Revisiting LaMnO3: A density functional theory study

arXiv:2503.11189 · cond-mat.str-el, cond-mat.mtrl-sci · Submitted 2025-03-14 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: We're looking at the paper: "Revisiting LaMnO3: A density functional theory study".

Mira: Here's the gist of the paper. Density functional theory has been widely applied to a variety of realistic materials but often fails to accurately describe the properties of correlated systems <ref:2503.11189#pg2>.

Kai: In this study, using the prototypical transition metal oxide LaMnO3 as an example, we systematically assess the performance of two distinct DFT + U methods: spin-polarized DFT + U (SDFT + U) and spin-unpolarized charge-only DFT + U (CDFT + U) <ref:2503.11189#pg2>.

Mira: We found that, while the Coulomb U acts similarly in both approaches, the Hund’s JH plays a fundamentally different role, particularly in determining the correct magnetic phases <ref:2503.11189#pg5>.

Kai: Our investigation demonstrates the active role of JH in the exchange splitting, leading to distinct magnetic ground states based on the different schemes <ref:2503.11189#pg5>.

Title and authors: Kai: So we're looking at this paper called "Revisiting LaMnO3: A density functional theory study." It’s by Ju Hyeon Lee and Bongjae Kim from Kyungpook National University. The title suggests they are re-examining how density functional theory handles this material because it often fails for correlated systems, right?

Mira: Exactly. We're looking at a transition metal oxide, LaMnO3, which is a classic example of something that standard density functional theory methods struggle to describe accurately. The authors are testing two different ways to use the DFT plus U method: spin-polarized DFT plus U and spin-unpolarized charge-only DFT plus U.

Lev: From an engineering standpoint, testing two different implementations of a correction method is a solid way to check if the physics you're seeing is actually robust or just an artifact of your setup.

Kai: Right, because the whole point here isn't just to run one calculation, but to see how those differences in implementation affect what we think we know about this material. It’s about checking the methodology itself.

Mira: The authors are setting up a comparison between SDFT plus U and CDFT plus U to see how they handle the physics of LaMnO3 differently, especially concerning magnetic ordering.

Lev: So if one method gets the magnetic phase wrong, it tells you that your choice of functional setup matters a lot for predicting things like what kind of magnetism is stable.

The paper's summary: Kai: The core finding they present in "Revisiting LaMnO3: A density functional theory study" is that while the Coulomb interaction U behaves similarly in both SDFT plus U and CDFT plus U, the Hund’s JH term plays a fundamentally different role. This is key for figuring out which magnetic phases are correct.

Mira: I mean, they show that the way Hund's JH acts in these two schemes determines the correct magnetic phases, which is a big deal because standard DFT approaches often get this wrong when it comes to LaMnO3.

Lev: So what they're saying is that you can’t just rely on one DFT plus U setup; you need to be careful about how you introduce those specific parameters, like JH, because it dictates the final magnetic structure.

Kai: They specifically point out that their investigation demonstrates the active role of JH in the exchange splitting, which leads to distinct magnetic ground states depending on whether they use the SDFT plus U scheme or the CDFT plus U scheme.

Mira: That means even though both methods are DFT plus U, they aren't interchangeable for this material; one setup favors one magnetic state while the other favors a different one.

Lev: It’s like you’re using two different tools to measure the same thing, and the measurement depends entirely on which tool you pick.

The paper's improvements: Kai: The paper suggests an improvement by systematically assessing these two distinct DFT plus U methods—SDFT plus U and CDFT plus U—as a way to move beyond just using one approach blindly for correlated systems like LaMnO3.

Mira: They are showing that the comparison between SDFT plus U and CDFT plus U is important because it highlights how the double-counting energy expressions for both schemes differ, which directly impacts the results.

Lev: The improvement here is gaining a clearer understanding of *why* some methods succeed where others fail, by isolating the effect of those specific terms like JH versus U.

Kai: They also look at other exchange–correlation functionals like SCAN and HSE06 as alternatives to DFT plus U, and they show that SCAN can give better characterization of the electronic structure, capturing spin splitting better than standard LDA or PBE functionals do.

Mira: But they note a caveat too—for LaMnO3 specifically, even with the improved SCAN functional, you still need proper values for both U and JH to get the correct magnetic phases.

Lev: So while they improve the theory by comparing methods, they also point out that experimental input is still crucial; the theory needs those right parameters to match reality.

