betaFeedback

WorksPhysics and Astronomy

Majorana neutrino mass constraints in the landscape of nuclear matrix elements

Abstract

We discuss up-to-date constraints on the Majorana neutrino mass mββm_{ββ} from neutrinoless double beta decay (0νββ0νββ) searches in experiments using different isotopes: KamLAND-Zen and EXO (136^{136}Xe), GERDA and MAJORANA (76^{76}Ge) and CUORE (130^{130}Te). Best fits and upper bounds on mββm_{ββ} are explored in the general landscape of nuclear matrix elements (NME), as well as for specific NME values obtained in representative nuclear models. By approximating the likelihood of 0νββ0νββ signals through quadratic forms, the analysis of separate and combined isotope data becomes exceedingly simple, and allows to clarify various aspects of multi-isotope data combinations. In particular, we analyze the relative impact of different data in setting upper bounds on mββm_{ββ}, as well as the conditions leading to nonzero mββm_{ββ} at best fit, for variable values of the NMEs. Detailed results on mββm_{ββ} from various combinations of data are reported in graphical and numerical form. Implications for future 0νββ0νββ data analyses and NME calculations are briefly discussed.

Introduction

The process of neutrinoless double beta decay (0νββ0\nu\beta\beta),

(Z,A)→(Z+2,A)+2e−(1)(Z,A)\to(Z+2,A)+2e^{-} \tag*{(1)}

expected to occur for some candidate isotopes (Z,A)(Z,A) if neutrinos are Majorana fermions, may be interpreted as a miniature event of leptonic matter creation or “Little Bang,” whose discovery would have profound implications for particle and nuclear physics and for cosmology [1].

In the standard three-neutrino paradigm [2], the process would be mediated by three Majorana neutrino mass states νi\nu_i (i=1,2,3i=1,2,3) mixed with the three known flavor states να\nu_\alpha (α=e,μ,τ\alpha=e,\mu,\tau) via a unitary mixing matrix UαiU_{\alpha i}, parametrized in terms of three mixing angles (θ12\theta_{12}, θ13\theta_{13}, θ23\theta_{23}), one (Dirac) phase δ\delta, and two (Majorana) phases φ1,2\varphi_{1,2}. The relevant particle physics parameter is the effective Majorana neutrino mass mββm_{\beta\beta}, defined as

mββ=∣Ue12m1+Ue22m2+Ue32m3∣(2)m_{\beta\beta}=\left|U_{e1}^{2}m_1+U_{e2}^{2}m_2+U_{e3}^{2}m_3\right| \tag*{(2)}

and related to the observable 0νββ0\nu\beta\beta decay half-life TiT_i in each isotope i=(Z,A)i=(Z,A) via

1Ti=GiMi2mββ2(3)\frac{1}{T_i}=G_iM_i^2m_{\beta\beta}^2 \tag*{(3)}

where GiG_i is the phase space, and MiM_i is the nuclear matrix element (NME) for the decay. It is useful to contrast the Majorana ν\nu mass mββm_{\beta\beta} with the sum of neutrino masses

Σ=m1+m2+m3(4)\Sigma=m_1+m_2+m_3 \tag*{(4)}

that, being a source of gravity, can produce observable cosmological effects [2]. Figure 1 shows the regions allowed in the (Σ,mββ)(\Sigma,m_{\beta\beta}) plane at the 2σ2\sigma level (Δχ2=4\Delta\chi^2=4) by a global analysis of neutrino oscillation data [3], for masses mim_i either in normal ordering (NO, m1,2<m3m_{1,2}<m_3) or in inverted ordering (IO, m3<m1,2m_3<m_{1,2}). For a given value of Σ\Sigma, the vertical spread of mββm_{\beta\beta} is mostly due to the unknown relative phases of the UeiU_{ei} matrix elements in Eq. (2).

Image not extracted

Figure 1. Majorana ν\nu mass mββm_{\beta\beta} versus the sum of ν\nu masses Σ\Sigma (both in units of 10−310^{-3} eV). The points inside the blue (NO) and red (IO) wedge-shaped regions are allowed at 2σ2\sigma by the global analysis of ν\nu oscillation data. Adapted from [3].

Current cosmological data provide typical upper bounds on Σ\Sigma at the level of O(100)O(100) meV, that are more easily accommodated in NO than in IO [3]; see also the overview of recent constraints on Σ\Sigma and their impact on ν\nu mass ordering in [4]. Several 0νββ0\nu\beta\beta decay searches are also exploring the O(100) meV range for mββm_{\beta\beta} [1, 2]; in particular, the latest constraints from KamLAND-Zen [5] (136Xe{}^{136}\mathrm{Xe}) can plunge into the region mββ∼m_{\beta\beta}\sim few ×10\times10 meV for favorable values of the NME.1 Other very sensitive 0νββ0\nu\beta\beta searches, all probing half-lives Ti>1025T_i>10^{25} y at 90%90\% C.L., have been performed by the experiments EXO [6] (136Xe{}^{136}\mathrm{Xe}), GERDA [7] and MAJORANA [8] (76Ge{}^{76}\mathrm{Ge}), and CUORE [9] (130Te{}^{130}\mathrm{Te}).

Building upon previous work [3], we discuss in detail how to combine current (Xe, Ge, Te) data for given NME values.2 The approach allows a do-it-yourself 0νββ0\nu\beta\beta global analysis in terms of χ2\chi^2 functions with (up to) quadratic dependence on the signal strength 1/Ti1/T_i, which is a good approximation to recent results [3, 10, 11]. In particular, we describe how to derive mββm_{\beta\beta} constraints at a given confidence level, using both separate and combined (Xe, Ge, Te) data, for generic values of the nuclear matrix elements (the “NME landscape”), as well as for representative NME values from different nuclear models. Our approach clarifies interesting aspects of the 0νββ0\nu\beta\beta data analysis, such as the relative importance of each isotope in determining (non)zero best fits and upper bounds for mββm_{\beta\beta}.

