Internal Coordinates

How do we describe the shape of a molecule? This simple question is central to chemistry, as a molecule's structure dictates its energy, stability, and reactivity. The most direct approach—listing the Cartesian (x, y, z) coordinates of every atom—is complete but deeply inefficient. It buries the essential information about molecular shape under irrelevant data about the molecule's overall position and orientation in space. To truly understand molecular behavior, we need a more intuitive and efficient language, one that focuses exclusively on the molecule's intrinsic geometry.

This article explores the powerful concept of internal coordinates, the chemist's native language for describing molecular structure. In the first section, ​​Principles and Mechanisms​​, we will explore how we move from the cumbersome 3N Cartesian coordinates to the elegant 3N-6 internal degrees of freedom. We will define the core components—bond lengths, angles, and dihedral angles—and uncover the profound trade-off between simplifying potential energy and the resulting complexity in kinetic energy. Following that, the section on ​​Applications and Interdisciplinary Connections​​ will reveal the surprising versatility of this concept, showing how the same principles used to model molecules are applied to design robotic arms, analyze planetary orbits, and decode the complex machinery of life itself. By mastering this framework, we can move from a clumsy list of positions to a powerful understanding of shape and function.

Imagine trying to describe a ballet dancer’s performance. You could, in principle, record the precise $(x, y, z)$ coordinates of every cell in their body at every millisecond. This would be a flawlessly complete description, yet it would also be utterly incomprehensible. It tells you everything and explains nothing. You would be drowning in data, missing the breathtaking beauty of the plié or the pirouette. The essence of the dance lies not in absolute positions, but in the relative arrangement of limbs, the angles of joints, the elegant curvature of the body—in short, the dancer's internal geometry.

The world of molecules is much the same. To understand a molecule, to predict its properties, its stability, or how it will react, we need to understand its energy. And a molecule’s potential energy depends not on where it is in space or how it’s oriented, but on its shape.

Let’s start with a simple molecule, like water, $\text{H}_2\text{O}$. It has three atoms ($N=3$). The most straightforward way to tell a computer where these atoms are is to list their Cartesian coordinates: $(x_1, y_1, z_1)$ for the first hydrogen, and so on. That’s $3N=9$ numbers in total. For a slightly more complex molecule like ethane, $\text{C}_2\text{H}_6$, with its eight atoms, you’d need $3 \times 8 = 24$ numbers to pin down its geometry. This seems manageable, but it's already hiding a deep inefficiency.

The potential energy of an isolated water molecule doesn’t change if we pick it up and move it to the other side of the room. It doesn’t care if we rotate it to face a different direction. These motions—three for translation (along the x, y, and z axes) and three for rotation (around the x, y, and z axes)—are "external" motions. They change the $3N$ Cartesian numbers, but they have absolutely no effect on the molecule's internal structure or its energy. Using Cartesian coordinates to describe molecular energy is like insisting on using global GPS coordinates to describe the dancer's pose; it’s correct, but it mixes the essential with the irrelevant.

The first step in truly understanding a physical system is to identify its symmetries and discard the irrelevant information. For a molecule, the irrelevance lies in its absolute position and orientation. We need a language that speaks only of the molecule's intrinsic shape.

What remains when we ignore overall translation and rotation? We start with $3N$ total degrees of freedom. We subtract three for translation and three for rotation (for any non-linear molecule). The result is the magic number that governs molecular shape: $\boldsymbol{3N-6}$. This is the number of ​​internal degrees of freedom​​—the true number of variables needed to define a molecule's shape and, therefore, its potential energy. For water ($N=3$), this is $3(3)-6=3$. For ethane ($N=8$), it's $3(8)-6=18$. We’ve already simplified the problem significantly, going from 24 coordinates down to 18 for ethane.

Chemists have long had an intuitive language for these $3N-6$ freedoms, a language built from the very connections that hold molecules together. We call these ​​internal coordinates​​:

• ​​Bond Lengths:​​ The distance between two bonded atoms. This is the most basic element of molecular shape.
• ​​Bond Angles:​​ The angle formed by three atoms ($A-B-C$). This tells us how the bonds are arranged relative to each other.
• ​​Dihedral (or Torsion) Angles:​​ The twist around a central bond, defined by a sequence of four atoms ($A-B-C-D$). This is a crucial, and often less intuitive, coordinate that describes the three-dimensional nature of molecules. It's the reason butane has different "conformers" (like anti and gauche), and it's what governs the folding of proteins. A dihedral angle of $0^\circ$ or $180^\circ$ means the four atoms lie in a single plane.