Conclusion: Kai: To wrap up on "Revisiting LaMnO3: A density functional theory study," the main implication is that accurately describing strongly correlated systems like LaMnO3 within DFT-based frameworks requires careful attention to how you implement the DFT plus U correction, especially regarding Hund's JH.

Mira: Essentially, they show that the choice between spin-polarized and spin-unpolarized charge-only DFT plus U methods creates different magnetic phase diagrams for LaMnO3, which is a significant finding because it shows implementation details matter.

Lev: From an error correction standpoint, this tells us that when we build real hardware simulations for these materials, we can't just pick any parameter set; the overhead of that choice directly influences whether your simulation settles on the right physical state.

Kai: Also, they give us a clearer picture of how electronic structure evolves with increasing Coulomb U and Hund's JH, showing how the majority eg orbital splits in both schemes consistently.

Mira: And they map out magnetic exchange couplings, showing that the sign of those exchanges changes at phase boundaries in ways that align with what we see experimentally.

Lev: So for future work, we need to ensure that whatever framework we use—SDFT plus U or CDFT plus U—it can correctly capture these subtle shifts in the exchange interactions as parameters change.

Kai: That's the gist of it. The paper "Revisiting LaMnO3: A density functional theory study" gives us a much deeper look into the nuances of DFT plus U for transition metal oxides like this one.

Mira: It really emphasizes that for complex materials, you have to test your implementation choices because they fundamentally change the magnetic picture you get.

Lev: So we’re going to keep an eye on how these different implementations perform when applied to more complex systems in real quantum hardware setups.

Kyungpook National University

cond-mat.str-el, cond-mat.mtrl-sci

Submitted: 2025-03-14

Updated: 2025-08-19

Journal ref: Curr. Appl. Phys. 78, 7-14 (2025)

DOI: 10.1016/j.cap.2025.06.010

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 81/100

The gist: Density functional theory has been widely applied to a variety of realistic materials but often fails to accurately describe the properties of correlated systems <ref:2503.11189#pg2>.

Key concepts

SDFT + U
This method uses spin-polarized DFT combined with the Hubbard U term. It considers both spin up and spin down electrons, which is crucial for describing magnetic properties in materials like LaMnO3.
CDFT + U
This method uses a charge-only DFT functional without explicit spin polarization, but still includes the Hubbard U term. It provides an alternative way to treat correlation effects by focusing on charge degrees of freedom.
Hund’s J (JH)
Hund's J describes the energy difference between different spin states within a single atom. It is found to play a fundamentally different role in determining magnetic phases compared to the Coulomb U interaction, specifically influencing exchange splitting.
Magnetic Phase Diagram
This visualization shows how the material's magnetic behavior changes as parameters like Coulomb U and Hund's J are varied. The study found that SDFT+U and CDFT+U predict different stable magnetic states (like A-AFM or FM) in this diagram.

Terminology

Summary

Density functional theory has been widely applied to a variety of realistic materials but often fails to accurately describe the properties of correlated systems <ref:2503. p. 0

Where the paper says it

Revisiting LaMnO3: A density functional theory study Ju Hyeon Lee and Bongjae Kim* Department of Physics, Kyungpook National University, Daegu 41566, Republic of Korea *E-mail: bongjae@knu. kr Abstract Density functional theory has been widely applied to a variety of realistic materials but often fails to accurately describe the properties of correlated systems.

11189#pg2>. In this study, using the prototypical transition metal oxide LaMnO3 as an example, we systematically assess the performance of two distinct DFT + U methods: spin-polarized DFT + U (SDFT + U) and spin-unpolarized charge-only DFT + U (CDFT + U) <ref:2503. p. 0
Where the paper says it

However, the methodological details of DFT + U and its specific impact on the electronic structure and magnetic properties of correlated systems remain incompletely understood. In this study, using the prototypical transition metal oxide LaMnO3 as an example, we systematically assess the performance of two distinct DFT + U methods: spin-polarized DFT + U (SDFT + U) and spin-unpolarized charge-only DFT + U (CDFT + U).

11189#pg2>. We found that, while the Coulomb U acts similarly in both approaches, the Hund’s JH plays a fundamentally different role, particularly in determining the correct magnetic phases <ref:2503. p. 0
Where the paper says it

In this study, using the prototypical transition metal oxide LaMnO3 as an example, we systematically assess the performance of two distinct DFT + U methods: spin-polarized DFT + U (SDFT + U) and spin-unpolarized charge-only DFT + U (CDFT + U). We found that, while the Coulomb U acts similarly in both approaches, the Hund’s JH plays a fundamentally different role, particularly in determining the correct magnetic phases.

