Osman Goni Ridwan

Lecture 4 · Part I: Quantum mechanics you need

Atomic orbitals, core and valence electrons

Where do s, p, d and f, and “core” versus “valence”, come from?

You will be able to explain

  • what n, ℓ and m label, and how they set an orbital's shape and nodes
  • why many-electron atoms fill their orbitals in the order they do, with exceptions such as Cr and Cu
  • what distinguishes core, semicore and valence electrons
  • what changes in a calculation when you choose Ti, Ti_pv or Ti_sv

Before this

The variational principle and basis sets

Time

About 55 minutes

DFT & VASP Course · Lecture 41 / 16

Every POTCAR you choose makes a statement about atoms: which electrons take part in bonding, and which are frozen in a core. To read that statement you need the language of atomic orbitals: n, ℓ and m, s, p, d and f, shells, shielding and the filling order. This lecture builds that language from the one atom we can solve exactly, hydrogen, and ends with the choice between Ti, Ti_pv and Ti_sv.

Press P to present one slide at a time, S for speaker notes.

Which titanium POTCAR should you use?

Ti, Ti_pv or Ti_sv: 4, 10 or 12 valence electrons. The difference is which shells count as core.

DatasetValence electronsValence configurationENMAX (eV)
Ti43d³ 4s¹178.330
Ti_pv103p⁶ 3d³ 4s¹222.335
Ti_sv (recommended)123s² 3p⁶ 3d³ 4s¹274.610

Standard PBE datasets of the potpaw.64 release.

DFT & VASP Course · Lecture 42 / 16

VASP provides three standard PBE datasets for titanium and recommends Ti_sv. They differ in which electrons are treated explicitly: "pv" adds the 3p shell to the valence, and "sv" adds the 3s shell as well. Each choice changes the number of electrons VASP handles and the cutoff it needs. To understand what is being chosen, and why it matters, we need to know what 3s, 3p, 3d and 4s are. That starts with hydrogen.

What labels the hydrogen atom's states?

Three quantum numbers, n, ℓ and m; the energy depends on n alone.

ψnℓm(r,θ,ϕ)=Rnℓ(r)  Yℓm(θ,ϕ),En=−13.6 eVn2\psi_{n\ell m}(r,\theta,\phi) = \eqterm{R}{R_{n\ell}(r)}\; \eqterm{Y}{Y_{\ell m}(\theta,\phi)}, \qquad \eqterm{E}{E_n = -\frac{13.6\ \text{eV}}{n^2}}
  • Radial part: set by n=1,2,3,…n = 1, 2, 3, \dots and ℓ\ell.
  • Angular part: ℓ=0,…,n−1\ell = 0, \dots, n-1 (s, p, d, f) and m=−ℓ,…,ℓm = -\ell, \dots, \ell.
  • Energy: −1-1 rydberg / n2/\,n^2; all n2n^2 states with the same nn share it.
  • s, p, d, fℓ=0,1,2,3\ell = 0, 1, 2, 3: 1, 3, 5, 7 orbitals each (2ℓ+12\ell + 1 values of m).
  • A shellAll orbitals with the same n: 1s; 2s 2p; 3s 3p 3d; …

DFT & VASP Course · Lecture 43 / 16

Hydrogen and hydrogen-like one-electron ions have analytic solutions in the nonrelativistic Coulomb model. Because the potential −1/r-1/r depends only on the distance from the nucleus, not on the direction, the wavefunction separates into a radial part Rnℓ(r)R_{n\ell}(r) and an angular part Yℓm(θ,ϕ)Y_{\ell m}(\theta, \phi), the spherical harmonics.

