betaFeedback

WorksPhysics and Astronomy

Many-Body Theory Calculations of Positron Scattering and Annihilation in H2_2, N2_2 and CH4_4

Abstract

The recently developed ab initio many-body theory of positron molecule binding [J. Hofierka et al., Nature, 606, 688 (2022)] is combined with the shifted pseudostates method [A. R. Swann and G. F. Gribakin, Phys. Rev. A 101, 022702 (2020)] to calculate positron scattering and annihilation rates on small molecules, namely H2_2, N2_2 and CH4_4. The important effects of positron-molecule correlations are delineated. The method provides uniformly good results for annihilation rates on all the targets, from the simplest (H2_2, for which only a sole previous calculation agrees with experiment), to larger targets, where high-quality calculations have not been available.

Developing fundamental knowledge of positron scattering and annihilation in molecules is essential to e.g., realize antimatter-based molecular spectroscopy [1, 2, 3] and next-generation antimatter traps [4, 5, 6, 7]; elucidate the process of molecular fragmentation [8, 9, 10, 11]; and properly understand how positrons propagate in and can act as probes of living tissue (relating to DNA damage and dosimetry in PET [12, 13, 14, 15] [16, 17]), the galaxy (e.g., to understand the galactic-centre annihilation signal [18, 19] and dark matter [20]), and materials [21].

The positron-molecule system is, however, characterized by strong positron-molecule correlations that are non-local and act over different length scales [22], and for molecules that bind the positron, spectacular resonance effects due to coupled electronic and vibrational dynamics [1]. They make the theoretical and computational description a challenging many-body problem. For positron scattering, RR-matrix [23–27], Schwinger multichannel [28–35], Kohn variational [36–40], model-potential [41, 42], and CCC (convergent close coupling) [43, 44] methods have been applied with considerable success to small molecules including H2_2, CH4_4, N2_2, CO2_2, CO, allene, formamide and pyrazine (see also [45]). Calculation of the positron-molecule annihilation rate — of chief interest in this work — is however, strikingly more difficult. For a gas of number density ngn_g the positron annihilation rate is parameterized as λ=πr02cngZeff\lambda= \pi r_0^2 c n_g Z_{\rm eff}, where r0r_0 is the classical electron radius, cc is the speed of light, and ZeffZ_{\rm eff} is the effective number of electrons that participate in the annihilation process. Formally ZeffZ_{\rm eff} is equal to the electron density at the positron,

Zeff=∫∑i=1Neδ(r−ri)∣Ψk(r1,…,rNe,r)∣2 d3r1…d3rNe dr,Z_{\rm eff} = \int\sum_{i=1}^{N_e} \delta\left(\mathbf{r}-\mathbf{r}_i\right) \lvert\Psi_k(\mathbf{r}_1,\ldots,\mathbf{r}_{N_e},\mathbf{r})\rvert^2 \,\mathrm{d}^3r_1\ldots\mathrm{d}^3r_{N_e}\,\mathrm{d}r,

where Ψk\Psi_k is the total wavefunction of the system, with electron coordinates ri\mathbf{r}_i and positron coordinate r\mathbf{r} [46]. It describes the scattering of positron of momentum k\mathbf{k} by the molecule, and is normalised asymptotically to the product of the ground-state target molecular wavefunction and positron plane wave. Accurate calculation of ZeffZ_{\rm eff} thus requires proper account of the scattering dynamics and positron-molecule correlations, including short-range electron-positron interactions. Even for the simplest molecule, H2_2, calculations of ZeffZ_{\rm eff} via sophisticated methods including R-matrix [25] and the Kohn-variational [39, 47] and Schwinger multichannel methods [48] disagree, all substantially underestimating experiment [49–51] (by ∼15–50%), to which only a stochastic variational method calculation [52] is compatible.

For N2_2, used ubiquitously as a buffer-gas in positron traps [4, 5], the Schwinger multichannel method (the only ab initio calculation we are aware of) underestimates experiment by a factor >3>3. Moreover, these methods cannot be easily scaled to larger molecules. Theoretical developments are demanded.

Many-body theory is a powerful method that can accurately account for strong positron and electron correlations with atoms, molecules and ions (see e.g., [53–66]). For atoms, a B-spline implementation provided a complete ab initio description of positron scattering, annihilation and cooling [62, 63, 67–69] and positronium (Ps) ‘pickoff’ annihilation [70] in (noble-gas) atoms. Most recently, we developed and successfully applied a multicentred Bethe-Salpeter Gaussian-orbital based many-body approach to positron binding in molecules, implemented in our EXCITON+ code [22].

