Hall effect on the joint cascades of magnetic energy and helicity in helical magnetohydrodynamic turbulence

Running Hu LHD, Institute of Mechanics, Chinese Academy of Sciences, Bei**g 100190, PR China School of Engineering Science, University of Chinese Academy of Sciences, Bei**g 100049, PR China    **-Han Xie Department of Mechanics and Engineering Science, College of Engineering, Peking University, Bei**g 100871, PR China State Key Laboratory of Turbulence and Complex Systems, College of Engineering, Peking University, Bei**g 100871, PR China    Xinliang Li LHD, Institute of Mechanics, Chinese Academy of Sciences, Bei**g 100190, PR China School of Engineering Science, University of Chinese Academy of Sciences, Bei**g 100049, PR China    Chang** Yu [email protected] LHD, Institute of Mechanics, Chinese Academy of Sciences, Bei**g 100190, PR China    Yuan Hu LHD, Institute of Mechanics, Chinese Academy of Sciences, Bei**g 100190, PR China    Jianchun Wang Department of Mechanics and Aerospace Engineering, Southern University of Science and Technology, Shenzhen, Guangdong 518055, PR China    Shiyi Chen Eastern Institute for Advanced Study, Ningbo 315200, PR China State Key Laboratory of Turbulence and Complex Systems, College of Engineering, Peking University, Bei**g 100871, PR China Department of Mechanics and Aerospace Engineering, Southern University of Science and Technology, Shenzhen, Guangdong 518055, PR China
(May 15, 2024)
Abstract

Helical magnetohydrodynamic turbulence with Hall effects is ubiquitous in heliophysics and plasma physics, such as star formation and solar activities, and its intrinsic mechanisms are still not clearly explained. Direct numerical simulations reveal that when the forcing scale is comparable to the ion inertial scale, Hall effects induce remarkable cross helicity. It then suppresses the inverse cascade efficiency, leading to the accumulation of large-scale magnetic energy and helicity. The process is accompanied by the breaking of current sheets via filaments along magnetic fields. Using the Ulysses data, the numerical findings are separately confirmed. These results suggest a novel mechanism wherein small-scale Hall effects could strongly affect large-scale magnetic fields through cross helicity.

preprint: APS/123-QED

In Hall magnetohydrodynamic (HMHD) turbulence, the ions and electrons decouple from each other, which can be modeled via the Hall term in the generalized Ohm’s law (Galtier, 2016). HMHD model is valid for the process with timescales shorter than the ion cyclotron period, such as star formation (Marchand et al., 2019; Norman and Heyvaerts, 1985; Wurster et al., 2021), dynamo action (Gómez et al., 2010; Mininni et al., 2005), planetary magnetosphere (Dorelli et al., 2015), and solar activities (Bhattacharjee, 2004; Cassak et al., 2005). In the solar corona, the strongly helical magnetic field collisionlessly reconnects, leading to the rapid eruption of plasma (Ebrahimi and Raman, 2015; Wan et al., 2015; Faganello et al., 2008; Servidio et al., 2009; Sun et al., 2022).

In solar wind observation, the turbulent nature can be verified through the von Kármán-Howarth (vKH) relations for magnetohydrodynamic (MHD) turbulence (De Karman and Howarth, 1938; Politano and Pouquet, 1998; Marino et al., 2011, 2012; Sorriso-Valvo et al., 2007). Based on vKH relations, Smith et al. (2009) discovered prevailing inverse energy cascades when the cross helicity is pronounced. Marino et al. (2012) found that cross helicity could affect the linear scaling ratio of the vKH relation. In MHD and HMHD turbulence, magnetic helicity is another conservative quantity besides energy, and could measure the magnetic reconnection process quantitatively (Galtier, 2016). Politano et al. (2003) derived the vKH relations for magnetic helicity in MHD turbulence. Banerjee and Galtier (2016) derived the balance for magnetic helicity in HMHD turbulence. Brandenburg et al. (2011) found the sign change of magnetic helicity at small- and large-scale solar wind.

In practice, even if solar wind can be described by MHD turbulence, the detailed mechanisms are hard to be investigated in actual space plasma due to its complex nature. High-precision direct numerical simulations (DNS) are necessary. DNS of the helical MHD turbulence showed that helicity injection could lead to the formation of a large-scale magnetic field (Müller et al., 2012; Richner et al., 2022). Mininni et al. (2005) found the faster growth rates of large-scale magnetic fields in HMHD turbulence. Dreher et al. (2005) studied the formation and disruption of Alfvénic filaments under Hall effects.

However, even if there have been separate studies on magnetic helicity and Hall effects, there remains limited researches on their coupling effects, which are significant in the collisionless reconnection of the solar corona (Ebrahimi and Raman, 2015). The vKH relation about the magnetic helicity in HMHD turbulence is still not addressed, and the existing related theoretical works are also not verified. In this letter, we will investigate the Hall effects on the cascades of magnetic energy and helicity, and then verify those findings based on vKH relations in actual solar wind.

The governing equations for the velocity 𝐯𝐯\bf vbold_v and the magnetic field 𝐛𝐛\bf bbold_b are (Marino and Sorriso-Valvo, 2023; Müller et al., 2012):

t𝐯subscript𝑡𝐯\displaystyle\partial_{t}{\bf v}∂ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT bold_v =(𝐯)𝐯p+𝐣×𝐛+λΔ1𝐯+νΔ𝐯+𝐟𝐯,absent𝐯𝐯𝑝𝐣𝐛𝜆superscriptΔ1𝐯𝜈Δ𝐯subscript𝐟𝐯\displaystyle=-({\bf v}\cdot{\bf\nabla}){\bf v}-{\bf\nabla}p+{\bf j}\times{\bf b% }+\lambda\Delta^{-1}{\bf v}+\nu\Delta{\bf v}+{\bf f_{v}},= - ( bold_v ⋅ ∇ ) bold_v - ∇ italic_p + bold_j × bold_b + italic_λ roman_Δ start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT bold_v + italic_ν roman_Δ bold_v + bold_f start_POSTSUBSCRIPT bold_v end_POSTSUBSCRIPT , (1)
t𝐛subscript𝑡𝐛\displaystyle\partial_{t}{\bf b}∂ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT bold_b =×(𝐔×𝐛)+λbΔ1𝐛+ηΔ𝐛+𝐟𝐛,absent𝐔𝐛subscript𝜆𝑏superscriptΔ1𝐛𝜂Δ𝐛subscript𝐟𝐛\displaystyle=\nabla\times({\bf U}\times{\bf b})+\lambda_{b}\Delta^{-1}{\bf b}% +\eta\Delta{\bf b}+{\bf f_{b}},= ∇ × ( bold_U × bold_b ) + italic_λ start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT roman_Δ start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT bold_b + italic_η roman_Δ bold_b + bold_f start_POSTSUBSCRIPT bold_b end_POSTSUBSCRIPT ,

