A and for an additive adjustment of vector-like terms via the small relativistic ss-NME.

Since the main quantitative information (apart from indications about the sign of ss) has been derived just from μ0\mu_0 and μ1\mu_1, we surmise that the SMM can be quite powerful even when the experimental data are less accurate than those used in this work for 113Cd{}^{113}\mathrm{Cd}. In this context, it should be noted that 113Cd{}^{113}\mathrm{Cd} is just one of several nuclei where forbidden β\beta decays occur with a significant spectral dependence on gAg_A [15, 24, 34, 35] and possibly on other parameters such as the ss-NME or similar ones.

In cases where the available spectral data are scarce, it should anyway be possible to derive, within a specified energy window we∈[wemin⁡,wemax⁡]w_e \in[w_e^{\min}, w_e^{\max}], at least the spectrum normalization (μ0\mu_0) and the average energy (μ1\mu_1) with reasonable approximation. By equating the experimental and theoretical values for these two moments, constraints on (r,s)(r, s) or equivalent parameters could be derived, without the need of complicated data fits. Among the forbidden β\beta-decay spectra with significant gAg_A dependence, of particular importance is the 115In{}^{115}\mathrm{In} spectrum [15], that was experimentally observed long ago in [36] and recently measured with a bolometric detector in [25]. A first data analysis with the spectrum-shape method (SSM), and using only gAg_A as a free parameter, suggests significant gAg_A quenching [25] but does not account for both normalization and shape at the same time. It remains to be seen if allowance for an extra parameter such as the ss-NME can provide a match to all data. The presently introduced SMM might allow a rapid check of this possibility, and will be applied to, e.g., existing 115In{}^{115}\mathrm{In} data in a separate work.

Acknowledgments

The work of E.L. and A.M. is partly supported the Italian Ministero dell’Università e Ricerca (MUR) through the research grant number 2017W4HA7S “NAT-NET: Neutrino and Astroparticle Theory Network” under the program PRIN 2017, and by the Istituto Nazionale di Fisica Nucleare (INFN) through the “Theoretical Astroparticle Physics” (TAsP) project. We are grateful to P. Belli, R. Bernabei, F. A. Danevich and V. I. Tretyak for providing us with a digitized form of the 113Cd{}^{113}\mathrm{Cd} β\beta-decay spectrum data [16] as updated in Fig. 29 of [28], and for useful correspondence.

Analysis of relevant NME

In the Table III–Table V we have listed the single-particle (s.p.) matrix elements corresponding to all relevant transitions in the vector and axial-vector NME of interest for our analysis, namely, the small s=VM431(0)s = {}^{V}\mathcal{M}_{431}^{(0)} and the large (and largely cancelling) terms VM440(0){}^{V}\mathcal{M}_{440}^{(0)} and AM441(0){}^{A}\mathcal{M}_{441}^{(0)}. The relevant pre-factors have been included so that these correspond to single-particle model NME and are thus comparable to the numerical NME values listed in [18] for 113Cd{}^{113}\mathrm{Cd}.

0f7/20f_{7/2}0f5/20f_{5/2}1p3/21p_{3/2}1p1/21p_{1/2}0g9/20g_{9/2}0g7/20g_{7/2}1d5/21d_{5/2}1d3/21d_{3/2}2s1/22s_{1/2}0h11/20h_{11/2}
0f7/20f_{7/2}0000000009.5
0f5/20f_{5/2}000000000-4.3
1p3/21p_{3/2}000000000-12.5
1p1/21p_{1/2}0000000000
0g9/20g_{9/2}0000000000
0g7/20g_{7/2}0000000000
1d5/21d_{5/2}0000000000
1d3/21d_{3/2}0000000000
2s1/22s_{1/2}0000000000
0h11/20h_{11/2}-9.5-4.312.50000000

Table III. The single-particle matrix elements VM431(0){}^{V}\mathcal{M}_{431}^{(0)} [fm3^{3}] for 113Cd{}^{113}\mathrm{Cd} for the relevant orbitals.