These are spherical coordinates: r is distance from the nucleus, θ the angle down from the z axis, and φ the angle around that axis. Spherical harmonics are the angular functions that solve the direction-dependent part of this central-potential problem. An orbital is one spatial one-electron wavefunction; its density describes a distribution, rather than a path around the nucleus. Three whole numbers label the solutions:

  • the principal quantum number n=1,2,3,…n = 1, 2, 3, \dots, which sets the energy and the overall size;
  • the angular-momentum quantum number ℓ=0,1,…,n−1\ell = 0, 1, \dots, n-1, which sets the shape; the letters s, p, d, f stand for ℓ=0,1,2,3\ell = 0, 1, 2, 3;
  • the magnetic quantum number m=−ℓ,…,ℓm = -\ell, \dots, \ell, which labels the component of orbital angular momentum along a chosen z axis, Lz=mℏL_z=m\hbar.

The energy zero is a separated electron and nucleus; a negative energy means a bound state. The energy is En=−Z2 Ry/n2E_n = -Z^2\,\mathrm{Ry}/n^2, and for a given nn all values of ℓ\ell and mm have the same energy: n2n^2 states per shell, 2n22n^2 with spin. This degeneracy in ℓ\ell is special to the pure 1/r1/r potential, and the next slides show how it breaks in other atoms. With an infinitely heavy nucleus the constant is one rydberg, 13.606 eV, half a hartree. The measured ionization energy of hydrogen is 13.598 eV; the small difference comes mostly from the finite mass of the proton, which multiplies the energy by 1/(1+me/mp)1/(1 + m_e/m_p).

How do you read a configuration such as 3p²?

3 names the shell, p the subshell, and ² the number of electrons distributed across its orbitals.

3p2 ↑ m = -1 ↑ m = 0 m = 1 3 spatial orbitals × 2 spin states = 6 places Boxes are states, not regions of space.
Orbital-box diagramEach box is one spatial orbital, with room for two opposite spin states (↑ and ↓). The arrows label spin states; they are not trajectories or orbiting particles.

DFT & VASP Course · Lecture 44 / 16

A shell contains all orbitals with the same n. A subshell fixes both n and ℓ: the 3p subshell contains three spatial orbitals, since ℓ = 1 gives m = −1, 0, +1. A superscript counts electrons in the entire subshell; 3p² means two electrons shared among those three orbitals, not two electrons in each orbital. The s, p, d, f subshells have 1, 3, 5, 7 spatial orbitals and capacities of 2, 6, 10, 14 electrons.

This is a preview of Lecture 5: a spatial orbital has two independent spin states and can hold at most one electron of each spin. For an isolated atom's equal-energy orbitals, Hund's first rule favors occupying different orbitals with parallel spin before pairing. The figure illustrates that usual filling pattern; actual configurations depend on the full electronic energy. Pauli exclusion supplies the capacity constraint, while energies and interactions decide the filling.

For example, [Ne] 3s² 3p² describes silicon: [Ne] abbreviates neon's ten core electrons, then 2 + 2 more electrons give Z = 14. [Ar] abbreviates argon's eighteen-electron configuration. The atom's Z is its nuclear charge and its electron count when neutral. POTCAR's ZVAL counts only the electrons treated explicitly as valence in that dataset.

What do s, p and d orbitals look like?

For the real hydrogen orbitals shown here, ℓ counts angular nodes and n − ℓ − 1 counts radial nodes.

nodenode ⟨r⟩ 01020 r (bohr) r²R² (radial probability) z x angular shape, m = 0 (blue +, orange −)
ExactHydrogen orbitals with m = 0. Left: the probability of finding the electron at distance r. Right: the angular shape in the xz plane.

DFT & VASP Course · Lecture 45 / 16

An orbital ψnℓm\psi_{n\ell m} has n−1n - 1 nodes in all: n−1−ℓn - 1 - \ell radial nodes, spheres where RnℓR_{n\ell} changes sign, and ℓ\ell angular nodes, planes or cones where YℓmY_{\ell m} does. So 1s, 2p, 3d and 4f, the first orbital of each ℓ\ell, have no radial node at all. Try 3s, 3p and 3d in the figure: the 3s orbital has two inner lobes before its main one, while 3d has none and sits further in on average (⟨r⟩\langle r\rangle = 10.5 bohr, against 13.5 for 3s).