Here, we extend the approach beyond binding, combining it with the recently-devised shifted-pseudostate-normalization method of Swann and Gribakin [42] to perform fixed-nuclei [71] calculations of low-energy positron scattering and (direct) annihilation rates on the small molecules H2_2 (the most abundant molecule in space), N2_2 (a key buffer-gas in positron traps), and CH4_4 (abundant in planetary atmospheres). We quantify the effects of positron-molecule correlations including positron-induced polarization, screening, virtual-Ps and positron-hole interactions, and compare with experiment and theory where available. For the annihilation rates, we find near perfect agreement with the benchmark stochastic variational calculation [52, 72] and experiment for H2_2, providing a consensus, and find excellent agreement with experiment for N2_2 and CH4_4.

Theory and numerical implementation.—The positron (quasiparticle) wavefunction ψε\psi_\varepsilon in the field of a many-electron target is found from the Dyson equation [22, 73]:

(H(0)+Σ^ε)ψε(r)=εψε(r),(1)\left(H^{(0)}+\hat{\Sigma}_\varepsilon\right)\psi_\varepsilon(\mathbf{r})=\varepsilon\psi_\varepsilon(\mathbf{r}), \tag*{(1)}

Here H(0)H^{(0)} is the zeroth-order Hamiltonian, which we take to be that of the positron in the Hartree-Fock (HF) field of the ground-state molecule, and Σ^ε\hat{\Sigma}_\varepsilon is the non-local, energy-dependent correlation potential (self-energy) [74]. In practice we calculate the matrix elements of Σ\Sigma via its diagrammatic expansion in the residual electron-electron and electron-positron interactions [75]. See Ref. [22] for full details. Briefly, we include three classes of infinite series in the expansion: Figure 1 (a) the ‘GWGW’ diagram (the product of the positron Green’s function GG and the screened Coulomb interaction WW, which we calculate at the Bethe-Salpeter-Equation level), describes the positron-induced polarization of the molecular electron cloud, the screening of it by the molecular electrons, and electron-hole attractions; (b) the electron-positron ladder series (‘Γ\Gamma block’) that describes the non-perturbative virtual-Ps formation process and; (c), the positron-hole ladder series (‘Λ\Lambda block’).

Main contributions to the positron-molecule self energy

Figure 1. The main contributions to the positron-molecule self energy: (a), the GWGW diagram, which describes polarisation, and screening and electron-hole interaction corrections to it; (b) and (c), the infinite ladder series of screened electron-positron interactions (‘Γ\Gamma-block’) and positron-hole interactions (‘Λ\Lambda-block’). Lines labelled ν\nu (μ\mu) ((nn)) are excited positron (electron) ((hole)) propagators; a single (double) wavy line denotes a bare (dressed) Coulomb interaction. See Extended Data Fig. 1 of [22] for full details.

We expand the electron and positron states in Gaussian basis sets (see below), transforming Eqn. (1) into a linear matrix equation. For a target that has no bound states for the positron, its solution yields a set of nn discrete positron continuum pseudostates and their corresponding energies εn\varepsilon_n (n=1,2,…n = 1, 2, \ldots). These pseudostates decay exponentially rather than oscillate at large positron-target separations, and are normalized to unity instead of to an asymptotic plane wave as required by a true continuum state. Moreover, the lack of spherical symmetry of the multicentered target means that the orbital angular momentum is not conserved. However, at low positron momenta (kRa≪1kR_a \ll1, where RaR_a is the radius of the target), the mixing between partial waves due to the noncentral nature of the potential is small or negligible, and one can identify (approximately) states with eigenvalues of the squared orbital angular momentum operator L2L^2 close to zero (ss-states), which are expected to dominate the low-energy scattering and annihilation. In this case we can obtain the appropriate normalization following Swann and Gribakin [42], comparing the energies of (approximate) ss-states against corresponding free positron pseudostate energies εn(0)\varepsilon_n^{(0)} (found by setting H0H_0 equal to the positron kinetic energy). We thus calculate the ss-wave phase shift for a positron of energy εn\varepsilon_n as δ0=[n−f−1(εn)]π\delta_0 = [n - f^{-1}(\varepsilon_n)]\pi, where nn is the number of the ss-wave pseudostate, and f(n)f(n) is a function of a continuous variable nn satisfying f(n)=εn(0)f(n) = \varepsilon_n^{(0)} [42]. We use the same procedure for pp- and dd-waves. Moreover, we make use of the shifted energies to approximate the annihilation rate as Zeff=4πδepA−2Z_{\rm eff} = 4\pi\delta_{\rm ep} A^{-2}, with normalisation factor A2=22επ dε/dnA^2 = 2\sqrt{2\varepsilon}\pi\,d\varepsilon/dn [42] and δep=2∑i=1Ne/2γi∫∣φi(r)∣2∣ψ(r)∣2dτ\delta_{\rm ep} = 2\sum_{i=1}^{N_e/2}\gamma_i\int|\varphi_i(\mathbf{r})|^2|\psi(\mathbf{r})|^2d\tau is the annihilation contact density summed over all occupied electronic MOs φi\varphi_i, including vertex enhancement factors