11189#pg5>. Our investigation demonstrates the active role of JH in the exchange splitting, leading to distinct magnetic ground states based on the different schemes <ref:2503. p. 0
Where the paper says it

We found that, while the Coulomb U acts similarly in both approaches, the Hund’s JH plays a fundamentally different role, particularly in determining the correct magnetic phases. Our investigation demonstrates the active role of JH in the exchange splitting, leading to distinct magnetic ground states based on the different schemes.

11189#pg5>.

Comparison of DFT + U Implementations

The study compared two distinct implementations of DFT + U: SDFT + U and CDFT + U <ref:2503.11189#pg3>. The differences between these schemes stem from how the Hubbard U term is applied to the spin-polarized charge functional in SDFT + U, whereas CDFT + U uses a spin-unpolarized charge functional <ref:2503. p. 2

Where the paper says it

In SDFT + U, the exchange–correlation energy depends on the spin-resolved charge densities, whereas in CDFT + U, it depends only on the total charge density. Accordingly, the charge density functionals in the two schemes differ: in SDFT + U, the Hubbard U term is added on top of the spin-polarized charge functional, while in CDFT + U, the spin-unpolarized charge functional is used.

11189#pg3>. These differences result in distinct double-counting energy expressions for both schemes <ref:2503. p. 2
Where the paper says it

The doublecounting energy is expressed as follows: 𝑬𝐝𝐜 𝐒𝐃𝐅𝐓 = 𝑼 𝟐 𝒏(𝒏 − 𝟏) − 𝑱𝑯 𝟐 +𝒏𝝈(𝒏𝝈 − 𝟏) ↑,↓ 𝝈 (𝟏) 𝑬𝒅𝒄 𝑪𝑫𝑭𝑻 = 𝑼 𝟐 𝒏(𝒏 − 𝟏) − 𝑱𝑯 𝟐 𝒏, 𝒏 𝟐 − 𝟏- (𝟐) here, 𝐸23 4567(9567) denotes the double-counting energy in the spin-(un)polarized functional, n is the occupation number (density matrix), and σ is the spin channel. Differences in the double counting correction, hence different energy corrections, separates two DFT + U schemes.

11189#pg3>. Furthermore, the role of Hund’s JH differs significantly between the two schemes, which is a key finding <ref:2503. p. 2
Where the paper says it

Differences in the double counting correction, hence different energy corrections, separates two DFT + U schemes. Also, the role of Hund’s JH differs significantly between the two schemes as we will discuss in detail below.

11189#pg5>.

Magnetic Phase Diagram Analysis

The magnetic phase diagrams obtained using SDFT + U and CDFT + U schemes reveal fundamental differences <ref:2503.11189#pg5>. For the SDFT + U scheme, the A-AFM ground state is stabilized in the small-U and small-JH regime <ref:2503. p. 3

Where the paper says it

The phase diagram reveals the fundamental differences between the two DFT + U schemes. In the case of SDFT + U scheme, the A-AFM ground state is stabilized in the small-U and small-JH regime [Fig.

11189#pg5>. Increasing Coulomb U drives the system toward a ferromagnetic (FM) ordering, while an increase in Hund’s JH stabilizes the G-type antiferromagnetic (G-AFM) ordering <ref:2503. p. 3
Where the paper says it

As U and JH increase, the magnetic configuration evolves in distinct ways. Increasing the Coulomb U drives the system toward a ferromagnetic (FM) ordering, while an increase in Hund’s JH stabilizes the G-type antiferromagnetic (G-AFM) ordering.

11189#pg5>. In contrast, the CDFT + U scheme exhibits a G-AFM ground state for small U and JH <ref:2503. p. 3
Where the paper says it

In contrast, the magnetic phase diagram obtained using the CDFT + U scheme is markedly different. In this case, the system exhibits a G-AFM ground state for small U and JH regime.

11189#pg5>. Increasing either U or JH promotes FM interactions between Mn sites, leading to a transition into an A-AFM phase <ref:2503. p. 3
Where the paper says it

In this case, the system exhibits a G-AFM ground state for small U and JH regime. Increasing either U or JH promotes FM interactions between Mn sites, leading to a transition into an A-AFM phase, in which in-plane interactions become ferromagnetic.