The left panel plots the radial probability r2Rnℓ2r^2R_{n\ell}^2, probability per unit radial distance, integrated over all directions. The probability of a thin shell from r to r + dr is this density times dr. The factor r2r^2 counts the area of a sphere of radius rr; it is why the radial density tends to zero as r → 0 even though ∣ψ1s∣2|\psi_{1s}|^2 is largest there.

The figure shows m=0m = 0 only. Combining +m+m and −m-m gives real orbitals: pxp_x and pyp_y have the same shape as pzp_z, pointed along x and y; for d they give four-leaf shapes (dxzd_{xz}, dyzd_{yz}, dxyd_{xy}, dx2−y2d_{x^2-y^2}) that differ from the m=0m = 0 orbital, dz2d_{z^2}. These real orbitals are convenient labels for the local orbital contributions reported by VASP.

For ℓ > 0, orbital angular momentum has magnitude ℓ(ℓ+1)ℏ\sqrt{\ell(\ell+1)}\hbar, while m labels only its z component. Thus a general complex spherical harmonic is not simply an orbital pointed in one classical direction. A node is a zero of the wavefunction; the quoted angular-node count describes these real-orbital representations, including the m = 0 functions plotted here.

How should you read a three-dimensional orbital picture?

Lobes show angular structure; their colors mark wavefunction sign, not positive and negative charge.

xyzp, ℓ = 1, m = 0Blue: positive amplitude · orange: negative amplitude
Angular shape, m = 0Drag to rotate. Surface radius is proportional to ∣Yℓ0∣2|Y_{\ell0}|^2; colors show the sign of Yℓ0Y_{\ell0}. This is an angular-shape surface, not a fixed-probability isosurface of the full orbital.

DFT & VASP Course · Lecture 46 / 16

The 2D xz-plane drawing on the previous slide cuts through a three-dimensional angular pattern. Here you can rotate that pattern and compare s, p, d and f. An s orbital is direction-independent. The pzp_z pattern has two opposite-sign lobes separated by a node in the xy plane. The dz2d_{z^2} pattern also has an equatorial belt; fz3f_{z^3} has additional angular nodes. The z direction is a chosen axis, not a preferred universal direction in an isolated atom.

Squaring the wavefunction removes its sign, so both colors contribute positively to probability density and both describe the same negative electron charge. Sign matters when orbitals are combined, as in interference and bonding. A typical full-orbital isosurface joins positions where a chosen amplitude or density value is constant; its appearance depends on that chosen value and on the radial function. This figure deliberately isolates the angular factor so those different kinds of pictures are not confused.

What changes when an atom has many electrons?

Inner electrons shield the nucleus, so orbitals with the same n no longer share one energy.

  • ShieldingElectrons in inner orbitals screen the nuclear charge seen by outer ones.
  • Penetrations orbitals reach the nucleus, so they feel more of its charge and drop lowest.
  • Centrifugal barrierHigh-ℓ orbitals are kept away from the nucleus and see less of it.

Result: within a shell, E(ns) < E(np) < E(nd). The hydrogen degeneracy is lifted.

DFT & VASP Course · Lecture 47 / 16

With more than one electron, the equation can no longer be solved exactly (Lecture 3), but the qualitative picture is clear. Think of one electron moving in the average field of the nucleus and all the other electrons. In a spherical-average approximation that field is spherical, so s, p and d labels remain useful, but the radial equation now feels an effective potential

Veff(r)=−Zeff(r)r+ℓ(ℓ+1)2r2.V_{\mathrm{eff}}(r) = -\frac{Z_{\mathrm{eff}}(r)}{r} + \frac{\ell(\ell+1)}{2r^2}.

The first term is the nuclear attraction, screened by the electrons inside radius rr: close to the nucleus the electron feels nearly the full charge ZZ, far outside the other electrons only a net charge of about 1. The second term, the centrifugal barrier, comes from the angular motion and pushes electrons with larger ℓ\ell away from the nucleus. Near the nucleus a radial function behaves as rℓr^\ell, so s orbitals have amplitude right at the nucleus, p orbitals start linearly, and d orbitals more slowly still.

