Unipolar Impedance Measurements

• Unipolar impedance measurements are techniques that use single-ended, two-terminal configurations to characterize complex device impedances while managing parasitic effects.
• They employ calibration and de-embedding strategies—such as open/short corrections—to subtract fixture parasitics and isolate the true response of the device under test.
• These methods are applied across cryogenic, tribological, and bioimpedance domains, with limitations arising from contact impedance and distributed parasitic errors.

Searching arXiv for papers on unipolar impedance measurement and closely related calibration/modeling work. arXiv search query: "unipolar impedance measurements LCR meter cryogenic two-wire site:arxiv.org" Unipolar impedance measurements are impedance measurements performed with a single-ended reference or a two-terminal geometry, so that the device under test is interrogated between one excitation node and one return node rather than through a fully differential multiport or tetrapolar arrangement. In one important usage, the term denotes the strictly two-wire, two-terminal operation of an LCR meter inside a cryostat; in adjacent literatures it can also denote single-ended excitation with differential readout, or be used only to describe a single-supply implementation rather than a true single-ended measurement geometry. Across these usages, the defining technical problem is the same: the measured complex impedance is inseparable from parasitic series elements, shunt admittances, contact impedances, fixture loading, and reference-node behavior unless those contributions are explicitly modeled or calibrated (Carpenter et al., 13 Jun 2025, Puchtler et al., 2024, Kusche et al., 2019, Li et al., 15 Oct 2025).

1. Terminological scope and measurement geometries

The literature uses “unipolar” in more than one sense. In cryogenic component metrology, it refers to a single-ended or two-terminal measurement made with an LCR meter: the instrument applies a single AC voltage between its “HI” and “LO” terminals and measures the resulting complex current, while the device under test remains a two-terminal element embedded in series and parallel parasitics (Carpenter et al., 13 Jun 2025). In AC bridge measurements of rolling bearings, excitation is likewise single-ended, with a sinusoidal source between ground and a driven node, but the bridge imbalance is measured differentially between two floating midpoints; this is unipolar excitation with differential sensing (Puchtler et al., 2024). In bioimpedance instrumentation, by contrast, “unipolar” commonly denotes a local electrode referenced to a distant common electrode or ground, and two recent device papers explicitly contrast that arrangement with bipolar or tetrapolar sensing because of electrode–skin impedance and common-mode interference (Kusche et al., 2019, Rodriguez et al., 2015). A further terminological shift appears in integrated sensor readout, where a system may run from a single $1.2\ \text{V}$ supply yet remain fully differential in its signal path, so that “unipolar” describes the supply rails but not the sensing topology (Li et al., 15 Oct 2025).

Context	Meaning of “unipolar”	Geometry
Cryogenic LCR metrology	Single-ended / two-terminal	Two-wire DUT in cryostat
AC Wheatstone bridge for bearings	Single-ended excitation	Ground-referenced source, differential bridge readout
Bioimpedance	Local electrode vs common reference	Two-electrode or monopolar reference geometry
CMOS sensor readout	Single supply, not single-ended sensing	Fully differential signals on $0$–$1.2\ \text{V}$ rails

This terminological spread matters because performance limits differ by geometry. In two-terminal cryogenic measurements, the dominant issue is de-embedding background impedance. In biomedical monopolar measurements, electrode interface impedance is often the dominant term. In unipolar-source bridge systems, the critical distinction is that common-mode rejection is recovered at the sensing stage even though excitation itself is single-ended. A plausible implication is that “unipolar impedance measurement” is not a single instrument class but a methodological family defined by how the reference node enters the measurement model.

2. Circuit representations and governing models

The most explicit unipolar model in the recent literature is the cryogenic two-terminal LCR configuration in which the unknown device impedance $\mathbb{Z}_x(\omega,T)$ is measured through a background consisting of series parasitics and shunt admittances. The series contribution is written as

$\mathbb{Z}_s = R_s + j\omega L_s,$

and the local shunt admittance near the DUT as

$\mathbb{Y}_o = G_o + j\omega C_o.$

A second shunt admittance near the meter, $\mathbb{Y}_p$, is introduced for the BNC cables and near-meter wiring, and the analysis is made analytically solvable by taking $\mathbb{Y}_p=\mathbb{Y}_o$. Measuring a short channel $\mathbb{Z}_{Sh}$, an open channel $\mathbb{Z}_{Op}$, and the DUT channel $0$0 yields the algebraic de-embedding relation

$0$1

which eliminates $0$2 and $0$3 point-by-point in frequency without introducing fit parameters (Carpenter et al., 13 Jun 2025).