0f7/20f_{7/2}0f5/20f_{5/2}1p3/21p_{3/2}1p1/21p_{1/2}0g9/20g_{9/2}0g7/20g_{7/2}1d5/21d_{5/2}1d3/21d_{3/2}2s1/22s_{1/2}0h11/20h_{11/2}
0f7/20f_{7/2}0500-160-47500000-301
0f5/20f_{5/2}5000-538000000615
1p3/21p_{3/2}160-5380000000794
1p1/21p_{1/2}-475000000000
0g9/20g_{9/2}00000774-283-6245650
0g7/20g_{7/2}00007740-7042226320
1d5/21d_{5/2}0000283-704091100
1d3/21d_{3/2}0000-624-222911000
2s1/22s_{1/2}0000-5656320000
0h11/20h_{11/2}301615-7940000000

Table V. The single-particle matrix elements AM441(0)\mathcal{A}\mathcal{M}_{441}^{(0)} [fm4^{4}] for 113^{113}Cd for the relevant orbitals.

Given that the initial and final states are ground states with spin-parities 1/2+1/2^+ and 9/2+9/2^+, respectively, the transition is most likely dominated by a transition between the neutron orbital 2s1/22s_{1/2} and the proton orbital 0g9/20g_{9/2}. This means VM431(0)(s.p.)=0 fm3{}^{V}\mathcal{M}_{431}^{(0)}(\mathrm{s.p.})=0\ \mathrm{fm}^{3}, VM440(0)(s.p.)=637 fm4{}^{V}\mathcal{M}_{440}^{(0)}(\mathrm{s.p.})=637\ \mathrm{fm}^{4}, and AM441(0)(s.p.)=565 fm4{}^{A}\mathcal{M}_{441}^{(0)}(\mathrm{s.p.})=565\ \mathrm{fm}^{4}. The nuclear models give VM440(0)=−317{}^{V}\mathcal{M}_{440}^{(0)}=-317–827 fm4827\ \mathrm{fm}^{4}, and AM441(0)=314{}^{A}\mathcal{M}_{441}^{(0)}=314–848 fm4848\ \mathrm{fm}^{4}, which are in reasonable agreement with the single-particle NME with the final values depending on contributions from other near-by orbitals. For VM431(0){}^{V}\mathcal{M}_{431}^{(0)} the situation is however very different, as no orbitals near the Fermi-surface have nonzero contributions. Furthermore, the CVC relation VM431(0)=0.0678(R−1)VM440(0){}^{V}\mathcal{M}_{431}^{(0)}=0.0678(R^{-1}){}^{V}\mathcal{M}_{440}^{(0)} mentioned in Sec. IID (which does not apply exactly when multiple configurations are allowed for the wave functions) suggests that VM431(0)(MQPM,CVC)=9.7 fm3{}^{V}\mathcal{M}_{431}^{(0)}(\mathrm{MQPM},\mathrm{CVC})=9.7\ \mathrm{fm}^{3}, VM431(0)(IBFM−2,CVC)=3.7 fm3{}^{V}\mathcal{M}_{431}^{(0)}(\mathrm{IBFM-2},\mathrm{CVC})=3.7\ \mathrm{fm}^{3}, and VM431(0)(ISM)=8.4 fm3{}^{V}\mathcal{M}_{431}^{(0)}(\mathrm{ISM})=8.4\ \mathrm{fm}^{3}, while the models give the values 0.37 fm30.37\ \mathrm{fm}^{3}, 0 fm30\ \mathrm{fm}^{3}, and 0 fm30\ \mathrm{fm}^{3}, respectively. For the CVC values for the smallest NME to hold, the transition would need to be basically a pure 0h11/20h_{11/2}–1p3/21p_{3/2} which does not seem reasonable based on the spin-parities, the facts that these are ground states, and the proton and neutron numbers of the nuclei.

0f7/20f_{7/2}0f5/20f_{5/2}1p3/21p_{3/2}1p1/21p_{1/2}0g9/20g_{9/2}0g7/20g_{7/2}1d5/21d_{5/2}1d3/21d_{3/2}2s1/22s_{1/2}0h11/20h_{11/2}
0f7/20f_{7/2}434319-363-42600000695
0f5/20f_{5/2}-319336484000000-306
1p3/21p_{3/2}-363-4840000000-907
1p1/21p_{1/2}426000000000
0g9/20g_{9/2}0000729382-644-3976370
0g7/20g_{7/2}0000-382621452-501-5700
1d5/21d_{5/2}0000-644-45258382000
1d3/21d_{3/2}0000397-501-820000
2s1/22s_{1/2}00006375700000
0h11/20h_{11/2}695306-9070000001111

Table IV. The single-particle matrix elements VM440(0)\mathcal{V}\mathcal{M}_{440}^{(0)} [fm4^{4}] for 113^{113}Cd for the relevant orbitals.