So s electrons penetrate under the inner shells and feel more of the nuclear charge, and their energy drops furthest; p electrons less so, d electrons least. The degeneracy of hydrogen is lifted, Ens<Enp<EndE_{ns} < E_{np} < E_{nd}. Lecture 6 makes the screening quantitative for helium: each electron sees an effective charge of 27/16 ≈ 1.69 instead of 2. Computing this shielding from the electrons' mutual repulsion, in a mean field, is the job of the Hartree–Fock method (Lecture 6).

In what order do atoms fill their orbitals?

Mostly in the aufbau order, 1s 2s 2p 3s 3p 4s 3d, with exceptions such as Cr and Cu.

AtomZGround configuration
Silicon14[Ne] 3s² 3p²
Potassium19[Ar] 4s
Scandium21[Ar] 3d 4s²
Titanium22[Ar] 3d² 4s²
Chromium24[Ar] 3d⁵ 4s (not 3d⁴ 4s²)
Copper29[Ar] 3d¹⁰ 4s (not 3d⁹ 4s²)
  • Why 4s before 3dIn potassium, the 4s orbital penetrates the argon core more than 3d does.
  • A surpriseIn the transition metals, the electron removed first is 4s, not 3d.

DFT & VASP Course · Lecture 48 / 16

Combine the ordering Ens<Enp<EndE_{ns} < E_{np} < E_{nd} with the Pauli exclusion principle, at most two electrons per orbital (Lecture 5), and fill the orbitals from the bottom: that is the aufbau, or building-up, rule. [Ne] and [Ar] abbreviate the filled shells of neon and argon.

Penetration explains the most famous oddity. In potassium, the 4s orbital has three inner lobes that reach through the argon core towards the nucleus. The nodeless 3d orbital lies almost entirely outside the core, where the nuclear charge is screened by all 18 core electrons. So 4s lies lower, potassium's last electron goes into 4s, and calcium is 4s². The configurations in the table are the measured ground states. Chromium and copper break the simple rule, taking only one 4s electron. The 3d and 4s energies are so close that the details of the repulsion between electrons decide the order, and here they favour a half-filled or filled d shell.

The order also does not survive ionization. Once the 3d shell starts to fill, the 3d electrons, sitting largely inside the 4s, shield the nucleus from the 4s electrons, and in scandium and the transition metals after it the electron removed first is 4s, not 3d. The lesson for calculations is that 3d and 4s lie close in energy, so their relative occupation responds to the chemical environment.

What are core and valence electrons?

Valence electrons make bonds; core electrons sit deep and barely change. Semicore lie in between.

Titanium, 22 electrons, from the nucleus outwards → 1s 2 e⁻ 2s 2 e⁻ 2p 6 e⁻ 3s 2 e⁻ 3p 6 e⁻ 3d 2 e⁻ 4s 2 e⁻ frozen core: 10 e⁻ valence: 12 e⁻ = ZVAL
Titanium's shells from the inside out, in its ground configuration. Choose a dataset to see which shells it treats as valence and which it freezes in the core.

DFT & VASP Course · Lecture 49 / 16

The valence electrons occupy the outermost, highest-energy orbitals and take part in bonding. The core electrons are tightly bound close to the nucleus, with binding energies of hundreds or thousands of eV, and hardly notice the neighbouring atoms. Between them, semicore states such as titanium's 3s and 3p lie deep enough to be core-like, tens of eV below the valence, but high enough to respond to short bonds, pressure or magnetism.

VASP's PAW datasets freeze the core: the core electrons keep the configuration they had in the atomic calculation that generated the dataset (the frozen-core approximation, Lecture 0). In the figure, Ti freezes 18 electrons, Ti_pv 12 and Ti_sv 10. The configuration listed for a dataset need not be the free atom's: Ti is listed as 3d³ 4s¹, while the atom's ground state is 3d² 4s². In the solid the valence electrons are redistributed self-consistently anyway; what the dataset fixes is how many there are and which shells are frozen.