This set of coordinates—lengths, angles, and dihedrals—forms a powerful toolkit. One systematic way to build a molecule's geometry from them is called a ​​Z-matrix​​. You place the first atom at the origin, the second along an axis, the third in a plane, and every subsequent atom is defined by one length, one angle, and one dihedral relative to atoms already placed. This procedure automatically strips away the six external degrees of freedom.

The great power of internal coordinates is their inherent invariance. If you take a molecule and translate it by some vector $\mathbf{a}$, the Cartesian coordinates of every atom change. But the distance between any two atoms, say atom $i$ and atom $j$, remains $d'_{ij} = \|(\mathbf{r}_j + \mathbf{a}) - (\mathbf{r}_i + \mathbf{a})\| = \|\mathbf{r}_j - \mathbf{r}_i\| = d_{ij}$. The distances, angles, and dihedrals are all unchanged. If you run a molecular dynamics simulation and only save the history of the internal coordinates, you have captured the complete story of the molecule's flexing, vibrating, and shape-shifting. The only information you’ve lost is the rather boring tale of where the whole molecule was drifting and how it was tumbling in space.

We have found a coordinate system that dramatically simplifies the description of potential energy. The landscape we need to explore to find a molecule's most stable structure is now only $3N-6$ dimensional, and the coordinates we use are chemically intuitive. This makes finding the minimum energy geometry much more efficient. So, is there a catch?

Of course there is, and it's a beautiful one. In physics, complexity rarely vanishes; it merely transforms. The price we pay for a simple potential energy description is a complicated ​​kinetic energy​​.

To see this, let's first construct the most beautiful description for kinetic energy. The classical kinetic energy is $T = \frac{1}{2} \sum m_i \dot{\mathbf{r}}_i^2$. This expression is complicated by the different masses $m_i$. We can simplify it by inventing ​​mass-weighted Cartesian coordinates​​, defined by the simple scaling $\mathbf{q}_i = \sqrt{m_i} \mathbf{r}_i$. In this clever coordinate system, the kinetic energy becomes a pristine sum of squares: $T = \frac{1}{2} \sum \dot{\mathbf{q}}_i^2$. Motion in this mass-weighted space is like motion in a perfectly flat, Euclidean world, where every particle effectively has a mass of one. The laws of motion are as simple as they can be.

Now, what happens when we switch from this beautifully simple Cartesian world to our chemically intuitive internal coordinates? The transformation is non-linear. A bond length, for instance, involves a square root: $r = \sqrt{(x_2-x_1)^2 + \dots}$. This non-linear mapping takes the flat, Euclidean space of mass-weighted coordinates and bends it. The $3N-6$ dimensional space described by internal coordinates is intrinsically ​​curved​​.

Imagine an ant living on the surface of a sphere. It thinks it's walking in a straight line, but from our 3D perspective, we see its path is curved. The ant's "internal" world has a non-Euclidean geometry. The same is true for molecules. In the curved space of internal coordinates, the expression for kinetic energy becomes a complicated affair. A "straight" motion along one internal coordinate (like a bond stretch) can induce motion in another (like an angle bend), not because of a force, but because of the curvature of the coordinate system itself. This is known as ​​kinetic coupling​​.

This coupling is mathematically captured by the ​​Wilson G-matrix​​, which can be thought of as the metric tensor that defines the geometry of this curved molecular configuration space. The kinetic energy operator, which is a simple Laplacian in mass-weighted Cartesians ($\hat{T} \propto \sum \frac{\partial^2}{\partial X_k^2}$), becomes the fearsome Laplace-Beltrami operator in internal coordinates, full of coordinate-dependent coefficients and cross-derivatives ($\frac{\partial^2}{\partial q_\alpha \partial q_\beta}$).

