Skip to content
LumenaryAcademy
Foundations 65 min

Chemical bonding

Ionic, covalent and metallic bonds; electronegativity and polarity; Lewis structures, formal charge and resonance; the shapes of molecules; and the energy in a bond.

Code tested · 6AI-draftedAwaiting review by a chemist

Your guides: Chloe · The Chemist, Margaret · The Analyst, Gideon · The Sceptic

MotivationH

What holds matter together?

Table salt shatters when you hit it with a hammer; copper flattens into a sheet; diamond scratches almost anything. Water, with molecules of mass 18 daltons, is a liquid at room temperature, while carbon dioxide, at 44 daltons, is a gas. Ammonia dissolves in water by the bucketful; methane hardly at all. These differences are not about which atoms are present so much as about how they are held together: by the bonds within molecules and crystals, and by the weaker forces between molecules that you will meet in Gases, liquids and solids.

In the previous lesson you saw that each atom has a few valence electrons in its outer shell, held more or less tightly depending on where it sits in the periodic table. Bonding is what happens to those valence electrons when atoms meet. This lesson gives you the working chemist's toolkit for it: electronegativity, Lewis structures, formal charge, resonance and VSEPR. These are models, simpler than the quantum mechanics underneath, and part of the skill is knowing where each one stops working.

IntuitionH

Why atoms bond: it is all energy

Two hydrogen atoms far apart do not interact. Bring them closer and each electron starts to feel the attraction of both nuclei; the energy of the pair falls. Push them too close and the two nuclei, both positive, repel strongly; the energy rises steeply. In between is a minimum, and that minimum is the bond.

bond energy(about 436–458 kJ/mol)74 pmseparate atoms: E = 0nuclei repeldistance between the nucleipotential energy
The potential energy of two hydrogen atoms as their nuclei approach. The minimum sits at the bond length, 74 pm; the depth of the well is the energy needed to pull the atoms apart again. The measured bond enthalpy, 436 kJ/mol at 25 °C, is a little smaller than the depth of the well (about 458 kJ/mol), because even in its lowest state the molecule keeps some vibrational energy.

A chemical bond exists wherever atoms stay together because their combined energy is lower than when apart. The bond length is where the energy is lowest, and the depth of the well is the bond energy. Everything else in this lesson is about how the energy gets lowered, and three limiting cases cover most of chemistry.

Ionic bonding. Sodium holds its single 3s electron loosely; chlorine is one electron short of a full shell. Moving an electron from Na to Cl costs energy for isolated atoms: 496 kJ/mol to ionise sodium, of which only 349 kJ/mol is recovered when chlorine accepts the electron. But a Na⁺ ion and a Cl⁻ ion attract each other strongly, and in a crystal each ion is surrounded by six of the opposite charge. Assembling gaseous ions into solid sodium chloride releases about 787 kJ/mol (the lattice energy), which more than pays for the transfer. Ionic solids are hard, brittle (a shift by one ion brings like charges face to face), melt at high temperatures (801 °C for NaCl) and conduct electricity only when molten or dissolved, when the ions can move.

Covalent bonding. When two atoms both hold their electrons tightly, as two chlorine atoms or carbon and hydrogen do, neither gives one up. Instead a pair of electrons is shared, concentrated between the nuclei and attracted by both. Atoms can share one, two or three pairs: single, double and triple bonds, which get shorter and stronger in that order (carbon–carbon: 154, 134 and 120 pm; about 348, 614 and 839 kJ/mol). Covalent bonds make discrete molecules (water, methane, sugar) and also giant networks (diamond, quartz), which is why "covalent" says nothing by itself about melting point.

Metallic bonding. In a metal, each atom contributes its valence electrons to a shared pool spread over the whole crystal. The positive ions sit in a sea of mobile electrons, which explains why metals conduct electricity and heat, reflect light and can be bent and hammered: the ions can slide past each other without breaking a directional bond. Bonding in solids in Lumenary Materials takes this much further.

Analogy, not a formal construction

Two families living far apart, each with a car, can save money by car-sharing if they move next door; that is a covalent bond. One family can also simply give its car to the other (ionic), or a whole town can pool its cars into a car club that anyone can use (metallic). The analogy is about who controls the electrons, and nothing else: electrons are not owned, and they are not cars.