where 𝐯=0𝐯0\nabla\cdot{\bf v}=0∇ ⋅ bold_v = 0 and 𝐛=0𝐛0\nabla\cdot{\bf b}=0∇ ⋅ bold_b = 0, p𝑝pitalic_p is the pressure, 𝐔=𝐯dI𝐣𝐔𝐯subscript𝑑𝐼𝐣{\bf U}={\bf v}-d_{I}{\bf j}bold_U = bold_v - italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT bold_j, dIsubscript𝑑𝐼d_{I}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT is the ion inertial length, 𝐣=×𝐛𝐣𝐛{\bf j}=\nabla\times{\bf b}bold_j = ∇ × bold_b is the current density, 𝐟𝐯subscript𝐟𝐯{\bf f_{v}}bold_f start_POSTSUBSCRIPT bold_v end_POSTSUBSCRIPT and 𝐟𝐛subscript𝐟𝐛{\bf f_{b}}bold_f start_POSTSUBSCRIPT bold_b end_POSTSUBSCRIPT are forcing terms at forcing wavenumbers kfsubscript𝑘𝑓k_{f}italic_k start_POSTSUBSCRIPT italic_f end_POSTSUBSCRIPT, λΔ1𝐯𝜆superscriptΔ1𝐯\lambda\Delta^{-1}{\bf v}italic_λ roman_Δ start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT bold_v and λbΔ1𝐛subscript𝜆𝑏superscriptΔ1𝐛\lambda_{b}\Delta^{-1}{\bf b}italic_λ start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT roman_Δ start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT bold_b are hypo-viscous terms, ν𝜈\nuitalic_ν and η𝜂\etaitalic_η are the kinematic viscosity and magnetic diffusivity, respectively.

Six DNSs are performed in this letter. The computational parameters are shown in TABLE 1. M3, M6 and M40 are the MHD cases with different kfsubscript𝑘𝑓k_{f}italic_k start_POSTSUBSCRIPT italic_f end_POSTSUBSCRIPT, while H3, H6 and H40 are corresponding HMHD cases. The random forcing schemes (𝐟𝐯subscript𝐟𝐯{\bf f_{v}}bold_f start_POSTSUBSCRIPT bold_v end_POSTSUBSCRIPT and 𝐟𝐛subscript𝐟𝐛{\bf f_{b}}bold_f start_POSTSUBSCRIPT bold_b end_POSTSUBSCRIPT) only inject maximum magnetic helicity artificially but do not inject any hydrodynamic or cross helicity. See the Supplemental Material (SM) (sup, ) for more numerical details.

Table 1: The computational configurations of flows. εvsubscript𝜀𝑣\varepsilon_{v}italic_ε start_POSTSUBSCRIPT italic_v end_POSTSUBSCRIPT and εbsubscript𝜀𝑏\varepsilon_{b}italic_ε start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT are injection rates of hydrodynamic and magnetic energy.
Case Grid dIsubscript𝑑𝐼d_{I}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ν=η𝜈𝜂\nu=\etaitalic_ν = italic_η λ=λb𝜆subscript𝜆𝑏\lambda=\lambda_{b}italic_λ = italic_λ start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT kfsubscript𝑘𝑓k_{f}italic_k start_POSTSUBSCRIPT italic_f end_POSTSUBSCRIPT εv=εbsubscript𝜀𝑣subscript𝜀𝑏\varepsilon_{v}=\varepsilon_{b}italic_ε start_POSTSUBSCRIPT italic_v end_POSTSUBSCRIPT = italic_ε start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT
M3 20483superscript204832048^{3}2048 start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT 4×1054superscript1054\times 10^{-5}4 × 10 start_POSTSUPERSCRIPT - 5 end_POSTSUPERSCRIPT 0.1 3 0.015
H3 20483superscript204832048^{3}2048 start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT 0.05 4×1054superscript1054\times 10^{-5}4 × 10 start_POSTSUPERSCRIPT - 5 end_POSTSUPERSCRIPT 0.1 3 0.015
M6 20483superscript204832048^{3}2048 start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT 4×1054superscript1054\times 10^{-5}4 × 10 start_POSTSUPERSCRIPT - 5 end_POSTSUPERSCRIPT 0.1 6 0.015
H6 20483superscript204832048^{3}2048 start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT 0.05 4×1054superscript1054\times 10^{-5}4 × 10 start_POSTSUPERSCRIPT - 5 end_POSTSUPERSCRIPT 0.1 6 0.015
M40 10243superscript102431024^{3}1024 start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT 104superscript10410^{-4}10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT 0.05 40-42 0.015
H40 10243superscript102431024^{3}1024 start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT 0.05 104superscript10410^{-4}10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT 0.05 40-42 0.015
Refer to caption
Figure 1: Compensated Spectra. (a) Eb(k)k5/3subscript𝐸𝑏𝑘superscript𝑘53E_{b}(k)k^{5/3}italic_E start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT ( italic_k ) italic_k start_POSTSUPERSCRIPT 5 / 3 end_POSTSUPERSCRIPT. (b) HM(k)k10/3subscript𝐻𝑀𝑘superscript𝑘103H_{M}(k)k^{10/3}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT ( italic_k ) italic_k start_POSTSUPERSCRIPT 10 / 3 end_POSTSUPERSCRIPT.