11189#pg5>. The phase space exhibiting the correct magnetic structure, A-AFM, is significantly narrower in the CDFT + U case—except for SCAN functional—than in the SDFT + U case <ref:2503. p. 3
Where the paper says it

As we discuss in detail below, the response of the electronic structure to Hund’s JH differs significantly between the two schemes, which accounts for the distinct phase diagrams. Importantly, the phase space exhibiting the correct magnetic structure, A-AFM, is significantly narrower in the CDFT + U case--except for SCAN functional--than in the SDFT + U case, illustrating the nontrivial magnetic behavior of LMO.

11189#pg5>.

Electronic Structure Evolution

The evolution of the electronic structure as a function of Coulomb U and Hund’s JH shows that increasing U enlarges the Mott-type gap in both schemes <ref:2503.11189#pg6>. The partially filled majority eg orbital progressively splits with increasing U, a behavior consistent across both SDFT + U and CDFT + U methods <ref:2503. p. 4

Where the paper says it

While there are minor differences in the projected density of states (PDOS), the overall electronic structure exhibits a similar trend: the unoccupied orbitals, particularly in the minority spin channel, move to higher energy levels. The partially filled majority eg orbital progressively splits with increasing U.

11189#pg7>. Regarding Hund’s JH, in the SDFT + U method, increasing JH effectively reduces the strength of the Coulomb interaction through the renormalization Ueff = U − JH <ref:2503. p. 5
Where the paper says it

In the SDFT + U method, the spin splitting is already pronounced, and the role of JH, which is expected to increase the total spin moment, is rather limited. Moreover, increasing JH effectively reduces the strength of the Coulomb interaction through the renormalization Ueff = U − JH.

11189#pg7>. In contrast, CDFT + U shows an active role for Hund’s JH, leading to a pronounced increase in spin splitting between majority and minority t2g states <ref:2503. p. 9
Where the paper says it

In CDFT + U, exchange splitting between majority and minority d-bands is small; even at U = 6 eV, without JH, the Mn d-states remain in a low-spin configuration [Fig. Upon inclusion of JH, we observe a clear increase in exchange splitting, highlighting the active role played by Hund’s JH in CDFT + U [15,16].

11189#pg7>.

Magnetic Exchange Couplings

Mapping the energetics of magnetic orders onto a Heisenberg-type spin Hamiltonian allows for the extraction of magnetic exchange interactions like Jab and Jc <ref:2503.11189#pg9>. The sign of the magnetic exchanges changes at around the phase boundaries, aligning well with transitions observed in Fig. p. 7

Where the paper says it

As U increases, regardless of JH, the interaction strengths generally become more negative, indicating stronger FM interaction. The sign of the exchange parameters change at around the phase boundaries, which aligns well with the transitions observed in Fig.

2 <ref:2503. p. 2
Where the paper says it

We found that CDFT scheme yields a slightly smaller volume compared with SDFT scheme. However, to avoid additional complications arising from the structural degree of freedom, we adopted the experimental structures for all our calculations, as reported in ref.

p. 2
Where the paper says it

We found that CDFT scheme yields a slightly smaller volume compared with SDFT scheme. However, to avoid additional complications arising from the structural degree of freedom, we adopted the experimental structures for all our calculations, as reported in ref.

p. 2
Where the paper says it

We found that CDFT scheme yields a slightly smaller volume compared with SDFT scheme. However, to avoid additional complications arising from the structural degree of freedom, we adopted the experimental structures for all our calculations, as reported in ref.

p. 2
Where the paper says it

We found that CDFT scheme yields a slightly smaller volume compared with SDFT scheme. However, to avoid additional complications arising from the structural degree of freedom, we adopted the experimental structures for all our calculations, as reported in ref.

11189#pg10>. In the SDFT + U case, increasing JH enhances AFM tendencies through overall positive signs of the exchanges <ref:2503. p. 8
Where the paper says it

2, the contrasting roles of JH are further confirmed in our analysis of exchange interactions. In the SDFT + U, increasing JH enhances AFM tendencies, as evident from the overall positive sign of the exchanges.

