Tripoli University

PhD. Applied Science Program, Physics, DFT

Density Functional Theory

(DFT)
By: Ibrahim Mustafa Abdulrahaman

Proof. Dr. Abdalla M. Al-kelsh

(Spring 2025)
 A Comprehensive Overview of DFT Theory and Applications

DFT remains a cornerstone of computational materials science in 2024

with publication volumes doubling every 5-6 years
Presentation Roadmap:
Foundations → Approximations → Applications → Challenges
 Before DFT: Thomas-Fermi Model and Its Limitations
The Thomas-Fermi model (1927) was the first attempt to
describe electronic structure using electron density rather than
wavefunctions.

E[n] = CF ∫ n(r) 5/3 dr + ∫ V ext (r) n(r) dr + ½ ∫∫ n(r)n(r')/|r-r'| dr

dr'
Thomas-Fermi Total Energy Functional

Historical Significance:

• First density-based approach to electronic structure

• Precursor to modern Density Functional Theory
• Introduced the concept of kinetic energy as a functional of density

Major Limitations:

• Fails catastrophically for molecular bonding

• ~50% error in cohesive energies
• Cannot describe chemical bonds or shell structure
• No exchange-correlation effects included
 Hohenberg-Kohn Theorems (1964) - First Theorem
HK Theorem I: Unique Correspondence
The external potential V ext (r) is uniquely determined by the ground-state electron density n(r).
V ext (r) n(r)

Proof Strategy (Levy-Lieb Constrained Search):

• Proof by contradiction: assume two different V ext give same n(r)

• Use variational principle to show this leads to contradiction
• Establishes uniqueness of V ext n(r) mapping

Fundamental Significance:

• Establishes electron density as the fundamental variable

• Reduces 3N-dimensional problem to 3-dimensional
• Foundation of all density functional theory

Reference: Hohenberg, P. & Kohn, W. Phys. Rev. 136, B864 (1964). DOI: 10.1103/PhysRev.136.B864
 Hohenberg-Kohn Theorems - Second Theorem
(Variational Principle)

HK Theorem II: Variational Principle

For any trial density ñ(r), the energy calculated from the universal functional F[n] satisfies:

E[ñ] ≥ E₀

where E₀ is the true ground-state energy

Universal Functional F[n]:

• F[n] = T[n] + V ee [n] (kinetic + electron-electron interaction)

• Independent of external potential V ext

• Same for all N-electron systems

• Total energy: E[n] = F[n] + ∫ V ext (r)n(r)dr

Practical Significance:

• Provides a method to find ground-state density

 From HK to KS: The Complete Theorem
The Challenge: While Hohenberg-Kohn theorems establish that the ground
state energy is a functional of density, they don't provide a practical way to
calculate it.

The Kohn-Sham Approach (1965)

Map the interacting many-body problem onto an auxiliary system of non-interacting

electrons moving in an effective potential, chosen to reproduce the same ground state
density.

Key Insight: The kinetic energy can be split into:

T[n] = Ts [n] + T c[n]

where Ts[n] is the kinetic energy of non-interacting electrons with density n(r),
and T c[n] is a correction term.

Proof of Existence: For any interacting system with density n(r), there exists a
unique non-interacting system with the same density, subject to an effective
potential V eff (r).

Practical Impact: Transforms the many-body problem into a set of single-

 Kohn-Sham Equations
Kohn and Sham (1965) introduced a practical approach to DFT by mapping the interacting
electron system to a non-interacting one with the same density [1]

[1] Kohn, W. & Sham, L. J. Phys. Rev. 140, A1133 (1965)

Self-consistent field cycle in Kohn-Sham DFT Slide 4

 Kohn-Sham Equations in Detail

[-½∇² + V eff(r)]φi(r) = εiφi(r)

Veff(r) = Vext(r) + VH(r) + Vxc(r)

Components of the Effective Potential:

• V ext (r): External potential (nuclei + applied fields)

• V H (r): Hartree potential (classical electron-electron)
VH (r) = ∫ n(r')/|r-r'| dr'
• V xc (r): Exchange-correlation potential
Vxc (r) = δE xc [n]/δn(r)

Self-Consistent Field (SCF) Algorithm:

1. Start with initial guess for n(r)

2. Construct V eff (r)
3. Solve KS equations for φ i(r)
4. Calculate new density: n(r) = Σ i |φ i(r)|²
5. If not converged, return to step 2

Reference: Kohn, W. & Sham, L.J. Phys. Rev. 140, A1133 (1965). DOI: 10.1103/PhysRev.140.A1133
 Total Energy Functional in DFT
The total energy functional in DFT combines several
components [1] :

