Reading water off the page: geometry, orbitals, acidity, and spectra

Water is the molecule we know best and understand least casually. Everyone can draw it — a fat oxygen with two hydrogen ears — but almost every interesting property of the substance is hiding in details that the cartoon leaves out: the exact angle between the bonds, the shape of the orbitals the lone pairs live in, the fact that pure water is both an acid and a base at once, and the reason a swimming pool is blue when a glassful is not.

This post takes the single molecule \(\mathrm{H_2O}\) and reads it off the page in four passes — geometry, orbitals, acid–base behavior, and spectra — because those four are not separate facts to memorize. Each one explains the next. The bend gives the dipole; the dipole comes from lone pairs; the lone pairs are the base sites; and the bonds whose stretches we will see in the infrared are the same bonds whose faint overtones turn the ocean blue.

1. The shape

Two numbers fix the geometry of an isolated water molecule in the gas phase: the O–H bond length, about 0.958 Å, and the H–O–H angle, about 104.5°.¹ The molecule is bent, and that bend is the whole story.

The standard explanation is VSEPR. Oxygen brings six valence electrons; two of them pair with the hydrogens to make bonds, and the remaining four sit as two lone pairs. That is four “electron domains” around the oxygen — two bonds and two lone pairs — and four domains arrange themselves tetrahedrally. We only see the two O–H bonds, so the molecule looks bent rather than tetrahedral, and the ideal tetrahedral angle of 109.5° is squeezed down to 104.5° because lone pairs are fatter than bonds and push harder.

That is the right intuition, but the precise angle is telling you something more specific. A pure \(sp^3\) picture predicts 109.5°; the real 104.5° sits between 109.5° and the 90° you would get from unhybridized \(p\) orbitals. The bonds, in other words, carry more p-character than an ideal \(sp^3\) hybrid, and the lone pairs carry correspondingly more s-character. We will see that asymmetry again in a moment, from a completely different direction, and it will turn out to be measurable.

One consequence is immediate. Because the molecule is bent, the two O–H bond dipoles do not cancel — they add to a net dipole moment of 1.85 D, pointing along the symmetry axis with the oxygen as the negative end. A linear water molecule would be nonpolar, would not hydrogen-bond the way it does, would not dissolve salt, and would not be a greenhouse gas. Almost everything water does follows from the fact that it does not lie flat.

2. The electrons

To go below VSEPR we need the molecular orbitals, and to get those cleanly we use the molecule’s symmetry. Bent water belongs to the point group \(C_{2v}\): it has a two-fold rotation axis (the bisector of the H–O–H angle) and two mirror planes (the molecular plane and the plane perpendicular to it through the axis). Put the \(C_2\) axis along \(z\) and the molecule in the \(yz\) plane.

Symmetry sorts the atomic orbitals into a small number of labelled boxes. The two hydrogen 1s orbitals combine into two symmetry-adapted combinations: a symmetric one (their sum, which is unchanged by every symmetry operation — label \(a_1\)) and an antisymmetric one (their difference — label \(b_2\)). The oxygen valence orbitals sort themselves too: the \(2s\) and \(2p_z\) are both \(a_1\); the \(2p_y\) is \(b_2\); and the \(2p_x\), which sticks straight out of the molecular plane, is \(b_1\) — and it is alone in its box, because no hydrogen combination has \(b_1\) symmetry.

Orbitals only mix with other orbitals of the same label. So:

• The \(a_1\) set — O\(\,2s\), O\(\,2p_z\), and the symmetric H combination — mix into a low bonding orbital (\(2a_1\)), a roughly nonbonding one (\(3a_1\)), and an empty antibonding one.
• The \(b_2\) set — O\(\,2p_y\) and the antisymmetric H combination — make a bonding orbital (\(1b_2\)) and an empty antibonding partner.
• The lonely \(b_1\) orbital (O\(\,2p_x\)) has nothing to mix with, so it stays a pure, nonbonding oxygen \(p\) orbital (\(1b_1\)), perpendicular to the molecular plane.