The paper is structured as follows: In § 2 we describe the ingredients of our analysis in terms of notation, parametrization of experimental results for (Xe, Ge, Te), and associated NME’s. In § 3 we discuss the main results of the analysis in terms of mββm_{\beta\beta} constraints, by considering two qualitatively different situations: (1) cases where, a priori, mββ=0m_{\beta\beta}=0 is preferred, and (2) more general cases where the best fit may be at mββ>0m_{\beta\beta}>0. Upper bounds on mββm_{\beta\beta} are explored both graphically and numerically in the NME landscape, by using separate and combined (Xe, Ge, Te) data. In § 4 we summarize our results and comment on further applications and perspectives.

Ingredients of the Analysis

In this Section we introduce the notation, the experimental results and their parametrization, the landscape of NME and the phase space related to the three isotopes Xe, Ge and Te.

Notation and units

Following [3], we introduce the inverse half-life

Si=1/Ti ,(5)S_i = 1/T_i \,, \tag*{(5)}

that represents, up to a constant factor, the observable decay rate or signal strength in each i=(Z,A)i=(Z,A) isotope. Equation (3) reads then

Si=GiMi2mββ2 .(6)S_i = G_i M_i^2 m_{\beta\beta}^2 \,. \tag*{(6)}

To keep the notation compact, we absorb in GiG_i terms as 1/me21/m_e^2 and gA4g_A^4 (where gA=1.276g_A=1.276 [12] is the bare value of the axial-vector coupling), that are factorized out in other conventions. In particular, we can make contact with the notation of [1], where 1/T=G01gA4M0ν2mββ2/me21/T=G_{01}g_A^4M_{0\nu}^2m_{\beta\beta}^2/m_e^2, by identifying G=G01gA4/me2G=G_{01}g_A^4/m_e^2 and M=M0νM=M_{0\nu} for each isotope. We also follow [1] by taking the MiM_i as positive real numbers, referred to the bare value of gAg_A (unless otherwise noticed). Qualitative effects of the so-called quenching of gAg_A in nuclear matter [13] are separately commented below.

Finally, the following units are adopted:

[mββ]=meV,(7)[m_{\beta\beta}] = \mathrm{meV}, \tag*{(7)}
[Ti]=1026 y,(8)[T_i] = 10^{26}\,y, \tag*{(8)}
[Si]=10−26 y−1,(9)[S_i] = 10^{-26}\,y^{-1}, \tag*{(9)}
[Gi]=10−26 y−1 (meV)−2.(10)[G_i] = 10^{-26}\,y^{-1}\,(\mathrm{meV})^{-2}. \tag*{(10)}

Experimental inputs and parametrizations

In principle, the 0νββ0\nu\beta\beta data analysis would be straightforward, if likelihood profiles were provided for the signal strength SiS_i (or for TiT_i) in each experiment, e.g., in terms of a function χi2(Si)\chi_i^2(S_i). Barring error correlations among independent experiments, one should sum up the χi2\chi_i^2 functions, express the SiS_i in terms of mββm_{\beta\beta} via Eq. (6) for a given set of MiM_i, and map the resulting best fits and allowed regions for mββm_{\beta\beta}. In practice, experimental papers often focus on a single point of the likelihood profile (e.g., the TiT_i bound at 90% C.L.), whereas its shape has to be derived from supplementary information.

A useful empirical fact, first noted in [10] and further elaborated in [3], is that the functions χi2(Si)\chi_i^2(S_i) are often well approximated by (up to) quadratic forms in SiS_i; see also the recent results in [11]. Such forms cover 0νββ0\nu\beta\beta decay searches ranging from zero background (with poissonian, linear dependence on SiS_i) to large backgrounds (with gaussian, quadratic dependence on SiS_i) [3]. In particular, we have checked that the quadratic approximation works very well also for the latest KamLAND-Zen data [5, 14], up to 3σ3\sigma level at least.

Each experimental result is thus parametrized through a function Δχi2(Si)\Delta\chi_i^2(S_i) of the form:

Δχi2(Si)=aiSi2+biSi+ci,(11)\Delta\chi_i^2(S_i) = a_i S_i^2 + b_i S_i + c_i, \tag*{(11)}

where the offset cic_i is set by the condition that the minimum value Δχi2=0\Delta\chi_i^2=0 is reached within the physical region Si≥0S_i \ge0, namely,

ci={0for bi≥0,bi2/4aifor bi<0.(12)c_i = \begin{cases} 0 & \text{for } b_i \ge0,\\ b_i^2/4a_i & \text{for } b_i < 0. \end{cases} \tag*{(12)}

For ai>0a_i > 0 the Δχi2\Delta\chi_i^2 functions are parabolic, with a vertex placed at either Si=0S_i = 0 (null result, bi=0b_i = 0), or Si<0S_i < 0 (negative fluctuation in the unphysical region, bi>0b_i > 0), or Si>0S_i > 0 (physical signal or positive fluctuation, bi<0b_i < 0). In the latter case, the offset cic_i guarantees Δχi2=0\Delta\chi_i^2 = 0 at Si=−bi/2aiS_i = -b_i/2a_i. For ai=0a_i = 0, the Δχi2\Delta\chi_i^2 functions are linear. For the same isotope, the results of independent experiments are combined by summing their Δχi2\Delta\chi_i^2’s, and readjusting the total offset as per Eq. (12). In all cases, 90% C.L. bounds on the half-life (T90=1/S90T_{90} = 1/S_{90}) are obtained by imposing Δχi2(S90)=2.706\Delta\chi_i^2(S_{90}) = 2.706.

Table I, updated from [3] with the inclusion of the latest KamLAND-Zen results [5], reports the coefficients of the parametrization in Eq. (11) and the T90T_{90} bounds for the most sensitive current experiments (T90>0.1T_{90} > 0.1), as well as for combinations of experiments using the same isotope. For later purposes, we also consider hypothetical CUORE results for an exactly null signal, denoted as CUORE* (or Te*).

Figure 2 shows the numerical information of Table I in graphical form; the left and right panels refer, respectively, to separate experiments and to same-isotope combinations (Xe, Ge, Te). A few remarks about these graphs and the numerics are in order. The case of linear Δχ2\Delta\chi^2 functions applies to current GERDA and MAJORANA results, whose combination (denoted as Ge) sets a bound T90=2.07T_{90}=2.07 stronger than for GERDA alone (T90=1.8T_{90}=1.8). All the other experiments are characterized by parabolic functions. KamLAND-Zen and EXO report, respectively, a negative and a positive fluctuation, that partly cancel in their combination (denoted as Xe). As a result, the Xe bound T90=2.26T_{90}=2.26 is slightly weaker (T90=2.26T_{90}=2.26) than for KamLAND-Zen alone (T90=2.3T_{90}=2.3). Note that, for both the Ge and Xe combinations, it is Δχi2=0\Delta\chi_i^2=0 at Si=0S_i=0.

