Natural Occupation Numbers

In the world of quantum chemistry, simple models that assign electrons to fixed orbitals provide a useful but incomplete picture of molecular reality. This simplified view, epitomized by the Hartree-Fock approximation, neglects the intricate, correlated dance that electrons perform to avoid one another—a phenomenon known as electron correlation. The failure to account for this correlation, especially in complex situations like bond breaking, represents a significant knowledge gap that limits the predictive power of basic theories. This article introduces natural occupation numbers as a powerful concept to bridge this gap. You will learn how to look beyond simple orbital diagrams and see the true electronic structure. The first chapter, "Principles and Mechanisms," will demystify what natural occupation numbers are, how they are derived from the one-particle reduced density matrix, and what their values reveal about a system's electronic nature. Following this, the "Applications and Interdisciplinary Connections" chapter will demonstrate how these numbers are used in practice as indispensable tools for diagnosing theoretical models, building sophisticated calculations, and choreographing the complex dance of chemical transformations.

Imagine you're trying to describe the population of a bustling city. A simple approach would be to get a census count for each district. "District A: 20,000 people. District B: 35,000 people." This is a useful, but static, snapshot. It tells you where people are, on average, but it tells you nothing about the flow of traffic between districts, the daily commutes, or the intricate social and economic connections that make the city a living, breathing entity.

In the quantum world of molecules, the simplest picture—the ​​Hartree-Fock approximation​​—is much like this census. It assigns electrons to specific regions called ​​orbitals​​ and gives us a tidy list: "Orbital 1: 2 electrons. Orbital 2: 2 electrons. Orbital 3: empty." For a system perfectly described by this picture, the ​​occupation numbers​​ are neat integers: 2 or 0 for a closed-shell molecule. If a calculation for a two-electron system yields occupation numbers of exactly {2, 0}, we know this simple, single-configuration picture tells the whole story.

But electrons, much like people, are not static and independent. They repel each other, and this mutual avoidance, a phenomenon we call ​​electron correlation​​, is the "social dynamic" that our simple census misses. An electron isn't just in an orbital; it is actively dancing to avoid its neighbors. This cooperative dance means that no single orbital assignment is ever perfectly correct. The true electronic state is a rich mixture, a quantum superposition, of many different configurations. To capture this dynamic reality, we need a more powerful tool—one that can map out the "traffic" and "connections" between our quantum districts.

That tool is the ​​one-particle reduced density matrix​​, or ​​1-RDM​​. Let's call it $\gamma$. You can think of it as a detailed matrix-form atlas of the electron population. The elements on its main diagonal, $\gamma_{ii}$, tell you the average number of electrons in orbital $i$, just like a simple census. But the crucial difference lies in the off-diagonal elements, $\gamma_{ij}$. These numbers measure the quantum mechanical "coherence" or "mixing"—the electron "traffic"—between orbital $i$ and orbital $j$. If these off-diagonal terms are zero, our orbitals are independent. If they are non-zero, it’s a sign that electrons are delocalized across them, and our chosen orbital "districts" are not the most natural way to view the system.

So, how do we find the most "natural" districts? We perform a mathematical maneuver that is one of the most elegant and powerful in quantum chemistry: we diagonalize the density matrix. Think of it like rotating a map. If you're trying to describe a city grid that is tilted relative to North, your directions will be complicated mixes of "northeast" and "northwest". But if you rotate your map to align with the city's streets, your description becomes simple: "three blocks along Main St., two blocks along 1st Ave."

Diagonalizing the 1-RDM is precisely this act of finding the best-aligned quantum map. The new, perfectly-aligned "districts" that emerge are called the ​​natural orbitals​​. In the basis of these special orbitals, the 1-RDM becomes diagonal—all the off-diagonal "traffic" terms are zero. The values that now appear on the diagonal are the ​​natural occupation numbers​​, $n_k$. Each $n_k$ tells us the average number of electrons that "live" in that specific natural orbital $k$.