γi=1+1.31/∣εi∣+(0.834/∣εi∣)2.15\gamma_i = 1 + \sqrt{1.31/|\varepsilon_i|} + (0.834/|\varepsilon_i|)^{2.15} for MO ii with energy εi<0\varepsilon_i < 0 that account for the effect of short-range electron-positron Coulomb attraction [63, 67].

We implement the above in our EXCITON+ Gaussian-basis code [22] using aug-cc-pVXZ (X=T or Q) basis sets on the atoms of the molecule and up to 20 ‘ghost’ centres away from the molecule to describe virtual-Ps formation, and a 19s17p16d15f even-tempered set on the molecular centre to help describe the long-range interactions; we assessed convergence and sensitivity to bond lengths (see Supplemental Material (SM) for full details) [76].

Results: effect of many-body correlations.—Figure 2 shows the phase shifts and normalized annihilation rate ZeffZ_{\rm eff} for H2_2 (representative of the three molecules considered) calculated at the Hartree-Fock (HF), Σ(2)\Sigma^{(2)}, GWGW, GW+ΓGW+\Gamma, and GW+Γ+ΛGW+\Gamma+\Lambda level for the correlation potential (see Figure 1). At the HF level the positron-molecule interaction is repulsive (corresponding to a negative phase shift, and small electron-positron overlap and thus annihilation rate); including the bare polarization Σ(2)\Sigma^{(2)} [92] produces an attractive interaction at low momenta (turning the phase shift positive and increasing the electron-positron overlap and thus ZeffZ_{\rm eff}), which is further enhanced by the inclusion of the dressed ring diagrams of the GWGW@BSE, i.e., the intra-ring BSE electron-hole attractions are larger than the repulsive screening effects from the random-phase approximation ring series. The additional inclusion of the virtual-Ps contribution (GW+ΓGW+\Gamma) further increases the attractive potential substantially, causing a factor of ∼3\sim3 increase in the phaseshift maximum and a more than doubling of ZeffZ_{\rm eff} at low momenta, but is tempered by the repulsive positron-hole (Λ\Lambda-block) contribution. The corresponding graphs for N2_2 and CH4_4 are presented in Figure S4 in the Supplemental Material. We now consider the scattering cross sections and annihilation rates, comparing with experiment and theory where available.

Phase shifts and normalized annihilation rates

Figure 2. The effects of positron-molecule many-body correlations shown via the calculated ss-wave scattering phase shift (left) and normalized annihilation rate ZeffZ_{\rm eff} (right) for H2_2 (representative of the three molecules considered in this work) in different approximations to the positron-molecule self energy (see Figure 1): Hartree Fock (black dotted); bare polarization Σ(2)\Sigma^{(2)} (black dot-dashed); GWGW (black dashed); GW+ΓGW+\Gamma (black dot-dot-dashed); and GW+Γ+ΛGW+\Gamma+\Lambda (solid line).

Partial wave contributions and sensitivity to the bond lengths for H₂, N₂, and CH₄

Figure S4. Partial wave contributions and sensitivity to the bond lengths. Calculated ss (red), pp (blue) and dd-wave (green) phase shifts and normalized annihilation rate ZeffZ_{\mathrm{eff}} for H₂, N₂, and CH₄. For H₂ calculations were performed using a bond length of R=1.40R = 1.40 (lines) and R=1.45R = 1.45 a.u. (symbols). For N₂ calculations were performed for bond lengths of R=2.014R = 2.014 a.u. (lines) and R=2.068R = 2.068 a.u. (symbols).

