Electron Configurations and the Aufbau Principle

Line the elements up by atomic number and plot how hard it is to tear off their outermost electron, and a saw-toothed curve appears: a steep climb to each noble gas, then a cliff down to the next alkali metal. That sawtooth is the periodic table drawn by an instrument instead of by a chemist, and learning to read it is what this article is about. Underneath it sits one organizing idea — a neutral atom's $Z$ electrons occupy hydrogen-like orbitals labelled by $n$ and $\ell$ , filled from the bottom up. Quantum mechanics supplies the machinery, and we take it as given: the antisymmetry that becomes the exclusion principle, capping each orbital at two electrons, is owned by aq-qm-pauli-exclusion-principle; the filling engine that orders orbitals by $n+\ell$ (the Madelung rule) and justifies it through penetration and screening is owned by aq-qm-aufbau-principle-electron-configurations; and the climbing effective charge that drives chemical periodicity is owned by aq-qm-periodic-table-quantum-mechanics. Here we turn instead to the spectroscopic evidence for that picture: the measured first-ionization energies that make shell structure visible directly in laboratory data, the ground configurations that atomic spectra actually report, and the handful of elements — chromium and copper the famous ones — where the measured configuration overrules the tidy rule. A configuration is not the end of the story but its start. It is the input a spectroscopist feeds into term analysis, the subject of the companion article LS Coupling and Term Symbols; get it right and the whole ladder of an atom's spectral lines becomes readable. The register here is gentle, but the numbers are real and every one is checked.

Key takeaways

An electron configuration lists how a neutral atom's $Z$ electrons fill hydrogen-like orbitals labelled by $n$ and $\ell$ — carbon is $1s^2\,2s^2\,2p^2$ , argon closes at $\dots3p^6$ .
Exclusion caps a subshell at $2(2\ell+1)$ electrons and a shell at $2n^2$ : $s,p,d,f$ hold $2,6,10,14$ and shells hold $2,8,18,32$ — the block widths of the table.
The filling order is the Madelung $(n+\ell)$ rule (recalled, not re-derived): $1s\,2s\,2p\,3s\,3p\,4s\,3d\,4p\,5s\dots$ , with $4s$ ahead of $3d$ .
The first-ionization sawtooth is the shell structure read off an instrument: noble-gas peaks (He $24.587$ eV, Ne $21.565$ , Ar $15.760$ ), alkali troughs (Li $5.392$ , Na $5.139$ , K $4.341$ ).
Two small dips — beryllium above boron ( $9.323 \to 8.298$ ) and nitrogen above oxygen ( $14.534 \to 13.618$ ) — are subshell-closure and exchange fingerprints, visible in the measured numbers.
Chromium $[\mathrm{Ar}]3d^5\,4s^1$ and copper $[\mathrm{Ar}]3d^{10}4s^1$ are measured exceptions: the simple order is not the true energy order and it does not survive ionization.
Key idea: fill hydrogen-like orbitals from the bottom up under the exclusion cap of two per orbital, in Madelung $(n+\ell)$ order — and then read the periodic law straight off the measured ionization-energy sawtooth, exceptions included.

The core insight

Fill hydrogen-like orbitals from the bottom up, two electrons per spatial orbital, in Madelung $(n+\ell)$ order; the capacities $2(2\ell+1)$ per subshell then generate the periodic table, and the measured first-ionization sawtooth — noble-gas peaks, alkali troughs, and the Be and N dips — is that shell structure made visible in laboratory data.

Where this lives today

Every atom's ground configuration and ground term is tabulated in the NIST Atomic Spectra Database, and it is not folklore — it is read from measured spectra, the ionization energies of the sawtooth being one of the cleanest inputs. Photoelectron and X-ray photoelectron spectroscopy (XPS) map orbital binding energies directly, seeing the $2s$ / $2p$ and $3s$ / $3p$ splittings that the screening picture predicts, and turning "which subshell" into a chemical fingerprint used across surface science and catalysis. Configurations also decide which materials become useful light sources: the open $4f$ subshells of neodymium and erbium ions, filled according to exactly this bookkeeping, supply the sharp transitions of the Nd:YAG laser and the erbium-doped fibre amplifier that carries internet traffic. And in astrophysics the ground configuration of each species sets the lines a star can emit, so measured stellar spectra are inverted through configuration and term analysis into abundances — the composition of the cosmos, read from the same rules that draw the periodic table.