DefinitionsH

Definitions

Definition— electronegativity

The electronegativity χ\chi of an element is the tendency of its atom, in a molecule, to attract shared electrons to itself. On Pauling's scale it ranges from about 0.8 (caesium) to 3.98 (fluorine). It is a property of an atom in a bond, not of a free atom.

Definition— bond polarity and dipole moment

A bond between atoms of different electronegativity is polar: the shared electrons sit closer to the more electronegative atom, which carries a partial negative charge δ−\delta-, leaving δ+\delta+ on the other. The dipole moment μ=qd\mu = q d of two charges ±q\pm q a distance dd apart is a vector pointing (in the chemistry convention) from δ+\delta+ to δ−\delta-. Its unit in chemistry is the debye, 1 D≈3.336×10−301\ \text{D} \approx 3.336 \times 10^{-30} C m.

Definition— Lewis structure

A Lewis structure shows a molecule's valence electrons as bonding pairs (lines between atoms) and lone pairs (dots), placed so that, where possible, each atom has a noble-gas configuration: eight electrons around most atoms (the octet rule), two around hydrogen.

Definition— formal charge

The formal charge of an atom in a Lewis structure is FC=V−L−12B\text{FC} = V - L - \tfrac12 B, where VV is the number of valence electrons of the free atom, LL the number of electrons in its lone pairs, and BB the number of electrons in its bonds. It is a bookkeeping device, not the atom's actual charge. The formal charges in a structure add up to the charge of the molecule or ion.

Definition— resonance

When a molecule cannot be described by a single Lewis structure, it is described by several resonance structures that differ only in where the electrons are placed, not the atoms. The real molecule is a single resonance hybrid, a weighted blend of them; it does not switch between them.

Definition— VSEPR model

The valence-shell electron-pair repulsion model predicts the shape of a molecule by assuming that the electron domains around a central atom (bonds, counting a double or triple bond as one domain, and lone pairs) arrange themselves to be as far apart as possible.

ExplanationH

Electronegativity and bond polarity

Ionic and covalent bonding are the two ends of a continuum, and the electronegativity difference Δχ\Delta\chi tells you where a bond sits along it. Two identical atoms share equally (Δχ=0\Delta\chi = 0). As Δχ\Delta\chi grows, the shared pair shifts towards one atom, and at large Δχ\Delta\chi the description "one atom has given its electron to the other" becomes the better one.

00.511.522.53Cl–ClC–HO–HH–FNa–ClCs–Fpolar covalentmostly ionicelectronegativity difference, Δχnon-polar
Bonds placed on a scale of electronegativity difference. The boundaries at 0.4 and 1.8 are a common convention, not laws of nature: textbooks use 0.5, 1.7 or 2.0, and some bonds (H–F with Δχ = 1.78, for one) sit right at a boundary.

Pauling also proposed an estimate of the "percentage ionic character" of a bond, 100 (1−e−Δχ2/4)100\,(1 - e^{-\Delta\chi^2/4}). It is an empirical formula, useful for comparisons rather than as a measurement. The program below classifies a dozen bonds.

python
import math

# Pauling electronegativities (Allred's 1961 revision of Pauling's scale)
EN = {"H": 2.20, "Li": 0.98, "C": 2.55, "N": 3.04, "O": 3.44, "F": 3.98, "Na": 0.93,
      "Mg": 1.31, "Si": 1.90, "Cl": 3.16, "K": 0.82, "Cs": 0.79}

def describe(a, b):
    d = abs(EN[a] - EN[b])
    ionic = 100 * (1 - math.exp(-d * d / 4))     # Pauling's estimate of "ionic character"
    kind = "non-polar covalent" if d < 0.4 else "polar covalent" if d < 1.8 else "mostly ionic"
    neg = a if EN[a] > EN[b] else b
    return d, ionic, kind, neg

print("bond    delta-chi  ionic %   type                 partial negative end")
for a, b in [("Cl", "Cl"), ("C", "H"), ("N", "H"), ("C", "O"), ("O", "H"), ("C", "F"), ("Si", "O"),
             ("H", "F"), ("Mg", "O"), ("Na", "Cl"), ("K", "F"), ("Cs", "F")]:
    d, ionic, kind, neg = describe(a, b)
    print(f"{a + '-' + b:7} {d:9.2f} {ionic:8.0f}   {kind:20} {neg if d else '-'}")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Three things in the output deserve comment. The C–H bond (Δχ=0.35\Delta\chi = 0.35, about 3% ionic) is nearly non-polar, which is why hydrocarbons such as petrol and candle wax do not mix with water. O–H and N–H bonds are strongly polar, with the hydrogen at the positive end; that is the origin of hydrogen bonding. And even caesium fluoride, the most ionic combination of stable elements, comes out at 92%, not 100%: no bond is perfectly ionic, because the electron cloud of an anion is always somewhat pulled towards its neighbours.