Cross sections.—Figure 3 shows the calculated elastic cross sections. For H2_2, for which the various calculations agree to ≲10%\lesssim10\% error, our calculation agrees best with (though is slightly larger than) the accurate convergent-close-coupling calculation [44], the model-potential calculation [42] (which underestimates the ZeffZ_{\mathrm{eff}}, see below) and measurements of Zecca et al. [77]. For N2_2 our result is compatible with the measurements of Zecca et al. [32]. For CH4_4 our results are consistent with the measurements of Zecca et al. [32] at ∼1\sim1 eV, but are considerably larger at low energies (where they are compatible with the model-potential calculation [42] that gives ZeffZ_{\mathrm{eff}} in agreement with experiment, see below). We also calculate the scattering length aa from fits of the effective range theory expansion to calculated phase shifts (see Supplementary Material Table I), finding them to be within ∼\sim5-20% error of previous calculations.

Figure 3. Many-body theory calculated scattering cross sections for H2_2, N2_2 and CH4_4: ss-wave (thin dashed black line), pp wave (thin dash-dotted black line), dd-wave (dotted) and total (thick solid black line). Results are shown for bond lengths of R=1.45R = 1.45, 2.014, and 2.06 a.u. for H2_2, N2_2 and (the C-H bond in) CH4_4. Also shown are measurements by Zecca et al. [32, 77, 78] (green triangles) and Charlton et al. [79] (orange triangles) for each molecule; the Schwinger multichannel [80] (orange line), Kohn variational [38] (blue diamonds), convergent-close-coupling [44] (red filled circles) calculations and modified-effective-range-theory fit of measured cross sections [81] (red squares) for H2_2; Schwinger multichannel [30] (orange line), local complex potential [82] (plus symbols) and correlation-polarization-model [83] (dashed red) calculations for N2_2; correlation-polarization-model-potential calculations of Franz [84] (orange dashed line), Jain and Gianturco [85] (blue dashed line), Swann and Gribakin [42] (blue solid line) and Dibyendu et al. [86] (magenta dashed line), Schwinger multichannel [32] (turquoise dashed), and measurements of Sueoka and Mori [87] (blue triangle) and Dababneh et al. [88] (red diamonds) for CH4_4.

Annihilation rate ZeffZ_{\mathrm{eff}}.—Of chief interest in this work is the annihilation rate ZeffZ_{\mathrm{eff}}, due to the challenge it poses for theory and lack of accurate methods.

theory and lack of accurate methods. Figure 4 shows our normalized annihilation rate Zeff(k)Z_{\mathrm{eff}}(k) as a function of positron momentum calculated in our most sophisticated approximation (GW+Γ+ΛGW+\Gamma+\Lambda self-energy and including vertex enhancement factors). We show the discrete data points calculated for the ss-wave, along with fits to the physically motivated form [62] Zeff(k)=F/(k2+k2+Ak4)+BZ_{\mathrm{eff}}(k)=F/(k^2+k^2+Ak^4)+B where FF, kk, AA and BB are constants [93]. We also show the calculated room-temperature Maxwellian average (open circle) Z‾eff=(2πkBT)−3/2∫0∞Zeff(k)exp⁡(−k2/2kBT)4πk2 dk\overline{Z}_{\mathrm{eff}}=(2\pi k_BT)^{-3/2}\int_0^\infty Z_{\mathrm{eff}}(k)\exp(-k^2/2k_BT)4\pi k^2\,dk. Table I gives the values of Z‾eff\overline{Z}_{\mathrm{eff}}: for H2_2 and N2_2 (CH4_4), we found it to be <1%<1\% (10%10\%) larger than Zeff(k)Z_{\mathrm{eff}}(k) at thermal k∼0.05k\sim0.05 a.u.

Figure 4. Many-body theory calculated annihilation rate Zeff(k)Z_{\mathrm{eff}}(k) (using the GW+Γ+ΛGW+\Gamma+\Lambda self-energy and including annihilation vertex enhancement factors) for H2_2, N2_2 and CH4_4: ss-wave (thin dashed black line), pp-wave (thin dashed-dotted), total s+p+ds+p+d (thick solid black line). Results are shown for bond lengths of R=1.4R = 1.4, 2.014, and 2.06 a.u. for H2_2, N2_2 and (the C-H bond in) CH4_4. Also shown are the room-temperature Maxwellian averaged Z‾eff\overline{Z}_{\mathrm{eff}} from our calculation (black open circle; for H2_2 black down triangle is for R=1.45R = 1.45 a.u. for comparison) and experiment (red triangles) for H2_2 [49, 50, 89], N2_2 [51], and CH4_4 [51], along with energy-resolved measurements for CH4_4 [90] (red diamonds). For H2_2 we also show the calculated room temperature values from the Kohn variational method (blue diamonds) [39, 47, 91], molecular R-matrix (magenta square) [25] and stochastic variational method at bond length of R=1.4R = 1.4 a.u. (green triangle down) [72], and the Schwinger multichannel method at k=0.05k = 0.05 a.u. (orange squares) [30, 48]. The latter is also shown for N2_2. For CH4_4 we also show the individual ss-wave contributions from the 2a12a_1 and one of the triply-degenerate t2t_2 HOMOs.