FIG. 1 shows the spectra of magnetic energy Eb(k)=|𝐤|=k{𝐛^(𝐤)𝐛^(𝐤)}/2subscript𝐸𝑏𝑘subscript𝐤𝑘^𝐛𝐤superscript^𝐛𝐤2E_{b}(k)=\sum_{|{\bf k}|=k}\Re\{\hat{\bf b}({\bf k})\cdot\hat{\bf b}^{\ast}({% \bf k})\}/2italic_E start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT ( italic_k ) = ∑ start_POSTSUBSCRIPT | bold_k | = italic_k end_POSTSUBSCRIPT roman_ℜ { over^ start_ARG bold_b end_ARG ( bold_k ) ⋅ over^ start_ARG bold_b end_ARG start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ( bold_k ) } / 2 and magnetic helicity HM(k)=|𝐤|=k{𝐚^(𝐤)𝐛^(𝐤)}subscript𝐻𝑀𝑘subscript𝐤𝑘^𝐚𝐤superscript^𝐛𝐤H_{M}(k)=\sum_{|{\bf k}|=k}\Re\{\hat{\bf a}({\bf k})\cdot\hat{\bf b}^{\ast}({% \bf k})\}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT ( italic_k ) = ∑ start_POSTSUBSCRIPT | bold_k | = italic_k end_POSTSUBSCRIPT roman_ℜ { over^ start_ARG bold_a end_ARG ( bold_k ) ⋅ over^ start_ARG bold_b end_ARG start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ( bold_k ) }, where ^^\hat{\cdot}over^ start_ARG ⋅ end_ARG represents the values in spectral space, and 𝐚^(𝐤)^𝐚𝐤\hat{\bf a}({\bf k})over^ start_ARG bold_a end_ARG ( bold_k ) is the magnetic potential. In FIG. 1 (a), Eb(k)k5/3similar-tosubscript𝐸𝑏𝑘superscript𝑘53E_{b}(k)\sim k^{-5/3}italic_E start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT ( italic_k ) ∼ italic_k start_POSTSUPERSCRIPT - 5 / 3 end_POSTSUPERSCRIPT in the inertial range (k(4,100)𝑘4100k\in(4,100)italic_k ∈ ( 4 , 100 )), but in the ion inertial scale (k1/dIsimilar-to𝑘1subscript𝑑𝐼k\sim 1/d_{I}italic_k ∼ 1 / italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT), Eb(k)k7/3similar-tosubscript𝐸𝑏𝑘superscript𝑘73E_{b}(k)\sim k^{-7/3}italic_E start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT ( italic_k ) ∼ italic_k start_POSTSUPERSCRIPT - 7 / 3 end_POSTSUPERSCRIPT for HMHD cases (H3 and H6). For the magnetic helicity in FIG. 1 (b), HM(k)k10/3similar-tosubscript𝐻𝑀𝑘superscript𝑘103H_{M}(k)\sim k^{-10/3}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT ( italic_k ) ∼ italic_k start_POSTSUPERSCRIPT - 10 / 3 end_POSTSUPERSCRIPT for M3 and M6, due to the Alfvénic balance (Mininni and Pouquet, 2009). For the inverse cascade ranges of H40 and M40, HM(k)k8/3Eb(k)/ksimilar-tosubscript𝐻𝑀𝑘superscript𝑘83similar-tosubscript𝐸𝑏𝑘𝑘H_{M}(k)\sim k^{-8/3}\sim E_{b}(k)/kitalic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT ( italic_k ) ∼ italic_k start_POSTSUPERSCRIPT - 8 / 3 end_POSTSUPERSCRIPT ∼ italic_E start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT ( italic_k ) / italic_k, implying a fully helical magnetic field. In the evolutionary process given in SM (sup, ), HM(k)k10/3similar-tosubscript𝐻𝑀𝑘superscript𝑘103H_{M}(k)\sim k^{-10/3}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT ( italic_k ) ∼ italic_k start_POSTSUPERSCRIPT - 10 / 3 end_POSTSUPERSCRIPT in the inverse cascade range (Müller et al., 2012; Linkmann and Dallas, 2017). Furthermore, the comparison of M40 and H40 shows that Hall effects enhance large-scale magnetic energy and helicity by nearly tenfold. This implies Hall effects could strongly affect large-scale magnetic fields, which is the main subject of this letter.

Then, the filtering approach (Camporeale et al., 2018; Manzini et al., 2022) is used for the details of cascades. The fluxes for the magnetic energy (Π~Eb1subscript~Π𝐸𝑏1\widetilde{\Pi}_{Eb1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 1 end_POSTSUBSCRIPT and Π~Eb2subscript~Π𝐸𝑏2\widetilde{\Pi}_{Eb2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 2 end_POSTSUBSCRIPT) and magnetic helicity (Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT and Π~HM2subscript~Π𝐻𝑀2\widetilde{\Pi}_{HM2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 2 end_POSTSUBSCRIPT) are considered:

Π~Eb1subscript~Π𝐸𝑏1\displaystyle\widetilde{\Pi}_{Eb1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 1 end_POSTSUBSCRIPT =ϵijkj~iτjk(𝐯,𝐛),absentsubscriptitalic-ϵ𝑖𝑗𝑘subscript~𝑗𝑖subscript𝜏𝑗𝑘𝐯𝐛\displaystyle=-\epsilon_{ijk}\tilde{j}_{i}\tau_{jk}({\bf v},{\bf{b}}),= - italic_ϵ start_POSTSUBSCRIPT italic_i italic_j italic_k end_POSTSUBSCRIPT over~ start_ARG italic_j end_ARG start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_τ start_POSTSUBSCRIPT italic_j italic_k end_POSTSUBSCRIPT ( bold_v , bold_b ) , (2a)
Π~Eb2subscript~Π𝐸𝑏2\displaystyle\widetilde{\Pi}_{Eb2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 2 end_POSTSUBSCRIPT =ϵijkdIj~iτjk(𝐣,𝐛),absentsubscriptitalic-ϵ𝑖𝑗𝑘subscript𝑑𝐼subscript~𝑗𝑖subscript𝜏𝑗𝑘𝐣𝐛\displaystyle=\epsilon_{ijk}d_{I}\tilde{j}_{i}\tau_{jk}({\bf j},{\bf b}),= italic_ϵ start_POSTSUBSCRIPT italic_i italic_j italic_k end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT over~ start_ARG italic_j end_ARG start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_τ start_POSTSUBSCRIPT italic_j italic_k end_POSTSUBSCRIPT ( bold_j , bold_b ) , (2b)
Π~HM1subscript~Π𝐻𝑀1\displaystyle\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT =2ϵijkb~iτjk(𝐯,𝐛),absent2subscriptitalic-ϵ𝑖𝑗𝑘subscript~𝑏𝑖subscript𝜏𝑗𝑘𝐯𝐛\displaystyle=-2\epsilon_{ijk}\tilde{b}_{i}\tau_{jk}({\bf v},{\bf b}),= - 2 italic_ϵ start_POSTSUBSCRIPT italic_i italic_j italic_k end_POSTSUBSCRIPT over~ start_ARG italic_b end_ARG start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_τ start_POSTSUBSCRIPT italic_j italic_k end_POSTSUBSCRIPT ( bold_v , bold_b ) , (2c)
Π~HM2subscript~Π𝐻𝑀2\displaystyle\widetilde{\Pi}_{HM2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 2 end_POSTSUBSCRIPT =2ϵijkdIb~iτjk(𝐣,𝐛),absent2subscriptitalic-ϵ𝑖𝑗𝑘subscript𝑑𝐼subscript~𝑏𝑖subscript𝜏𝑗𝑘𝐣𝐛\displaystyle=2\epsilon_{ijk}d_{I}\tilde{b}_{i}\tau_{jk}({\bf j},{\bf b}),= 2 italic_ϵ start_POSTSUBSCRIPT italic_i italic_j italic_k end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT over~ start_ARG italic_b end_ARG start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_τ start_POSTSUBSCRIPT italic_j italic_k end_POSTSUBSCRIPT ( bold_j , bold_b ) , (2d)

