Diffusion Capacitance¶
Forward-injected carrier charge in a pn junction gives a bias-dependent incremental capacitance distinct from depletion capacitance.
Core Idea¶
Diffusion capacitance is the small-signal charge response of forward-injected carriers in a pn diode. Forward bias injects excess minority carriers into the quasi-neutral regions. Their stored charge changes with terminal voltage; the incremental capacitance at a chosen operating point is \(C_d=dQ_{\mathrm{stored}}/dV\). This is distinct from depletion or junction capacitance \(C_j\), which comes from modulation of charge in the space-charge region. MIT's diode lecture derives the two charge stores separately and places \(C_d\) alongside conductance and \(C_j\) in the small-signal circuit.[1]
For a narrow-base ideal model with fixed transport times, MIT gives hole-side \(Q_p=\tau_{Tp}I_p\) and electron-side \(Q_n=\tau_{Tn}I_n\). Under a forward exponential current law, their incremental contributions add: \(C_d\simeq(q/kT)(\tau_{Tp}I_p+\tau_{Tn}I_n)\). A shorthand \(C_d=\tau_T g_d\) is valid when the effective storage time and model assumptions make \(Q=\tau_T I\) differentiable with fixed \(\tau_T\) at that bias. It is not an unconditional law for “some charge always in transit” in every device. UCLA's derivation also distinguishes a wide-base lifetime parameter from a narrow-base transit time; the particular time scale is geometry-dependent.[1][2]
Structural Signature¶
Sig role-phrases: injected minority charge; quasi-neutral storage; operating-point voltage derivative; transport or lifetime scale; parallel small-signal contribution; depletion contrast.
- Carrier store: excess holes in the n-side quasi-neutral region and excess electrons in the p-side region, not just static space-charge at the junction.
- Current–charge relation: in the stated model each side has \(Q_i=\tau_i I_i\); \(\tau_i\) may be a lifetime for a wide-base idealization or a transit time for a narrow-base one.[2]
- Perturbation: a sufficiently small and slow \(dV\) changes the carrier distribution so \(dQ_i/dV\) represents an incremental circuit capacitance.
- Combination: the side contributions add in the equivalent circuit; at strong forward bias they may exceed \(C_j\), while at reverse/small forward bias depletion capacitance dominates in the cited lecture's model.[1]
- Failure limit: if \(\tau_i\) varies materially with bias, the derivative is \(d(\tau_i I_i)/dV\), including the \(I_i d\tau_i/dV\) term. At high frequency one must use a dynamic diffusion admittance rather than assume a constant lumped capacitance.
What It Is Not¶
Not an ordinary parallel-plate capacitor inserted in the diode. The small-signal branch represents changing stored carrier population. Not the same as \(C_j\): that tracks depletion-layer charge, and can dominate under reverse bias. Not a claim that all junction current is one carrier stream or that one universal \(\tau\) describes both sides. Not a complete switching-time prediction by itself: resistance, circuit load and charge removal dynamics also matter. Shockley's original 1949 pn-junction theory describes carrier diffusion and frequency-dependent admittance, but its accessible abstract is not cited as proof of the later compact \(C_d=\tau g\) formula.[3][1]
Scope of Application¶
The clearest habitat is a forward-biased pn diode under a quasistatic small-signal approximation at a specified temperature, geometry and operating current. MIT's lecture shows both minority-carrier contributions in the quasi-neutral regions and notes that diffusion capacitance rises with forward current, whereas junction capacitance dominates reverse bias and small forward bias.[1] The concept may appear in other semiconductor charge-control models, but transfer requires identifying the actual stored charge and transport relation. Calling every device's transit charge “diffusion capacitance” would erase that physical boundary.
Clarity¶
State what \(Q\) counts. In MIT's notation \(qP_n\) is hole charge in the n-side neutral region, with \(\tau_{Tp}=(W_n-x_n)^2/(2D_p)\) in the narrow-base calculation; similarly the electron contribution has its own time. \(C_d\) is the derivative of the total relevant stored charge, not the ratio \(Q/V\) and not the depletion term. At a forward bias where \(I\gg I_0\), the ideal diode law gives \(g_d\simeq qI/kT\), so the time–conductance shortcut follows if the effective time is fixed. UCLA's wide-base derivation instead identifies \(Q_p=I_p\tau_p\) using carrier lifetime. Mixing those time scales without declaring geometry would turn a useful abbreviation into a wrong physical assertion.[1][2]
Manages Complexity¶
The abstraction turns a distributed diffusion profile into a circuit element around an operating point. Rather than solve a time-dependent transport equation for every small perturbation, a designer uses stored charge and its bias derivative in a parallel small-signal equivalent circuit. The simplification preserves how forward bias increases storage. It loses information about spatial profiles and high-frequency phase delay, so its use must be limited to the regime where the carrier profile can follow the perturbation. Shockley's original paper already connects diffusion transport with frequency-dependent admittance; the simple capacitance is a later/local approximation to that physics.[3][1]
Abstract Reasoning¶
Choose the diode model and operating point. Calculate excess charge on each neutral side, then differentiate it with respect to voltage. If the current on a side is exponential and its transport time is fixed, write \(C_{di}=\tau_i dI_i/dV\). Add both contributions and compare them with \(C_j\) at the same bias. If one side is negligible, a one-sided approximation is justified by that asymmetry, not by the general definition. For fast signals or bias-dependent transport parameters, return to the dynamic diffusion model instead of extending the shortcut.