So we face a fundamental choice: a simple kinetic energy and a complex, high-dimensional potential energy (Cartesians), or a simple, low-dimensional potential energy and a complex kinetic energy (internals). For many problems, like finding a molecule's stable shape, simplifying the potential energy landscape is the winning strategy.

Internal coordinates are a powerful tool, but a master craftsperson knows their tool's limitations. There are situations where this elegant machinery breaks down. What happens if three atoms in a row, say in a molecule like $\text{CO}_2$, become perfectly linear? The bond angle is $180^\circ$. A dihedral angle that relies on these three atoms as a reference becomes ill-defined, just as the concept of longitude becomes meaningless at the North and South Poles. At these "singularities," the math falls apart.

For systems that are naturally linear, or for weakly-bound clusters of atoms (like a trio of argon atoms) where there are no fixed "bonds" to define, or for the repeating lattice of a crystal, forcing a description based on a single bonding pattern is unnatural and fragile. In these cases, the robustness of Cartesian coordinates makes them the superior choice.

What about the most exciting parts of chemistry—chemical reactions where bonds break and form? Consider the tautomerization of 2-pyridone, where a proton hops from a nitrogen atom to an oxygen atom. A minimal set of internal coordinates defined for the reactant, which includes an N-H bond length, becomes hopelessly ill-suited to describe the product, which has an O-H bond instead.

Here, chemists employ a wonderfully counter-intuitive trick: they use more coordinates than they need. By defining a ​​redundant set of internal coordinates​​ that includes variables for all potential bonds—both the N-H bond that is breaking and the O-H bond that is forming—we create a more flexible and robust description. It seems this would over-constrain the system, but with the right mathematical machinery (using the Moore-Penrose pseudo-inverse to handle the redundancy), the optimization algorithm can intelligently navigate the reaction path. It distributes the geometric changes across all the relevant degrees of freedom, smoothly transforming from reactant to product. This turns a logical problem (redundancy) into a powerful practical solution.

Ultimately, the journey from the raw $3N$ Cartesian coordinates to a refined set of internal coordinates is a story of abstraction and simplification. By focusing on the intrinsic geometry, we discard the irrelevant distractions of a molecule's place in the world. This allows us to build a more intuitive and efficient picture of the potential energy that governs its behavior. The price we pay is a more complex description of kinetic energy, a complexity that reveals a deep and beautiful truth: the configuration space of molecules is not flat, but a curved manifold whose geometry dictates the dynamics of atoms. By learning the rules of this geometry, and by knowing when to bend them, we can unlock the secrets of molecular structure and reactivity.

This is just the first step. By applying further transformations based on molecular symmetry, we can arrive at ​​symmetry coordinates​​, and finally, at ​​normal coordinates​​, which represent the pure, uncoupled vibrations of the molecule—the notes in its symphony. Each step is a move away from a literal, clumsy description towards a more essential and powerful understanding. It is the very essence of the scientific endeavor.

Now that we have learned the alphabet and grammar of internal coordinates, what sorts of stories can we tell? We have seen that describing a molecule by its bond lengths, angles, and torsions is a more natural language than a sterile list of $x, y, z$ positions. This is not merely a matter of mathematical taste. As we shall see, choosing the right way to look at a system is often the most crucial step in understanding it. This choice transforms intractable problems into simple ones and reveals deep, surprising connections between fields that, on the surface, have nothing to do with one another. This is where the real fun begins.

Let’s start in the natural home of internal coordinates: computational chemistry. Suppose we want to find the most stable shape of a molecule, say methane, $\text{CH}_4$. We can think of this as a ball rolling on a hilly landscape—the potential energy surface—trying to find the bottom of a valley. Our job is to tell the computer how to describe this landscape. If we use Cartesian coordinates, we have to keep track of $3N$ variables, one for each $x, y, z$ position of each of the $N$ atoms. For methane, with $N=5$ atoms, this is $15$ variables. But we know that the energy of the molecule doesn't change if we pick it up and move it, or if we turn it around. These are the $6$ degrees of freedom of overall translation and rotation. They are "useless" coordinates for describing the molecule's shape.