The physical picture

Think of the atom as an apartment block. Floors are shells (fixed $n$ ); each floor has rooms of increasing size labelled $s,p,d,f$ (fixed $\ell$ ); every room sleeps exactly two tenants of opposite spin. A third tenant is impossible, so you fill from the ground floor up — that is the building-up, or Aufbau, idea. The twist is that some upper-floor rooms are cheaper than lower ones: an $s$ electron has no centrifugal barrier and dives through the screening cloud of the others toward the bare nucleus, so a high-floor $4s$ room can undercut the lower $3d$ room. The Madelung combination $n+\ell$ is the bookkeeping for that competition, and the first figure is the staircase it draws.

Figure

The Madelung staircase: orbitals fill by increasing n+ℓ, ties broken toward smaller n — so 4s (n+ℓ = 4) fills before 3d (n+ℓ = 5), the inversion that opens the transition metals

Why isn't every room on a floor the same energy, as in bare hydrogen? Because the other electrons form a screen, and low- $\ell$ orbitals penetrate it while high- $\ell$ orbitals are held out by their centrifugal barrier — the mechanism whose derivation belongs to aq-qm-aufbau-principle-electron-configurations and aq-qm-periodic-table-quantum-mechanics. We do not need the derivation to see its consequence in data. Stack the elements and measure the energy to remove the loosest electron, and the shells announce themselves: a hard-to-ionize noble gas at the top of each period, an easy alkali metal at the bottom of the next. The second figure is that measurement — the periodic law read straight off an instrument, with the real numbers on the axis.

Figure

First ionization energy against atomic number Z = 1–20 (NIST values, eV): each period climbs from an alkali trough to a noble-gas peak, then crashes. He 24.587 eV is the hardest to ionize; the small dips Be→B (9.323→8.298) and N→O (14.534→13.618) are subshell-closure and half-filled-shell fingerprints

The core equations

A configuration is a piece of notation with a hard combinatorial limit built in. The exponent on each subshell cannot exceed the number of distinct spin-orbitals it offers, which is where the capacities come from.

n\ell^{\,q}\qquad q\le 2(2\ell+1) \qquad (1)

where: $n$ — principal quantum number ( $1,2,3,\dots$ ); $\ell$ — orbital angular-momentum quantum number ( $0,1,\dots,n-1$ , written $s,p,d,f$ ); $q$ — number of electrons in that subshell; $2(2\ell+1)$ — its capacity, the $(2\ell+1)$ values of $m_\ell$ times the two spin states. Physically the exponents are a census: they count occupied one-electron states, nothing more. The cap is the exclusion principle, which we state and use but do not derive — the two-line proof from wavefunction antisymmetry is owned by aq-qm-pauli-exclusion-principle.

\text{no two electrons share all four of }(n,\ell,m_\ell,m_s) \qquad (2)

where: $m_\ell$ — magnetic quantum number, $-\ell\le m_\ell\le\ell$ ; $m_s=\pm\tfrac12$ — spin projection. Reading: each complete label names one seat, and a seat holds one electron. Summing the seats gives the capacities of equation (1). In a many-electron atom the single-electron energy depends on both $n$ and $\ell$ , not on $n$ alone as in hydrogen, because each electron moves in a screened effective charge rather than the bare nucleus.

E_{n\ell}\approx-\frac{Z_{\mathrm{eff}}^2}{n^2}\,\mathrm{Ry},\qquad \mathrm{Ry}=13.6057\ \mathrm{eV} \qquad (3)

where: $Z_{\mathrm{eff}}=Z-\sigma$ — effective nuclear charge, the bare $Z$ reduced by the averaged screening $\sigma$ of the inner electrons; $\mathrm{Ry}$ — Rydberg energy. The screening that makes $\sigma$ , and its climb across a period, are quantified by aq-qm-periodic-table-quantum-mechanics; here $Z_{\mathrm{eff}}$ is the single number that turns "which shell" into "how tightly bound." Because low- $\ell$ orbitals penetrate the screen and bind more tightly, the neutral-atom filling order is the empirical Madelung rule, which we recall as an input rather than prove.