Worked examplesH

Lewis structures, formal charge and resonance

Lewis structures are drawn by a routine that works for most molecules made of main-group elements:

  1. Count the valence electrons of all the atoms, adding one for each negative charge and subtracting one for each positive charge.
  2. Write the skeleton: usually the least electronegative atom in the middle (never hydrogen, which forms one bond), joined to the others by single bonds.
  3. Complete the octets of the outer atoms with lone pairs, then put any electrons left over on the central atom.
  4. If the central atom has fewer than eight electrons, turn lone pairs on outer atoms into extra bonds (double or triple).
  5. Check formal charges, and prefer the structure in which they are smallest, with any negative formal charge on the most electronegative atom.

Example— Water, carbon dioxide and the ammonium ion

Water. 2 × 1 + 6 = 8 valence electrons. Skeleton H–O–H uses 4; the remaining 4 go on oxygen as two lone pairs. Oxygen has its octet, each hydrogen its pair, and every formal charge is zero.

Carbon dioxide. 4 + 2 × 6 = 16 electrons. O–C–O uses 4; completing the oxygen octets uses 12 more, leaving carbon with only four electrons. Turning one lone pair from each oxygen into a second bond gives O=C=O: every atom has an octet and every formal charge is zero. The alternative O≡C–O also gives every atom an octet, but with formal charges +1 and −1 on the oxygens, so O=C=O is preferred, and it matches the experiment: both C–O bonds in CO₂ have the same length, 116 pm.

Ammonium, NH₄⁺. 5 + 4 × 1 − 1 = 8 electrons, four N–H bonds, no lone pairs. Nitrogen has V=5V = 5, L=0L = 0, B=8B = 8: formal charge 5−0−4=+15 - 0 - 4 = +1, which is the charge of the ion.

Formal charges are tedious by hand and easy by computer. This program compares candidate structures for four molecules and ions, each atom described by its lone-pair electrons and its number of bonds:

python
VALENCE = {"H": 1, "C": 4, "N": 5, "O": 6, "S": 6, "Cl": 7}

def formal_charges(atoms):
    """atoms: list of (element, lone-pair electrons, bonds counted as shared pairs)."""
    return [VALENCE[el] - lone - bonds for el, lone, bonds in atoms]

def assess(name, atoms):
    fc = formal_charges(atoms)
    # a good Lewis structure: small formal charges, and any negative one on an electronegative atom
    negative_on = [el for (el, _, _), q in zip(atoms, fc) if q < 0]
    labels = " ".join(f"{el}({q:+d})" if q else el for (el, _, _), q in zip(atoms, fc))
    print(f"{name:12} {labels:22} total {sum(fc):d}   sum|FC| = {sum(map(abs, fc))}"
          f"   negative on: {', '.join(negative_on) or '-'}")

print("Carbon dioxide")
assess("O=C=O", [("O", 4, 2), ("C", 0, 4), ("O", 4, 2)])
assess("O≡C-O", [("O", 2, 3), ("C", 0, 4), ("O", 6, 1)])
print("Dinitrogen oxide, N2O (atoms in the order N, N, O)")
assess("N=N=O", [("N", 4, 2), ("N", 0, 4), ("O", 4, 2)])
assess("N≡N-O", [("N", 2, 3), ("N", 0, 4), ("O", 6, 1)])
assess("N-N≡O", [("N", 6, 1), ("N", 0, 4), ("O", 2, 3)])
print("Nitrate ion, NO3- (one of three equivalent structures)")
assess("O=N(-O)-O", [("N", 0, 4), ("O", 4, 2), ("O", 6, 1), ("O", 6, 1)])
print("Thiocyanate ion, SCN- (atoms in the order S, C, N)")
assess("S=C=N", [("S", 4, 2), ("C", 0, 4), ("N", 4, 2)])
assess("S-C≡N", [("S", 6, 1), ("C", 0, 4), ("N", 2, 3)])
assess("S≡C-N", [("S", 2, 3), ("C", 0, 4), ("N", 6, 1)])
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

For dinitrogen oxide (N₂O, used as an anaesthetic in medicine and dentistry), two structures tie on total formal charge, N=N=O and N≡N–O, and the third is clearly worse. Neither of the two good ones is "the" structure: the molecule is a hybrid of both, with the N–N bond between a double and a triple bond in length, and the N–O bond between a single and a double. Thiocyanate, SCN⁻, is similar: S=C=N⁻ (charge on the more electronegative nitrogen) and ⁻S–C≡N both contribute, which is why thiocyanate can bind to a metal ion through either end.