Considering comparison with other theory and experiment, for H2_2 a number of sophisticated calculations of ZeffZ_{\mathrm{eff}}, namely

H2_2N2_2CH4_4
Present MBTa^{a}14.9, 15.7b^{b}30.5c^{c}119.2d^{d}
SMCe^{e} [30, 48]7.708.96–
R-matrix [25]10.4––
Kohn var. [39, 47]12.6f^{f}––
SVM [94]14.6, 15.7b^{b}––
Corr. pol. [95, 96]–44±444 \pm499.5g^{g}
LCAOh^{h} [97]14.6––
Model-pot. [42]10.629.8163
Experiment14.7±0.214.7 \pm0.2 [89]30.8±0.230.8 \pm0.2 [51]140±0.8140 \pm0.8 [51]
16.0±0.216.0 \pm0.2 [49, 51]i^{i}
14.6±0.114.6 \pm0.1 [50]

Table 1. Maxwellian-averaged annihilation rate Z‾eff\overline{Z}_{\mathrm{eff}}.

the Schwinger variational (7.7), R-matrix (10.4) and Kohn-variational (12.6), are in considerable disagreement, and moreover, all substantially underestimated experiment (14.6 – 16 [49–51, 89], with 16 the recommended value [51]). The only compatible calculation to date is the stochastic variational calculation of Zhang and Mitroy [52] (14.6 for a bond length of R=1.4R=1.4 a.u., and 15.7 for a bond length of R=1.45R=1.45 a.u. Our respective calculations of 14.9 and 15.7 are in near perfect agreement with the stochastic variational method and the experiment, providing a consensus, and demonstrate that the many-body theory accurately describes the correlations. The scattering length a∼1/2κ=−2.72a \sim1/2\kappa=-2.72 determined effectively from the fit to Zeff(k)Z_{\mathrm{eff}}(k) is in <10%<10\% error compared to the accurate convergent close coupling and stochastic variational calculations (see Supplemental Material Table I).

For N2_2, the only ab initio calculation we are aware of is the Schwinger multichannel calculation, which finds Zeff=8.96Z_{\mathrm{eff}}=8.96 [30, 48], compared to the recommended measured value of 30.8±0.230.8 \pm0.2 [51]. In contrast, our calculated value of 30.5 is in excellent agreement with experiment, indicating proper account of the correlations that act to enhance ZeffZ_{\mathrm{eff}}, and with the recent model-potential calculation [42]. We found that a 2% increase in the bond length leads to a ∼5%\sim5\% increase of Z‾eff\overline{Z}_{\mathrm{eff}}. We found that the fractional contribution to the ss-wave ZeffZ_{\mathrm{eff}} from the highest 5 MOs a1ga2ua1g2eua_{1g}a_{2u}a_{1g}2e_u of HF ionization energies 40.67 eV, 20.91 eV, 17.34 eV, 17.10 eV and 17.10 eV to be 0.06, 0.24, 0.28, 0.21, and 0.21, respectively, i.e., a non-negligible fraction of annihilation occurs on MOs below the HOMO due to their favourable overlap with the positron wavefunction that is maximum around the N atoms. This pattern was observed for bound states [22] and in the extensive fragmentation patterns [8, 9, 10, 11] of polyatomic molecules.