E[n] = Ts [n] + E H [n] + E xc [n] + ∫V ext(r)n(r)dr

T s [n]: Kinetic energy of non-interacting electrons

E H [n]: Hartree (Coulomb) energy

E xc [n]: Exchange-correlation energy

∫V ext (r)n(r)dr: External potential energy

The exchange-correlation term E xc [n] contains all many-body

effects and is the only term requiring approximation [2] .
Relative contributions of energy components in silicon crystal
(Data from Martin, R. M. Electronic Structure, Cambridge University Press, 2004)
[1] Kohn, W. & Sham, L. J. Phys. Rev. 140, A1133 (1965)
[2] Jones, R. O. & Gunnarsson, O. Rev. Mod. Phys. 61, 689 (1989)

Slide 5
 The Exact Definition of E xc

Exchange-Correlation Energy Definition

The exchange-correlation energy E xc accounts for all many-body effects not
captured by the non-interacting kinetic energy and classical electrostatic
terms:

Exc [n] = (T[n] - T s [n]) + (E ee [n] - E H[n])

Kinetic correlation + Non-classical electron-electron interaction

Interpretation via Fluctuation-Dissipation Theorem:

• Links Excto the response of the system to external perturbations

• Provides rigorous framework for developing approximations
• Connects to frequency-dependent response functions

Components of E xc :
• Exchange: Arises from Pauli exclusion principle
• Correlation: Accounts for instantaneous electron-electron interactions
• Together they create the exchange-correlation hole around each electron
Reference: Burke, K. et al. Rev. Mod. Phys. 87, 897 (2015). DOI: 10.1103/RevModPhys.87.897
 Jacob's Ladder of Approximations
Exchange-correlation functionals in DFT are organized in a
hierarchy of increasing accuracy and complexity [1] :

LDA - Local Density Approximation

1
Depends only on density n(r)

GGA - Generalized Gradient Approximation

2
Adds density gradient ∇n(r) (e.g., PBE)

meta-GGA - Meta-GGA
3
Adds kinetic energy density (e.g., SCAN)

Hybrids - Hybrid Functionals

4
Include exact exchange (e.g., PBE0, HSE06)

Jacob's Ladder of DFT approximations showing increasing accuracy and computational

Beyond - Advanced Methods
5 cost [2]
RPA, double hybrids, van der Waals, TDDFT

[1] Perdew, J. P. & Schmidt, K. AIP Conference Proceedings 577, 1 (2001)

[2] Burke, K. J. Chem. Phys. 136, 150901 (2012)
Slide 6
 LDA: Limitations and Successes
Local Density Approximation (LDA): The exchange-correlation energy
depends only on the local value of the electron density.

Exc LDA [n] = ∫ n(r)ε xc(n(r)) dr

Perdew-Zunger (PZ81) parameterization based on quantum Monte Carlo calculations

Successes:

• Accurate lattice constants for simple metals and semiconductors (<5% error)
• Good description of equilibrium geometries
• Computationally efficient and robust

Limitations:

• Overestimates binding energies (overbinding)

• Underestimates band gaps in semiconductors and insulators
• Poor description of weakly bound systems
• Cannot describe van der Waals interactions

Reference: Perdew, J.P. & Zunger, A. Phys. Rev. B 23, 5048 (1981). DOI: 10.1103/PhysRevB.23.5048
 GGA: PBE & PW91 & revPBE

Generalized Gradient Approximation (GGA)

GGAs improve upon LDA by incorporating the gradient of the electron density:

Exc GGA [n] = ∫ n(r) ε xc (n(r), ∇n(r)) dr

Key GGA Functionals:

• PW91 (Perdew-Wang 1991): First widely used GGA

• PBE (Perdew-Burke-Ernzerhof 1996): Simplified version of PW91, satisfies more exact
conditions
GGA Key Features:
• revPBE : Revised PBE with improved adsorption energetics
Includes density gradient ∇n(r)
Improvements over LDA: Semi-local approximation
Computationally efficient
• Better description of binding energies and bond lengths Good balance of accuracy vs. cost
• Improved atomization energies (errors reduced by ~50%)
• More accurate structural properties of solids
• Still underestimates band gaps, but less severely than LDA
 meta-GGA & SCAN
meta-GGA Functionals: The next step up Jacob's ladder of DFT approximations,
incorporating the kinetic energy density (τ) of the Kohn-Sham orbitals in addition
to the electron density (n) and its gradient (∇n).

SCAN (Strongly Constrained and Appropriately Normed)

The first meta-GGA that satisfies all 17 known exact constraints that a meta-GGA can satisfy,
developed by Sun, Ruzsinszky, and Perdew (2015).