Knowledge Transfer¶
The hole-side and electron-side calculations transfer within pn-junction modeling because the same stored-charge derivative acts on two unlike carrier populations. A wide-base diode keeps the derivative logic but changes the relevant storage time from simple traversal to recombination lifetime; that is a substantive model change.[2] The phrase diffusion capacitance does not transfer literally to every delayed flow in mechanics or software; an analogy to “in-transit inventory” lacks semiconductor carriers, quasi-neutrality and electrical voltage derivative.
Examples¶
Hole storage in the n-side region. MIT lecture 16 derives \(Q_p=\tau_{Tp}I_p\) for excess holes and \(\tau_{Tp}=(W_n-x_n)^2/(2D_p)\) in its narrow-base model. Under the forward exponential law, \(C_{dn}\simeq(q/kT)\tau_{Tp}I_p\). Mapped back: injected holes are the store; \(V\) changes \(I_p\); fixed transport time converts that change to charge derivative; the result is one diffusion-capacitance branch.[1]
Electron storage on the p-side. The same lecture separately gives \(Q_n=\tau_{Tn}I_n\) for electrons and \(C_{dp}\simeq(q/kT)\tau_{Tn}I_n\). Its total \(C_d=C_{dn}+C_{dp}\) uses the two branches in parallel. These are different carrier streams, diffusion regions, widths and diffusivities within the same diode, not two measured devices. Mapped back: change the carrier side and geometry, preserve charge derivative, then sum—not an unjustified single-carrier result. UCLA's equations likewise distinguish the two currents and wide/narrow-base times.[1][2]
Reverse-bias boundary. In the cited MIT bias comparison, \(C_j\) dominates reverse bias and small forward bias; \(C_d\) dominates only in strong forward bias. Mapped back: the fact that the device remains a pn diode is insufficient to classify its measured capacitance as diffusion-dominated. The injected-charge store is the discriminator.[1]
Structural Tensions¶
Forward conduction versus stored-charge burden. Stronger forward injection raises current and small-signal conductance, which is useful for conducting operation; in the same model it raises the stored carrier population and \(C_d\), increasing the charge that must be rearranged when bias changes. Seeking a low-charge fast response therefore conflicts with relying on high injected current in this operating regime. The diode equations and bias comparison support both sides; this is an engineering operating-point tradeoff, not a claim that \(C_d\) alone predicts a whole circuit's switching speed.[1][2]
Diagnostic: At the intended forward bias, how much of the desired conduction comes with minority-carrier storage, and does its charge derivative dominate the small-signal or switching response?
Structural–Framed Character¶
The core is structural within physics: a stored quantity changes with a control variable, so its derivative enters a local response model. The specific identity is framed by semiconductor transport, pn geometry, carrier species, temperature and bias. It has no moral evaluation; “good” or “bad” capacitance is relative to an engineer's conduction and speed goals. Human practice chooses operating point and equivalent-circuit approximation, while excess carriers and charge conservation are physical irrespective of naming.
The vocabulary arose in semiconductor-device theory, with Shockley's 1949 original diffusion account and later educational derivations documenting the lineage.[3][1] It can travel to other electronic devices only after their charge-control equations are established. Importing the same symbol \(\tau g\) into an arbitrary delayed process is not recognition of diffusion capacitance; it is a formal analogy lacking the pn carrier mechanism. Its character: structural charge-response law tightly framed by semiconductor transport and model regime.
Structural Core vs. Domain Accent¶
The skeletal relation is derivative of stored inventory with respect to a control input. The domain-bound mechanism is voltage-controlled minority-carrier charge in the quasi-neutral parts of a junction, with transport or lifetime linking charge to current and an equivalent small-signal branch. Remove those and one may retain a generic dynamic-storage analogy but not this device quantity. The named entry fails the prime bar because voltage, injected carriers, bias regime and diffusion physics are constitutive, not mere illustrative accents. It is provisionally unparented because the current graph has no generic electrical-capacitance genus; a portable response-derivative skeleton would require separate admission.
Instantiates / Related Primes¶
This entry is a provisional unparented root pending a defensible electrical-capacitance genus. Pn-junction small-signal models are related, and depletion capacitance is a contrasting charge store, not a synonym.
Neighborhood in Abstraction Space¶
Diffusion Capacitance sits in a sparse region of the domain-specific corpus (68th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Condensed Matter & Physical Chemistry Models (26 abstractions)
Nearest neighbors
- Su–Schrieffer–Heeger model — 0.85
- Mott–Schottky Equation — 0.84
- Dynamic logic (digital electronics) — 0.84
- Single Vegetative Obstruction Model — 0.83
- Rooted product of graphs — 0.83
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
- Junction/depletion capacitance: voltage changes charge in the space-charge region; it is often the leading reverse-bias term.[1]
- A fixed physical capacitor: the element is a local equivalent response of carriers.
- One universal transit time: both carrier sides and wide/narrow-base geometries can differ.[2]
- High-frequency exact admittance: the quasistatic derivative is a bounded approximation to transport dynamics.[3]
References¶
[1] J. del Alamo, MIT 6.012 Lecture 16, The pn Junction Diode (II) (2005), pp.11–18, charge and two-capacitance derivation. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m
[2] UCLA, Basic Semiconductor Devices for Electrical Engineers, chapter 3, equations 3.127–3.137, wide- and narrow-base diode charge. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g
[3] William Shockley, “The Theory of p–n Junctions in Semiconductors and p–n Junction Transistors,” Bell System Technical Journal (1949), original abstract on diffusion and dynamic admittance. registry ↩a ↩b ↩c ↩d