where τij(𝐩,𝐪)=piqj~pi~qj~subscript𝜏𝑖𝑗𝐩𝐪~subscript𝑝𝑖subscript𝑞𝑗~subscript𝑝𝑖~subscript𝑞𝑗{\bf\tau}_{ij}({\bf p},{\bf q})=\widetilde{{p_{i}q_{j}}}-\widetilde{{p}_{i}}% \widetilde{{q}_{j}}italic_τ start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT ( bold_p , bold_q ) = over~ start_ARG italic_p start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_q start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_ARG - over~ start_ARG italic_p start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_ARG over~ start_ARG italic_q start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT end_ARG for arbitrary vectors 𝐩𝐩\bf pbold_p and 𝐪𝐪\bf qbold_q, and the superscript ~~\widetilde{\cdot}over~ start_ARG ⋅ end_ARG represents the filtered quantities. Π~Eb1subscript~Π𝐸𝑏1\widetilde{\Pi}_{Eb1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 1 end_POSTSUBSCRIPT and Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT both appear in MHD and HMHD turbulence, while Π~Eb2subscript~Π𝐸𝑏2\widetilde{\Pi}_{Eb2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 2 end_POSTSUBSCRIPT and Π~HM2subscript~Π𝐻𝑀2\widetilde{\Pi}_{HM2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 2 end_POSTSUBSCRIPT only appear in HMHD turbulence. The overall magnetic energy Π~Eb=Π~Eb1+Π~Eb2subscript~Π𝐸𝑏subscript~Π𝐸𝑏1subscript~Π𝐸𝑏2\widetilde{\Pi}_{Eb}=\widetilde{\Pi}_{Eb1}+\widetilde{\Pi}_{Eb2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b end_POSTSUBSCRIPT = over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 1 end_POSTSUBSCRIPT + over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 2 end_POSTSUBSCRIPT, and the overall magnetic helicity Π~HM=Π~HM1+Π~HM2subscript~Π𝐻𝑀subscript~Π𝐻𝑀1subscript~Π𝐻𝑀2\widetilde{\Pi}_{HM}=\widetilde{\Pi}_{HM1}+\widetilde{\Pi}_{HM2}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT = over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT + over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 2 end_POSTSUBSCRIPT. For filtering approaches, different filters lead to consistent qualitative results (Yan et al., 2020; Kuzzay et al., 2019), and the fourth-order Butterworth filter (Camporeale et al., 2018) is used in this study.

Refer to caption
Figure 2: Fluxes versus the filter width ΔΔ\Deltaroman_Δ, where L=2π𝐿2𝜋L=2\piitalic_L = 2 italic_π. (a) Magnetic energy fluxes; (b) Magnetic helicity fluxes. Solid lines: H40; dashed lines: M40.

FIG. 2 (a) shows the magnetic energy fluxes of H40 and M40, where delimited-⟨⟩\left\langle\cdot\right\rangle⟨ ⋅ ⟩ represents the ensemble average. As shown, most magnetic energy cascades forward towards small scales. Weak inverse cascades also appear and have been identified as the α𝛼\alphaitalic_α-effects (Müller et al., 2012; Linkmann and Dallas, 2017; Pouquet et al., 2019). FIG. 2 (b) shows the magnetic helicity fluxes. Π~HM1delimited-⟨⟩subscript~Π𝐻𝑀1\left\langle\widetilde{\Pi}_{HM1}\right\rangle⟨ over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT ⟩ is the dominant term, and most magnetic helicity cascades inversely towards large scales. Hall effects slightly reduce the inverse cascade rates, attributed to the far greater large-scale magnetic helicity and hypo-viscosity. In summary, Hall effects minimally affect the inverse cascades of magnetic energy and helicity in FIG. 2, but strongly enhance the large-scale magnetic energy and helicity in FIG. 1. Therefore, it can be inferred that the enhanced large-scale magnetic field may be related to the inefficiency of cascades.

With large scales governed by the inverse magnetic helicity cascades, the dominant term Π~HM1delimited-⟨⟩subscript~Π𝐻𝑀1\left\langle\widetilde{\Pi}_{HM1}\right\rangle⟨ over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT ⟩ is the focus henceforth. According to Eq. (1) and (2c), the cascade is directly related to cross helicity. FIG. 3 (a) shows the cross helicity spectra HC(k)=|𝐤|=k{𝐯^(𝐤)𝐛^(𝐤)}subscript𝐻𝐶𝑘subscript𝐤𝑘^𝐯𝐤superscript^𝐛𝐤H_{C}(k)=\sum_{|{\bf k}|=k}\Re\{\hat{\bf v}({\bf k})\cdot\hat{\bf b}^{\ast}({% \bf k})\}italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT ( italic_k ) = ∑ start_POSTSUBSCRIPT | bold_k | = italic_k end_POSTSUBSCRIPT roman_ℜ { over^ start_ARG bold_v end_ARG ( bold_k ) ⋅ over^ start_ARG bold_b end_ARG start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ( bold_k ) }. As shown, HC(k)subscript𝐻𝐶𝑘H_{C}(k)italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT ( italic_k ) is approximately zero for MHD cases but negative for HMHD cases. In fact, for MHD cases, cross helicity is conservative (Galtier, 2016). No cross helicity is injected, making it inherently zero. In contrast, for HMHD cases, generalized helicity and magnetic helicity are conservative (Galtier, 2016). The generalized helicity can be decomposed into the sum of magnetic, hydrodynamic and cross helicity weighted by dIsubscript𝑑𝐼d_{I}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT. Therefore, the weighted sum of hydrodynamic and cross helicity is conservative. The hydrodynamic helicity is transformed to cross helicity through the Hall term (dI×(𝐣×𝐛)subscript𝑑𝐼𝐣𝐛d_{I}\nabla\times({\bf j}\times{\bf b})italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ∇ × ( bold_j × bold_b )). For the physical mechanisms, Hall effects are in fact the ion cyclotron effects on the magnetic field induced by the Lorentz force. In MHD cases, only the ion cyclotron effects on the velocity field are considered. In HMHD cases, the introduction of ion cyclotron effects on the magnetic field, which couples the chiralities of velocity and magnetic fields, produces non-zero cross helicity.

Refer to caption
Figure 3: Cross helicity distributions and magnetic helicity cascades. (a) HC(k)subscript𝐻𝐶𝑘H_{C}(k)italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT ( italic_k ). (b) P.D.F. of |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG |, with Δ1=0.038LsubscriptΔ10.038𝐿\Delta_{1}=0.038Lroman_Δ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = 0.038 italic_L and Δ2=0.013LsubscriptΔ20.013𝐿\Delta_{2}=0.013Lroman_Δ start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = 0.013 italic_L. Solid lines: H40; Dashed lines: M40. (c) Joint P.D.F. of Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT and |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG | with Δ=0.038LΔ0.038𝐿\Delta=0.038Lroman_Δ = 0.038 italic_L. (d) Conditional average of ΓHM1subscriptΓ𝐻𝑀1\Gamma_{HM1}roman_Γ start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT versus |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG |, with Δ=0.038LΔ0.038𝐿\Delta=0.038Lroman_Δ = 0.038 italic_L.