Key Features of SCAN:

• Incorporates the Laplacian τ(r) = ½Σ i|∇φ i|²

• Satisfies all 17 known exact constraints for meta-GGAs
• Appropriately normed to reference systems ("appropriate norms")
• Significant improvement in surface energies

Performance Improvements:

• Remarkable accuracy for systems with localized exchange-correlation holes

• Improved lattice constants compared to GGAs
• Better description of weak interactions
• Enhanced accuracy for diverse bonding situations
 Hybrid Functionals (HSE06, PBE0)

Hybrid Functional Approach

Hybrid functionals incorporate a portion of exact Hartree-Fock (HF)

exchange with density functional approximations:

Exc hybrid = αE x HF + (1-α)E x DFT + E c DFT

α is the mixing parameter (typically 0.25)

Key Hybrid Functionals:

• HSE06: Screened hybrid that applies HF exchange only for short-range interactions
• PBE0: Global hybrid with 25% HF exchange mixed with 75% PBE exchange

Advantages:

• Significantly reduces band gap error to <0.3 eV

• Better describes localized states (d and f electrons)
• Improves description of charge transfer and excitations
• Partially corrects self-interaction error

Reference: Heyd, J., Scuseria, G.E. & Ernzerhof, M. J. Chem. Phys. 118, 8207 (2003). DOI: 10.1063/1.1564060
 DFT Accuracy Metrics: 50 Materials Dataset
Benchmark Dataset Overview

A comprehensive evaluation of DFT accuracy using a carefully

curated dataset of 50 materials, focusing on four critical properties
that determine material behavior and applications.

Key Properties Evaluated:

a₀ (Lattice Constant): Fundamental structural parameter

B (Bulk Modulus): Mechanical response to pressure

E_coh (Cohesive Energy): Thermodynamic stability

Reproducibility analysis across different DFT codes and functionals
E_gap (Band Gap): Electronic properties

Mean Absolute Error (MAE) Results:

The benchmark reveals significant variations in accuracy across different exchange-

correlation functionals, with hybrid functionals generally outperforming GGA and LDA for
band gaps, while GGA functionals show excellent performance for structural properties.

Reference: DOI: 10.1126/science.aad3000 - "Reproducibility in density functional theory calculations of solids"

 Example: Band Gap in MoS₂
Monolayer MoS₂ is a direct band gap semiconductor with
promising applications in nanoelectronics. DFT calculations
with different approximations yield varying results [1] :

Why GGA underestimates band gaps:

• Incorrect treatment of electron self-interaction
• Missing derivative discontinuity in exchange-correlation Band structure of monolayer MoS₂ showing direct gap at K point

potential
• Delocalization error in the functional [2]
Slide 7
 Electronic Band Structures
Comparing Simple Metal (Cu)
Copper (Cu): Simple Metal

Fermi Surface
Band structure showing free-electron-like behavior

Density of States

Key Features: DFT accurately captures s-p hybridized bands, d-bands below Fermi
level, and characteristic Fermi surface.
 Self-interaction & Derivative Discontinuity
Self-Interaction Error (SIE) Derivative Discontinuity

In standard DFT functionals, an electron incorrectly interacts with The exact exchange-correlation potential exhibits a discontinuous
its own contribution to the electron density, leading to derivative when the electron number crosses an integer value,
delocalization errors and underestimated band gaps. which is missing in standard functionals.

Perdew-Zunger Correction:

The Perdew-Zunger approach subtracts the self-interaction term

for each orbital, significantly improving the description of localized
states.

SIE Effects on Properties:

Property LDA LDA+SIC Improvement

Band Gap
0.2 eV 3.7 eV +3.5 eV
(NiO) Comparison of exact vs. approximate exchange-correlation potentials

Ionization Near
Underestimated Significant Modern Solutions:
Energy Exact
Hybrid Functionals: Partial exact exchange reduces SIE
 Case Study: Monolayer MoS₂

Crystal Structure & Properties Electronic Properties Comparison

Crystal System: Hexagonal

Space Group: P6/mmm

DFT-PBE GW BSE
Lattice Parameter: 3.18 Å
1.64 2.86 1.88
eV eV eV (Exciton)
Thickness: 6.5 Å

Cohesion Energy: -4.86 eV

Direct Band Gap: Yes (K-point)

 Selected References

Key Textbooks & Reviews Hot Topics in DFT Research (2024)

Foundational Textbooks
Machine Learning Enhanced DFT

Electronic Structure: Basic Theory and Practical Methods

Richard M. Martin Keywords: Neural networks, ML potentials, Functional optimization
Cambridge University Press (2020), 2nd Edition

2D Materials & van der Waals Systems

Electronic Structure: Basic Theory and Practical Methods
Richard M. Martin
Cambridge University Press (2004), 1st Edition
Keywords: vdW corrections, Moiré patterns, Twisted bilayers

Seminal Papers
Real-Time TDDFT

Inhomogeneous Electron Gas

P. Hohenberg, W. Kohn Keywords: Pump-probe, Attosecond dynamics, Non-adiabatic effects
Physical Review 136, B864 (1964)
DOI: 10.1103/PhysRev.136.B864

Optical properties of monolayer MoS₂

Self-Consistent Equations Including Exchange and Correlation Effects

Phys. Rev. B 88, 245309 (2013)
W. Kohn, L. J. Sham
DOI: 10.1103/PhysRevB.88.245309
Physical Review 140, A1133 (1965)
DOI: 10.1103/PhysRev.140.A1133