For ISM and IBFM-2 the value of VM431(0)\mathcal{V}\mathcal{M}_{431}^{(0)} is systematically zero, as in the ISM calculations the orbital 0h11/20h_{11/2} was kept empty to reduce the large computational burden, and in IBFM-2 the initial state with spin-parity 1/2+1/2^{+} cannot be formed by pairing 0+0^{+} and 2+2^{+} bosons with a fermion with spin-parity 11/2−11/2^{-}. For MQPM the contributions between 0h11/20h_{11/2} and the lower proton orbitals are included, but the contributions for the higher orbitals are not very reliable as the parameter tuning can be reasonably done only for the lowest orbitals.

Based on these arguments, we surmise that the s-NME is not well described by the nuclear models while the other NME do not suffer from the same problems. Therefore, it makes sense to take the s-NME as a tuning parameter instead of the large matrix elements. Concerning the sign of ss, the option s>0s > 0 is theoretically regarded as being more reasonable [19], since all the above estimates—despite being largely uncertain—typically provide s≥0s \ge0. However, one cannot exclude a priori that s<0s < 0, also because some single-particle contributions may be negative (see Table III).

Quadratic forms in the (r,s)(r,s) parameters and conic sections

As mentioned in Sec. II D and discussed in [19], in the (r,s)(r,s) region where theoretical and experimental spectra match, subtle cancellations in StS^{t} occur among large NME, with residuals modulated by smaller terms. Despite the complexity of the full StS^{t} expression at next-to-leading order [18], some insights can be gained by elaborating upon Eqs. (25, 26), and by recalling that the moments are associated with quadratic forms in both rr and ss.

The leading NME cancellation operating in StS^{t}, as well as in its integral μ0t\mu_{0}^{t}, takes the form of a square of a large linear term in rr, modulated by a smaller term in ss. In first approximation, we thus expect the 0th0^{\mathrm{th}} moment to take the form μ0t≃(a−br−cs)2+d\mu_{0}^{t} \simeq(a-br-cs)^{2}+d, with aa and bb much larger than cc and dd. In the plane charted by the (r,s)(r,s) parameters, the equation μ0t=μ0e\mu_{0}^{t}=\mu_{0}^{e} is then solved by two parallel straight lines with a small slope −c/b-c/b. In reality, due to subleading NME terms, the perfect square is altered into a more general quadratic form in (r,s)(r,s). Correspondingly, the two straight lines are just the degenerate limit of a conic section, which is actually an elongated and slanted ellipse.

Figure 5 (upper panel) reports graphically the above qualitative considerations in the plane charted by the rr and ss parameters. The elongated solid ellipse is the locus of points where μ0t=μ0e\mu_{0}^{t}=\mu_{0}^{e}, i.e., where the normalization (event rate above threshold) of the experimental and theoretical spectra match, irrespective of their shapes that may be quite different. The dashed straight lines are a degenerate approximation of the ellipse.

Image not extracted

Figure 5. Qualitative expectations for the conic sections obtained by equating the theoretical and experimental moments in the (r,s)(r, s) plane (in arbitrary scales). Upper panel: ellipse associated to μ0\mu_0, together with the degenerate limit (straight parallel lines). Middle panel: hyperbola associated to μ1\mu_1, together with it asymptotes (roughly aligned along the vertical axis and the ellipse major axis); the hyperbola crosses the ellipse in four points (solutions). Lower panel: bundle of hyperbolae associated to μ1\mu_1 and higher moments, crossing each other and the μ0\mu_0 ellipse in a single point (marked with a dot).

While the spectrum normalization is associated to μ0\mu_{0}, the spectrum shape is associated to higher moments, starting from μ1\mu_{1}. Considering that μ1\mu_{1} is a ratio of quadratic forms, the constraint μ1t(r,s)=μ1e\mu_{1}^{t}(r,s)=\mu_{1}^{e} leads to a quadratic equation as well, identifying another conic section that turns out to be a hyperbola. The appearance of a hyperbola for μ1\mu_{1} can be qualitatively understood as follows.