11189#pg10>. Conversely, in CDFT + U, increasing JH drives the exchanges toward negative values, favoring FM <ref:2503. p. 8
Where the paper says it

In the SDFT + U, increasing JH enhances AFM tendencies, as evident from the overall positive sign of the exchanges. This is the opposite for CDFT + U, where increasing JH drives the exchanges toward negative values, favoring FM.

11189#pg10>.

Comparison with Advanced Functionals

The study compared the performance of SCAN and HSE06 approaches as alternatives to DFT + U <ref:2503.11189#pg5>. The SCAN functional shows enhanced characterization of the electronic structure, better capturing spin splitting than LDA, PBE, and PBEsol functionals <ref:2503. p. 2

Where the paper says it

For the exchange–correlation functional, the LDA and two types of GGA functionals, Perdew–Burke– Ernzerhof (PBE) [18] and PBE revised for solids (PBEsol) [19] were employed. We also included the strongly constrained and appropriately normed (SCAN) meta-GGA functional [20] and the Heyd– Scuseria–Ernzerhof (HSE06) hybrid functional [21].

11189#pg9>. However, for LMO, the SCAN functional still requires proper values of U and JH for the correct description of magnetic phases <ref:2503. p. 3
Where the paper says it

Importantly, the phase space exhibiting the correct magnetic structure, A-AFM, is significantly narrower in the CDFT + U case--except for SCAN functional--than in the SDFT + U case, illustrating the nontrivial magnetic behavior of LMO. This highlights the challenges in the accurate description of the magnetism of LMO within DFT framework [11-14].

11189#pg5>. Hybrid functionals like HSE06 observe a very strong spin splitting, corresponding to a U value of 7. p. 2
Where the paper says it

For the exchange–correlation functional, the LDA and two types of GGA functionals, Perdew–Burke– Ernzerhof (PBE) [18] and PBE revised for solids (PBEsol) [19] were employed. We also included the strongly constrained and appropriately normed (SCAN) meta-GGA functional [20] and the Heyd– Scuseria–Ernzerhof (HSE06) hybrid functional [21].

0 eV or more within the CDFT + U scheme <ref:2503. p. 2
Where the paper says it

We found that CDFT scheme yields a slightly smaller volume compared with SDFT scheme. However, to avoid additional complications arising from the structural degree of freedom, we adopted the experimental structures for all our calculations, as reported in ref.

11189#pg5>.

Conclusion

Overall, our study highlights the challenges in accurately describing strongly correlated systems such as LMO within DFT-based frameworks <ref:2503.11189#pg5>. The findings emphasize the nontrivial nature of these systems and provide valuable insight for future computational studies of related materials <ref:2503.11189#pg5>.

Acknowledgments

This work was supported by Kyungpook National University Research Fund, 2023 <ref:2503.11189#pg5>.

--- Page 6 ---

Fig. 6 shows the exchange couplings as functions of Coulomb U for both the SDFT + U and CDFT + U schemes <ref:2503. p. 7

Where the paper says it

2 from exchange couplings. 6 shows the exchange couplings as functions of Coulomb U for both the SDFT + U and CDFT + U schemes.

11189#pg10>. The sign of the exchange parameters change at around the phase boundaries, which aligns well with transitions observed in Fig. p. 7
Where the paper says it

As U increases, regardless of JH, the interaction strengths generally become more negative, indicating stronger FM interaction. The sign of the exchange parameters change at around the phase boundaries, which aligns well with the transitions observed in Fig.

2 <ref:2503.11189#pg10>. In the SDFT + U case, we observe a crossover of Jc from positive to negative near U = 2. p. 7
Where the paper says it

The sign of the exchange parameters change at around the phase boundaries, which aligns well with the transitions observed in Fig. In the SDFT + U case, we observe a crossover of Jc from positive to negative near U = 2.

5 eV for JH = 0 eV (see red curve), which corresponds exactly to the A-AFM to FM transition in Fig. p. 7
Where the paper says it

In the SDFT + U case, we observe a crossover of Jc from positive to negative near U = 2. 5 eV for JH = 0 eV (see red curve), which corresponds exactly to the A-AFM to FM transition in Fig.

2 <ref:2503.11189#pg10>. The critical U for the Jc sign change increases systematically with JH, mirroring the trend seen in the SDFT + U phase diagram <ref:2503. p. 7
Where the paper says it

5 eV for JH = 0 eV (see red curve), which corresponds exactly to the A-AFM to FM transition in Fig. The critical U for the Jc sign change increases systematically with JH, mirroring the trend seen in the SDFT + U phase diagram.