A different but related unipolar formulation appears in the unbalanced AC Wheatstone bridge for rolling bearings. There the source voltage $0$4 is applied between node $0$5 and node $0$6, the unknown impedance $0$7 occupies one bridge arm, and the bridge voltage $0$8 is measured between the midpoints. Without probe loading, the bridge relation is

$0$9

and the full inversion includes the finite probe impedance $1.2\ \text{V}$0 between the midpoint nodes (Puchtler et al., 2024). In this sense, a unipolar excitation need not imply single-ended sensing of the response.

For electrode-based measurements, the physically realistic forward model is the complete electrode model. With electrode voltages $1.2\ \text{V}$1, contact impedances $1.2\ \text{V}$2, and conductivity field $1.2\ \text{V}$3, the boundary law on each electrode is

$1.2\ \text{V}$4

The shunt model is obtained in the ideal-contact limit $1.2\ \text{V}$5, so that the electrode surface potential equals the electrode voltage exactly. The paper on electrode modelling shows that replacing the complete model by the shunt model causes only an almost linear error with respect to the contact impedances in modelling absolute current-to-voltage measurements, but it also changes the Sobolev regularity of the potential and therefore the convergence rate of finite-element approximations (Staboulis et al., 2013). For unipolar measurements that depend on a common reference electrode, this distinction is often decisive.

3. Calibration, de-embedding, and traceability

Unipolar measurements are rarely accurate in raw form, and the recent literature converges on a layered calibration strategy. In the cryogenic LCR method, calibration begins before the cryostat is attached: the B&K Precision 891 bench LCR meter is open-calibrated with the BNC leads open-circuited and short-calibrated with the BNC leads shorted together. After cooldown, parasitics that belong to the cryostat and PCB are handled not at the instrument but at the sample carrier, using designated open channels with no component soldered and short channels with a $1.2\ \text{V}$6 jumper. At each temperature, $1.2\ \text{V}$7, $1.2\ \text{V}$8, and $1.2\ \text{V}$9 are measured on a common frequency grid, with interpolation in $\mathbb{Z}_x(\omega,T)$0 and $\mathbb{Z}_x(\omega,T)$1 if the meter auto-scaling produces different grids, and then combined through the algebraic correction above (Carpenter et al., 13 Jun 2025).

The bridge literature uses an analogous open–short logic. For the rolling-bearing AC Wheatstone bridge, the raw unknown $\mathbb{Z}_x(\omega,T)$2 obtained from the bridge inversion is corrected by measuring an open state $\mathbb{Z}_x(\omega,T)$3 and a short state $\mathbb{Z}_x(\omega,T)$4, then applying

$\mathbb{Z}_x(\omega,T)$5

Because the assembled bridge contains cable and probe parasitics, the known bridge impedances are not taken from nominal component values alone; instead, six pairwise terminal impedances are measured and a nonlinear system is solved numerically with the trust-region dogleg algorithm to recover the embedded impedances $\mathbb{Z}_x(\omega,T)$6, $\mathbb{Z}_x(\omega,T)$7, $\mathbb{Z}_x(\omega,T)$8, and $\mathbb{Z}_x(\omega,T)$9 (Puchtler et al., 2024).

At the highest metrological level, the three-arm current comparator bridge generalizes the same idea from fixture correction to traceable complex comparison. Three admittances $\mathbb{Z}_s = R_s + j\omega L_s,$0, $\mathbb{Z}_s = R_s + j\omega L_s,$1, and $\mathbb{Z}_s = R_s + j\omega L_s,$2 share a common excitation $\mathbb{Z}_s = R_s + j\omega L_s,$3, while an injection arm $\mathbb{Z}_s = R_s + j\omega L_s,$4 driven by $\mathbb{Z}_s = R_s + j\omega L_s,$5 closes the ampere-turn balance: $\mathbb{Z}_s = R_s + j\omega L_s,$6 Solving for the unknown gives

$\mathbb{Z}_s = R_s + j\omega L_s,$7

The bridge was demonstrated on an air-core inductor and a parallel RC network against decade resistance and capacitance standards, with relative deviations in the $\mathbb{Z}_s = R_s + j\omega L_s,$8–$\mathbb{Z}_s = R_s + j\omega L_s,$9 range (Callegaro et al., 2014). This does not define “unipolar” in the usual instrument sense, but it establishes a useful reference point: unipolar measurements become metrologically credible only when the reference path, parasitics, and complex balance condition are controlled as rigorously as the DUT itself.

4. Frequency selection, parameter extraction, and measurable range

Unipolar measurements are intrinsically frequency-conditioned. In the cryogenic LCR study, the excitation amplitude was fixed at $\mathbb{Y}_o = G_o + j\omega C_o.$0 and the sweep covered $\mathbb{Y}_o = G_o + j\omega C_o.$1 to $\mathbb{Y}_o = G_o + j\omega C_o.$2. Parameter extraction was performed only in bands where the corrected impedance behaved approximately like a pure $\mathbb{Y}_o = G_o + j\omega C_o.$3 or $\mathbb{Y}_o = G_o + j\omega C_o.$4. For capacitive DUTs, the extracted capacitance was computed as

$\mathbb{Y}_o = G_o + j\omega C_o.$5

and for resistive DUTs the extracted resistance was

$\mathbb{Y}_o = G_o + j\omega C_o.$6

The reported practice was to select a flat frequency region, average over it, and quote the standard deviation as a statistical uncertainty (Carpenter et al., 13 Jun 2025).

The same paper shows that frequency-domain de-embedding can extend apparent instrument range. The B&K 891 has specified accuracy only up to about $\mathbb{Y}_o = G_o + j\omega C_o.$7 across most frequencies and about $\mathbb{Y}_o = G_o + j\omega C_o.$8 up to around $\mathbb{Y}_o = G_o + j\omega C_o.$9, but the relevant constraint is on the measured impedance $\mathbb{Y}_p$0, not necessarily on the standalone DUT impedance $\mathbb{Y}_p$1. A $\mathbb{Y}_p$2 resistor at $\mathbb{Y}_p$3 produced raw $\mathbb{Y}_p$4 values below $\mathbb{Y}_p$5, often below $\mathbb{Y}_p$6, yet after correction the inferred $\mathbb{Y}_p$7 was about $\mathbb{Y}_p$8 at $\mathbb{Y}_p$9, about $\mathbb{Y}_p=\mathbb{Y}_o$0 at $\mathbb{Y}_p=\mathbb{Y}_o$1, and about $\mathbb{Y}_p=\mathbb{Y}_o$2 at $\mathbb{Y}_p=\mathbb{Y}_o$3, albeit with large uncertainty in the lowest-temperature case (Carpenter et al., 13 Jun 2025).

In dynamic unipolar bridge measurements, carrier frequency must also exceed the bandwidth of the DUT variation. For the bearing bridge, the relative deviation of a measured capacitance extreme was found to follow

$\mathbb{Y}_p=\mathbb{Y}_o$4

where $\mathbb{Y}_p=\mathbb{Y}_o$5 is the modulation frequency of the capacitance and $\mathbb{Y}_p=\mathbb{Y}_o$6 is the bridge excitation frequency. The reported design rules were $\mathbb{Y}_p=\mathbb{Y}_o$7 for a maximum error of $\mathbb{Y}_p=\mathbb{Y}_o$8 and $\mathbb{Y}_p=\mathbb{Y}_o$9 for a maximum error of $\mathbb{Z}_{Sh}$0 (Puchtler et al., 2024). This makes explicit that a unipolar measurement is not defined only by topology; it is also defined by a timescale separation between carrier, demodulation, and DUT dynamics.

Integrated readout architectures push this logic further. A $\mathbb{Z}_{Sh}$1, $\mathbb{Z}_{Sh}$2 time-to-digital impedance measurement IC scanned $\mathbb{Z}_{Sh}$3 piezo-resistive sensors in $\mathbb{Z}_{Sh}$4 at $\mathbb{Z}_{Sh}$5, covering $\mathbb{Z}_{Sh}$6 to $\mathbb{Z}_{Sh}$7 with $\mathbb{Z}_{Sh}$8 error and up to $\mathbb{Z}_{Sh}$9 SNR, but it did so with a fully differential excitation/readout front-end rather than a strict single-ended geometry (Li et al., 15 Oct 2025). This suggests that when throughput and power become dominant constraints, differential demodulation is often retained even if the system is supply-unipolar.

5. Experimental domains and representative results

The clearest recent demonstration of unipolar two-terminal metrology is cryogenic component characterization. A $\mathbb{Z}_{Op}$0-channel multilayer PCB was mounted at the base stage of a $\mathbb{Z}_{Op}$1He cryostat, with twisted-pair constantan and NbTi wiring from room temperature to the cold board and footprints for common SMD sizes. Using the corrected two-terminal procedure, the study unambiguously identified a $\mathbb{Z}_{Op}$2 drop in capacitance for $\mathbb{Z}_{Op}$3 5XR multilayer ceramic capacitors cooled from $\mathbb{Z}_{Op}$4 to $\mathbb{Z}_{Op}$5: about $\mathbb{Z}_{Op}$6 at $\mathbb{Z}_{Op}$7, about $\mathbb{Z}_{Op}$8 at $\mathbb{Z}_{Op}$9, and about $0$00 at $0$01. The same method measured a $0$02 SiO$0$03/SiON thin-film RF capacitor as essentially temperature-stable, while a nominal $0$04 device was extracted as only about $0$05, which the authors treated as a limitation of the method when the DUT is only a few percent of the parasitic background. Four $0$06 thick-film resistors increased to about $0$07–$0$08 at $0$09 and to about $0$10–$0$11 at $0$12, with one part increasing nearly an order of magnitude (Carpenter et al., 13 Jun 2025).