For CH4_4, we find excellent agreement with the positron-momentum-dependent Zeff(k)Z_{\mathrm{eff}}(k) measurements of Marler et al. [90] across the full momentum range including up to k∼0.4k \sim0.4 a.u., where the pp-wave contributes (though we do not resolve the structure around 0.17 a.u.). Our thermalized value Z‾eff=119.2\overline{Z}_{\mathrm{eff}}=119.2 is lower than the measurement 140±0.8140 \pm0.8 [49, 51], and the model-potential calculation [42] (which uses adjustable parameters), especially at small kk. We found the fractional contribution from the 1a11a_1, 2a12a_1 and each of the t2t_2 orbitals, of ionization energies 304.92 eV, 25.66 eV, 14.83 eV, to be 0.0025, 0.15, 0.281. The large scattering length in CH4_4 makes Z‾eff\overline{Z}_{\mathrm{eff}} very sensitive to the correlation potential strength at low momenta (since κ∼1/2a≪1\kappa\sim1/2a \ll1). We assessed convergence of the basis set, increasing from 12 to 20 ghosts, and from TZ to QZ functions, finding only a 5% increase (see SM Figure S2). We include angular momenta functions up to ℓ=4\ell=4: whilst the basis functions from different centres combine to provide effectively higher angular momenta [98], this may be insufficient to converge the virtual-P5_5 diagram [99]. Moreover, the annihilation-vertex enhancement factors (determined from ab initio calculations for atoms [63, 67]) may underestimate the true short-range enhancement for delocalized MOs, especially since the positron can probe electron density in interstitial regions where the nuclear repulsion is reduced [100]. Calculation of the vertex enhancement for molecules is extremely challenging, and beyond the scope of this work. Further theoretical and experimental work on CH4_4 is warranted.

Summary and Outlook.—The accurate ab initio calculation of the positron-molecule annihilation rate has proved to be a formidable problem, thwarting the efforts of quantum chemistry methods for all but the simplest molecule, H2_2, for which only a sole (stochastic variational) calculation agrees with experiment. In this work, many-body theory was developed and applied to calculate positron scattering properties and annihilation rates in H2_2, N2_2 and CH4_4. The effects of correlations were elucidated. For the annihilation rates ZeffZ_{\mathrm{eff}}, for H2_2, the power of the approach was demonstrated by reproducing the benchmark stochastic variational result, thus providing a consensus with experiment. Moreover, overall excellent agreement with experiment was also found for N2_2 and CH4_4 (though further theoretical and experimental work on the latter was called for).

The positron-molecule correlation potential the many-body approach provides can be by incorporated in a TT-matrix (see e.g., [57, 101–104] or Schwinger implementation [28–35] approach to enable calculations on larger molecules, and should provide uniformly good accuracy. The framework provides a foundation for a many-body description of annihilation γ\gamma spectra, and of inelastic scattering (including Ps formation) [57, 105, 106], and to describe the coupled electronic and vibrational degrees of freedom [107] (we note in this regard the impressive recent success of the close-convergent-coupling approach in electron scattering on H2_2 [108]) for ab initio calculation of resonant annihilation scattering in positron-binding molecules [1].

Acknowledgments

Acknowledgments.—We thank Mike Charlton, Cliff Surko, James Danielson, Jack Cassidy and Sarah Gregg for useful discussions, and Gleb Gribakin and Andrew Swann for useful comments on the manuscript. This work was supported by the European Research Council grant 804383 “ANTI-ATOM”, and used the Northern Ireland High Performance Computing service funded by EPSRC (EP/T022175) and the ARCHER2 UK National Supercomputing Service.

References

interpolate to the energy of the pseudostate (see Supplementary Information Fig. 1).

Numerical Implementation: Details of Basis Sets

We construct the self energy and solve the Dyson equation in our EXCITON+ code [1] [2]. We use aug-cc-pVQZ Dunning basis sets [3, 4] centred on the atoms, enabling accurate determination of the electronic structure including polarizabilities. To more accurately describe the long-range positron-induced polarisation we place an additional basis set for the electron and positron at the molecular centre: for H2_2 and N2_2 we use an aug-cc-pVTZ basis set of hydrogen for the electronic basis and a diffuse even-tempered 19s17p16d15f basis for the positron, with exponents for the jj-th Gaussian for each angular momentum ll given as ζj=ζ0βj−1\zeta_j = \zeta_0\beta^{j-1}, with β=2\beta= 2, ζ0=10−5\zeta_0 = 10^{-5} for l=0−1l = 0 - 1 and ζ0=10−4\zeta_0 = 10^{-4} for l=2−3l = 2 - 3. For CH4_4 we place the diffuse even-tempered basis on the central carbon atom. Moreover, to more accurately describe the virtual-Ps formation process, which takes place a few a.u. away from the molecule and requires large angular momentum (to resolve the electron-positron distance and describe excited Ps), we place additional aug-cc-pVTZ basis sets of hydrogen for both electron and positron functions on ‘ghost’ centres surrounding the molecule. For H2_2 and N2_2 we used 12 on the vertices of a regular icosahedron with a circumscribed sphere of radius 2 and 3 a.u. from the centre of the molecule. For CH4_4 we additionally considered 20 ghosts placed on the vertices of a regular dodecahedron at 3 a.u. distance from the centre of the molecule (shown in Figure S1). Moreover, for CH4_4 we also considered an alternative square-pyramidal cap arrangement, with each ghost 1 Å away from the hydrogens, giving only a 0.5% improvement. Finally, we also upgraded the aug-cc-pVTZ basis set on the 20 ghost centres to QZ with only about 1% increase in the Zˉeff\bar{Z}_{\mathrm{eff}} result. Convergence w.r.t. number of virtual states was assessed: see e.g., Supplemental Figure S2, which shows the convergence of ZeffZ_{\mathrm{eff}} and Zˉeff\bar{Z}_{\mathrm{eff}} as a function of the electron and positron basis sizes for N2_2 and CH4_4.