IsotopeExperiment or combinationaia_ibib_icic_iT90/1026 yT_{90}/10^{26}\,yT90T_{90} (expt.)Reference
136Xe^{136}\mathrm{Xe}KamLAND-Zen5.1573.9780.0002.3002.3[?]
136Xe^{136}\mathrm{Xe}EXO0.440−0.3380.0650.3500.35[?]
76Ge^{76}\mathrm{Ge}GERDA0.0004.8670.0001.8001.8[?]
76Ge^{76}\mathrm{Ge}MAJORANA0.0000.7310.0000.2700.27[?]
130Te^{130}\mathrm{Te}CUORE0.245−0.6370.4140.2160.22[?]
136Xe^{136}\mathrm{Xe}Xe (KamLAND-Zen + EXO)5.5973.6400.0002.260——
76Ge^{76}\mathrm{Ge}Ge (GERDA + MAJORANA)0.0005.5980.0002.070——
130Te^{130}\mathrm{Te}Te (CUORE data as above)0.245−0.6370.4140.2160.22[?]
130Te^{130}\mathrm{Te}Te* (CUORE*, sensitivity)0.2450.0000.0000.3010.28[?]

Table I. Coefficients of the quadratic parametrization of Δχi2\Delta\chi_i^2 in terms of the signal strength Si=1/TiS_i = 1/T_i. The first two columns report the isotope and the names of the experiments or their combination. The next three columns report our evaluation of the coefficients (aia_i, bib_i, cic_i) for the various experiments (upper five rows) and for their combinations in the same isotope (lower rows). The bottom row refers to the case of CUORE sensitivity for null result (tagged by *). The sixth column reports our 90% C.L. (Δχ2=2.706\Delta\chi^2 = 2.706) half-life limits T90T_{90}, to be compared with the experimentally quoted one in the seventh column (as taken from the reference in the eighth column, when applicable).

Figure 2. Δχ2\Delta\chi^2 functions in terms of the half-life TT (top abscissa) and of the signal strength S=1/TS=1/T (bottom abscissa). Left and right panels: separate experiments and their combinations for the same isotope, respectively. Dotted horizontal lines intersect the curves at 90% C.L. See the text for details.

Results for the Te isotope depend on the single CUORE experiment, which shows a positive fluctuation at the level of 0.64σ=[Δχ2(0)]1/20.64\sigma=[\Delta\chi^2(0)]^{1/2}. As anticipated we consider, besides the real Te results, also hypothetical Te* results, where this fluctuation is canceled by setting bi=0b_i=0 (and thus also ci=0c_i=0). The half-time limit for Te* (T90=0.301T_{90}=0.301) is in reasonable agreement with the median sensitivity quoted by the CUORE experiment for null result (T90=0.28T_{90}=0.28). In § 3, the combination of Xe, Ge, and Te* results (all with Δχi2=0\Delta\chi_i^2=0 at Si=0S_i=0) will provide a simple starting point, before discussing the full combination of Xe, Ge and Te constraints on mββm_{\beta\beta}.

Landscape of nuclear matrix elements

In order to study the combination of 0νββ0\nu\beta\beta results in full generality, we consider unconstrained values of the nuclear matrix elements MXeM_{\mathrm{Xe}}, MGeM_{\mathrm{Ge}}, MTeM_{\mathrm{Te}} in the numerical range Mi∈[0.2,20]M_i\in[0.2,20]. Within this landscape, we also consider representative MiM_i values from four different approaches to nuclear modeling, including the nuclear shell model (SM), the quasi-particle random phase approximation (QRPA), the energy-density functional theory (EDF), and the interacting boson model (IBM). The NME values are taken from a recent compilation of results [25, 26] [15, 16, 17, 18, 19, 20, 21, 22, 23, 24] as reported in [1] for the bare value of gAg_A (see Table I therein), and are listed in Table II for the sake of completeness.

Figure 3 shows the NME landscape in each of the three planes charted by pairs (Mi,Mj)(M_i,M_j), together with the representative MiM_i values reported in Table II, which refer to the bare gAg_A. The issue of the effective gAg_A value to be used in nuclear matter, either bare or quenched by a factor qq (gA→qgAg_A\to qg_A with q<1q<1), is largely debated and model-dependent [1, 13]. For NMEs dominated by the axial-vector (Gamow-Teller) component (as it is often the case), the leading quenching effect would amount to rescaling the product GiMi2G_iM_i^2 by a factor q4∼gA4q^4\sim g_A^4, that can be assumed to operate on MiM_i only (Mi→q2MiM_i\to q^2M_i) if GiG_i is kept constant.

Figure 3. Landscape of nuclear matrix elements Mi=(MXe,MGe,MTe)M_i = (M_{\mathrm{Xe}}, M_{\mathrm{Ge}}, M_{\mathrm{Te}}), in each of the three planes charted by pairs (Mi,Mj)(M_i, M_j). Also shown are the representative MiM_i values reported in Table II from different models (SM, QRPA, EDF, IBM). The arrows show the effect of rescaling each MiM_i as Mi/gA2M_i/g_A^2 (representing typical quenching effects). See the text for details.

As a representative quenching effect one may consider, e.g., the typical case qgA≃1qg_A\simeq1, namely, q∼1/gAq\sim1/g_A, leading to an approximate rescaling Mi→Mi/gA2M_i\to M_i/g_A^2, as shown in each panel of Figure 3 by an arrow (applicable to any marked point). Stronger quenching would be associated to longer arrows (not shown). On the other hand, quenching effects would be weaker for sizeable NME vector (Fermi) components, not scaling with q2q^2. Moreover, some MiM_i calculations may exhibit a milder dependence on qgAqg_A for different reasons. In some QRPA calculations, e.g., 2νββ2\nu\beta\beta data are used to adjust the particle-particle parameter gppg_{pp}, partly trading the effect of quenching gAg_A from its bare value to unity [27]. In the same approach, large statistical covariances are observed among the MiM_i values for different isotopes, inducing noticeable effects on mββm_{\beta\beta} constraints [27], as recently discussed in [3]. The marked points in Figure 3 also seem to suggest an overall positive correlation (possibly enhanced by quenching effects) but, since they do not represent a statistical distribution, their covariances (if any) will be ignored.