Furthermore, the normalized cross helicity can be defined as cosθ~=𝐯~𝐛~/[|𝐯~||𝐛~|]~𝜃~𝐯~𝐛delimited-[]~𝐯~𝐛\widetilde{\cos\theta}={\widetilde{{\bf v}}\cdot\widetilde{{\bf b}}}/[{|% \widetilde{{\bf v}}||\widetilde{{\bf b}}|}]over~ start_ARG roman_cos italic_θ end_ARG = over~ start_ARG bold_v end_ARG ⋅ over~ start_ARG bold_b end_ARG / [ | over~ start_ARG bold_v end_ARG | | over~ start_ARG bold_b end_ARG | ], which evaluates the angle between velocity and the magnetic fields. The normalized cross helicity directly affects the cascade rates through the term ×(𝐯×𝐛)𝐯𝐛\nabla\times({\bf v}\times{\bf b})∇ × ( bold_v × bold_b ) in Eq. (1) and Π~Eb1subscript~Π𝐸𝑏1\widetilde{\Pi}_{Eb1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 1 end_POSTSUBSCRIPT,Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT in Eq. (2). As |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG | increases, the velocity and magnetic fields align with each other, which could reduce the efficiency of Π~Eb1subscript~Π𝐸𝑏1\widetilde{\Pi}_{Eb1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_E italic_b 1 end_POSTSUBSCRIPT,Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT. FIG. 3 (b) shows the probability distribution functions (P.D.F.) of |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG | with two filter widths Δ1=0.038LsubscriptΔ10.038𝐿\Delta_{1}=0.038Lroman_Δ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = 0.038 italic_L and Δ2=0.013LsubscriptΔ20.013𝐿\Delta_{2}=0.013Lroman_Δ start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = 0.013 italic_L. As the filter width increases or the Hall effects are introduced, |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG | centers around 1.0, indicating enhanced alignments between 𝐯~~𝐯\bf\widetilde{v}over~ start_ARG bold_v end_ARG and 𝐛~~𝐛\bf\widetilde{b}over~ start_ARG bold_b end_ARG. FIG. 3 (c) compares the joint P.D.F. of Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT and cosθ~~𝜃\widetilde{\cos\theta}over~ start_ARG roman_cos italic_θ end_ARG between H40 and M40 with the filter width Δ=0.038LΔ0.038𝐿\Delta=0.038Lroman_Δ = 0.038 italic_L. For M40, Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT are almost independent with |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG |. In contrast, as Hall effects are introduced (H40), Π~HM1subscript~Π𝐻𝑀1\widetilde{\Pi}_{HM1}over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT primarily concentrates on the region with significant |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG |, indicating a potential decrease in efficiency. For a quantitative estimation, a direct definition of efficiency ΓHM1subscriptΓ𝐻𝑀1\Gamma_{HM1}roman_Γ start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT is introduced as

ΓHM1=|Π~HM1|2|𝐛~||ϵijkτjk(𝐯,𝐛)𝐞i||𝐯~×𝐛~||𝐯~||𝐛~|,subscriptΓ𝐻𝑀1subscript~Π𝐻𝑀12~𝐛subscriptitalic-ϵ𝑖𝑗𝑘subscript𝜏𝑗𝑘𝐯𝐛subscript𝐞𝑖~𝐯~𝐛~𝐯~𝐛\displaystyle\Gamma_{HM1}=\frac{|\widetilde{\Pi}_{HM1}|}{2|\widetilde{\bf b}||% \epsilon_{ijk}\tau_{jk}({\bf v},{\bf b}){\bf e}_{i}|}\frac{|\widetilde{\bf v}% \times\widetilde{\bf b}|}{|\widetilde{\bf v}|\,|\widetilde{\bf b}|},roman_Γ start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT = divide start_ARG | over~ start_ARG roman_Π end_ARG start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT | end_ARG start_ARG 2 | over~ start_ARG bold_b end_ARG | | italic_ϵ start_POSTSUBSCRIPT italic_i italic_j italic_k end_POSTSUBSCRIPT italic_τ start_POSTSUBSCRIPT italic_j italic_k end_POSTSUBSCRIPT ( bold_v , bold_b ) bold_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT | end_ARG divide start_ARG | over~ start_ARG bold_v end_ARG × over~ start_ARG bold_b end_ARG | end_ARG start_ARG | over~ start_ARG bold_v end_ARG | | over~ start_ARG bold_b end_ARG | end_ARG , (3)

where 𝐞isubscript𝐞𝑖{\bf e}_{i}bold_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT is the unit vector on the i𝑖iitalic_i-th direction, the first fraction evaluates the efficiency loss induced by the angle between the two vectors 𝐛~~𝐛\widetilde{\bf b}over~ start_ARG bold_b end_ARG and ϵijkτjk(𝐯,𝐛)𝐞isubscriptitalic-ϵ𝑖𝑗𝑘subscript𝜏𝑗𝑘𝐯𝐛subscript𝐞𝑖\epsilon_{ijk}\tau_{jk}({\bf v},{\bf b}){\bf e}_{i}italic_ϵ start_POSTSUBSCRIPT italic_i italic_j italic_k end_POSTSUBSCRIPT italic_τ start_POSTSUBSCRIPT italic_j italic_k end_POSTSUBSCRIPT ( bold_v , bold_b ) bold_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT, and the second fraction evaluates the direct efficiency loss induced by the angle between 𝐯~~𝐯\widetilde{\bf v}over~ start_ARG bold_v end_ARG and 𝐛~~𝐛\widetilde{\bf b}over~ start_ARG bold_b end_ARG. FIG. 3 (d) shows the conditional average ΓHM1||cosθ~|\left\langle\Gamma_{HM1}||\widetilde{\cos\theta}|\right\rangle⟨ roman_Γ start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT | | over~ start_ARG roman_cos italic_θ end_ARG | ⟩ with Δ=0.038LΔ0.038𝐿\Delta=0.038Lroman_Δ = 0.038 italic_L. ΓHM1subscriptΓ𝐻𝑀1\Gamma_{HM1}roman_Γ start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT is inversely proportional to |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG |. In addition, H40 has a relatively lower efficiency than M40. Specifically, when Δ=0.038LΔ0.038𝐿\Delta=0.038Lroman_Δ = 0.038 italic_L, Hall effects reduce the averaged efficiency ΓHM1delimited-⟨⟩subscriptΓ𝐻𝑀1\left\langle\Gamma_{HM1}\right\rangle⟨ roman_Γ start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT ⟩ by 53.2%.