Filling the eight valence electrons from the bottom gives the ground configuration \((2a_1)^2(1b_2)^2(3a_1)^2(1b_1)^2\):

Here is where the orbital picture earns its keep, and where the cartoon of “two identical lone pairs” quietly breaks. In the canonical molecular orbitals above, the two lone pairs are not equivalent. One of them, \(1b_1\), is a pure oxygen \(p\) orbital pointing out of the molecular plane. The other, \(3a_1\), lies in the plane and is a mix of oxygen \(s\) and \(p\). They have different shapes and different energies — and that difference is something you can measure.

Photoelectron spectroscopy shoots photons at water vapor and records the energy needed to knock each electron out. If the two lone pairs were identical, you would see a single lone-pair band. Instead you see two clearly separated ionization energies — about 12.6 eV for \(1b_1\) and 14.7 eV for \(3a_1\) — followed by \(1b_2\) near 18.5 eV and the deep \(2a_1\) near 32 eV.² The spectrum is the receipt: the “two equivalent rabbit-ear lone pairs” you may have drawn in an organic-chemistry class are a localized description — a perfectly valid mathematical re-mixing of the canonical orbitals that is handy for thinking about hydrogen bonding — but it is not what the electrons individually look like, and the photoelectron spectrum knows the difference. Both pictures describe the same total electron density; only one of them has orbitals you can ionize one at a time.

3. Acid and base at once

Water’s most underappreciated trick is that it reacts with itself. In any glass of pure water, a tiny fraction of the molecules are constantly trading a proton:

\[2\,\mathrm{H_2O} \;\rightleftharpoons\; \mathrm{H_3O^+} + \mathrm{OH^-}\]

One water molecule donates a proton (acting as a Brønsted acid, leaving behind hydroxide \(\mathrm{OH^-}\)); the other accepts it (acting as a base, becoming hydronium \(\mathrm{H_3O^+}\)). A substance that is both acid and base is amphoteric, and the proton it accepts lands exactly on one of the lone pairs from the previous section. The orbital story and the acid–base story are the same story.

The position of this equilibrium is the autoionization constant,

\[K_w = [\mathrm{H_3O^+}][\mathrm{OH^-}] = 1.0\times10^{-14}\quad(\text{at }25\,^{\circ}\mathrm{C}).\]

At 25 °C the two concentrations are equal at \(10^{-7}\,\mathrm{M}\), which is where “neutral = pH 7” comes from. But \(K_w\) is an equilibrium constant, and like any equilibrium it shifts with temperature. Autoionization is endothermic, so warming water pushes it to the right: more ions, larger \(K_w\), smaller \(\mathrm{p}K_w\). Crucially, the neutral pH — where \([\mathrm{H_3O^+}] = [\mathrm{OH^-}]\) — is always \(\mathrm{p}K_w/2\), so it slides too:

So a pot of pure water at 100 °C is perfectly neutral at pH ≈ 6.1, not 7 — neutrality is about \([\mathrm{H_3O^+}] = [\mathrm{OH^-}]\), not about a magic number. This is the kind of fact that the “pH 7 is neutral” shorthand quietly hides.

How acidic is water, taken as an acid? It depends entirely on the convention you use, which is worth stating plainly because the textbooks disagree. In the autoionization convention above, \(\mathrm{p}K_a(\mathrm{H_2O}) = \mathrm{p}K_w = 14.0\). If you instead fold water’s own molar concentration (about 55.5 M) into the constant, you get the often-quoted \(\mathrm{p}K_a \approx 15.7\). Both numbers are “right”; they are answers to slightly different questions, and arguments about water’s pKa are almost always arguments about which one you meant.