TABLE II: Representative nuclear matrix elements (MXeM_{\mathrm{Xe}}, MGeM_{\mathrm{Ge}}, MTeM_{\mathrm{Te}}) computed within four different models (SM, QRPA, EDF, and IMB) for bare value of gAg_A. Adapted from [1].

MXeM_{\mathrm{Xe}}MGeM_{\mathrm{Ge}}MTeM_{\mathrm{Te}}ReferenceModel
12.282.892.76[15]
22.453.072.96[15]
31.633.371.79[16]SM
41.763.571.93[16]
52.392.663.16[17]
61.555.091.37[18]
72.915.264.00[19]
82.724.854.67[20]QRPA
91.113.122.90[21]
101.183.403.22[21]
114.204.605.13[22]
124.775.556.41[23]EDF
134.246.044.89[24]
143.255.143.96[25]
153.406.344.15[26]IBM

Table 2.

Summarizing, the NME landscape in Figure 3 is meant to cover a wide and continuous range of MiM_i values, either unquenched or arbitrarily quenched. Graphical results will be shown for unconstrained MiM_i values in this landscape. Marked points in Figure 3 are meant to represent typical unquenched MiM_i (central values) as taken from the literature (see Table II), while the arrows provide visual guidance for typical quenching effects (qgA∼1q g_A \sim1).

Phase space

The last ingredient is represented by the phase space GiG_i for 0νββ0\nu\beta\beta decay in i=Xe,Gei = \mathrm{Xe}, \mathrm{Ge} and Te\mathrm{Te}, that we take from the calculation in [26]. In our notation and units:

GXe=14.78×10−6 ,(13)G_{\mathrm{Xe}} = 14.78 \times10^{-6}\,, \tag*{(13)}
GGe=2.40×10−6 ,(14)G_{\mathrm{Ge}} = 2.40 \times10^{-6}\,, \tag*{(14)}
GTe=14.42×10−6 .(15)G_{\mathrm{Te}} = 14.42 \times10^{-6}\,. \tag*{(15)}

The phase space uncertainties [28] are much smaller than those related to 0νββ0\nu\beta\beta data and are not considered herein.

Constraints on the Majorana neutrino mass

The previously discussed functions Δχi2(Si)\Delta\chi_i^2(S_i) can be recast in terms of quadratic functions of mββ2m_{\beta\beta}^2 through Eq. (6):

Δχi2=αimββ4+βimββ2+γi ,(16)\Delta\chi_i^2 = \alpha_i m_{\beta\beta}^4 + \beta_i m_{\beta\beta}^2 + \gamma_i\,, \tag*{(16)}

where the offset γi\gamma_i is set by

γi={0for βi≥0 ,βi2/4αifor βi<0 .(17)\gamma_i = \begin{cases} 0 & \text{for } \beta_i \ge0\,, \\ \beta_i^2/4\alpha_i & \text{for } \beta_i < 0\,. \end{cases} \tag*{(17)}

The best-fit value of mββm_{\beta\beta} is set by:

Δχi2=0→mββ={0for βi≥0 ,(−βi/2αi)1/2for βi<0 .(18)\Delta\chi_i^2 = 0 \to m_{\beta\beta} = \begin{cases} 0 & \text{for } \beta_i \ge0\,, \\ (-\beta_i/2\alpha_i)^{1/2} & \text{for } \beta_i < 0\,. \end{cases} \tag*{(18)}

Table III reports the parametric coefficients (αi,βi,γi)(\alpha_i, \beta_i, \gamma_i) for the Xe\mathrm{Xe}, Ge\mathrm{Ge}, Te\mathrm{Te} and Te∗\mathrm{Te}^* cases, with their explicit dependence on the matrix elements MXeM_{\mathrm{Xe}}, MGeM_{\mathrm{Ge}}, MTeM_{\mathrm{Te}}.

Caseαi\alpha_iβi\beta_iγi\gamma_i
Xe1.223×10−9MXe41.223 \times10^{-9}M_{\mathrm{Xe}}^45.380×10−5MXe25.380 \times10^{-5}M_{\mathrm{Xe}}^20
Ge01.344×10−5MGe21.344 \times10^{-5}M_{\mathrm{Ge}}^20
Te5.094×10−11MTe45.094 \times10^{-11}M_{\mathrm{Te}}^4−9.186×10−6MTe2-9.186 \times10^{-6}M_{\mathrm{Te}}^20.414
Te*5.094×10−11MTe45.094 \times10^{-11}M_{\mathrm{Te}}^400

Table III. Coefficients of the quadratic parametrization Δχi2=αimββ4+βimββ2+γi\Delta\chi_i^2 = \alpha_i m_{\beta\beta}^4 + \beta_i m_{\beta\beta}^2 + \gamma_i for the cases Xe, Ge, Te and Te*.

Constraints on mββm_{\beta\beta} from two or more isotopes are obtained by summing the corresponding Δχi2\Delta\chi_i^2, and by adjusting the offset so that it obeys Eq. (17), namely: Δχ2=αmββ4+βmββ2+γ\Delta\chi^2 = \alpha m_{\beta\beta}^4 + \beta m_{\beta\beta}^2 + \gamma, where α=∑iαi\alpha= \sum_i \alpha_i, β=∑iβi\beta= \sum_i \beta_i, and γ=0\gamma= 0 for β≥0\beta\ge0 (γ=β2/4α\gamma= \beta^2/4\alpha otherwise). Bounds on mββm_{\beta\beta} at a given confidence level are obtained by solving

Δχ2(mββ)=ΔCL ,(19)\Delta\chi^2(m_{\beta\beta}) = \Delta_{\mathrm{CL}}\,, \tag*{(19)}

where, e.g., ΔCL=2.706\Delta_{\mathrm{CL}} = 2.706, 4 and 9 for limits at 90% C.L., 2σ2\sigma and 3σ3\sigma, respectively.