In fact, both hydrodynamic helicity in hydrodynamic turbulence (Chen et al., 2003; Yan et al., 2020) and cross helicity in HMHD turbulence could affect cascade efficiencies. It could be of interest to investigate the disparities between the effects induced by the two helicity. Generally, hydrodynamic or cross helicity has two sources: geometric and phase alignments (Milanese et al., 2021). Taking the cross helicity as an example, geometric alignment means an increase in |cosθ~|~𝜃|\widetilde{\cos\theta}|| over~ start_ARG roman_cos italic_θ end_ARG |, while phase alignment denotes a transformation from negative cosθ~~𝜃\widetilde{\cos\theta}over~ start_ARG roman_cos italic_θ end_ARG to positive cosθ~~𝜃\widetilde{\cos\theta}over~ start_ARG roman_cos italic_θ end_ARG. Milanese et al. (2021) found that in hydrodynamic turbulence, hydrodynamic helicity cascades forward with dynamic phase alignments, while no geometry alignment is detected. In comparison, FIG. 3 (b) reveals that geometric alignments are the key processes in HMHD turbulence. The two different alignments lead to different effects on cascades. As shown in FIG. 3 (b-d), the geometric alignment induced by Hall effects strongly reduces the cascade efficiency.

Refer to caption
Figure 4: Structures with |𝐣|>2.2mean{|𝐣|}𝐣2.2mean𝐣|{\bf j}|>2.2\,\textrm{mean}\{|{\bf j}|\}| bold_j | > 2.2 mean { | bold_j | }. The background slice shows the contour of |𝐣|𝐣|{\bf j}|| bold_j |. (a) M3; (b) H3; (c) M40; (d) H40. Pink lines: magnetic lines; Green lines: streamlines.

FIG. 4 shows the current density |𝐣|𝐣|{\bf j}|| bold_j |, where the pink and green lines give the magnetic lines and streamlines, respectively. The comparison of FIG. 4 (a) M3 and (b) H3 reveals that when the forcing scale is significantly larger than dIsubscript𝑑𝐼d_{I}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT, Hall effects induce filaments along the magnetic field (Miura and Hori, 2009), attributed to the ion cyclotron and whistler modes (Banerjee and Galtier, 2016; Meyrand and Galtier, 2012) and the transverse instabilities (Dreher et al., 2005) in HMHD. The comparison of FIG. 4 (c) M40 and (d) H40 shows that as the forcing scale is close to dIsubscript𝑑𝐼d_{I}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT, current sheets are totally broken by these filaments along the magnetic field. The current density distribution of H40 is irregular and coarse. Moreover, alignments between the streamlines and magnetic lines are more pronounced in H40 than in M40.

Then, experimental evidence is necessary to verify our numerical findings. Since only single-point time series of solar wind can be obtained by spacecraft, vKH relations are needed (Politano and Pouquet, 1998). Exact vKH relations for the magnetic helicity in isotropic HMHD turbulence are derived here, where the bidirectional transfer model (Xie and Bühler, 2019) is applied. After lengthy derivations in SM (sup, ), the first vKH relation for the magnetic helicity can be written as

VHM1(r)subscript𝑉𝐻𝑀1𝑟\displaystyle V_{HM1}(r)italic_V start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT ( italic_r ) =r2δ(U3b2U2b3)δbLabsent𝑟2delimited-⟨⟩𝛿subscript𝑈3subscript𝑏2subscript𝑈2subscript𝑏3𝛿subscript𝑏𝐿\displaystyle=\frac{r}{2}\langle\delta(U_{3}b_{2}-U_{2}b_{3})\delta b_{L}\rangle= divide start_ARG italic_r end_ARG start_ARG 2 end_ARG ⟨ italic_δ ( italic_U start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT - italic_U start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) italic_δ italic_b start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT ⟩ (4)
=13εHMIrK{kf}εHM(K)sinKrKrcosKrK3r2,absent13subscript𝜀𝐻𝑀𝐼𝑟subscript𝐾subscript𝑘𝑓subscript𝜀𝐻𝑀𝐾𝐾𝑟𝐾𝑟𝐾𝑟superscript𝐾3superscript𝑟2\displaystyle=\frac{1}{3}\varepsilon_{HMI}\,r-\sum_{K\in\{k_{f}\}}\varepsilon_% {HM}(K)\frac{\sin Kr-Kr\cos Kr}{K^{3}r^{2}},= divide start_ARG 1 end_ARG start_ARG 3 end_ARG italic_ε start_POSTSUBSCRIPT italic_H italic_M italic_I end_POSTSUBSCRIPT italic_r - ∑ start_POSTSUBSCRIPT italic_K ∈ { italic_k start_POSTSUBSCRIPT italic_f end_POSTSUBSCRIPT } end_POSTSUBSCRIPT italic_ε start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT ( italic_K ) divide start_ARG roman_sin italic_K italic_r - italic_K italic_r roman_cos italic_K italic_r end_ARG start_ARG italic_K start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG ,
={εHMFr/3,rkf1,εHMIr/3,rkf1,\displaystyle=\left\{\begin{aligned} -\varepsilon_{HMF}\,r/3,rk_{f}\ll 1,\\ \varepsilon_{HMI}\,r/3,rk_{f}\gg 1,\\ \end{aligned}\right.= { start_ROW start_CELL - italic_ε start_POSTSUBSCRIPT italic_H italic_M italic_F end_POSTSUBSCRIPT italic_r / 3 , italic_r italic_k start_POSTSUBSCRIPT italic_f end_POSTSUBSCRIPT ≪ 1 , end_CELL end_ROW start_ROW start_CELL italic_ε start_POSTSUBSCRIPT italic_H italic_M italic_I end_POSTSUBSCRIPT italic_r / 3 , italic_r italic_k start_POSTSUBSCRIPT italic_f end_POSTSUBSCRIPT ≫ 1 , end_CELL end_ROW

where δqi(𝐫)=qi(𝐱)qi(𝐱)𝛿subscript𝑞𝑖𝐫subscript𝑞𝑖superscript𝐱subscript𝑞𝑖𝐱\delta q_{i}({\bf r})=q_{i}({\bf x}^{\prime})-q_{i}({\bf x})italic_δ italic_q start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( bold_r ) = italic_q start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( bold_x start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ) - italic_q start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( bold_x ) is the two-point increment for an arbitrary vector 𝐪𝐪{\bf q}bold_q; 𝐫=𝐱𝐱𝐫superscript𝐱𝐱{\bf r=x^{\prime}-x}bold_r = bold_x start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT - bold_x is the displacement between two positions 𝐱𝐱\bf xbold_x and 𝐱superscript𝐱\bf x^{\prime}bold_x start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT with a distance r=|𝐫|𝑟𝐫r=|{\bf r}|italic_r = | bold_r |; the subscript L𝐿Litalic_L refers to the longitudinal direction (along 𝐫𝐫\bf rbold_r), and the subscripts 2 and 3 refer to the two remaining transverse directions; εHMIsubscript𝜀𝐻𝑀𝐼\varepsilon_{HMI}italic_ε start_POSTSUBSCRIPT italic_H italic_M italic_I end_POSTSUBSCRIPT and εHMFsubscript𝜀𝐻𝑀𝐹\varepsilon_{HMF}italic_ε start_POSTSUBSCRIPT italic_H italic_M italic_F end_POSTSUBSCRIPT are the inverse and forward cascade rates, εHM(K)subscript𝜀𝐻𝑀𝐾\varepsilon_{HM}(K)italic_ε start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT ( italic_K ) is the injection rate at the forcing wavenumber K{kf}𝐾subscript𝑘𝑓K\in\{k_{f}\}italic_K ∈ { italic_k start_POSTSUBSCRIPT italic_f end_POSTSUBSCRIPT }. This equation is the isotropic form of the results derived by Banerjee and Galtier (2016). Using the isotropic condition, VHM1(r)subscript𝑉𝐻𝑀1𝑟V_{HM1}(r)italic_V start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT ( italic_r ) is equivalent to