\text{fill by increasing }(n+\ell);\ \text{ties: smaller }n\ \text{first} \qquad (4)

where: $n+\ell$ — the sorting key that reproduces $1s,2s,2p,3s,3p,4s,3d,4p,5s,\dots$ ; the classic outlier is $4s$ ( $n+\ell=4$ ) before $3d$ ( $n+\ell=5$ ). The penetration-and-screening argument that justifies (4), and its variational reading as an energy minimum, are owned by aq-qm-aufbau-principle-electron-configurations.

Derivation ASubshell and shell capacities from the quantum-number census

The goal is to derive the numbers everyone memorizes — $s,p,d,f$ holding $2,6,10,14$ electrons and shells holding $2,8,18,32$ — as a pure count of distinct spin-orbitals, so that the shape of the periodic table (two $s$ -columns, six $p$ , ten $d$ , fourteen $f$ ) is exposed as bookkeeping rather than decree. We take the exclusion rule that one electron fills one complete label $(n,\ell,m_\ell,m_s)$ from aq-qm-pauli-exclusion-principle and simply enumerate the labels, first at fixed $\ell$ , then summed over a shell.

Step 1. For a fixed $\ell$ there are $2\ell+1$ values of $m_\ell$ , running from $-\ell$ to $+\ell$ . Why: these are the orientations of the orbital angular momentum, and they are the spatial orbitals available in the subshell.

Step 2. Each spatial orbital admits two electrons, $m_s=\pm\tfrac12$ , so the subshell capacity is $g_\ell=2(2\ell+1)$ , giving $s\to2$ , $p\to6$ , $d\to10$ , $f\to14$ . Why: the factor of two is the only thing spin contributes to the count, and it is what makes every subshell hold an even number.

Step 3. Sum over the subshells of a shell using the identity $\sum_{\ell=0}^{n-1}(2\ell+1)=n^2$ (the odd numbers $1,3,5,\dots$ sum to a square), so $g_n=2n^2 = 2,8,18,32$ for $n=1,2,3,4$ . Why: this is the shell capacity, and folded through the Madelung order it becomes the period lengths $2,8,8,18,18,32$ of the table. *Verified in verify_aq-optics-electron-configurations-aufbau.py: $2(2\ell+1)=2,6,10,14$ and $\sum(2\ell+1)=n^2$ giving $2n^2=2,8,18,32$ .*

Go deeper · Derivation B · 2 minReading the ionization-energy sawtooth

Derivation BReading the ionization-energy sawtooth

The goal here is the reverse of the usual textbook move: instead of predicting ionization energies from theory, we treat the measured first-ionization curve of the light elements as raw evidence and read the atom's shell structure out of it, peak by peak and dip by dip. Everything below is a statement about NIST numbers; the underlying reason the effective charge climbs across a period, and why the two small dips occur, is quantified by aq-qm-periodic-table-quantum-mechanics and recalled by name.

Step 1. Across each period the energy rises overall — Li $5.392 \to$ Ne $21.565$ eV, Na $5.139 \to$ Ar $15.760$ eV — because adding protons while filling the same shell tightens every electron's binding. Why: this rising ramp is the direct fingerprint of $Z_{\mathrm{eff}}$ increasing along a row, the periodic table's steepest trend.

Step 2. Each period ends at a noble-gas peak (He $24.587$ , Ne $21.565$ , Ar $15.760$ eV) and the next begins at an alkali trough (Li, Na, K at $5.392, 5.139, 4.341$ eV). Why: a closed shell is uniquely stable, and the next electron starts a new shell far from the nucleus and outside the closed core, so it is loosely held — the tooth spacing is set by the shell capacities.