When should semicore electrons be valence?

When bonding, magnetism or the method can disturb them.

  • BondingShort bonds or high pressure push neighbouring atoms into the semicore region.
  • MagnetismSpin polarization of the valence can polarize the semicore shells too.
  • Hybrid functionalsExact exchange couples valence and semicore states: use _pv or _sv whenever possible.
  • The price, titaniumNELECT up to 3 times larger (4 → 12 per atom) and ENMAX from 178 to 275 eV.
  • The ruleStart from the recommended dataset; test another if your property depends on it.

DFT & VASP Course · Lecture 410 / 16

A frozen core is exact only if the core really does not change. Three situations disturb semicore states:

  • Bonding. When atoms come close, under pressure or in short bonds, the neighbours' orbitals overlap the semicore shells, which then take part in the bonding.
  • Magnetism. A spin-polarized valence shell exerts a different exchange field on up and down semicore electrons, and frozen semicore shells cannot respond.
  • Hybrid functionals. Exact exchange (Lecture 6) couples occupied orbitals of matching spin, including semicore states, so freezing them is a worse approximation; VASP strongly recommends the _pv and _sv datasets for hybrids.

For the early 3d metals, VASP's recommended datasets are Ti_sv, V_sv and Cr_pv; the plain Ti, V and Cr datasets can give reasonable results but should be used with great care. In most cases, though, freezing the semicore makes little difference, and testing is the way to know.

Semicore electrons cost. More electrons means more bands to compute, and the deeper, more compact orbitals need more plane waves, which shows up as a larger ENMAX. In a compound the default cutoff is the largest ENMAX of all species, so the price depends on the partners: with oxygen (ENMAX 400 eV), the cutoff does not change between Ti and Ti_sv; only the number of electrons does.

Why are 3d and 4f electrons special?

They are the first d and f shells: no radial node, so they stay compact even as valence electrons.

  • No radial node1s, 2p, 3d, 4f are the first orbitals of their ℓ, and nodeless in r.
  • CompactIn hydrogen, ⟨r⟩ = 10.5 bohr for 3d against 13.5 for 3s; in titanium, the 3d shields the 4s from the nucleus.
  • ConsequenceCompounds with partly filled d and f shells include the strongly correlated materials that common functionals often describe poorly.
⟨r⟩nℓ=3n2−ℓ(ℓ+1)2 a0\langle r \rangle_{n\ell} = \frac{3n^2 - \ell(\ell+1)}{2}\ a_0
  • Mean distance from the nucleus for hydrogen; smaller for larger ℓ\ell at the same nn.

DFT & VASP Course · Lecture 411 / 16

Why does a radial node matter? Orbitals with the same ℓ\ell must be orthogonal to each other. The 2s orbital must be orthogonal to 1s, which forces it to change sign once, a node, and pushes its main lobe outward; 3s must be orthogonal to both 1s and 2s, and so on. But 3d is the first d orbital: there is no lower d orbital to be orthogonal to, so it has no node and is free to stay compact. The same reasoning applies to the first allowed orbital of each angular type, 2p and 4f. It is a useful hydrogenic guide, rather than a universal predictor of orbital size across different atoms.

The formula on the slide, exact for hydrogen, puts numbers on it: at n=3n = 3, ⟨r⟩\langle r\rangle is 13.5 bohr for 3s, 12.5 for 3p and 10.5 for 3d. In a transition metal the 3d shell lies largely inside the 4s, with which it shares the valence, and it shields the nucleus from the 4s, which is why 4s ionizes first.