By switching to internal coordinates, we throw away this useless information from the start. We describe the molecule only by what defines its shape: its bond lengths and angles. The number of variables we need to track drops to $3N-6$, which for methane is $15-6=9$. In one stroke, we have simplified the problem by focusing only on the coordinates that matter for the molecule's internal energy. For large molecules, this efficiency gain is enormous.

But the real beauty is not just about efficiency; it's about intuition. Internal coordinates are the native language of chemistry. A chemist doesn't think about the absolute Cartesian position of a chlorine atom; they think about whether it is on the same side (cis) or the opposite side (trans) of a double bond. This chemical concept maps perfectly onto a single internal coordinate: a dihedral angle. To change a molecule from its cis form to its trans form in a computer model, you don't need to perform some complicated series of Cartesian shifts. You simply change one number—the value of the key dihedral angle—from $0^\circ$ to $180^\circ$. The language of internal coordinates directly mirrors the logic of chemical transformations.

Of course, nature is never quite so simple. For a long, flexible molecule like an alkane chain, the dance of atoms is complex. The stiff, high-energy vibrations of bond stretching are very different from the soft, low-energy rotations of dihedral angles. Cartesian coordinates mix them all together, creating a horribly complicated energy landscape with long, narrow, winding valleys that are difficult for optimization algorithms to navigate. Internal coordinates help to separate these motions, making the landscape smoother and the path to the energy minimum much shorter. However, this convenience comes with its own perils. If three atoms in a chain straighten into a line (a bond angle of $180^\circ$), the definition of the next dihedral angle becomes mathematically unstable—a singularity. This is like the gimbal lock that plagued early spacecraft. Modern computational chemistry often uses a clever compromise called Redundant Internal Coordinates (RICs), which provides the chemical intuition of internal coordinates while avoiding their mathematical pitfalls, giving us the best of both worlds.

Here is where our story takes a surprising turn. The mathematical framework we developed for molecules is, in fact, a universal language for describing any system of connected objects. What does a molecule of pentane have in common with a satellite deploying its solar panels? Everything! Both can be seen as a kinematic chain: a series of rigid bodies connected by hinges.

Imagine a simplified solar array as a chain of five segments. If the length of each segment and the angle at each hinge are fixed, the only way the array can move is by twisting at the hinges. These twists are nothing more than dihedral angles. The entire complex motion of the array unfolding from its stowed to its deployed configuration can be described by changes in just two or three of these torsional angles. The same set of equations that governs the flexing of a molecule can be used by an engineer to design the motion of a robotic arm or a complex mechanical linkage. For a simple, non-cyclic chain of $N$ parts with fixed segment lengths and hinge angles, the number of internal degrees of freedom is simply $N-3$, the number of independent torsional twists. This is a beautiful example of the unity of scientific principles.

The analogy extends to even grander scales. Consider the heavens. In celestial mechanics, the motion of a planet around a star (a two-body problem) is described by a set of Keplerian orbital elements. How does this relate to a simple triatomic molecule like water? At first, the question seems absurd. But let's look at how we break down the problem in each case.

For a water molecule, we first remove the overall translation and rotation, leaving us with $3N-6=3$ internal coordinates that define its intrinsic shape—two bond lengths and one bond angle. For the planet, we first separate out the motion of the center of mass of the two-body system. We are left with describing the relative orbit. This orbit is an ellipse in space. Its intrinsic shape is defined by just two parameters: the semi-major axis $a$ (its size) and the eccentricity $e$ (its roundness). Its orientation in space is defined by three angles (inclination $i$, longitude of ascending node $\Omega$, and argument of periapsis $\omega$).

Do you see the parallel? The bond lengths and angle of the water molecule are the "internal shape parameters," analogous to the orbital shape parameters $a$ and $e$. The three Euler angles that orient the water molecule in space are analogous to the three angles $i, \Omega, \omega$ that orient the orbit in space. The way we decompose a molecule's degrees of freedom and the way we decompose an orbit's degrees of freedom are structurally identical. Nature, it seems, uses the same organizational principles for the microscopic and the cosmic.

Nowhere is the power of internal coordinates more apparent than in the bewilderingly complex world of biology. A protein can have tens of thousands of atoms. A full description in Cartesian coordinates would involve hundreds of thousands of variables—a hopeless mess. To make sense of this, we must focus on the few coordinates that truly matter for the protein's function.