Since the denominator and numerator of μ1\mu_{1} differ only by a wew_{e} integrand of O(1)O(1) in the latter, [see Eq. (8)], the associated quadratic forms turn out to have nearly proportional coefficients (as confirmed by numerical inspection of Tab. I), corresponding to geometrically similar ellipses upon rescaling. The scaling factor turns out to be slightly more pronounced along the abscissa rr, that governs the main NME cancellation term. Roughly speaking, the equation μ1t(r,s)=μ1e\mu_1^t(r, s) = \mu_1^e imposes that the ratio between two very similar elliptic forms (slightly differing along the abscissa) is close to unity. If one form is written as x2+y2−2ρxyx^2 + y^2 - 2\rho xy (where xx and yy are generic coordinates), the other is thus obtained by scaling xx as (1+δ)x(1 + \delta)x, where δ≪1\delta\ll1. It is easy to check that, at first order in δ\delta, the ratio of these two forms is unity for either x=0x = 0 (corresponding to a vertical line) or for x=ρyx = \rho y (a slanted line, roughly along the ellipse major axis). These two lines define a degenerate hyperbola, coincident with its two asymptotes: a vertical one and a slanted one. In general, the equation μ1t(r,s)=μ1e\mu_1^t(r, s) = \mu_1^e entails a less simplistic situation: the hyperbola defined by this equation is not exactly degenerate, and its asymptotes may be slightly tilted with respect to the above expectations.

Figure 5 (middle panel) reports graphically the typical locus of points where μ1t(r,s)=μ1e\mu_1^t(r, s) = \mu_1^e, i.e., where the average energies of the experimental and theoretical spectra do match. The locus is a hyperbola (blue solid curve), whose branches are relatively close to the degenerate limit (asymptotes, dashed lines). The hyperbola and the ellipse will cross in four points, i.e., there will be four different (r,s)(r, s) solutions to the coupled equations μ0t=μ0e\mu_0^t = \mu_0^e and μ1t=μ1e\mu_1^t = \mu_1^e. In general, the extreme solutions in rr (red dots) correspond to unphysically low or high values of gAg_A, and can be discarded a priori; the remaining two solutions (blue dots) typically correspond to negative and positive values of ss.

With a reasoning similar to μ1\mu_1, also the conic sections defined by the second or higher moments are expected to be hyperbolae. One then gets an ellipse for n=0n = 0 (related to the event rate) and a bundle of hyperbolas for n≥1n \ge1 (related to the spectrum shape). If a perfect match between theory and data can be achieved, all these conic sections must intersect in just one of the four previous points, and diverge to some extent in the others. This case is shown in the lower panel of Figure 5, assuming the solution marked by a dot. In reality, the theory-data match is never perfect, and the intersections will be slightly separated even in the best possible case.

The above discussion allows to understand some features of previous results reported in [19] by using the so-called revised SSM. This approach conflates the half-life and spectrum-shape methods (previously conflicting when using only rr as a free parameter, see e.g., [18]), by treating ss as an additional degree of freedom. In particular, let us consider Figure 2 of [19], showing the results of fits to an independent 13^{13}C β\beta-decay data set [17] in the plane charted by gAg_A and the ss-NMA, for the same three nuclear models considered in our work. In that figure, while the half-life fit leads to an elliptical solution (akin to the ellipse determined by μ0\mu_0 in our formalism), the spectrum shape fit leads to a band that crosses the ellipse. We can interpret such a band as the path taken by the fit algorithm in [19] to follow (and to fuzzily jump across) the two half-branches of the bundle of higher-moment hyperbolae, the other two halves (leading to unphysical values of gAg_A) being discarded by construction (the fit was also restricted to s>0s > 0 therein). With the spectral moment method discussed herein, the fit results in [19] can be understood in a unified picture.