CH4 showing the positions of the atoms and 20 ghost centres

Figure S1. CH4_4 showing the positions of the atoms and 20 ‘ghost’ centres (coloured light red) on which we place basis sets. The ghost centres are placed on the vertices of a regular dodecahedron with a circumscribed sphere of radius 3 a.u.

Effects of Correlations on Phase Shifts, Elastic Scattering Cross Sections and ZeffZ_{\mathrm{eff}}

The effects of the various diagrams describing the positron-molecule correlations, similar to Figure 1 of the main text, is shown for the phase shifts, cross sections and ZeffZ_{\mathrm{eff}} for all of the molecules. One point to note is that for H2_2 and CH4_4 the effect of the attractive electron-hole interactions within the bubbles, which are included in the GW@BSE calculation dominates over the repulsive electron-screening for N2_2, meaning that the correlation potential is more attractive compared to the bare self energy Σ2\Sigma^2. The opposite is found in N2_2.

Convergence of $Z_{\mathrm{eff}}$ in $\mathrm{N}_2$ and $\mathrm{CH}_4$ with respect to electron and positron basis sizes

Figure Supplemental Fig. 2. Convergence of ZeffZ_{\mathrm{eff}} in N2\mathrm{N}_2 (top) and CH4\mathrm{CH}_4 (bottom) with respect to electron and positron basis sizes. Values of ZeffZ_{\mathrm{eff}} (k=0.053k = 0.053) (circles) and Zˉeff\bar{Z}_{\mathrm{eff}} (diamonds) calculated with the positron-molecule self energy at the GW@BSE+ΓGW@BSE+\Gamma level, i.e., including only Figure 1(a) and Figure 1(b) from the main paper; convergence of the Λ\Lambda block is faster) are shown with respect to the number of electron (blue) and positron (red) Hartree-Fock molecular orbitals included in the basis. The number of electron (positron) states is kept at maximum when varying positron (electron) basis size. Empty symbols and dotted lines are results of calculations using a single ghost atom (at the centre of mass), striped symbols and dashed lines are from calculations with 6 additional ghost atoms (3 a.u. away from center of mass) and solid lines and symbols are from calculations with 12 ghost atoms (with hydrogen aug-cc-pVTZ basis). For CH4\mathrm{CH}_4 an additional result using 20 ghost basis centres with hydrogen aug-cc-pVQZ basis functions is shown as dotted symbols (ZeffZ_{\mathrm{eff}}) and thick lines (energies of the states). The final calculation used 850 positron states and 595 electron states. This corresponds to diagonalising a symmetric matrix of dimension 505,750, and was performed using 40 high-memory nodes of ARCHER2 using (a maximum of) 17 TB RAM.

Nine-panel plots of phase shifts, cross sections, and normalized annihilation rates for H₂, N₂, and CH₄

Figure 3 (PDF p. 10). Effect of positron-molecule correlations. Calculated ss-wave phase shifts, cross sections and normalized annihilation rate ZeffZ_{\mathrm{eff}} for H₂, N₂, and CH₄ in different approximations to the positron-molecule self energy: Hartree Fock (black dotted line); many-body theory at the Σ(2)\Sigma^{(2)} (black dot-dashed); GWGW (at BSE level), black dashed); GW+ΓGW+\Gamma (black dot-dot-dashed); and GW+Γ+ΛGW+\Gamma+\Lambda levels (thick solid line; black filled circles denote the discrete data. For the ZeffZ_{\mathrm{eff}} the solid lines are the fit to the the functional form described in the main text.