There is one more place the simple equation lies, and it is a beautiful lie. We write the product of autoionization as \(\mathrm{H_3O^+}\), but a bare hydronium ion is not really what floats around in liquid water. The excess proton is shared and smeared across a little cluster of molecules — the Eigen cation \(\mathrm{H_9O_4^+}\) (a hydronium hydrogen-bonded to three more waters) and the Zundel cation \(\mathrm{H_5O_2^+}\) (a proton shared equally between two waters), interconverting on a femtosecond timescale.³ That delocalization is why protons move through water far faster than any ion has any right to: rather than a single heavy ion shouldering its way through, a proton hops down a chain of hydrogen bonds — bond breaks here, forms there — so that the charge travels much farther than any one nucleus does. This is the Grotthuss mechanism, and it is why acids conduct electricity so well. The lone pairs we drew in Section 2 are the hand-off points.

4. Talking to light

Finally, how does water interact with light — and what does that tell us back about the geometry? A bent triatomic has exactly the right number of moving parts to be interesting. With \(N = 3\) atoms, there are \(3N = 9\) degrees of freedom; subtract three for translation and three for rotation (water is nonlinear), and \(3N - 6 = 3\) vibrational modes remain:

Symmetry labels the modes just as it labelled the orbitals: the symmetric stretch and the bend keep the molecule’s full symmetry (\(A_1\)), while the antisymmetric stretch swaps the two bonds (\(B_2\)). In \(C_{2v}\), a vibration is infrared-active if it changes the dipole moment — which means it must transform like \(x\), \(y\), or \(z\). Both \(A_1\) (like \(z\)) and \(B_2\) (like \(y\)) qualify, so all three modes are infrared-active.¹ That is why water vapor is such a voracious infrared absorber, and — together with its permanent dipole, which also makes it a strongly absorbing asymmetric-top rotor in the microwave and far-infrared — why \(\mathrm{H_2O}\) is the dominant natural greenhouse gas. The bend from Section 1 is doing the absorbing.

Now climb up in energy to visible and ultraviolet light, where transitions move electrons rather than nuclei. Water’s first electronic absorption promotes a lone-pair electron (out of that \(1b_1\) orbital) and does not switch on until about 7 eV — roughly 175 nm, deep in the vacuum ultraviolet.⁴ There is simply no electronic transition anywhere in the 400–700 nm visible band. That is why water is colorless: a glassful has nothing to absorb.

And yet a deep enough body of water is unmistakably blue, and not because of the sky. The blue is intrinsic, and its origin is the one loose end tying this whole post together. Liquid water absorbs very weakly in the red, around 660–700 nm, through overtones and combinations of the O–H stretching vibrations — the same \(\nu_1\) and \(\nu_3\) from the diagram above, stacked several quanta high.⁵ Each individual absorption is feeble, which is why a glassful looks clear, but over meters of ocean or the depth of a glacier the red end of sunlight is quietly removed and what comes back to your eye is blue. Water is, as far as anyone knows, the only common substance whose color comes from vibrational rather than electronic transitions — its color is, quite literally, the sound of its bonds, shifted up into the visible.

The molecule reads as one piece

Run back through the four passes and notice that nothing was independent. The bend (§1) exists because oxygen carries two lone pairs, and those lone pairs are not the identical ears of the cartoon but two distinct orbitals you can ionize separately (§2). Those same lone pairs are the base sites that catch a proton during autoionization, and the structure of the proton they catch is what makes water conduct (§3). The O–H bonds whose bending and stretching we counted as three infrared modes are, several overtones up, exactly the bonds whose absorption makes the ocean blue (§4).

The cartoon of an oxygen with two hydrogen ears is not wrong. It is just the title page. Everything water does for a living is written in the parts the cartoon leaves out — the precise angle, the asymmetric lone pairs, the self-ionization, and the faint red overtones — and once you can read those, the most familiar molecule stops being familiar in the dull sense and starts being familiar in the way an old machine is familiar to the person who can take it apart.

References