A bond is two clouds overlapping
Lewis dots and VSEPR tell us *that* a bond exists and *what shape* results, but they stay silent on what a bond actually is. The first real answer is valence bond theory. Its picture is wonderfully physical: each atom has electrons living in fuzzy clouds — its atomic orbitals — and a covalent bond forms when an orbital from one atom *overlaps* with an orbital from the other, letting a shared pair of electrons sit in the overlap region between the two nuclei.
More overlap means more electron density pooled between the nuclei, which means a stronger bond. So in valence bond theory a covalent bond is a tangible thing: it is the place where two orbitals merge. This is a big step up from a dash on paper — it tells us bonds have direction (orbitals point in particular directions) and strength (set by how much they overlap).
Carbon breaks the simple picture
Now a puzzle. Carbon's electron configuration gives it valence electrons in two different kinds of orbital — one spherical (called s) and a few dumbbell-shaped ones pointing along different axes (called p). These orbitals have different shapes, different directions, and different energies. So if carbon used them raw, its four bonds in methane (CH₄) should be unequal — some short, some long, pointing every which way.
But experiment is stubborn: methane's four C–H bonds are *identical*, all pointing to the corners of a perfect tetrahedron at 109.5°. The raw orbitals simply do not match what we measure. Something must reconcile the neat tetrahedral geometry VSEPR predicted with the messy mixed orbitals carbon actually starts with.
Reshuffling the deck: hybridization
The fix is an idea called hybridization. Before bonding, an atom blends its mismatched s and p orbitals into a fresh set of *identical* orbitals, all the same shape and energy, pointing in exactly the directions the bonds need. Carbon mixes its one s and three p orbitals into four matching 'sp³' hybrids, each aimed at a tetrahedron corner — and now its four equal bonds make perfect sense.
Two flavors of bond: sigma and pi
Valence bond theory also explains why a double bond is different from a single one — and the answer is that orbitals can overlap in two distinct ways. These are the sigma and pi bonds. A sigma (σ) bond forms when orbitals overlap head-on, directly along the line joining the two nuclei. The electron density piles up right between the atoms, making sigma bonds strong and letting the atoms freely twist around them. Every single bond is a sigma bond.
A pi (π) bond is sideways overlap: two dumbbell-shaped p orbitals lie parallel and overlap above and below the bond axis, like two planks slapped together face to face. Pi overlap is weaker than head-on sigma overlap, and crucially it *locks* the atoms — you cannot twist around a pi bond without breaking it. A double bond is one sigma plus one pi; a triple bond is one sigma plus two pi. That sideways pi bond is exactly why a C=C double bond is rigid while a C–C single bond spins freely.