The nitrate ion needs three equivalent structures:

NOOONOOONOOO−−+−−+−−+
Three resonance structures of the nitrate ion, NO₃⁻. Each has one N=O double bond and two N–O single bonds, with formal charges +1 on nitrogen and −1 on each singly bonded oxygen. The double-headed arrow means "these are descriptions of one species", not an equilibrium.

Resonance is also why benzene's six C–C bonds are all the same length (140 pm, between 154 and 134 pm) and why the carbonate ion, CO₃²⁻, has three identical bonds. It is not an exotic exception but the normal state of most molecules with double bonds next to lone pairs or other double bonds, and it stabilises them: the hybrid is lower in energy than any single structure would be. Molecular orbitals replace the bookkeeping with a single picture in which the electrons are spread over the whole ion from the start.

When the octet rule fails. Three kinds of exception are common:

  • Too few electrons. In BF₃, boron has only six valence electrons around it. It readily accepts a lone pair from another molecule (ammonia, for instance), which makes it a typical Lewis acid.
  • An odd number of electrons. Nitrogen monoxide, NO, has 11 valence electrons, so some atom must have an unpaired electron. Such molecules, radicals, tend to be reactive.
  • More than eight electrons around the central atom. Sulfur hexafluoride, SF₆, has six S–F bonds, twelve electrons around sulfur. Elements from period 3 down can be surrounded by more than four electron domains; period 2 elements cannot.
ExplanationH

The shapes of molecules

A Lewis structure shows which atoms are connected, not how they are arranged in space. The VSEPR model fills the gap with one idea: electron domains around a central atom repel each other and spread out as far as possible.

DomainsArrangementLone pairsShapeExampleAngle
2linear0linearCO₂, BeCl₂ (gas)180°
3trigonal planar0trigonal planarBF₃, NO₃⁻120°
3trigonal planar1bentSO₂, O₃about 117–119°
4tetrahedral0tetrahedralCH₄, NH₄⁺109.5°
4tetrahedral1trigonal pyramidalNH₃107°
4tetrahedral2bentH₂O104.5°
5trigonal bipyramidal0trigonal bipyramidalPCl₅90°, 120°
5trigonal bipyramidal3linearXeF₂180°
6octahedral0octahedralSF₆90°
6octahedral2square planarXeF₄90°
COOBFFFCHHHHNHHHOHHCO₂linear180°BF₃trigonal planar120°CH₄tetrahedral109.5°NH₃trigonal pyramidal107°H₂Obent104.5°
VSEPR shapes of five small molecules. A solid wedge is a bond pointing towards you, a hashed wedge a bond pointing away; shaded lobes are lone pairs. Lone pairs take up more room than bonding pairs, so the H–X–H angle closes from 109.5° in methane to 107° in ammonia and 104.5° in water.

The progression from methane to water is the model's best-known success. All three have four electron domains arranged roughly tetrahedrally, but a lone pair is held by only one nucleus and spreads out closer to it, so it pushes the bonding pairs together. Count the domains, and you have the arrangement; then name the shape from the positions of the atoms only. Water's electron domains are tetrahedral, but the molecule is bent.

VSEPR's predictions are qualitative. It says water's angle is "a bit less than 109.5°" but cannot say 104.5°, and it is silent on hydrogen sulfide, whose angle (92°) is far smaller than water's although the model treats the two identically. It also does poorly for transition-metal compounds, whose shapes depend on d electrons. For quantitative geometries chemists compute them, as Computational chemistry shows.

ComputationC

Polar bonds, polar molecules

A molecule's dipole moment is the vector sum of its bond dipoles (and of contributions from lone pairs). Shape therefore decides whether polar bonds make a polar molecule. In CO₂ the two C=O dipoles point in opposite directions and cancel; in water the two O–H dipoles point down the two arms of a V and add.