Step 3. Two small dips interrupt the rise. Beryllium sits above boron ( $9.323 \to 8.298$ eV): boron's new electron enters the higher $2p$ subshell just after $2s$ closes. Nitrogen sits above oxygen ( $14.534 \to 13.618$ eV): oxygen must pair a second electron into a $2p$ orbital, breaking nitrogen's stable half-filled $2p^3$ . Why: these are subshell-closure and half-filled-shell effects, the exchange stabilization owned by the recalled QM articles, made visible as a $\sim1$ eV wobble in the measured curve.

Step 4. The single deepest trough in the whole table is caesium at $3.894$ eV — one $6s$ electron far outside a large xenon core. Why: it confirms the pattern to its logical end, the loosest valence electron in nature, and it is why caesium anchors the photoelectric and atomic-clock traditions. *Verified in verify_aq-optics-electron-configurations-aufbau.py: He max $24.587$ , Cs min $3.894$ ; Be $>$ B, N $>$ O, Mg $>$ Al, P $>$ S dips; alkali troughs below their preceding noble gas.*

Go deeper · Derivation C · 1 minThe Madelung order and the chromium and copper exceptions as spectroscopic facts

Derivation CThe Madelung order and the chromium and copper exceptions as spectroscopic facts

The goal is to run the filling engine by hand to produce ground configurations, and then to confront it honestly with the two textbook elements where the measured configuration disagrees — chromium and copper — treating those as spectroscopic data, not as a theory to be rebuilt. The $(n+\ell)$ ordering, its penetration justification, and the exchange stabilization behind the exceptions are all recalled from aq-qm-aufbau-principle-electron-configurations; what we own here is the confrontation with the NIST configurations.

Step 1. Pour $Z$ electrons into sequence (4), filling each subshell to capacity. Carbon ( $Z=6$ ) gives $1s^2\,2s^2\,2p^2$ ; argon ( $Z=18$ ) closes at $\dots3p^6$ ; potassium ( $Z=19$ ) adds its last electron to $4s$ , not $3d$ , because $4s$ comes first. Why: the $4s$ -before- $3d$ position is the one place the naive shell picture would go wrong, and it is exactly what builds the transition-metal block.

Step 2. Chromium ( $Z=24$ ) is measured as $[\mathrm{Ar}]\,3d^5\,4s^1$ , not the $3d^4\,4s^2$ the rule predicts, and copper ( $Z=29$ ) as $[\mathrm{Ar}]\,3d^{10}4s^1$ , not $3d^9\,4s^2$ . Why: with $3d$ and $4s$ nearly degenerate at these $Z$ , the extra stability of a half-filled ( $d^5$ ) or filled ( $d^{10}$ ) subshell — an exchange effect recalled by name above — is enough to poach one $4s$ electron, and the atomic spectra confirm it.

Step 3. The electron counts still balance: $24 = 18+5+1$ and $29 = 18+10+1$ . Why: the exception rearranges which subshells hold the outer electrons without changing their number, so it is a reordering, not a violation of the capacity census.

Step 4. These configurations are not a fixed property of an orbital-energy diagram. Once $3d$ is occupied it drops below $4s$ , so ions lose $4s$ first: $\mathrm{Fe}^{2+}$ is $[\mathrm{Ar}]3d^6$ , not $[\mathrm{Ar}]3d^4\,4s^2$ . Why: the neutral-atom filling order and the true one-electron energy order are different questions, and the recalled QM articles make that distinction precise. *Verified in verify_aq-optics-electron-configurations-aufbau.py: Madelung sequence first nine terms; Cr and Cu electron counts $18+5+1$ and $18+10+1$ .*

Limiting cases

Case	Result	Meaning
Hydrogenic, one electron	$E_n=-\dfrac{Z^2}{n^2}\,\mathrm{Ry}$	No screening, so energy depends only on $n$ ; the $\ell$ -degeneracy is exact and the $n+\ell$ rule is irrelevant.
Alkali, one valence electron	$E_{n\ell}=-\dfrac{\mathrm{Ry}}{(n-\delta_\ell)^2}$	The quantum defect $\delta_\ell$ (developed in Multi-electron Atoms and the Central-Field Approximation) shifts penetrating low- $\ell$ terms strongly: $\delta_s>\delta_p>\delta_d$ .
Neutral atom, ground filling	$(n+\ell)$ ascending, ties by smaller $n$	Madelung holds for most elements but fails at Cr, Cu, Nb, Mo, Pd, Ag and several lanthanides.
Ion or occupied $d$ -block	$E_{3d}<E_{4s}$ once $3d$ is filled	The filling order is not the true energy order; cations lose $4s$ first.