This has two consequences you will meet in VASP. First, compact spatial features tend to need more plane waves. However, ENMAX also depends on how the PAW dataset was constructed, including its smoothing region and projectors; a radial-node count alone cannot predict the cutoff. The listed oxygen and titanium values are properties of those particular datasets. Second, compact d and f orbitals keep their electrons close together, so the repulsion between them is strong. Many compounds with partly filled d or f shells are "strongly correlated", meaning that local repulsion competes strongly with the energy gained by spreading electrons between sites. Common local-density and generalized-gradient functionals (LDA and GGA) may describe that balance poorly; f elements are especially hard, because self-interaction errors are largest in strongly localized orbitals. One model correction, an on-site Hubbard U, adds an energy penalty for particular localized orbital occupations (Lecture 25).

Where do orbitals and shells appear in VASP?

The POTCAR supplies the valence count (ZVAL) and suggested cutoff (ENMAX); LORBIT projects bands onto s, p, d.

ConceptIn VASPWhere
Valence electrons per atomZVAL of each dataset; their total is NELECTPOTCAR header, OUTCAR
Frozen coreFixed when the dataset was generatedChoice of POTCAR (Lecture 18)
How compact the orbitals areENMAX, the suggested cutoff; the largest one is the default ENCUTPOTCAR header
s, p, d character of a bandLORBIT projects orbitals onto local ℓ,m\ell, mPROCAR, DOSCAR (Lecture 29)

DFT & VASP Course · Lecture 412 / 16

NELECT is set from the POTCARs for a neutral cell: ZVAL times the number of atoms of each species. ENMAX is each dataset's suggested cutoff, and the default ENCUT is the largest ENMAX among the POTCARs.

LORBIT asks VASP to project every Kohn–Sham orbital onto spherical harmonics around each atom, giving its s, p, d and f contributions there. There is no unique way to divide an orbital among atoms, so the numbers are qualitative, but they are usually a good guide to the character of a band. That is how a band structure gets labelled "Ti 3d" or "O 2p" in Lecture 29.

Lab 4: what do your POTCARs say?

Read ZVAL and ENMAX from three titanium datasets, then predict NELECT for a TiO₂ cell.

Read (no run)

  1. In your licensed potpaw directory, for Si, Ti, Ti_pv, Ti_sv and O: grep -E "ZVAL|ENMAX" POTCAR.
  2. Check Lab 0's OUTCAR: is NELECT = 2 × ZVAL(Si)?

Predict

  1. ZVAL 4, 10, 12 for Ti, Ti_pv, Ti_sv; 6 for O.
  2. A cell with 2 Ti and 4 O: NELECT = 32 with Ti, 48 with Ti_sv.
  3. Default ENCUT (default PREC): 400 eV either way, set by oxygen.

DFT & VASP Course · Lecture 413 / 16

The predictions use the standard potpaw.64 values: Ti, Ti_pv and Ti_sv have 4, 10 and 12 valence electrons and ENMAX 178.330, 222.335 and 274.610 eV; O has 6 and 400.0 eV. With 2 titanium and 4 oxygen atoms, as in a rutile cell, NELECT is 2×4+4×6=322 \times 4 + 4 \times 6 = 32 with Ti and 2×12+24=482 \times 12 + 24 = 48 with Ti_sv, and the default cutoff is the largest ENMAX, 400 eV, in both cases. If your POTCARs come from a different release, the numbers may differ slightly: record the first line of each, as in Lab 0. This site never shows POTCAR contents.

Which misconceptions should you drop now?

Five beliefs about orbitals and POTCARs. Click a card to turn it over.

“All orbitals of a shell have the same energy.”

Only in hydrogen and other one-electron ions. In many-electron atoms shielding puts ns below np below nd.

“4s fills before 3d, so it must also empty last.”

4s fills first in K and Ca, but in transition metals 4s electrons are removed first.

“The aufbau rule always gives the ground state.”

Cr is 3d⁵ 4s and Cu 3d¹⁰ 4s: the simple rule fails for them.

“More valence electrons is always better.”

It costs electrons and cutoff. Use the recommended dataset and test when the property depends on it.

“The POTCAR configuration is the atom's ground state.”

Ti is listed as 3d³ 4s¹; the free atom is 3d² 4s². The dataset fixes the count, not the occupations.