VHM2=2δ(v2bLvLb2)a2dIδ(bLbi)bi,subscript𝑉𝐻𝑀22delimited-⟨⟩𝛿subscript𝑣2subscript𝑏𝐿subscript𝑣𝐿subscript𝑏2superscriptsubscript𝑎2subscript𝑑𝐼delimited-⟨⟩𝛿subscript𝑏𝐿subscript𝑏𝑖superscriptsubscript𝑏𝑖\displaystyle V_{HM2}=2\left\langle\delta(v_{2}b_{L}-v_{L}b_{2})a_{2}^{\ast}% \right\rangle-{d_{I}}\left\langle\delta(b_{L}b_{i})b_{i}^{\ast}\right\rangle,italic_V start_POSTSUBSCRIPT italic_H italic_M 2 end_POSTSUBSCRIPT = 2 ⟨ italic_δ ( italic_v start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT - italic_v start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ⟩ - italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ⟨ italic_δ ( italic_b start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) italic_b start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ⟩ , (5)

where qi(𝐫)=[qi(𝐱)+qi(𝐱)]/2superscriptsubscript𝑞𝑖𝐫delimited-[]subscript𝑞𝑖𝐱subscript𝑞𝑖superscript𝐱2q_{i}^{\ast}({\bf r})=[q_{i}({\bf x})+q_{i}({\bf x}^{\prime})]/2italic_q start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ( bold_r ) = [ italic_q start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( bold_x ) + italic_q start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( bold_x start_POSTSUPERSCRIPT ′ end_POSTSUPERSCRIPT ) ] / 2 for an arbitrary vector 𝐪𝐪\bf qbold_q.

Refer to caption
Figure 5: Evidence of solar wind data. (a) Numerical verifications of VHM1subscript𝑉𝐻𝑀1V_{HM1}italic_V start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT and VHM2subscript𝑉𝐻𝑀2V_{HM2}italic_V start_POSTSUBSCRIPT italic_H italic_M 2 end_POSTSUBSCRIPT. Solid lines: positive values; dashed lines: negative values. (b) Example of the linear scaling in solar wind. (c) Conditional average of the cross helicity versus the magnetic helicity. (d) The normalized cascade rates versus the normalized cross helicity.

FIG. 5 (a) verifies the vKH relations using the results of H40, where black curves are predictions in Eq. (4) with εHMI=4.4×104subscript𝜀𝐻𝑀𝐼4.4superscript104\varepsilon_{HMI}=4.4\times 10^{-4}italic_ε start_POSTSUBSCRIPT italic_H italic_M italic_I end_POSTSUBSCRIPT = 4.4 × 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT and εHMF=1.0×104subscript𝜀𝐻𝑀𝐹1.0superscript104\varepsilon_{HMF}=1.0\times 10^{-4}italic_ε start_POSTSUBSCRIPT italic_H italic_M italic_F end_POSTSUBSCRIPT = 1.0 × 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT. The equivalent cascade rates are estimated through the fluxes over corresponding wavenumber ranges in SM (sup, ). As shown, VHM1subscript𝑉𝐻𝑀1V_{HM1}italic_V start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT and VHM2subscript𝑉𝐻𝑀2V_{HM2}italic_V start_POSTSUBSCRIPT italic_H italic_M 2 end_POSTSUBSCRIPT are almost the same. At the scale r/L<0.2𝑟𝐿0.2r/L<0.2italic_r / italic_L < 0.2, they both fit well with the predictions. In contrast, at the scale r/L0.2greater-than-or-approximately-equals𝑟𝐿0.2r/L\gtrapprox 0.2italic_r / italic_L ⪆ 0.2, the results deviate from the prediction, attributed to the simple inertial range assumption (Xie and Bühler, 2019). To further verify the vKH relations, 8 minute averaged time series for solar wind during day 209/1993 - day 253/1994, day 212/1995 - day 224/1996 and day 20/2006 - day 23/2007 of the Ulysses spacecraft are evaluated (Balogh et al., 1995; Smith et al., 1995). The three periods are very close to solar activity minima, and the heliocentric distance ranges from 2.5 to 4.5 AU (Marino et al., 2012). Using the Taylor hypothesis (Taylor, 1938), the vKH relation for magnetic helicity VHM(τ)=VLτδ(v3b2v2b3)δbL/2dIδ(bLbi)bit=εHMVLτ/3subscript𝑉𝐻𝑀𝜏subscriptdelimited-⟨⟩subscript𝑉𝐿𝜏𝛿subscript𝑣3subscript𝑏2subscript𝑣2subscript𝑏3𝛿subscript𝑏𝐿2subscript𝑑𝐼𝛿subscript𝑏𝐿subscript𝑏𝑖superscriptsubscript𝑏𝑖𝑡subscript𝜀𝐻𝑀subscript𝑉𝐿𝜏3V_{HM}(\tau)=\left\langle-V_{L}\tau\delta(v_{3}b_{2}-v_{2}b_{3})\delta b_{L}/2% -d_{I}\delta(b_{L}b_{i})b_{i}^{\ast}\right\rangle_{t}=-\varepsilon_{HM}V_{L}% \tau/3italic_V start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT ( italic_τ ) = ⟨ - italic_V start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT italic_τ italic_δ ( italic_v start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT - italic_v start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) italic_δ italic_b start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT / 2 - italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT italic_δ ( italic_b start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) italic_b start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT ⟩ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT = - italic_ε start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT italic_V start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT italic_τ / 3, where VLsubscript𝑉𝐿V_{L}italic_V start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT is the averaged longitudinal velocity, τ𝜏\tauitalic_τ is the time lag, tsubscriptdelimited-⟨⟩𝑡\left\langle\cdot\right\rangle_{t}⟨ ⋅ ⟩ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT represents the time average, bisubscript𝑏𝑖b_{i}italic_b start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT is normalized by (4πρ)1/2superscript4𝜋𝜌12(4\pi\rho)^{-1/2}( 4 italic_π italic_ρ ) start_POSTSUPERSCRIPT - 1 / 2 end_POSTSUPERSCRIPT, and ρ𝜌\rhoitalic_ρ is the mass density. FIG. 5 (b) gives one example of linear scaling related to VHM1(τ)subscript𝑉𝐻𝑀1𝜏V_{HM1}(\tau)italic_V start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT ( italic_τ ), using data of days 49-59 in 2006. As shown, VHM1(τ)subscript𝑉𝐻𝑀1𝜏V_{HM1}(\tau)italic_V start_POSTSUBSCRIPT italic_H italic_M 1 end_POSTSUBSCRIPT ( italic_τ ) is linear among nearly two decades with εHM=8.9×1011Jm/(kgs)\varepsilon_{HM}=-8.9\times 10^{11}\textrm{J}\cdot\textrm{m/(kg}\cdot\textrm{s})italic_ε start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT = - 8.9 × 10 start_POSTSUPERSCRIPT 11 end_POSTSUPERSCRIPT J ⋅ m/(kg ⋅ s ).