This program builds unit vectors for each VSEPR shape with NumPy, gives each bond a dipole proportional to its electronegativity difference (a crude but instructive model), and adds them up:

python
import numpy as np

EN = {"H": 2.20, "B": 2.04, "C": 2.55, "N": 3.04, "O": 3.44, "F": 3.98, "Cl": 3.16}

def unit(v):
    v = np.asarray(v, dtype=float)
    return v / np.linalg.norm(v)

def pyramid(angle_deg):
    """Three unit vectors, pointing down from the apex, with the given angle between each pair."""
    sin2 = (1 - np.cos(np.radians(angle_deg))) / 1.5
    st, ct = np.sqrt(sin2), np.sqrt(1 - sin2)
    return [np.array([st * np.cos(p), st * np.sin(p), -ct]) for p in np.radians([0, 120, 240])]

half = np.radians(104.5 / 2)
GEOMETRY = {   # unit vectors from the central atom to each outer atom (VSEPR shapes)
    "linear": [unit([0, 0, 1]), unit([0, 0, -1])],
    "trigonal planar": [unit([np.cos(a), np.sin(a), 0]) for a in np.radians([90, 210, 330])],
    "tetrahedral": [unit(v) for v in ([1, 1, 1], [1, -1, -1], [-1, 1, -1], [-1, -1, 1])],
    "pyramidal (107 deg)": pyramid(107.0),
    "bent (104.5 deg)": [unit([np.sin(half), 0, -np.cos(half)]), unit([-np.sin(half), 0, -np.cos(half)])],
}

def bond_dipole_sum(centre, outer, shape):
    """Sum of bond dipoles, each taken as (EN difference) along the bond, pointing to the negative end."""
    total = np.zeros(3)
    for atom, direction in zip(outer, GEOMETRY[shape]):
        total += (EN[atom] - EN[centre]) * direction      # positive: the outer atom is the negative end
    return np.linalg.norm(total)

for name, centre, outer, shape in [
        ("CO2", "C", ["O", "O"], "linear"),
        ("BF3", "B", ["F", "F", "F"], "trigonal planar"),
        ("CH4", "C", ["H", "H", "H", "H"], "tetrahedral"),
        ("CCl4", "C", ["Cl", "Cl", "Cl", "Cl"], "tetrahedral"),
        ("CH3Cl", "C", ["Cl", "H", "H", "H"], "tetrahedral"),
        ("CHCl3", "C", ["H", "Cl", "Cl", "Cl"], "tetrahedral"),
        ("NH3", "N", ["H", "H", "H"], "pyramidal (107 deg)"),
        ("H2O", "O", ["H", "H"], "bent (104.5 deg)")]:
    m = bond_dipole_sum(centre, outer, shape)
    print(f"{name:6} {shape:20} net bond-dipole sum {m:5.2f}  -> {'polar' if m > 1e-9 else 'non-polar'}")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Symmetry does the work: CO₂, BF₃, CH₄ and CCl₄ come out exactly non-polar, because for each of these shapes the unit vectors sum to zero. Replace one atom of a symmetric molecule, as in chloromethane (CH₃Cl), and the cancellation fails.

Polarity has consequences everywhere. Polar molecules attract each other, which raises boiling points (see Gases, liquids and solids); they dissolve ionic and polar substances (see Solutions); and a microwave oven heats food because the oscillating field keeps turning water's dipoles back and forth.

ComputationC

Energy in bonds: estimating reaction enthalpies

Breaking a bond always costs energy; forming one always releases it. If we know how much energy each bond is worth, we can estimate the energy change of a reaction without doing an experiment:

ΔrH≈∑(bond enthalpies of bonds broken)−∑(bond enthalpies of bonds formed).\Delta_r H \approx \sum (\text{bond enthalpies of bonds broken}) - \sum (\text{bond enthalpies of bonds formed}).