For instance, we might start with a 1-RDM in a basis of two arbitrary orthonormal orbitals that looks like this:

The non-zero off-diagonal element $0.2$ tells us our initial orbitals are "tilted" with respect to the true electronic structure. By finding the eigenvalues of this matrix, we discover the natural occupation numbers, which in this case are $n_1 \approx 1.63$ and $n_2 \approx 0.37$. These numbers, free from the complication of off-diagonal terms, give us a direct and physically meaningful measure of the electron distribution. The eigenvectors of the matrix would give us the precise recipe for constructing the new natural orbitals from our old ones.

The natural occupation numbers are not just a mathematical curiosity; they are a profound diagnostic tool. They obey a strict set of rules, and the way they deviate from the simple integer picture tells a rich story about the molecule's electronic character.

First, there are two fundamental laws they must obey:

1. ​​Conservation of Electrons​​: The sum of all the natural occupation numbers must equal the total number of electrons, $N$. That is, $\sum_k n_k = N$. This is an essential self-consistency check. No matter how complex the electron correlation, electrons are never created or destroyed. This property is mathematically guaranteed and holds for any valid wavefunction, from the simplest to the most complex.

2. ​​The Pauli Constraint​​: An individual occupation number is not completely free. The Pauli exclusion principle dictates that you cannot put more than one electron into the same spin-orbital. This translates into a strict rule for the occupation numbers of spin-orbitals: $0 \le n_k \le 1$. An occupation number can be 0 (empty), 1 (full), or any fraction in between, but never negative and never greater than 1. Any theoretical model that produces an occupation number of, say, $1.1$ for a spin-orbital is signaling a violation of fundamental physics and must be rejected. For the spin-summed spatial orbitals we often use, this limit becomes $0 \le n_k \le 2$.

With these rules in hand, we can interpret the spectrum of occupation numbers:

• ​​Near-Integer Occupations (e.g., 1.99, 0.01)​​: When occupation numbers are very close to 2 or 0, it means the simple Hartree-Fock picture is a very good approximation. Electron correlation is present, but it's a minor correction—what is often called ​​dynamic correlation​​. It’s like small, constant ripples on the surface of a lake.

• ​​Strongly Fractional Occupations (e.g., 1.15, 0.85)​​: This is the real smoking gun. When you find one or more occupation numbers that are far from 0 or 2—especially if they are close to 1—it is a definitive sign of ​​strong static correlation​​. It means that our simple picture of "one orbital filled, another empty" has completely broken down. The system cannot be described by a single configuration; at least two (or more) electronic configurations are so close in energy that the system is a true, inseparable hybrid of them. A set of occupations like {1.99, 1.98, 1.15, 0.85, 0.01} immediately tells a chemist that while four of the electrons are behaving simply, two of them are involved in a very strong interaction that requires a ​​multi-reference​​ theoretical description.

There is no better place to see the power and beauty of natural occupation numbers than in the breaking of a chemical bond. Let's take the simplest molecule, $\mathrm{H}_2$.

At its normal equilibrium bond length, $\mathrm{H}_2$ is well-described by placing both electrons in a single, sausage-shaped bonding molecular orbital. The corresponding antibonding orbital is empty. The natural occupation numbers are close to {2, 0}.

Now, let's start pulling the two hydrogen atoms apart. As we stretch the bond, the simple molecular orbital picture begins to fail spectacularly. The electrons, once happily shared in the bond, become increasingly localized on their respective atoms. The system gains what we call ​​diradical character​​—it behaves less like a single molecule and more like two independent hydrogen atoms.

How do natural occupation numbers capture this dramatic story? Beautifully. We can model this system with a wavefunction that is a mixture of two main configurations: the primary one with both electrons in the bonding orbital ($\phi_{bond}$), and a secondary one with both electrons in the antibonding orbital ($\phi_{anti}$). The wavefunction looks something like $\Psi = c_0 \Phi(\phi_{bond}^2) + c_1 \Phi(\phi_{anti}^2)$. The natural occupation numbers for this system are found to be $n_{bond} = 2c_0^2$ and $n_{anti} = 2c_1^2$.