Based on vKH relations, the turbulent space plasma can be recognized, and the magnetic helicity dissipation rates can be measured. Limited by the data resolution, the Hall effects on the inverse cascades cannot be directly recognized. Therefore, we decompose the numerical finding into two parts: the cross helicity generation, and the cascade efficiency suppression by cross helicity. Notably, in the actual solar wind, the cross helicity could also originate from other effects. In FIG. 5 (c), the conditional average of the cross helicity (HCsubscript𝐻𝐶H_{C}italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT) versus the magnetic helicity (HMsubscript𝐻𝑀H_{M}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT) is evaluated. Averaged over every 11 days, HMsubscript𝐻𝑀H_{M}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT and HCsubscript𝐻𝐶H_{C}italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT at the timescale δt(8min,4h)𝛿𝑡8min4h\delta t\in(8\,\textrm{min},4\,\textrm{h})italic_δ italic_t ∈ ( 8 min , 4 h ) are obtained by the Fourier expansion (Brandenburg et al., 2011; Matthaeus and Goldstein, 1982). The results at other timescales are consistent with our main findings and are given in SM (sup, ). As shown in FIG. 5 (c), when the Hall effects are relatively important (dIdItsubscript𝑑𝐼subscriptdelimited-⟨⟩subscript𝑑𝐼𝑡d_{I}\geq\left\langle d_{I}\right\rangle_{t}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ≥ ⟨ italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ⟩ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT), HCsubscript𝐻𝐶H_{C}italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT is strongly anticorrelated with HMsubscript𝐻𝑀H_{M}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT. In contrast, when dI<dItsubscript𝑑𝐼subscriptdelimited-⟨⟩subscript𝑑𝐼𝑡d_{I}<\left\langle d_{I}\right\rangle_{t}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT < ⟨ italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ⟩ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT, HCsubscript𝐻𝐶H_{C}italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT is relatively independent with HMsubscript𝐻𝑀H_{M}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT, except for the extreme event (HM/σ(HM)2greater-than-or-approximately-equalssubscript𝐻𝑀𝜎subscript𝐻𝑀2H_{M}/\sigma(H_{M})\gtrapprox 2italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT / italic_σ ( italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT ) ⪆ 2). Even if the data with an 8-min resolution is insufficient to resolve the Hall effects, the remarkable anticorrelation between HCsubscript𝐻𝐶H_{C}italic_H start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT and HMsubscript𝐻𝑀H_{M}italic_H start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT when dIdItsubscript𝑑𝐼subscriptdelimited-⟨⟩subscript𝑑𝐼𝑡d_{I}\geq\left\langle d_{I}\right\rangle_{t}italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ≥ ⟨ italic_d start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ⟩ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT imply that Hall effects and magnetic helicity could induce non-zero cross helicity. FIG. 5 (d) shows the conditional average of the normalized cascade rates ΓHM=|εHM|/[|δ𝐯||δ𝐛|2VLτ]subscriptΓ𝐻𝑀subscript𝜀𝐻𝑀delimited-[]𝛿𝐯superscript𝛿𝐛2subscript𝑉𝐿𝜏\varGamma_{HM}=|\varepsilon_{HM}|/[|\delta{\bf v}||\delta{\bf b}|^{2}V_{L}\tau]roman_Γ start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT = | italic_ε start_POSTSUBSCRIPT italic_H italic_M end_POSTSUBSCRIPT | / [ | italic_δ bold_v | | italic_δ bold_b | start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_V start_POSTSUBSCRIPT italic_L end_POSTSUBSCRIPT italic_τ ] versus the normalized cross helicity |cosθt|=|δ𝐯δ𝐛|/|δ𝐯||δ𝐛|subscript𝜃𝑡𝛿𝐯𝛿𝐛𝛿𝐯𝛿𝐛|\cos\theta_{t}|=|\delta{\bf v}\cdot\delta{\bf b}|/|\delta{\bf v}||\delta{\bf b}|| roman_cos italic_θ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT | = | italic_δ bold_v ⋅ italic_δ bold_b | / | italic_δ bold_v | | italic_δ bold_b |, where the distance r𝑟{r}italic_r is given by the middle scales with linear scaling. As shown, the normalized cascade rates are anticorrelated with |cosθt|subscript𝜃𝑡|\cos\theta_{t}|| roman_cos italic_θ start_POSTSUBSCRIPT italic_t end_POSTSUBSCRIPT |, consistent with our numerical findings.

In this letter, we perform six DNSs with grid points up to 20483superscript204832048^{3}2048 start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT to investigate Hall effects on helical MHD turbulence. We find that Hall effects could induce strong cross helicity by geometric alignments, which then strongly reduces inverse cascade efficiencies and leads to the accumulation of magnetic energy and helicity at large scales. The process is associated with the breaking of current sheets by filaments along magnetic fields. Then, we derive two vKH relations and verify them using the numerical and Ulysses data. The numerical findings about the cross helicity generation and the cascade efficiency suppression are then separately confirmed by the solar wind results. In the heliosphere, partial extreme solar wind events are generated by the eruption of the solar corona, where magnetic helicity and Hall effects both matter. Generally, one may believe that Hall effects are negligible in large-scale solar wind (Marino and Sorriso-Valvo, 2023). However, this letter gives a possible mechanism in solar wind turbulence that Hall effects at small scales could strongly affect large-scale magnetic fields through cross helicity.

Acknowledgements.
This work was supported by the National Key Research and Development Program of China (Grant Nos. 2019YFA0405300 and 2020YFA0711800) and NSFC Projects (Grant Nos. 91852203, 12072349, 92052102 and 12272006). The authors thank the Ulysses project for providing the solar wind data used in this study.

References