Scattering lengths

In Supplemental Table I we present the calculated scattering length aa determined by fitting the k→0k \to0 phase shifts to the effective-range-theory expansion [5]

tan⁡δ0=−ak[1−παk3a−4αk23ln⁡(Cαk4)]−1(2)\tan\delta_{0} = -ak \left[1 - \frac{\pi\alpha k}{3a} - \frac{4\alpha k^{2}}{3}\ln\left(\frac{C\sqrt{\alpha k}}{4}\right)\right]^{-1} \tag*{(2)}

where α\alpha is the molecular dipole polarizability (calculated here as 5.05, 11.08 and 15.72 for H₂, N₂ and CH₄, respectively).

Eqn. (1)a∼1/2κa \sim1/2\kappaOther calculations
H₂-3.07-2.72-2.39b, -2.49c, -2.57d, -2.06e, -2.2f
N₂-4.89-4.31-4.6 ± 0.1b, -4.64g
CH₄-13.4-8.9-17 ± 1b, -7.4h, -13.0i

Table S1. Calculated scattering lengths (a.u.) from Eqn. (1).

References

  1. [1]J. Hofierka, B. Cunningham, C. M. Rawlins, C. H. Patterson, and D. G. Green, Many-body theory of positron binding to polyatomic molecules, Nature 606, 688 (2022).DOI
  2. [2]We verified the veracity of the code by reproducing B-spline based many-body results for scattering on noble-gas atoms [15].
  3. [3]T. H. Dunning, Gaussian basis sets for use in correlated molecular calculations. I. The atoms boron through neon and hydrogen, J. Chem. Phys. 90, 1007 (1989).DOI
  4. [4]R. A. Kendall, T. H. Dunning, and R. J. Harrison, Electron affinities of the first-row atoms revisited. Systematic basis sets and wave functions, J. Chem. Phys. 96, 6796 (1992).DOI
  5. [5]L. Spruch, T. F. O’Malley, and L. Rosenberg, Modification of effective-range theory in the presence of a long-range potential, Phys. Rev. Lett. 5, 375 (1960).
  6. [6]A. R. Swann and G. F. Gribakin, Model-potential calculations of positron binding, scattering, and annihilation for atoms and small molecules using a gaussian basis, Phys. Rev. A 101, 022702 (2020).DOI
  7. [7]M. C. Zammit, D. V. Fursa, J. S. Savage, I. Bray, L. Chiari, A. Zecca, and M. J. Brunger, Adiabatic-nuclei calculations of positron scattering from molecular hydrogen, Phys. Rev. A 95, 022707 (2017).DOI
  8. [8]J. Y. Zhang and J. Mitroy, Stochastic variational calculation of zero-energy positron scattering from H, He, and H₂, Phys. Rev. A 83, 022711 (2011).DOI
  9. [9]R. Zhang, K. L. Baluja, J. Franz, and J. Tennyson, Positron collisions with molecular hydrogen: cross sections and annihilation parameters calculated using the R-matrix with pseudo-states method, J. Phys. B 44, 035203 (2011).
  10. [10]E. A. G. Armour and D. J. Baker, An improved theoretical value for Z_eff for low-energy positron-hydrogen-molecule scattering, J. Phys. B 19, L871 (1986).DOI
  11. [11]J. N. Cooper, E. A. G. Armour, and M. Plummer, The importance of an accurate target wavefunction in variational calculations for (e⁺–H₂) scattering, J. Phys. B 41, 245201 (2008).DOI
  12. [12]Z. Idziaszek and G. Karwasz, Applicability of modified effective-range theory to positron-atom and positron-molecule scattering, Phys. Rev. A 73, 064701 (2006).DOI
  13. [13]A. Zecca, L. Chiari, E. Trainotti, A. Sarkar, S. d’A. Sanchez, M. H. F. Bettega, M. T. do N. Varella, M. A. P. Lima, and M. J. Brunger, Positron scattering from methane, Phys. Rev. A 85, 012707 (2012).DOI
  14. [14]Y. Frongillo, B. Plenkiewicz, J.-P. Jay-Gerin, and A. Jain, Comparison of quasifree excess electron and positron states in simple molecular fluids: Methane and silane, Phys. Rev. E 50, 4754 (1994).pubmed.ncbi.nlm.nih.gov/9962555
  15. [15]J. Hofierka, C. M. Rawlins, B. Cunningham, C. H. Patterson, and D. G. Green, Many-body theory calculations of scattering and annihilation on noble-gas atoms using a gaussian basis in EXCITON+, Unpublished (2023).

Paper details

vlnc:W_z9sxrnby
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