A classic tool in structural biology is the Ramachandran plot. It's a simple 2D map showing the allowed combinations of the two most important backbone dihedral angles, $\phi$ and $\psi$, for the amino acids in a protein. But what does this map really represent? It is a radical projection of the protein's entire, ($3N-6$)-dimensional potential energy landscape onto just two coordinates. When we look at this plot, we are ignoring the state of all the other $3N-8$ internal degrees of freedom—the side-chain torsions, the bond angles, the neighboring residues, and so on. A particular $(\phi, \psi)$ point on the map doesn't correspond to a single structure, but to a vast family of structures that differ in all those "hidden" coordinates. This is why some regions of the plot might seem accessible for a tiny peptide but become forbidden in the context of a folded protein; a side chain, whose coordinates are not shown on the plot, would clash with another part of the protein. The Ramachandran plot is a powerful simplification, and understanding it as a projection of internal coordinate space is key to interpreting it correctly.

We can also use internal coordinates to build powerful descriptions of large-scale biological transformations. Consider the remarkable ability of a DNA double helix to switch from its common right-handed "B-form" to a rare left-handed "Z-form." To study this transition, we don't need to track every single atom. Instead, we can design a few clever collective variables built from key internal coordinates. The essential changes are: (1) an inversion of the helix's handedness, which can be captured by a generalized dihedral angle called the helical twist; (2) a flipping of the purine bases, captured by their glycosidic torsion angles; and (3) a change in the sugar ring's conformation, or "pucker," which is described by internal angles within the ring. By focusing on just these few crucial coordinates, we can create a low-dimensional map that describes the entire, complex transition from B-DNA to Z-DNA.

The concept is even more flexible than that. What if we want to describe the architecture of a protein made of large, rigid chunks like $\alpha$-helices? We can invent generalized internal coordinates that operate at a higher level of abstraction. We can define a "bond length" as the distance between the centers of mass of two helices, and "angles" and "dihedrals" to describe their relative orientation. This is no longer a standard Z-matrix built on covalent bonds, but it uses the same powerful idea: to describe a complex system's internal structure using a set of distances and angles that are independent of the system's overall position in space.

The journey ends at its most abstract and powerful. The idea of an "internal degree of freedom" can be pushed beyond simple geometry. Imagine watching a computer simulation of a molecule vibrating and twisting. The raw data is a blur of atomic motion. How can we find the dominant, meaningful patterns? One way is through a statistical technique called Principal Component Analysis (PCA). PCA finds the collective motions that account for the most variance in the trajectory. But the motions it finds depend entirely on the coordinates we use to describe the system. If we use mass-weighted Cartesian coordinates, PCA identifies the collective vibrations of the molecule, which are related to its normal modes. If, however, we first convert the trajectory into internal coordinates and then perform PCA, we find the dominant motions described in the language of chemistry—concerted changes in bond lengths, angles, and torsions. How we choose to describe the world fundamentally determines what we see in it.

The ultimate generalization of this idea takes us to materials science and thermodynamics. Consider a crystalline solid at a given temperature and pressure. Its state is not just defined by the positions of its atoms. It might have other "internal degrees of freedom," such as the concentration of vacancies or defects in its crystal lattice, or an overall magnetic ordering. These are not geometric coordinates in the simple sense, but they are parameters that describe the internal state of the material. Just as a molecule settles into the conformation that minimizes its potential energy, the material settles into the state that minimizes its appropriate thermodynamic potential—in this case, the Gibbs free energy, $g$. The equilibrium defect concentration is found not by moving atoms, but by minimizing $g$ with respect to the concentration $c$, satisfying the condition $(\partial g / \partial c)_{T,P} = 0$.

From a simple way to define a molecule's shape, we have arrived at a universal principle. The concept of an internal degree of freedom is a powerful tool for simplifying complexity. It allows us to parameterize the essential state of a system—be it a molecule, a protein, a solar panel, or a block of steel—and then apply the fundamental laws of physics and chemistry to find its most stable configuration. Choosing the right coordinates is the art of seeing the joints in nature's construction, and it is the key to understanding how it moves.