Two qualitatively different cases arise from current results: (a) for Xe\mathrm{Xe}, Ge\mathrm{Ge} and Te∗\mathrm{Te}^*, either separately or in combination, there is no offset γ\gamma and the Δχ2\Delta\chi^2 function is zeroed at mββ=0m_{\beta\beta} = 0; (b) for Te\mathrm{Te} results, characterized by γi>0\gamma_i > 0 (positive fluctuation), the combination with Xe\mathrm{Xe} or Ge\mathrm{Ge} (or both) may lead to γ>0\gamma> 0, implying a nonzero Majorana neutrino mass at best fit: mββ=[−β/2α]1/2m_{\beta\beta} = [-\beta/2\alpha]^{1/2}. We discuss separately these two cases below.

Combination of Xe, Ge, Te* constraints

In this section we consider current constraints from Xe\mathrm{Xe}, Ge\mathrm{Ge}, and from the pseudo-experiment Te∗\mathrm{Te}^* (corresponding to null signal in CUORE). In this case the analysis is straightforward, since the best fit is mββ=0m_{\beta\beta} = 0 for all isotopes and their combinations (independently of the NME). The analysis including real Te\mathrm{Te} data will bring forward cases with mββ>0m_{\beta\beta} > 0 at best fit, as discussed in the next Section.

123456789101112131415
Xe86.980.9121.6112.682.9127.968.172.9178.6168.047.241.646.761.058.3
Ge188.8177.7161.9152.8205.1107.2103.7112.5174.9160.5118.698.390.3106.286.1
Te*191.8178.8295.7274.3167.5386.4132.3113.4182.5164.4103.282.6108.3133.7127.6
Xe+Ge81.075.5101.995.078.586.859.363.7132.0122.744.839.242.854.850.4
Xe+Te*85.779.8120.4111.681.3127.466.669.2147.2135.346.540.746.260.157.5
Ge+Te*149.7140.2155.6146.5142.2106.991.388.8140.7127.885.869.577.592.979.4
Xe+Ge+Te*80.174.7101.394.477.286.758.461.5120.4110.744.238.542.454.249.9

Table 4. Bounds on mββm_{\beta\beta}/meV at 2σ2\sigma level from Xe, Ge and Te* results, both separately and in combination, for each of the 15 representative NME calculations listed in Tab. II. The best-fit value is mββ=0m_{\beta\beta}=0 in all cases.

We consider both separate and combined Xe, Ge and Te* constraints, as obtained by summing up the corresponding functions defined in tbl:3, Δχ2=∑iΔχi2=αmββ4+βmββ2\Delta\chi^2=\sum_i\Delta\chi_i^2=\alpha m_{\beta\beta}^4+\beta m_{\beta\beta}^2. For a given choice of MiM_i, upper limits on mββm_{\beta\beta} are obtained by solving ΔCL=αmββ4+βmββ2\Delta_{\mathrm{CL}}=\alpha m_{\beta\beta}^4+\beta m_{\beta\beta}^2. For definiteness we set ΔCL=4\Delta_{\mathrm{CL}}=4 (2σ2\sigma bounds), unless otherwise specified.

Table IV reports the upper bounds on mββm_{\beta\beta}, for each of the representative (MXe,MGe,MTe)(M_{\mathrm{Xe}}, M_{\mathrm{Ge}}, M_{\mathrm{Te}}) calculations listed in Table II. Concerning constraints from single isotopes, in most cases Xe sets the strongest bound, followed by weaker ones from Ge and Te*. However, for the cases numbered as 9 and 10 (QRPA), the bounds from Xe, Ge and Te* are comparable to each other, and for case 6 (QRPA) the Ge bound actually prevails over the Xe (and Te*) bound. Notice that such a hierarchy of mββm_{\beta\beta} constraints may change at different confidence levels, since the SiS_i bounds scale up at different rates (see Figure 2). In Table IV the combination of pairs of constraints improves appreciably upon each separate constraint; the relative balance in each pair is highlighted below. Finally, the total combination Xe+Ge+Te* provides even stronger bounds on mββm_{\beta\beta}, that range from a minimum of 38.5 meV (case 12, EDF) to a maximum of 120.4 meV (case 9, QRPA) at 2σ2\sigma.

Figure 4 shows isolines of the 2σ2\sigma bounds on mββm_{\beta\beta}, as derived by combining any two pairs among Xe, Ge and Te*, for unconstrained values of the NME. In each panel, single-isotope bounds are asymptotically recovered along each axis,

Image not extracted

Figure 4. Isolines of mββm_{\beta\beta} bounds (at 2σ2\sigma level) in the landscape of nuclear matrix elements, obtained from the combination of any two results among Xe, Ge and Te*. In each panel, the bounds are dominated by the isotope on the yy (xx) axis, in the region above (below) the dashed line. See the text for details.

for vanishing matrix element on the other axis. When both matrix elements are sizeable, the joint bound improves upon separate ones. In particular, at each marked point, the bounds in Table IV are recovered for the corresponding NME and pair of isotopes.

For each (Mx,My)(M_x, M_y) panel and (x,y)(x, y) isotope pair, the condition for the dominance of one isotope constraint over the other is easily derived. The two isotopes contribute equally to Δχx2=ΔCL/2=Δχy2\Delta\chi_x^2 = \Delta_{\mathrm{CL}}/2 = \Delta\chi_y^2. The solutions to these equations read Mxmββ=ξxM_xm_{\beta\beta} = \xi_x and Mymββ=ξyM_ym_{\beta\beta} = \xi_y, where ξx,y\xi_{x,y} are positive numbers. For ΔCL=4\Delta_{\mathrm{CL}} = 4 it is MTe/MGe=1.154M_{\mathrm{Te}}/M_{\mathrm{Ge}} = 1.154, MGe/MXe=2.488M_{\mathrm{Ge}}/M_{\mathrm{Xe}} = 2.488, and MTe/MXe=2.871M_{\mathrm{Te}}/M_{\mathrm{Xe}} = 2.871, shown as a dashed line in each panel of Figure 4. Along the dashed line, the two isotopes contribute with equal strength to the 2σ2\sigma upper bound; above the dashed line, the yy-axis isotope dominates over the xx-axis one, and viceversa. In this way one gets a graphical interpretation of the hierarchy of bounds for different NME, that was inferred from numerical inspection of Table IV.