• ​​At equilibrium​​: The bonding configuration dominates. $c_0$ is close to 1, and $c_1$ is very small. The occupations are therefore close to {2, 0}.
• ​​At dissociation​​: As the atoms move infinitely far apart, the energy of the two configurations becomes equal. To describe two neutral hydrogen atoms, the wavefunction must be an equal mixture, meaning $|c_0| \approx |c_1| \approx 1/\sqrt{2}$. Plugging this in, we find the occupation numbers become $n_{bond} \approx 2(1/\sqrt{2})^2 = 1$ and $n_{anti} \approx 2(1/\sqrt{2})^2 = 1$.

The occupation numbers have smoothly transitioned from {2, 0} to {1, 1}! This is a profound result. The numbers are telling us that at dissociation, the system is perfectly correlated. There is exactly one electron in the "bonding" natural orbital and one electron in the "antibonding" natural orbital. These two orbitals themselves have transformed, at this great distance, into orbitals localized on each atom. The language of natural orbitals has perfectly and quantitatively described the process of a covalent bond transforming into two independent radical atoms.

Even more elegantly, the Coulson-Fischer theory, which describes the bond using overlapping but non-orthogonal atomic orbitals, gives a similar result. It shows that as the overlap $S$ between the two atomic orbitals goes to zero, the occupation numbers of the two natural orbitals each approach 1, telling the exact same story of bond dissociation.

In the end, natural occupation numbers are far more than just numbers. They are a lens. Through them, we can peer past the convenient fictions of simple orbital diagrams and see the intricate, correlated dance of electrons that lies at the very heart of chemical structure, reactivity, and life itself.

We have spent some time understanding the machinery behind natural orbitals and their occupation numbers, these eigenvalues that pop out from the one-particle density matrix. At first glance, they might seem like a rather abstract piece of quantum bookkeeping. But to think of them this way is to miss the point entirely. These numbers are not just bookkeeping; they are a story. They are a secret language that electrons use to describe their collective lives, and learning to read this language opens up a spectacular new window onto the chemical world.

Imagine the electronic wavefunction of a molecule is a piece of music. In the simplest, and often quite boring, cases, the music is a single, sustained chord, where some notes are played fully and others are silent. This corresponds to the simple picture where some electron orbitals are completely full (occupation of 2) and others are completely empty (occupation of 0). The natural occupation numbers (NOONs) simply report these integer values. But where is the fun in that? The interesting music of chemistry—the reactions, the light absorption, the breaking and forming of bonds—happens when this simple chord dissolves into a complex, evolving harmony. It happens when electrons are not confined to one orbital but are shared, correlated, and delocalized across several. And the NOONs, by departing from their integer values, tell us exactly which notes are being played in this rich quantum symphony, and how loudly. Let us listen to some of the stories they tell.

Before you can solve a problem, you must first know you have one. In quantum chemistry, we have a wonderful collection of theoretical models, starting with the workhorse Hartree-Fock theory. These models often give us a single, neat electronic configuration for a molecule. But how do we know if this simple picture is telling the truth, or if it is a dangerously misleading caricature?

This is where natural occupation numbers serve as our indispensable compass. A well-behaved molecule, happy in its stable ground state, will typically have NOONs that are very close to integers, say $1.98$ for a strongly occupied orbital and $0.02$ for a weakly occupied one. These small deviations are the gentle hum of what we call dynamic correlation—the ever-present tendency of electrons to avoid bumping into each other. Our standard high-accuracy methods are brilliant at handling this.

But sometimes, the NOONs tell a more alarming story. Suppose we find an orbital that should have been occupied with an occupation of only $1.8$, and another that should have been empty with an occupation of $0.2$. This is a red flag. The compass needle is spinning wildly. It signals a severe case of static correlation, a condition where two or more electronic configurations are nearly equal in energy and the molecule simply cannot make up its mind. The single-configuration picture is not just approximate; it is qualitatively wrong. It is a "sick" theory for that particular problem.

Computational chemists have developed practical rules-of-thumb, guided by these principles. A common practice is to become highly suspicious of any single-reference calculation if any of the frontier orbitals show an occupation number between, say, $0.1$ and $1.9$. This diagnostic capability is not merely academic. It is a crucial tool for avoiding wasted effort on fantastically expensive calculations. Before deploying a powerful but sensitive method like CCSD(T), which is often called the "gold standard" of quantum chemistry, a savvy chemist will first run a cheaper calculation to inspect the system's NOONs. If the numbers indicate strong static correlation, they know that the very foundation of the CCSD(T) method is shaky, and its results are likely to be nonsense. The NOONs, along with related measures like the $T_1$ diagnostic, are the first line of defense against producing beautifully precise, yet utterly meaningless, numbers.