A negative result means more energy is released by the new bonds than was spent on the old ones: an exothermic reaction. For a diatomic molecule the bond enthalpy is a single measured number (436 kJ/mol for H–H). For a bond such as C–H, which occurs in thousands of molecules, a table gives an average, and the C–H bond in any particular molecule may differ from it by 10 to 20 kJ/mol. So the method gives estimates, and the program below tests it against reaction enthalpies measured by calorimetry, computed from standard enthalpies of formation in the NIST Chemistry WebBook:

python
# Average bond enthalpies in kJ/mol (typical textbook averages; tables differ by a few kJ/mol)
BOND = {"H-H": 436, "C-H": 413, "C-C": 348, "C=C": 614, "O=O": 498, "C=O (CO2)": 799,
        "O-H": 463, "N#N": 945, "N-H": 391, "Cl-Cl": 242, "H-Cl": 431}

# Standard enthalpies of formation at 298.15 K, kJ/mol (NIST Chemistry WebBook / CODATA)
DFH = {"CH4": -74.87, "O2": 0.0, "CO2": -393.51, "H2O(g)": -241.83, "H2": 0.0, "Cl2": 0.0,
       "HCl": -92.31, "N2": 0.0, "NH3": -45.94, "C2H4": 52.47, "C2H6": -84.0}

REACTIONS = [
    # name, bonds broken, bonds formed, reactants, products (as {species: coefficient})
    ("CH4 + 2 O2 -> CO2 + 2 H2O(g)", {"C-H": 4, "O=O": 2}, {"C=O (CO2)": 2, "O-H": 4},
     {"CH4": 1, "O2": 2}, {"CO2": 1, "H2O(g)": 2}),
    ("H2 + Cl2 -> 2 HCl", {"H-H": 1, "Cl-Cl": 1}, {"H-Cl": 2},
     {"H2": 1, "Cl2": 1}, {"HCl": 2}),
    ("N2 + 3 H2 -> 2 NH3", {"N#N": 1, "H-H": 3}, {"N-H": 6},
     {"N2": 1, "H2": 3}, {"NH3": 2}),
    ("C2H4 + H2 -> C2H6", {"C=C": 1, "H-H": 1}, {"C-C": 1, "C-H": 2},
     {"C2H4": 1, "H2": 1}, {"C2H6": 1}),
]

def enthalpy_from_bonds(broken, formed):
    """Energy in to break bonds, minus energy out when new bonds form."""
    return sum(BOND[b] * k for b, k in broken.items()) - sum(BOND[b] * k for b, k in formed.items())

def enthalpy_from_formation(reactants, products):
    return sum(DFH[s] * k for s, k in products.items()) - sum(DFH[s] * k for s, k in reactants.items())

print(f"{'reaction':32} {'bonds':>8} {'measured':>9} {'error':>7}   (kJ/mol)")
for name, broken, formed, reac, prod in REACTIONS:
    est = enthalpy_from_bonds(broken, formed)
    true = enthalpy_from_formation(reac, prod)
    print(f"{name:32} {est:8.0f} {true:9.1f} {est - true:7.1f}")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

The combustion of methane comes out at −802 kJ/mol from bonds and −802.3 kJ/mol from calorimetry; the hydrogen–chlorine and ammonia reactions are within about 1 kJ/mol. Do not be too impressed: those reactions use bonds whose "average" values come almost entirely from these very molecules (H–H, Cl–Cl and H–Cl are diatomic; the O–H value is essentially water's; the C=O value is specific to CO₂). The hydrogenation of ethene is the honest test, and there the estimate is off by 12.5 kJ/mol, or about 9%: typical for the method.

Thermochemistry makes this exact: instead of averages, it uses enthalpies of formation and Hess's law, which is what the "measured" column above already does.

ExplanationH

The professional view

Measuring bonds. Bond lengths and angles come from X-ray and neutron crystallography for solids and from microwave spectroscopy and electron diffraction for gases; the Cambridge Structural Database holds well over a million crystal structures of organic and metal–organic compounds. Dipole moments are measured from how a molecule's rotational spectrum shifts in an electric field. NIST's CCCBDB collects these measurements for small molecules side by side with computed values, and is the place to check a number before you use it.

Computing bonds. Lewis structures and VSEPR are what chemists sketch on a whiteboard; when numbers matter, they compute the electronic structure with quantum chemistry (density functional theory or wavefunction methods), which gives bond lengths typically within a picometre or two and, with the best methods, reaction energies within a few kJ/mol. Those methods are the subject of the Quantum level of this course.