Common trapThe aufbau order is the atom's true orbital-energy ordering

It is not. The Madelung $(n+\ell)$ rule orders the sequence in which orbitals are filled in neutral atoms near their ground state — nothing more. Once electrons occupy the $3d$ shell, $3d$ actually drops below $4s$ , which is why transition-metal cations lose $4s$ electrons first: $\mathrm{Fe}^{2+}$ is $[\mathrm{Ar}]3d^6$ , not $[\mathrm{Ar}]3d^4\,4s^2$ . And the rule is only approximate even for neutral atoms: chromium and copper break it outright because a half-filled or filled $d$ -subshell wins a near-degenerate contest. A configuration is not a fixed label on an orbital-energy diagram; it depends on $Z$ and on ionization state, and the honest source is the measured spectrum.

Worked examples

1 · Silicon: from atomic number to a measured ground term

Silicon has $Z=14$ . Write its ground configuration, identify the subshell that carries its spectrum, and state its measured ground term.

Step 1. Fill sequence (4): $1s^2\,2s^2\,2p^6\,3s^2\,3p^2$ , i.e. $[\mathrm{Ne}]\,3s^2\,3p^2$ . Check the count: $[\mathrm{Ne}]=10$ , plus $2+2=4$ , gives $14$ . The neon core is closed and spectroscopically silent. Silicon is $[\mathrm{Ne}]\,3s^2\,3p^2$ .

Step 2. Only the open $3p^2$ subshell shapes the spectrum. It offers $2(2\cdot1+1)=6$ seats, and placing two electrons among them gives $\binom{6}{2}=15$ distinct microstates — the raw material of term analysis. The two $3p$ electrons generate 15 microstates.

Step 3. Those 15 states organize into terms, and the measured ground term of silicon (NIST) is $^3P_0$ — the same term structure as carbon's $2p^2$ , since $3p^2$ and $2p^2$ share the open-subshell pattern. Decoding a configuration into its allowed terms $^1S_0,\,^3P_{0,1,2},\,^1D_2$ and knowing which lies lowest is the job of the companion articles LS Coupling and Term Symbols and Hund's Rules; here the configuration is the input and the ground term is a spectroscopic fact. Silicon's ground term is $^3P_0$ , read from its spectrum.

2 · Two roads to the ionization energy of sodium

Estimate the first ionization energy of sodium from its single valence electron, and compare with the measured $5.139$ eV.

Step 1. Sodium is $[\mathrm{Ne}]\,3s^1$ : one electron outside a closed core, so a hydrogen-like estimate applies with a quantum defect absorbing the penetration. Using the sodium $3s$ quantum defect $\delta_s=1.373$ (the central-field description of alkali spectra is the subject of Multi-electron Atoms and the Central-Field Approximation), the binding energy is $\mathrm{Ry}/(n-\delta_s)^2$ .

Step 2. Insert $n=3$ : $E = 13.6057/(3-1.373)^2 = 13.6057/(1.627)^2 = 13.6057/2.647 = \mathbf{5.14\ eV}$ . The quantum-defect estimate gives 5.14 eV.

Step 3. The measured first ionization energy of sodium is $5.139$ eV — agreement to three figures. The two descriptions meet: the configuration $[\mathrm{Ne}]3s^1$ says which electron leaves, and the quantum defect says how tightly it is held. This is the same $5.139$ eV that marks the sodium trough in the sawtooth figure. Configuration and quantum defect agree with the measured 5.139 eV.

Physics of the portrait

The electron configuration is a shell model built from three ideas that arrived over half a century — an empirical table, a filling principle, and a veto — before quantum mechanics fused them.