References

  1. [1]C. Adams, K. Alfonso, C. Andreoiu, E. Angelico, I. J. Arnquist, J. A. A. Asaad, F. T. Avignone, S. N. Axani, A. S. Barabash and P. S. Barbeau, et al. “Neutrinoless Double Beta Decay,” [arXiv:2212.11099 [nucl-ex]].arxiv.org/abs/2212.11099
  2. [2]M. Agostini, G. Benato, J. A. Detwiler, J. Menéndez and F. Vissani, “Toward the discovery of matter creation with neutrinoless double-beta decay,” [arXiv:2202.01787 [hep-ex]].DOI
  3. [3]V. Cirigliano, Z. Davoudi, W. Dekens, J. de Vries, J. Engel, X. Feng, J. Gehrlein, M. L. Graesser, L. Gráf and H. Hergert, et al. “Neutrinoless Double-Beta Decay: A Roadmap for Matching Theory to Experiment,” [arXiv:2203.12169 [hep-ph]].arxiv.org/abs/2203.12169
  4. [4]J. T. Suhonen, “Value of the Axial-Vector Coupling Strength in β and ββ Decays: A Review,” Front. in Phys. 5, 55 (2017) [arXiv:1712.01565 [nucl-th]].DOI
  5. [5]J. Suhonen and J. Kostensalo, “Double β Decay and the Axial Strength,” Front. in Phys. 7, 29 (2019)DOI
  6. [6]J. Liu et al. [UCNA], “Determination of the Axial-Vector Weak Coupling Constant with Ultracold Neutrons,” Phys. Rev. Lett. 105, 181803 (2010) [arXiv:1007.3790 [nucl-ex]].DOI
  7. [7]D. Mund, B. Maerkisch, M. Deissenroth, J. Krempel, M. Schumann, H. Abele, A. Petoukhov and T. Soldner, “Determination of the Weak Axial Vector Coupling λ = gA/gV from a Measurement of the β-Asymmetry Parameter A in Neutron Beta Decay,” Phys. Rev. Lett. 110, 12502 (2013) [arXiv:1204.0013 [hep-ex]].
  8. [8]W. T. Chou, E. K. Warburton and B. A. Brown, “Gamow-Teller beta-decay rates for A ≤ 18 nuclei,” Phys. Rev. C 47, 163-177 (1993)pubmed.ncbi.nlm.nih.gov/9968421
  9. [9]A. Faessler, G. L. Fogli, E. Lisi, V. Rodin, A. M. Rottunno and F. Simkovic, “Overconstrained estimates of neutrinoless double beta decay within the QRPA,” J. Phys. G 35, 075104 (2008) [arXiv:0711.3996 [nucl-th]].DOI
  10. [10]J. Barea, J. Kotila and F. Iachello, “Nuclear matrix elements for double-β decay,” Phys. Rev. C 87, no.1, 014315 (2013) [arXiv:1301.4203 [nucl-th]].DOI
  11. [11]H. Ejiri, “Nuclear Matrix Elements for β and ββ Decays and Quenching of the Weak Coupling gA in QRPA,” Front. in Phys. 7, 30 (2019)DOI
  12. [12]H. Ejiri, J. Suhonen and K. Zuber, “Neutrino–nuclear responses for astro-neutrinos, single beta decays and double beta decays,” Phys. Rept. 797, 1-102 (2019)DOI
  13. [13]P. Gysbers, G. Hagen, J. D. Holt, G. R. Jansen, T. D. Morris, P. Navrátil, T. Papenbrock, S. Quaglioni, A. Schwenk and S. R. Stroberg, et al. “Discrepancy between experimental and theoretical β-decay rates resolved from first principles,” Nature Phys. 15, no.5, 428-431 (2019) [arXiv:1903.00047 [nucl-th]].DOI
  14. [14]V. Cirigliano, Z. Davoudi, J. Engel, R. J. Furnstahl, G. Hagen, U. Heinz, H. Hergert, M. Horoi, C. W. Johnson and A. Lovato, et al. “Towards precise and accurate calculations of neutrinoless double-beta decay,” J. Phys. G 49, no.12, 120502 (2022) [arXiv:2207.01085 [nucl-th]].DOI
  15. [15]M. Haaranen, P. C. Srivastava and J. Suhonen, “Forbidden nonunique β decays and effective values of weak coupling constants,” Phys. Rev. C 93, no.3, 034308 (2016)DOI
  16. [16]P. Belli, R. Bernabei, N. Bukilic, F. Cappella, R. Cerulli, C. J. Dai, F. A. Danevich, J. R. d. Laeter, A. Incicchitti and V. V. Kobychev, et al. “Investigation of β decay of 113Cd,” Phys. Rev. C 76, 064603 (2007)
  17. [17]L. Bodenstein-Dresler et al. [COBRA], “Quenching of gA deduced from the β-spectrum shape of 113Cd measured with the COBRA experiment,” Phys. Lett. B 800, 135092 (2020) [arXiv:1806.02254 [nucl-ex]].
  18. [18]M. Haaranen, J. Kotila and J. Suhonen, “Spectrum-shape method and the next-to-leading-order terms of the β-decay shape factor,” Phys. Rev. C 95, no.2, 024327 (2017)DOI