Combination of Xe, Ge, Te constraints

In this section we consider real Te data, as opposed to the previous cases including Te* pseudo-data. The slight preference of Te data from CUORE for a nonzero signal (as compared with Xe, Ge and Te*, see Figure 2) brings forward new features of multi-isotope data constraints, although still at embryonic stages.

In general one may expect that, for relatively small values of MTeM_{\mathrm{Te}} (with respect to MXeM_{\mathrm{Xe}} and MGeM_{\mathrm{Ge}}), the Xe+Ge results will dominate over Te, keeping the best fit at mββ=0m_{\beta\beta} = 0. However, for increasing MTeM_{\mathrm{Te}}, the Te results will eventually prevail and set mββ>0m_{\beta\beta} > 0 at Δχ2=0\Delta\chi^2 = 0, affecting also upper bounds at some value ΔCL\Delta_{\mathrm{CL}}.

This situation anticipates what could happen with future and more accurate 0νββ0\nu\beta\beta data: their combination may (or may not) be consistent with some indications for nonzero mββm_{\beta\beta}, depending on both the data and the NME values. A future preference for mββ>0m_{\beta\beta} > 0 might even lead to lower bounds on mββm_{\beta\beta}, either separately or in combination, depending in part on (un)favorable values of the NME. Eventually, precise multi-isotope data might even test specific NME’s by selecting allowed ratios Mx/MyM_x/M_y [29, 30], namely, slanted allowed stripes in the NME landscape of Figure 3.3

In our approach, the occurrence of mββ>0m_{\beta\beta} > 0 at best fit is simply signaled, for a single isotope, by a coefficient βi<0\beta_i < 0 in the Δχ2\Delta\chi^2 function (currently occurring only for Te) and, for any combination of multi-isotope data, by a negative coefficient β=∑iβi<0\beta= \sum_i\beta_i < 0. The best-fit value of mββm_{\beta\beta} is then mββ=(−β/2α)1/2m_{\beta\beta} = (-\beta/2\alpha)^{1/2} with α=∑iαi\alpha= \sum_i\alpha_i, and its specific value depends on the NME’s. For β<0\beta< 0, the offset must be taken as γ=β2/4α\gamma= \beta^2/4\alpha. In all cases, upper bounds at a chosen confidence levels are set by Δχ2=ΔCL\Delta\chi^2 = \Delta_{\mathrm{CL}}. Table V shows the numerical results from current Xe, Ge and Te data, regarding the 2σ2\sigma limits (lower part) and the best-fit values (lower half) of mββm_{\beta\beta}. In the upper half, the rows corresponding the Xe, Ge and Ge+Xe are unchanged with respect to Table IV, but are repeated for completeness.

123456789101112131415
Xe86.980.9121.6112.682.9127.968.172.9178.6168.047.241.646.761.058.3
Ge188.8177.7161.9152.8205.1107.2103.7112.5174.9160.5118.698.390.3106.286.1
Te220.5205.6340.0315.3192.6444.2152.1130.3209.9189.0118.694.9124.5153.7146.6
Xe+Ge81.075.5101.995.078.586.859.363.7132.0122.744.839.242.854.850.4
Xe+Te89.683.4124.9115.785.6130.470.475.0167.5154.548.642.948.162.960.1
Ge+Te174.1163.2170.1160.5165.6109.6104.4103.3163.4148.6100.381.389.2106.589.2
Xe+Ge+Te83.577.9104.597.481.288.061.466.2135.0124.746.240.544.156.551.9
Xe000000000000000
Ge000000000000000
Te108.8101.4167.7155.695.0219.275.164.3103.593.258.546.861.475.872.4
Xe+Ge00000000000000
Xe+Te0000000031.736.000000
Ge+Te00000000000000
Xe+Ge+Te00000000000000

Table V. TABLE V: Upper half: Bounds on mββ/meVm_{\beta\beta}/\mathrm{meV} at 2σ2\sigma level from Xe, Ge and Te results, both separately and in combination, for each of the 15 representative NME calculations listed in Tab. II. Lower half: Corresponding best-fit values of mββm_{\beta\beta}.

Let us first comment on the mββm_{\beta\beta} best fits in the lower half of Table V. Of course, the Te row displays nonzero results, with mββm_{\beta\beta} scaling as 1/MTe1/M_{\mathrm{Te}}. In almost all cases including Te (combined with Xe or Ge or both), the positive contributions to β=∑iβi\beta= \sum_i\beta_i from i=i = Xe and Ge are never erased by the negative contribution from Te, and mββ=0m_{\beta\beta} = 0 is preferred. Only for cases 9 and 10 (QRPA), it turns out that the large ratio MTe≃2.6MXeM_{\mathrm{Te}} \simeq2.6M_{\mathrm{Xe}} makes Te prevail over Xe in the corresponding Xe+Te combinations, which show nonzero best fits.

Concerning the upper half of Table V, the combination of Te with Xe (or Ge) does not necessarily improve upon the separate 2σ2\sigma bounds. Roughly speaking, when the best-fit value of mββm_{\beta\beta} in Te is comparable or larger than the upper bound from Xe (Ge) alone, the joint bound from Xe+Te (Ge+Te) is weakened, as a result of the slight tension between the two isotopic data. A slight weakening also occurs whenever when Te is added to Xe+Ge in the global combination Xe+Ge+Te.4 The 2σ2\sigma bounds on mββm_{\beta\beta} compiled in Table V are contained in the following range:

mββ∈[40.5,135.0] meV (Xe+Ge+Te).(20)m_{\beta\beta} \in[40.5, 135.0]\ \mathrm{meV}\ (\mathrm{Xe}+\mathrm{Ge}+\mathrm{Te}) . \tag*{(20)}

The lowest (most optimistic) edge of this range would significantly cut from above the IO and NO allowed regions in Figure 1, setting also an upper limit on Σ\Sigma at the level of ∼400 meV\sim400\ \mathrm{meV}.