11189#pg10>. In contrast, for CDFT + U, when JH = 0 eV [Fig. p. 5
Where the paper says it

In contrast, CDFT + U scheme produces minimal spin splitting. These differences are clearly visible in the JH = 0 eV cases shown in Fig.

4(b)], the positive Jab (blue and black curves) and Jc (red curve), which support the G-AFM phase, decrease but remain positive up to U = 8 eV <ref:2503. p. 7
Where the paper says it

In contrast, for CDFT + U, when JH = 0 eV [Fig. 4(b)], the positive Jab (blue and black curves) and Jc (red curve), which support the G-AFM phase, decrease but remain positive up to U = 8 eV.

11189#pg10>. The system remains FM across the entire range, consistent with Fig. p. 7
Where the paper says it

4(b)], the positive Jab (blue and black curves) and Jc (red curve), which support the G-AFM phase, decrease but remain positive up to U = 8 eV. The system remains FM across the entire range, consistent with Fig.

2 <ref:2503.11189#pg10>. For JH = 0.5 eV, the sign of Jab changes to negative at around U = 4. p. 7
Where the paper says it

As U increases, regardless of JH, the interaction strengths generally become more negative, indicating stronger FM interaction. The sign of the exchange parameters change at around the phase boundaries, which aligns well with the transitions observed in Fig.

5 eV, and one obtains an A-AFM phase <ref:2503. p. 8
Where the paper says it

negative at around U = 4. 5 eV, and one obtains an A-AFM phase.

11189#pg10>. Jc also decreases but remains positive, so the system is A-AFM even for large U <ref:2503. p. 8
Where the paper says it

5 eV, and one obtains an A-AFM phase. Jc also decreases but remains positive, so the system is A-AFM even for large U.

11189#pg10>. For JH = 1.0 eV, Jab is negative throughout, and Jc becomes negative at around U = 2. p. 8
Where the paper says it

For JH = 1. 0 eV, Jab is negative throughout, and Jc becomes negative at around U = 2.

5 eV, resulting in a transition to G-AFM <ref:2503. p. 3
Where the paper says it

As U increases, the out-of-plane Mn–Mn interactions favor ferromagnetism, causing a transition from A-AFM to FM. In contrast, increasing JH progressively stabilizes antiferromagnetic interactions between nearestneighbor Mn sites, resulting in a transition from FM through A-AFM to G-AFM.

11189#pg10>. For even higher JH, the system remains as FM in the CDFT + U scheme <ref:2503. p. 8
Where the paper says it

5 eV, resulting in a transition to G-AFM. For even higher JH, the system remains as FM in the CDFT + U scheme.

11189#pg10>.

--- Page 7 ---

Fig. 7 shows the spin-polarized charge density is plotted over the energy range from 0 eV to −1 eV, with the Fermi energy set to 0 eV <ref:2503. p. 2

Where the paper says it

In SDFT + U, the exchange–correlation energy depends on the spin-resolved charge densities, whereas in CDFT + U, it depends only on the total charge density. Accordingly, the charge density functionals in the two schemes differ: in SDFT + U, the Hubbard U term is added on top of the spin-polarized charge functional, while in CDFT + U, the spin-unpolarized charge functional is used.

11189#pg7>. The Coulomb interaction U was set to 6 eV, and Hund’s JH was set to 1 eV <ref:2503. p. 3
Where the paper says it

First, we discuss the phase diagrams obtained using two distinct DFT + U schemes. 2 presents the ground-state magnetic phase diagrams for four different exchange-correlation functionals—LDA, PBE, PBEsol, and SCAN—constructed as based on the two Hubbard interaction parameters: Coulomb U and Hund’s JH.

11189#pg7>.

--- Page 8 ---

Fig. 3 shows the Projected density of states calculated using the PBE functional is shown for SDFT + U (a) and (a) SDFT + U (b) CDFT + U <ref:2503. p. 5

Where the paper says it

This behavior is markedly different to the case of SDFT + U. Projected density of states calculated using the PBE functional is shown for SDFT + U (a) and (a) SDFT + U (b) CDFT + U

11189#pg7>.