DFT & VASP Course · Lecture 414 / 16

Check yourself

One question per outcome. Answer aloud, then click to compare.

How many radial and angular nodes does a 4d orbital have?

Angular: ℓ = 2. Radial: n − ℓ − 1 = 1. Three nodes in all, n − 1.

Why does potassium's last electron go into 4s, not 3d?

4s penetrates the argon core and feels more of the nuclear charge, so it lies lower than 3d there.

What makes titanium's 3s and 3p semicore?

They lie deep, below the 3d and 4s valence, but are disturbed enough by bonding or magnetism that VASP recommends treating them as valence.

What changes in a calculation when you switch from Ti to Ti_sv?

8 more valence electrons per Ti (ZVAL 4 → 12), more bands, and a higher ENMAX; the frozen core shrinks from 18 to 10 electrons.

DFT & VASP Course · Lecture 415 / 16

What should you remember?

Atoms are hydrogen's orbitals, reordered by shielding and split into core and valence.

  • n, ℓ, m label hydrogen's states; ℓ gives s, p, d, f; there are n − ℓ − 1 radial and ℓ angular nodes.
  • Shielding and penetration put ns below np below nd; aufbau gives the filling order, with exceptions such as Cr and Cu.
  • Core electrons are frozen; valence electrons bond; semicore states sit between and can matter.
  • A POTCAR lists ZVAL and the suggested cutoff ENMAX: Ti, Ti_pv, Ti_sv have 4, 10, 12 valence electrons.

Next: Lecture 5 asks why many electrons are so much harder than one, and meets the Pauli principle properly.

DFT & VASP Course · Lecture 416 / 16

Key terms

Shell, subshell and orbital
A shell fixes n; a subshell fixes n and ℓ; its 2ℓ+12\ell+1 spatial orbitals have different magnetic labels m. Each spatial orbital has two spin states. A configuration superscript counts electrons in the entire subshell.
Valence electrons
The electrons in an atom's outermost, highest-energy orbitals, which take part in bonding. In VASP, the electrons a POTCAR treats explicitly; their number per atom is ZVAL.
Core electrons
Electrons in deep, tightly bound orbitals close to the nucleus, little affected by neighbouring atoms. A PAW dataset freezes them in the configuration used to generate it.
Semicore states
Orbitals that lie below the valence but high enough to respond to bonding, pressure or magnetism, such as titanium's 3s and 3p. VASP's _pv and _sv datasets treat them as valence.
Frozen-core approximation
The approximation that the core electrons keep the configuration of the isolated atom used to generate the PAW dataset.

VASP tags in this lecture

NELECT
Number of electrons. By default it is set from the POTCARs (ZVAL of each species times its number of atoms) for a neutral cell; set it only for a charged calculation.Default: the valence electrons of the POTCARs (neutral cell) · VASP Wiki: NELECT
ENCUT
Cutoff energy of the plane-wave basis set, in eV. The VASP Wiki strongly recommends setting it explicitly in every INCAR, and checking the convergence of the quantity you report.Default: the largest ENMAX among the POTCAR files · VASP Wiki: ENCUT
LORBIT
Selects a method to project the Kohn–Sham orbitals onto local quantum numbers (ℓ, m) around each atom, giving s-, p-, d- and f-resolved information in PROCAR and DOSCAR.Default: 0 (which needs RWIGS; from LORBIT = 10 on, RWIGS is ignored) · VASP Wiki: LORBIT

Further reading

  1. R. W. Field, 5.61 Physical Chemistry, Lecture 21: “Hydrogen atom II”, MIT OpenCourseWare (Fall 2017), CC BY-NC-SA. · MIT OCW
  2. R. W. Field, 5.61 Physical Chemistry, Lecture 23: “Many-electron atoms”, MIT OpenCourseWare (Fall 2017), CC BY-NC-SA. · MIT OCW
  3. VASP Wiki, “Choosing pseudopotentials” (advice, and the table of recommended PBE potentials, potpaw.64). · vasp.at/wiki