Origins · 1 minelectron configuration and how this was discovered

electron configuration

the shell picture that draws the periodic table · Mendeleev → Bohr → Stoner → Pauli

The configuration idea has no single author; it is the residue of four insights. Dmitri Mendeleev built the periodic table in 1869 as pure empirical pattern, ordering the elements by mass and leaving gaps for the undiscovered — a chart that worked for sixty years before anyone knew why. Niels Bohr supplied the mechanism: from 1913 his quantum atom placed electrons in shells, and by the early 1920s his "Aufbauprinzip" filled those shells in a definite order, a picture bold enough to predict that element 72 would be a transition metal chemically like zirconium, not a rare earth. When Dirk Coster and George de Hevesy found hafnium in exactly that slot in 1923, the shell picture had made a chemical prediction and won. Edmund Stoner sharpened it in 1924 by splitting Bohr's shells into subshells and counting their capacities as $2(2\ell+1)$ — the $2,6,10,14$ that shape the table. What none of them had was the reason for the cap, and Wolfgang Pauli provided it in 1925 with the exclusion principle: no two electrons share all their quantum numbers. Strikingly, Pauli's rule came a year before spin was named — his fourth quantum number was a mysterious "two-valuedness" that only later became $m_s=\pm\tfrac12$ . Together, Mendeleev's pattern, Bohr's filling, Stoner's counting, and Pauli's veto are the electron configuration.

History

1869

— Mendeleev arranges the elements into the periodic table

1913

— Bohr's quantum atom introduces electron shells

1921

— Bohr states the Aufbauprinzip, the building-up of shells in order

1923

— Coster and Hevesy identify hafnium in Bohr's predicted slot

1924

— Stoner gives the subshell capacities 2(2ℓ+1)

1925

— Pauli publishes the exclusion principle

1936

— Madelung states the n+ℓ filling rule

Going further

Texts. Foot, Atomic Physics, ch. 4 (central-field approximation, configurations) · Demtröder, Atoms, Molecules and Photons, ch. on many-electron atoms · Bransden & Joachain, Physics of Atoms and Molecules, ch. 8 · Haken & Wolf, The Physics of Atoms and Quanta, ch. 19.

Provenance. Ground configurations and terms (C: $^3P_0$ ; Si: $^3P_0$ ; Cr: $[\mathrm{Ar}]3d^5\,4s^1$ ; Cu: $[\mathrm{Ar}]3d^{10}4s^1$ ) are standard and verifiable against the NIST Atomic Spectra Database and Foot ch. 4. Subshell and shell capacities $2(2\ell+1)$ and $2n^2$ follow from elementary exclusion counting (owned by aq-qm-pauli-exclusion-principle). The Madelung $(n+\ell)$ ordering and its penetration/screening justification are owned by aq-qm-aufbau-principle-electron-configurations; the ionization-energy trends and $Z_{\mathrm{eff}}$ climb by aq-qm-periodic-table-quantum-mechanics. First ionization energies (eV) are NIST values; the anchors He $24.587$ , Ne $21.565$ , Ar $15.760$ , Li $5.392$ , Na $5.139$ , K $4.341$ , Cs $3.894$ , and the Be $\to$ B and N $\to$ O dips are reproduced in the verifier. The sodium quantum defect $\delta_s=1.373$ is the standard value.

Prerequisites. Solution of the Schrödinger Equation for Hydrogen · Quantum Numbers and Orbital Angular Momentum · Electron Spin and the Spin Magnetic Moment · Hydrogenic Wavefunctions and Radial Distributions. Leads to. LS Coupling and Term Symbols · Hund's Rules · Fine Structure and Spin-Orbit Coupling · The Hydrogen Spectrum and Spectral Series.

Is the aufbau filling order the same as the atom's true orbital-energy order?

No, and conflating them is the commonest trap. Aufbau tells you the order in which orbitals are filled as you build up neutral atoms in their ground state, and that order is captured by the Madelung $(n+\ell)$ rule. But it is not a statement that $4s$ always lies below $3d$ in energy. Once the $3d$ shell is occupied, mutual screening among the $n=3$ electrons raises $4s$ above $3d$ , so in an assembled transition metal $3d$ is the lower orbital — which is exactly why ionization removes the $4s$ electrons first. The filling order and the one-electron energy order answer two different questions, and the difference is what makes $\mathrm{Fe}^{2+}$ come out as $[\mathrm{Ar}]3d^6$ .