  19. [19]J. Kostensalo, J. Suhonen, J. Volkmer, S. Zatschler and K. Zuber, “Confirmation of gA quenching using the revised spectrum-shape method for the analysis of the 113Cd β-decay as measured with the COBRA demonstrator,” Phys. Lett. B 822, 136652 (2021) [arXiv:2011.11061 [nucl-ex]].DOI
  20. [20]H. Behrens and W. Bühring, “Electron Radial Wave Functions and Nuclear Beta-decay” (Clarendon Press, 1982), 626 pages.
  21. [21]O. S. Kirsebom, S. Jones, D. F. Strömberg, G. Martínez-Pinedo, K. Langanke, F. K. Röpke, B. A. Brown, T. Eronen, H. O. U. Fynbo and M. Hukkanen, et al. “Discovery of an Exceptionally Strong β-Decay Transition of 20F and Implications for the Fate of Intermediate-Mass Stars,” Phys. Rev. Lett. 123, no.26, 267201 (2019) [arXiv:1905.09407 [astro-ph.SR]].DOI
  22. [22]A. Kumar, P. C. Srivastava, J. Kostensalo and J. Suhonen, “Second-forbidden nonunique β-decays of 24Na and 36Cl assessed by the nuclear shell model,” Phys. Rev. C 101, no.6, 064304 (2020) [arXiv:2007.08122 [nucl-th]].DOI
  23. [23]J. V. Dawson, C. Reeve, J. R. Wilson, K. Zuber, M. Junker, C. Gossling, T. Kottig, D. Munstermann, S. Rajek and O. Schulz, “An investigation into the 113Cd beta decay spectrum using a CdZnTe array,” Nucl. Phys. A 818, 264-278 (2009) [arXiv:0901.0996 [nucl-ex]].DOI
  24. [24]J. Kostensalo, M. Haaranen and J. Suhonen, “Electron spectra in forbidden β decays and the quenching of the weak axial-vector coupling constant gA,” Phys. Rev. C 95, no.4, 044313 (2017)DOI
  25. [25]A. F. Leder, D. Mayer, J. L. Ouellet, F. A. Danevich, L. Dumoulin, A. Giuliani, J. Kostensalo, J. Kotila, P. de Marcillac and C. Nones, et al. “Determining gA/gV with High-Resolution Spectral Measurements Using a LiInSe2 Bolometer,” Phys. Rev. Lett. 129, no.23, 232502 (2022) [arXiv:2206.06559 [nucl-ex]].
  26. [26]W. Feller, An Introduction to Probability Theory and Its Applications, Volume II (Wiley, 2nd edition, 1991), 704 pages.
  27. [27]J. A. Shohat and J. D. Tamarkin, The Problem of Moments, Mathematical Surveys and Monographs, Volume I (American Society of Mathematics, 1970), 144 pages.
  28. [28]P. Belli, R. Bernabei, F. A. Danevich, A. Incicchitti and V. I. Tretyak, “Experimental searches for rare alpha and beta decays,” Eur. Phys. J. A 55, no.8, 140 (2019) [arXiv:1908.11458 [nucl-ex]].DOI
  29. [29]K. Schmüdgen, The Moment Problem, Graduate Texts in Mathematics (Springer International Publishing, 2017), 548 pages.DOI
  30. [30]N. I. Akhiezer, The Classical Moment Problem: And Some Related Questions in Analysis (Dover Publ., 2020), 253 pages.DOI
  31. [31]G. Talenti, “Recovering a function from a finite number of moments,” Inverse Problems 3, 501 (1987).DOI
  32. [32]R. Askey, I. J. Schoenberg and A. Sharma, “Hausdorff’s Moment problem and Expansions in Legendre Polynomials,” Journal of Mathematical Analysis and Applications 86, 237 (1982).DOI
  33. [33]M. Wang, G. Audi, F. G. Kondev, W. J. Huang, S. Naimi and X. Xu, “The AME2016 atomic mass evaluation” Chin. Phys. C 41, 030003 (2017).
  34. [34]J. Kostensalo and J. Suhonen, “gA-driven shapes of electron spectra of forbidden β decays in the nuclear shell model,” Phys. Rev. C 96, no.2, 024317 (2017)DOI
  35. [35]A. Kumar, P. C. Srivastava and J. Suhonen, “Second-forbidden nonunique β− decays of 59,60Fe: possible candidates for gA sensitive electron spectral-shape measurements,” Eur. Phys. J. A 57, no.7, 225 (2021) [arXiv:2101.03046 [nucl-th]].DOI
  36. [36]G. B. Beard and W. H. Kelly, “Beta Decay of Naturally Radioactive In-115,” Phys. Rev. 122, 1576 (1961).DOI

Paper details

vlnc:W_ejksfhvg
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents