MOS capacitor characterization often relies on measuring capacitance and conductance across gate voltage and frequency. These measurements can reveal how the semiconductor responds to an applied signal, as well as information about oxide capacitance, carrier lifetime and interface defects. A useful way to interpret the data is to compare two frequency-dependent functions: conductance divided by angular frequency, G/ω, and the negative frequency derivative of capacitance, −ωdC/dω.
For an MOS capacitor, these functions are related across the bias range, from accumulation through depletion to inversion. They exhibit peaks at the same frequency, and the peak values are equal. This connection offers a way to check the consistency of measured C-V and G-V data and, when the measurement frequency range is suitable, to estimate important device parameters.
MOS capacitor characterization through matched peaks
In multi-frequency measurements, the capacitance and conductance of a MOS structure change as the gate voltage and measurement frequency vary. The capacitance reflects how charge responds to the signal; conductance represents the associated dissipative response. Examining either quantity alone may leave important behavior difficult to distinguish. The paired functions G/ω and −ωdC/dω provide a complementary view of that response.
The relationship is not limited to a particular bias point or to ideal devices. Simulations show that the peak values and their frequency positions coincide from accumulation through inversion, including structures with interface states. In inversion, a peak associated with the supply of minority carriers can be especially useful. Other peaks may be associated with majority-carrier dielectric relaxation at much higher frequencies.
For a simplified equivalent circuit, the semiconductor is represented by a capacitance, CS, in parallel with a conductance, GS, and the oxide by its capacitance, COX. The peak frequency is determined by GS and the combined capacitance CS + COX. The shared maximum value of the two functions is determined by the relationship between COX and CS. This circuit interpretation links features in frequency-domain data to physical properties of the MOS system.
What the peak frequency and magnitude can reveal
Peak position and peak height respond to different material and device parameters. Their interpretation therefore depends on the bias region and on which response is producing the peak.
- Minority-carrier lifetime: In inversion, the peak frequency is related to the minority-carrier generation-recombination time, τg. As the lifetime increases, the peak frequency eventually saturates when minority-carrier supply from the quasi-neutral region exceeds the Shockley-Read-Hall generation rate.
- Interface states: In depletion, where interface traps can dominate the response, the peak magnitude of G/ω depends linearly on the peak interface-state density over the stated range below COX/2. The peak frequency also varies with the traps’ capture cross section.
- Majority-carrier relaxation: A high-frequency peak, including one in accumulation, is associated with the majority-carrier dielectric relaxation time. This behavior is relevant when measurements extend into very high-frequency ranges.
Interface traps can distort the frequency-dependent C-V and G-V response. Even so, simulations with a Gaussian distribution of donor-type interface states show that the matching relationship between the two functions remains. In the modeled case, the traps affect the lower-frequency response while the high-frequency peaks above 1012 Hz are not influenced. The practical importance of each feature depends on the device and the accessible measurement range.
Using multi-frequency data to extract MOS parameters
A practical extraction approach compares measured and simulated G/ω and −ωdC/dω curves over frequency. In the demonstrated InGaAs MOS capacitor example, data were collected from 1 kHz to 1 MHz. The inversion region was suitable for analysis because the relevant peak fell within that measurement window. The parameter set that minimized the difference between experiment and simulation was used to identify oxide capacitance, substrate doping and minority-carrier lifetime.
This is a joint-fit strategy rather than an interpretation based on a single curve or one peak in isolation. The matching peak values and positions provide constraints, while the equivalent-circuit model connects those features to the semiconductor and oxide parameters. In the reported example, the extracted values were consistent with results from other characterization approaches. The method’s usefulness, however, is limited by whether the relevant peak frequencies fall within the instrument’s practical measurement range.
In strong inversion, the analysis can often focus on the minority-carrier response. The study treated the effect of border traps as negligible for this specific analysis and used a circuit topology that assumes a trap-free oxide. If border traps significantly affect the response, a distributed resistance-capacitance network may be needed to represent them. That distinction matters when selecting a model for non-ideal high-k MOS structures.
Practical considerations for engineers
When applying the method, engineers and device teams should consider the following:
- Measure capacitance and conductance across a sufficiently broad frequency range to capture the peak relevant to the bias region of interest.
- Compare both frequency-dependent functions, rather than relying on capacitance or conductance alone.
- Interpret peak frequency and magnitude in the context of bias, carrier response, interface-state parameters and the equivalent circuit.
- Account for non-idealities such as interface states and border traps when deciding whether the simplified circuit is appropriate.
Conclusion
The relationship between G/ω and −ωdC/dω turns multi-frequency MOS capacitor measurements into a connected set of diagnostic features. Their coincident peaks can help verify data interpretation and support extraction of oxide capacitance, semiconductor doping and minority-carrier lifetime. For engineers evaluating MOS structures, the key is to combine both functions with an appropriate device model and a measurement range that captures the response of interest. The same peak behavior can also provide insight into interface traps and carrier relaxation, while reminding analysts to account for oxide-related non-idealities where relevant.