Why does the ionization energy dip at boron and again at oxygen?

Both dips are small breaks in the general rise across a period, and each has a clean cause. Beryllium closes the $2s$ subshell ( $2s^2$ ); boron's next electron must go into the higher $2p$ subshell, so it is easier to remove and the energy drops from $9.323$ to $8.298$ eV. Nitrogen has a half-filled $2p^3$ with one electron in each $p$ orbital, an unusually stable arrangement; oxygen is forced to put a second electron into an already-occupied $2p$ orbital, and that pairing raises the electron's energy, so oxygen ionizes more easily than nitrogen ( $14.534$ versus $13.618$ eV). The underlying reason a half-filled subshell is stable is an exchange effect derived in the recalled quantum-mechanics articles; here it shows up as a measurable $\sim1$ eV notch.

Do electron configurations change for ions and excited atoms?

Yes. The neutral ground configuration is only one entry in a much larger table. Ionizing an atom removes the least-bound electrons, and because the filling order and the true energy order differ, the electrons removed are not always the last ones added — transition metals lose $4s$ before $3d$ . Exciting an atom promotes an electron to a higher orbital, producing an excited configuration with its own set of terms and its own spectrum. This is why "the configuration of iron" is really shorthand for the neutral ground state; $\mathrm{Fe}^+$ , $\mathrm{Fe}^{2+}$ , and excited neutral iron each have different configurations, and astrophysical spectra routinely show lines from all of them.

Why do chromium and copper break the pattern, but not iron?

Because the exception needs a near-degeneracy plus a special target. At chromium and copper the $3d$ and $4s$ orbitals are close enough in energy that a small extra stabilization can tip the balance — and promoting one $4s$ electron to $3d$ reaches a half-filled $d^5$ (chromium) or a filled $d^{10}$ (copper), both unusually stable configurations. Iron sits at $3d^6\,4s^2$ : promoting a $4s$ electron would give $3d^7\,4s^1$ , which is neither half-filled nor filled, so there is no comparable prize and the ordinary order stands. The exceptions are not random; they cluster exactly where a single electron move completes or half-completes a $d$ or $f$ subshell.

What is the difference between a shell and a subshell?

A shell is the set of all orbitals with the same principal number $n$ ; a subshell is the finer division within a shell by orbital angular momentum $\ell$ . The $n=3$ shell contains the $3s$ , $3p$ , and $3d$ subshells. In bare hydrogen every subshell of a shell has the same energy — the $\ell$ -degeneracy — so the distinction is cosmetic. In a many-electron atom screening splits the subshells apart, with lower $\ell$ lying lower, and the split can be large enough that a subshell of one shell drops below a subshell of the next ( $4s$ below $3d$ ). That splitting is the whole reason the periodic table is not just a simple $2,8,18,32$ staircase.

How do we actually know an atom's ground configuration?

From its spectrum, not from the rule. The rule predicts, but the ground configuration and ground term listed in the NIST database are inferred from measured line positions, fine-structure intervals, and Zeeman patterns. When the prediction and the spectrum disagree — as at chromium and copper — the spectrum wins, and the tables record the measured configuration. This is why the exceptions are known so precisely: they are not theoretical curiosities but observations. The ionization energies plotted in the sawtooth are among the most direct pieces of this evidence, since each one marks the energy at which a specific outer electron leaves.

Are there exceptions I actually need to memorize?

For a first pass, two: chromium $[\mathrm{Ar}]3d^5\,4s^1$ and copper $[\mathrm{Ar}]3d^{10}4s^1$ , because they are the cleanest and most cited. Beyond those, the same half-filled and filled-subshell logic produces a scattered list — niobium, molybdenum, ruthenium, rhodium, palladium, silver, and several lanthanides and actinides — but the point is not to memorize the list. The point is to recognize the mechanism: wherever $s$ and $d$ (or $s$ and $f$ ) orbitals are nearly degenerate, a single electron move that completes or half-completes the higher- $\ell$ subshell can lower the energy, and only the measured spectrum settles it. Memorize the two flagships and the reason, and you can reconstruct the rest.