Crosstalk is the term given to the situation where energy from a signal on one line is transferred to a neighboring line by electromagnetic means. In general, both capacitive and inductive coupling exist. At the chip level, however, the currents through the signal lines are usually too small to induce magnetic coupling, so that parasitic inductance is ignored here.

Capacitive coupling, on the other hand, depends on the line-to-line spacing S as illustrated in the general situation portrayed in Figure. Since capacitive coupling between two conducting lines is inversely proportional to the distance between the two lines, a small value of S implies a large coupling capacitance $C_c$ exists. Because of this dependence, it is not uncommon to find a minimum layout spacing design rule for critical lines that is actually larger than which could be created in the processing line. Also, the capacitive coupling increases with the length of the interaction, so it is important that the interconnects not be placed close to one another for any extended distance.

Let us use the geometry in Figure to estimate the coupling capacitance. This cross-sectional view shows the spacing S between two identical interconnect lines. An empirical formula that provides a reasonable estimate for the coupling capacitance $C_c’$ per unit length is given by

$C~'_c=\epsilon_{ox}[0.03(\frac{w}{X_{ox}})+0.83(\frac{h}{X_{ox}})-0.07(\frac{h}{X_{ox}})^{0.222}](\frac{X_{ox}}{S})^{4/3}$

in units of F/cm which can be applied directly to the geometry. The total coupling capacitance in farads of a line that has a length d is calculated from

$C_c = C_c' . d$

This shows explicitly the fact that $C_c$ increases as the separation distance S decreases.

The importance of $C_c$ becomes evident when we examine how two circuits can interact via electric field coupling. Consider the situation shown in Figure where two independent lines interact through a coupling field $E_c$. Line 1 is at a voltage $V_1(t)$ at the input to inverter B, while line 2 has a voltage $V_2(t)$ which is the input of inverter D. The field is supported by the difference in voltages $V_1(t) - V_2(t)$. At the circuit level, we analyze the situation by introducing lumped-equivalent transmission line models as in Figure. The electric field interaction is included through the coupling capacitor $C_c$. The placement of $C_c$ in the circuit corresponds to the simplest type of single-capacitor coupling model; a more accurate analysis might add two capacitors, one on each side of the resistors. The current through the capacitor is calculated from the relation

$i_c=C_c\frac{dV_c}{dt} =C_c\frac{d(V_1-V_2)}{dt}$

and is assumed to flow from line 1 to line 2 by the choice of voltages. If the difference $V_1(t) - V_2(t)$ changes in time, then the two lines become electrically coupled and the voltages are different from the case where they are independent.