Bonds in software. Cheminformatics toolkits store molecules as graphs of atoms and bonds, with formal charges on the atoms, often written as SMILES strings: nitrate is [O-][N+](=O)[O-], one of its resonance structures. The open-source RDKit is widely used for this in industry and research. The following snippet is shown for illustration and is not run here, because RDKit is not available in the browser:

python
# Not run here: needs RDKit (pip install rdkit)
from rdkit import Chem

nitrate = Chem.MolFromSmiles("[O-][N+](=O)[O-]")
for atom in nitrate.GetAtoms():
    print(atom.GetSymbol(), atom.GetFormalCharge(), atom.GetDegree())

Software has to choose one resonance structure to store, which is a famous source of subtle bugs: two programs may write the same molecule differently, and comparing molecules reliably needs a canonical form. The open InChI identifier was designed by IUPAC and the InChI Trust partly to solve this.

Common misconceptionsH

Common misconceptions

Misconception

"Breaking bonds releases energy." It never does; breaking a bond always costs energy. Energy is released when new, stronger bonds form. Fuels and food store energy only in the sense that their reactions with oxygen form stronger bonds than they break.

Misconception

"Ionic and covalent bonds are two different kinds of thing." They are two ends of one continuum, and most real bonds are in between. Even sodium chloride has a little covalent character, and even C–H bonds are slightly polar.

Misconception

"Resonance structures flip back and forth." The molecule is one structure all the time, the hybrid. Resonance structures are like describing a rhinoceros as a cross between a dragon and a unicorn: the rhinoceros is a real animal, and it is not alternating between the two imaginary ones.

Misconception

"A molecule with polar bonds is a polar molecule." Only if the bond dipoles fail to cancel. CO₂ and CCl₄ have polar bonds and no dipole moment.

Misconception

"Formal charge is where the charge really is." Formal charge assumes perfectly equal sharing in every bond. The real charge distribution, which can be computed or inferred from experiments, may differ a lot; in the ammonium ion the +1 is spread over the hydrogens as much as on the nitrogen.

ExplanationH

Something safe to try at home

With an adult, turn a kitchen tap on so that a thin, smooth stream of water flows. Rub a plastic comb or an inflated balloon on dry hair or a wool jumper, and bring it close to the stream without touching it: the stream bends towards it. The usual explanation is that water's polar molecules turn and are drawn towards the charged comb. Careful experiments have shown that the main cause is electric charge induced in the stream, which a polar liquid such as water carries far more readily than a non-polar one: so it is a demonstration of polarity, just not quite in the way it is usually told. Keep the comb and your hands dry, and keep anything electrical well away from the water.

Exercises

Exercises

1. Chemical bonding

Warm-upHuman

Using Pauling electronegativities (H 2.20, C 2.55, N 3.04, O 3.44, F 3.98), which of these bonds is the most polar?

Choose one

2. Chemical bonding

CoreHuman

Carbon monoxide is usually drawn as a triple bond with one lone pair on each atom: :C≡O:. What is the formal charge on carbon in this structure?

3. Chemical bonding

CoreHuman

What shape does VSEPR predict for chlorine trifluoride, ClF₃? (Chlorine is the central atom; all four atoms are halogens with 7 valence electrons.)

Choose one

4. Chemical bonding

CoreHuman

Which of these molecules are polar (have a non-zero dipole moment)? Type their formulas separated by commas.

H2O, CO2, NH3, CH4, CHCl3, BF3, SO2, CCl4, HCN

Write elements separated by commas. Order and repeats don’t matter — just like in a set!

5. Chemical bonding

CoreHuman

Use the bond enthalpies H–H 436, F–F 158 and H–F 567 kJ/mol to estimate ΔrH\Delta_r H for H₂(g) + F₂(g) → 2 HF(g), in kJ/mol. Give a whole number, with its sign.

6. Chemical bonding

CoreHuman

Ozone, O₃, is drawn as two resonance structures, O=O–O and O–O=O. Which statement is correct?

Choose one

7. Chemical bonding

CoreComputational

Write estimate_dH(broken, formed), which returns the estimated reaction enthalpy in kJ/mol from the average bond enthalpies in BOND. Each argument is a dict mapping a bond name to how many of those bonds are broken (or formed), for example {"C-H": 4, "O=O": 2}. If a bond is missing from the table, let the KeyError propagate.

Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

8. Chemical bonding

StretchComputational

Write is_polar(bonds), where bonds is a list of (direction, size) pairs: direction is a vector (three numbers, not necessarily of length 1) from the central atom along a bond, and size is that bond's dipole, positive when the outer atom is the negative end. Return True if the bond dipoles do not cancel (the length of their vector sum exceeds 1e-6), else False. Use NumPy.

Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Further reading and resources