In tribological sensing, the unbalanced AC Wheatstone bridge was proposed as a reliable way of measuring impedances with low phase angles at sampling rates in the kHz range. In a hybrid bearing containing a single steel ball, with $0$13, $0$14, $0$15, $0$16, and $0$17, the processed capacitance exhibited one clear peak per cage revolution as the steel ball entered the load zone. Outside the load zone, the capacitance was about $0$18. In fatigue tests on $0$19 and $0$20 bearings, frequency-domain features derived from the measured complex impedance separated run-in, normal operation, and failure phases, and the changes were reported hours before vibration-based monitoring exceeded failure thresholds (Puchtler et al., 2024).

Biomedical instrumentation provides the most explicit counterexample to unipolar practice. A multichannel pulse-wave device injects $0$21 AC in the $0$22–$0$23 range and acquires up to four channels simultaneously with $0$24–$0$25 SNR, but it uses separate current and voltage electrodes, differential routing, analog rectification, and simultaneous-sampling ADCs precisely to avoid a unipolar or two-electrode geometry (Kusche et al., 2019). A batteryless implantable bio-impedance ASIC likewise performs accurate 4-point complex impedance extraction from $0$26 to $0$27, with around $0$28 resolution, because the authors explicitly regarded 2-terminal measurements as dominated by electrode–tissue interface impedances (Rodriguez et al., 2015). These examples do not invalidate unipolar methods; rather, they delimit the domains in which unipolar measurements remain attractive: passive components, fixture-bound devices, and systems where the reference path is mechanically and electrically reproducible.

6. Limitations, error mechanisms, and methodological boundaries

The dominant misconception is that a unipolar measurement fails only because of poor instrument resolution. The recent literature shows instead that the limiting errors are usually structural. In cryogenic two-terminal measurements, the principal failure modes are systematic meter phase offset, the approximation $0$29, the need to subtract a small DUT response from a much larger parasitic background, and the onset of DUT self-resonances or distributed parasitics above about $0$30. The $0$31 thin-film capacitor measured against an open-channel background of order $0$32 is the canonical example: the method still works formally, but small systematic errors in $0$33 and phase calibration dominate the extracted result (Carpenter et al., 13 Jun 2025).

For electrode-based measurements, the corresponding structural term is contact impedance. The complete electrode model treats each electrode as a constant-potential conductor attached through a finite contact impedance $0$34, while the shunt model assumes perfect contact. The modelling paper proves that the error in absolute current-to-voltage measurements produced by replacing the complete electrode model with the shunt model is almost linear in the contact impedances, and that the two models have genuinely different Sobolev regularity properties. It also shows numerically that the CEM stiffness matrix becomes ill-conditioned as $0$35 (Staboulis et al., 2013). Thus, “ignoring the contact” is not merely an approximation in the metrological sense; it changes the PDE class and the numerical behavior of the forward problem.

A further boundary case appears in the HiPIMS literature, where “unipolar” refers to target pulse configuration rather than impedance instrumentation. That paper does not perform classical impedance spectroscopy, but it models an insulated surface as a capacitor to ground with charging time

$0$36

The observed dependence of energy flux on floating, $0$37, $0$38, $0$39, diode-clamped, and grounded surfaces is explicitly described as capacitive charging and discharge during chopped pulses (Farahani et al., 2024). This suggests a useful conceptual extension: even when no impedance is reported as $0$40, unipolar systems often behave as time-domain impedance networks whose effective response is governed by capacitance, leakage, and reset paths.

Taken together, these results define the proper scope of unipolar impedance measurements. They are most effective when the DUT is a passive two-terminal element or a fixture-constrained load, when open and short backgrounds can be measured in situ, when the reference path is reproducible, and when the extracted parameter is taken only from frequency ranges that remain flat and model-consistent. They become progressively less reliable as the measured quantity becomes the difference between two large complex backgrounds, as contact or electrode impedances become unstable, or as the geometry demands localization that only tetrapolar or fully differential sensing can supply.