Figure 5 shows isolines of the best-fit value of mββm_{\beta\beta} in the left and right panels charted by (Xe, Te) and (Ge, Te), respectively. In the left panel, the condition ∑iβi>0\sum_i \beta_i>0 for mββ>0m_{\beta\beta}>0 implies MTe/MXe>2.42M_{\mathrm{Te}}/M_{\mathrm{Xe}}>2.42, satisfied only by two marked points (corresponding to the QRPA cases 9 and 10 in Table II) close to the isoline at mββ=30 meVm_{\beta\beta}=30\ \mathrm{meV}. In the right panel, none of the marked points falls in the analogous region MTe/MGe>1.21M_{\mathrm{Te}}/M_{\mathrm{Ge}}>1.21, although three of them are very close to its border. If future 0νββ0\nu\beta\beta experiments will show some indications for a signal, plots like these will help to

Figure 5. Isolines of best-fit values of mββ/meVm_{\beta\beta}/\mathrm{meV} in the planes charted by (Xe, Te) and (Ge, Te) nuclear matrix elements. The best fit is zero in the lower-right parts of each panel, as well as in the whole (Xe, Ge) panel (not shown).

90%2σ2\sigma3σ3\sigmaModel
172.783.5109.4
267.877.9101.9
389.3104.5141.5SM
483.397.4131.7
571.081.2105.4
673.688.0125.3
753.061.481.6
857.666.286.7QRPA
9117.7135.0176.3
10108.9124.7162.3
1140.346.260.1
1235.440.552.6EDF
1338.244.158.1
1448.856.574.7
1544.651.969.5IBM

Table 6. Upper bounds on mββm_{\beta\beta}/meV at 90% C.L., 2σ2\sigma, and 3σ3\sigma, from the combination of current Xe+Ge+Te data, for the representative NME calculations considered in this work.

locate the best-fit values of mββm_{\beta\beta} as a function of the NME, for each isotope pair. The (in)consistency of the best fits in different pairs will provide interesting clues about the interpretation of data in terms of light Majorana neutrinos.

Figure 6 shows isolines of the mββm_{\beta\beta} upper bounds at 2σ2\sigma, in the same planes of Figure 5. In comparison with the lower panels of Figure 4, a slight weakening of the bounds can be appreciated. Note that, in the presence of subregions where mββ>0m_{\beta\beta} > 0 at best fit, the offset γ\gamma depends on information coming from both isotopes, whose χ2\chi^2 contributions cannot be separated in the combination. The condition of equal contributions to ΔCL\Delta_{\mathrm{CL}} cannot be defined in general terms, and the dashed lines of Figure 4 are thus absent in Figure 6.

Figure 6. Isolines of mββm_{\beta\beta} bounds (at 2σ2\sigma level) in the planes charted by (Xe, Te) and (Ge, Te) nuclear matrix elements. The (Xe, Ge) panel (not shown) is unchanged with respect to Figure 4.

We conclude this Section by discussing the constraints on mββm_{\beta\beta} at various confidence levels, as derived from the global combination of current (Xe+Ge+Te) data, using the representative NME values in Table II. Since the best fit is mββ=0m_{\beta\beta} = 0 in all Xe+Ge+Te cases (see Table V), only upper bounds need to be quoted.

Table VI reports the 90% C.L., 2σ2\sigma, and 3σ3\sigma upper limits on mββm_{\beta\beta} (in meV). Figure 7 shows the Nσ=(Δχ2)1/2N_\sigma= (\Delta\chi^2)^{1/2} bounds as continuous functions of mββm_{\beta\beta}. Qualitatively, the strongest limits in are obtained using NMEs from the EDF and IBM models, followed by QRPA and SM cases in mixed order. These results can be generalized to any other choice of NME calculations, using the information provided in this paper.

Image not extracted

Figure 7. Significance of upper limits on mββm_{\beta\beta} in terms of Nσ=(Δχ2)1/2N_\sigma= (\Delta\chi^2)^{1/2}, from the combination of current Xe+Ge+Te data, for the representative NME calculations considered in this work.

In Figure 7, one of the NσN_\sigma curves labelled as QRPA shows a a markedly different (almost linear) slope, intersecting three SM curves. This peculiar curve corresponds to the lowest QRPA point in both panels of Figure 6 (case 6 in Table II), that is characterized by a rather large value of MGeM_{\rm Ge} as compared with MXe,TeM_{\rm Xe,Te}. As a result, Ge data prevail over Xe and Te in the combination, and the leading dependence is Δχ2∝S∝mββ2\Delta\chi^2 \propto S \propto m_{\beta\beta}^2 (rather than Δχ2∝S2∝mββ4\Delta\chi^2 \propto S^2 \propto m_{\beta\beta}^4), implying a roughly linear Nσ(mββ)N_\sigma(m_{\beta\beta}) function. Once more, this observation shows the importance of considering the full likelihood profile of the experimental results (e.g., in terms of S=1/TS = 1/T), rather than pointlike information (such as the 90% C.L. limit on TT).

As a final step, one could include a joint probability distribution or a Δχ2\Delta\chi^2 penalty defined over the NME landscape (MXeM_{\rm Xe}, MGeM_{\rm Ge}, MTeM_{\rm Te}), and numerically minimize the total Δχ2\Delta\chi^2 function. This exercise was performed in [3] by assuming a conservative characterization of the NME and their correlated uncertainties, derived within QRPA calculations [27]. A limit mββ<110m_{\beta\beta} < 110 meV was obtained at 2σ2\sigma [3]. By repeating the same exercise with the updated (Xe+Ge+Te) combination considered herein, we get the following marginalized bounds (in meV): mββ<79.5m_{\beta\beta} < 79.5 at 90% C.L., mββ<99.8m_{\beta\beta} < 99.8 at 2σ2\sigma, and mββ<169m_{\beta\beta} < 169 at 3σ3\sigma. Roughly speaking, from these results and from the summary in Eq. (20) one can state that the combination of current 0νββ0\nu\beta\beta experiments sets 2σ2\sigma upper bounds on mββm_{\beta\beta} at the level of ∼90\sim90–100 meV for “average” NME values, possibly lowered to ∼40\sim40–50 meV for favorable NME values.

A final remark is in order. In principle, one should replace the QRPA input from [27] with more general and up-to-date estimates of the NME’s and their uncertainties, characterizing also the spread among different models and calculations. However, no consensus estimates exist yet for NME fiducial values and covariances, although relevant work is in progress toward this goal [31] [1]. Part of the planned strategy involves benchmarking nuclear models for 0νββ0\nu\beta\beta decay against a variety of data, coming from related electroweak and strong interaction processes or from nuclear structure [32, 33, 34, 35].