Once our NOON compass tells us we are in dangerous territory, what do we do? We need a better map. We need to build a more sophisticated model, a multireference model, that embraces the molecule's indecisive nature from the start. The most powerful of these is the Complete Active Space Self-Consistent Field (CASSCF) method.

The core idea of CASSCF is to admit that we cannot possibly keep track of the correlated motion of all electrons in a molecule simultaneously—that would be computationally impossible. So, we make a compromise. We divide the orbitals into three groups: a set of "inactive" core orbitals that are always full, a set of "virtual" orbitals that are always empty, and a small, crucial set of "active" orbitals where the real drama is unfolding. Within this "active space," we solve the problem exactly, allowing the electrons to arrange themselves in all possible ways.

But how do we choose this active space? It is here that the NOONs transform from a diagnostic compass into an architect's blueprint. The NOONs from a preliminary, less expensive calculation tell us exactly which orbitals are "active." Is there an orbital with an occupation of $1.1$? It must go into the active space. Is there another with an occupation of $0.9$? It's active, too. What about an orbital with an occupation of $1.998$? While not exactly $2$, this orbital is clearly behaving itself. It is essentially a spectator. Including it in our active space would be a waste of computational effort; we can safely relegate it to the inactive set, confident that we are not missing the essential physics.

Therefore, the general strategy is to identify all orbitals whose occupation numbers deviate substantially from $0$ or $2$—say, those in the range $[0.02, 1.98]$—and designate them as the active space. This allows us to focus our most powerful computational microscope precisely on the region of the molecule where the interesting, multiconfigurational chemistry is happening.

With the ability to diagnose problems and build better models, we can now use NOONs to do something truly exciting: watch chemistry happen. Let's take the simple ethylene molecule, $\mathrm{C}_2\mathrm{H}_4$. In its planar ground state, it has a strong carbon-carbon double bond. A CASSCF calculation for this geometry tells a simple story: the $\pi$ bonding orbital has an occupation of about $1.98$, and the $\pi^*$ antibonding orbital has an occupation of $0.02$. It is a classic closed-shell molecule.

Now, let's grab one end of the molecule and twist it. As we rotate towards a $90^\circ$ angle, the overlap that creates the $\pi$ bond is destroyed. The bond breaks. What do the NOONs do? They beautifully choreograph this entire process. At $45^\circ$, the occupations might be $1.50$ and $0.50$. The bond is clearly weakened. At $90^\circ$, where the bond is fully broken, the NOONs become nearly $1.0$ and $1.0$.

What does this $\{1.0, 1.0\}$ pattern mean? It is the unambiguous signature of a singlet biradical—a species with two electrons on two different orbitals, whose spins are paired up to give an overall singlet state. The NOONs give us a continuous, quantitative measure of "biradical character." They trace the molecule's journey from a stable double bond to a reactive biradical intermediate.

This story is not unique to ethylene. It is the story of every covalent bond that breaks homolytically. In the triple bond of the $\mathrm{N}_2$ molecule, for instance, we can watch the six occupation numbers corresponding to the bonding and antibonding orbitals march from a pattern near $\{2, 2, 2, 0, 0, 0\}$ at equilibrium to $\{1, 1, 1, 1, 1, 1\}$ at dissociation. The point along this path where the occupations make a marked departure from integer values is the precise point where our simple textbook pictures of bonding break down and the full, complex quantum reality takes over. In fact, for simple "two-electron, two-orbital" models, one can show mathematically that the occupation numbers change smoothly as a function of geometry, perfectly capturing the transition from a non-correlated to a fully correlated state. We can even use the full spectrum of fractional occupation numbers to compute intuitive quantities like the "number of effectively unpaired electrons," which can be estimated using the formula $N_u = \sum_i n_i(2-n_i)$, where the sum runs over all spatial orbitals.