--- Page 9 ---

Fig. 4 shows the Projected density of states calculated using the PBE functional for SDFT + U (a) and CDFT + U (b) <ref:2503. p. 5

Where the paper says it

This behavior is markedly different to the case of SDFT + U. Projected density of states calculated using the PBE functional is shown for SDFT + U (a) and (a) SDFT + U (b) CDFT + U

11189#pg8>. The Coulomb interaction parameter U was fixed at 6 eV <ref:2503. p. 6
Where the paper says it

Projected density of states calculated using the PBE functional for SDFT + U (a) and CDFT + U (b). The Coulomb interaction parameter U was fixed at 6 eV.

p. 6
Where the paper says it

Projected density of states calculated using the PBE functional for SDFT + U (a) and CDFT + U (b). The Coulomb interaction parameter U was fixed at 6 eV.

11189#pg8>.

--- Page 9 ---

Fig. 5 shows the Projected density of states calculated within the SDFT + Ueff scheme (Ueff = U − JH) using the PBE functional <ref:2503. p. 6

Where the paper says it

The color coding is the same as in Fig. Projected density of states calculated within the SDFT + Ueff scheme (Ueff = U − JH) using the PBE functional.

11189#pg7>. The Coulomb interaction parameter U was fixed at 6 eV <ref:2503.11189#pg8>.

--- Page 9 ---

Fig. 8 shows the Projected density of states calculated using the LDA, PBE, PBEsol, SCAN, and HSE06 functionals <ref:2503. p. 2

Where the paper says it

For the exchange–correlation functional, the LDA and two types of GGA functionals, Perdew–Burke– Ernzerhof (PBE) [18] and PBE revised for solids (PBEsol) [19] were employed. We also included the strongly constrained and appropriately normed (SCAN) meta-GGA functional [20] and the Heyd– Scuseria–Ernzerhof (HSE06) hybrid functional [21].

11189#pg5>. Red and blue lines represent the Mn-eg and t2g orbitals, while the O-p orbital is shown in purple <ref:2503. p. 6
Where the paper says it

The Hund’s JH was fixed at 1 eV. Red and blue lines represent the Mn-eg and t2g orbitals, respectively, while the O-p orbital is shown in purple.

11189#pg5>.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved handling of correlated systems through DFT U methods allows AI to accurately predict magnetic ground states in transition metal oxides like LaMnO3 by systematically comparing spin-polarized and spin-unpolarized charge-only approaches.

  2. The ability to map magnetic exchange onto a Heisenberg type Hamiltonian, using equations (4) and (5), enables the AI system to extract effective exchange parameters Jab and Jc from calculated total energies, leading to better understanding of magnetic interactions.

  3. The comparison between CDFT U and SDFT U phase diagrams reveals that the CDFT U scheme exhibits a G-AFM ground state for small U and JH regime, allowing the AI to distinguish subtle magnetic phase stability differences based on the implementation details of DFT+U.

  4. The analysis of electronic structure evolution, specifically how unoccupied Mn-3d orbitals shift with increasing Coulomb U, allows the AI to predict the enlargement of Mott-type gaps in both SDFT U and CDFT U schemes consistently across different physical scenarios.

  5. The comparison between exchange coupling parameters from different magnetic configuration sets shows that Jab and Jc derived from different configurations (e.g., C-AFM and FM, or G-AFM and A-AFM) yield slightly different values, enabling the AI to identify potential limitations in simplified magnetic models.

  6. The assessment of SCAN and HSE functionals demonstrates that the correct magnetic configuration is consistently found over a much wider range of U values when using sizable Hund's JH, indicating the AI can prioritize or select functionals based on their robustness across parameter variations.

Abstract

Density functional theory (DFT) has been widely applied to a variety of realistic materials but often struggles to explain the properties of correlated systems. The DFT + U method, which introduces a Hubbard U correction to the DFT, has been instrumental in providing the treatment of systems such as transition metal oxide. The methodological details of DFT + U and their specifics on the electronic structures and magnetic properties of the correlated systems remain incompletely understood. In this study, taking the prototypical transition metal oxide system, LaMnO3, as an example, we systematically assess the performance of the two distinct DFT + U methods, spin-polarized DFT (SDFT + U) and spin-un-polarized DFT (CDFT + U). We found that while Coulomb U acts similarly for the two approaches, Hund JH plays a fundamentally different role, particularly in the determination of the magnetic phases. Our investigation shows the active role of Hund JH on the exchange splitting, leading to distinct magnetic ground configurations in the two methods. We further investigate the associated magnetic exchange interactions and compare our results with so-called beyond-DFT methods.

Related papers