Conclusions and perspectives

We have discussed an approach to the analysis of neutrinoless double beta decay experiments, in terms of Δχ2\Delta\chi^2 profiles for the signal strength SiS_i (inverse of the half-life TiT_i) in the isotopes i=Xe,Gei = {\rm Xe, Ge} and Te, building upon previous work [3]. The approach becomes exceedingly simple for quadratic approximations to such profiles, implying quadratic (in)equalities in the landscape of nuclear matrix elements MiM_i that connect the SiS_i to the Majorana mass mββm_{\beta\beta}. For convenience, some results have been discussed in terms of pseudo data for null signal in Te (dubbed Te*). Simple relations among the MiM_i have been derived to gauge the relative contributions of different isotope data in setting upper limits to mββm_{\beta\beta} (for null best fits in Xe, Ge and Te*), and to identify the conditions leading to a preference for nonzero mββm_{\beta\beta} (for generic Xe, Ge and Te data). Using the latest available 0νββ0\nu\beta\beta data, as well as representative values of the NME from different models, we have discussed current constraints on mββm_{\beta\beta} at several confidence levels and in various combinations, both numerically and graphically. Global 2σ2\sigma upper limits on mββm_{\beta\beta} are found in the range from 40.5 to 135 meV, depending on the NME.

The approach can be easily extended to nonstandard processes for 0νββ0\nu\beta\beta decay [36, 37] [1] by replacing the relation Si=GiMi2mββ2S_i = G_i M_i^2 m_{\beta\beta}^2 with the appropriate phase space, NME and particle physics parameter characterizing the process. Also, the approach can be extended to generic Δχ2(Si)\Delta\chi^2(S_i) functions, with a modest price to pay in terms of numerical (rather than analytical) solutions. We invite the experimental collaborations involved in 0νββ0\nu\beta\beta decay searches to publicly provide such Δχ2(Si)\Delta\chi^2(S_i) functions or equivalent ones, as they contain much more information than the usually quoted 90% C.L. limits on TiT_i. Indeed, the relative impact of such limits and of the resulting bounds on mββm_{\beta\beta} in a multi-isotope combination depend sensitively on the likelihood profiles of SiS_i (or TiT_i), and not only on the relative size of the MiM_i.

Our approach to the multi-isotope data analysis would be complete if one could also assign joint probability densities to the MiM_i, whose variations could then be treated as nuisance parameters and marginalized. So far, detailed results for the NME central values and covariances, including gAg_A quenching uncertainties, have been obtained in a specific (QRPA) model [27]. In perspective, it would be important to extend such investigations to other nuclear models, eventually reaching consensus values for the MiM_i and for their correlated (and possibly reduced) uncertainties.

In this sense, the combined analysis of 0νββ0\nu\beta\beta results is proceeding through to the same steps that have characterized similar fields (e.g., solar neutrinos) in the past: from low-statistics data and theoretical models with large uncertainties, to a wealth of accurate experimental results interpreted in increasingly refined and constrained models. Our work aims at providing one methodological step along this path.

Acknowledgments

This work is partly supported by 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 thank K. Inoue for useful information about the latest KamLAND-Zen data release [14].

References

References

  1. [25]J. Barea, J. Kotila and F. Iachello, “0νββ and 2νββ nuclear matrix elements in the interacting boson model with isospin restoration,” Phys. Rev. C 91, no.3, 034304 (2015) [arXiv:1506.08530 [nucl-th]].DOI
  2. [26]F. F. Deppisch, L. Graf, F. Iachello and J. Kotila, “Analysis of light neutrino exchange and short-range mechanisms in 0νββ decay,” Phys. Rev. D 102, no.9, 095016 (2020) [arXiv:2009.10119 [hep-ph]].DOI
  3. [27]A. Faessler, G. L. Fogli, E. Lisi, V. Rodin, A. M. Rotunno and F. Simkovic, “QRPA uncertainties and their correlations in the analysis of 0νββ decay,” Phys. Rev. D 79, 053001 (2009)DOI
  4. [28]S. Stoica and M. Mirea, “Phase Space Factors for Double-Beta Decays,” Front. in Phys. 7, 12 (2019)DOI
  5. [29]S. M. Bilenky and J. A. Grifols, “The Possible test of the calculations of nuclear matrix elements of the (ββ)₀ν decay,” Phys. Lett. B 550, 154-159 (2002) [arXiv:hep-ph/0211101 [hep-ph]].DOI
  6. [30]S. M. Bilenky and S. T. Petcov, “Nuclear matrix elements of 0νββ decay: Possible test of the calculations,” [arXiv:hep-ph/0405237 [hep-ph]].DOI
  7. [31]J. Engel and J. Menéndez, “Status and Future of Nuclear Matrix Elements for Neutrinoless Double-Beta Decay: A Review,” Rept. Prog. Phys. 80, no.4, 046301 (2017) [arXiv:1610.06548 [nucl-th]].DOI
  8. [32]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
  9. [33]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
  10. [34]M. Horoi, A. Neacsu and S. Stoica, “A Statistical Analysis for the Neutrinoless Double-Beta Decay Matrix element of ⁴⁸Ca,” [arXiv:2203.10577 [nucl-th]].DOI
  11. [35]H. Ejiri, L. Jokiniemi and J. Suhonen, “Nuclear matrix elements for neutrinoless ββ decays and spin-dipole giant resonances,” Phys. Rev. C 105, no.2, L022501 (2022) [arXiv:2202.00361 [nucl-th]].DOI
  12. [36]M. J. Dolinski, A. W. P. Poon and W. Rodejohann, “Neutrinoless Double-Beta Decay: Status and Prospects,” Ann. Rev. Nucl. Part. Sci. 69, 219-251 (2019) [arXiv:1902.04097 [nucl-ex]].DOI
  13. [37]W. Rodejohann, “Neutrino-less Double Beta Decay and Particle Physics,” Int. J. Mod. Phys. E 20, 1833-1930 (2011) [arXiv:1106.1334 [hep-ph]].DOI

Paper details

vlnc:W_y7ucr22a
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