Directional Point-Contact Spectroscopy

Updated 12 July 2026

The technique utilizes precise alignment of microcontacts along crystallographic axes to reveal superconducting gap amplitudes, multiband features, and anisotropy.
It employs diverse contact architectures—soft, needle-anvil, and shear-adjusted—that enable both Andreev-reflection and electron-phonon spectral analysis under cryogenic conditions.
In situ adjustments allow real-time control over junction resistance and spatial mapping, enhancing the resolution of multiband and anisotropic superconducting properties.

In situ directional point-contact spectroscopy is a class of cryogenic transport measurements in which a metallic microcontact is created, adjusted, and interrogated inside the experimental environment while the current-injection direction is selected with respect to crystallographic axes. In superconductors, the method most commonly appears as point-contact Andreev-reflection spectroscopy (PCAR), where the bias dependence of the differential conductance is analyzed to extract superconducting gap amplitudes, multiband structure, anisotropy, and gap symmetry; in the broader point-contact-spectroscopy sense it also includes second-derivative measurements that map the electron-phonon interaction spectral function and can reveal collective charge-density-wave modes. Across Fe-based superconductors, layered dichalcogenides, and kagome superconductors, the technique has evolved from simple needle-anvil contacts to soft contacts, in situ shear-adjusted junctions, and piezo-positioned lateral-scanning probes, while the interpretive framework has expanded from 1D BTK analysis to directional multiband, anisotropic, and Eliashberg-based descriptions (Daghero et al., 2011, Tortello et al., 2016, Bobrov et al., 2016, Zhao et al., 7 Sep 2025).

1. Conceptual basis and directional selectivity

The essential object in point-contact spectroscopy is a metallic constriction sufficiently small that transport is ballistic or Sharvin-like. In soft point-contact measurements on CsV $_3$ Sb $_5$ , contact resistances $R_J \sim 0.5$ – $10\,\Omega$ were taken to ensure ballistic or Sharvin-like conduction, whereas in a needle-anvil rig the stepping procedure was stopped when $R_J$ reached $10$– $100\,\Omega$ , identified there as the typical Sharvin condition (Zhao et al., 7 Sep 2025, Tortello et al., 2016). In superconducting contacts, the low-bias conductance is governed by Andreev reflection and is therefore sensitive to the superconducting order parameter. In ballistic normal-state contacts, the second derivative $d^2V/dI^2(V)$ is proportional to the Eliashberg electron-phonon spectral function $\alpha^2F(eV)$ , so bias-dependent structure can be assigned to phonon modes or, in special cases, to collective excitations (Bobrov et al., 2016).

Directional sensitivity is obtained by controlling which crystallographic face is presented to the counterelectrode. In Fe-based superconductors, directional PCAR uses single crystals mounted so that either the $ab$ plane or the $_5$ 0 axis is normal to the tip, ensuring directional current injection; in more advanced setups, a rotatable stage allows fine adjustment of the angle between crystal axes and the tip (Daghero et al., 2011). In CsV $_5$ 1Sb $_5$ 2, a wire bonded onto a freshly cleaved $_5$ 3-plane surface probes current flowing predominantly along $_5$ 4, whereas bonding onto a polished $_5$ 5 or $_5$ 6 face injects current parallel to the $_5$ 7 plane; the reported alignment error was kept below $_5$ 8 (Zhao et al., 7 Sep 2025). In layered 2H-NbSe $_5$ 9, c-axis and basal-plane conduction were selected by presenting either the basal plane or a plate edge to the counterelectrode, with the mechanical compliance chosen to avoid crushing the van der Waals-bonded layers (Bobrov et al., 2016).

The term in situ denotes that the contact is formed or tuned inside the cryogenic apparatus rather than assembled ex situ and then cooled. In the Fe-based PCAR review, both needle-anvil and soft contacts are described as implementable inside a cryostat so that contacts are formed and left undisturbed while cooling (Daghero et al., 2011). In the lateral-scanning rig, in situ capability is literal and mechanical: the tip can move along both the vertical $R_J \sim 0.5$ 0 and horizontal $R_J \sim 0.5$ 1 directions via piezoelectric nano-positioners, allowing the probe to contact different regions of a sample without warming the system (Tortello et al., 2016). In the older NbSe $R_J \sim 0.5$ 2 configuration, a torsion screw outside the cryostat shears the electrodes against each other until a metallic microcontact forms, which likewise permits in situ control of contact force and orientation (Bobrov et al., 2016).

2. Contact architectures and cryogenic implementations

Three experimentally distinct implementations recur in the literature: needle-anvil contacts, soft contacts, and shear-adjusted contacts. Each realizes directional control differently, and each imposes a different balance between transparency, mechanical stability, and angular definition.

Implementation	Direction selection	Representative features
Soft SPCAR	Junction on $R_J \sim 0.5$ 3 face for $R_J \sim 0.5$ 4-axis current; junction on $R_J \sim 0.5$ 5 face for in-plane current	16 $R_J \sim 0.5$ 6m Pt wire, Ag-paint particles 50–100 nm, $R_J \sim 0.5$ 7– $R_J \sim 0.5$ 8 nano-junctions, no external spring-loaded pressure head
Needle-anvil with lateral scan	Tip axis normal to selected face; $R_J \sim 0.5$ 9 and $10\,\Omega$ 0 piezos reposition the tip in situ	Detachable head of 26.5 mm diameter, $10\,\Omega$ 1 mm, $10\,\Omega$ 2 mm, Helmholtz coil, optional semi-rigid coaxial cable
Shear/torsion microcontact	Basal plane vs plate edge selects $10\,\Omega$ 3-axis vs basal-plane conduction	$10\,\Omega$ 4-shaped Be–Cu damper, torsion-screw adjustment outside cryostat, rigid vs soft support to tailor compliance

Soft contacts replace direct metallic-tip pressure with an Ag-paint-mediated ensemble of nanojunctions. In CsV $10\,\Omega$ 5Sb $10\,\Omega$ 6, a 16 $10\,\Omega$ 7m-diameter platinum wire was dipped in Ag-paint with particle size 50–100 nm and then gently pressed onto the selected crystal face. The Ag-paint provides a large ensemble, approximately $10\,\Omega$ 8– $10\,\Omega$ 9, of nano-junctions without mechanical scratching or tip plunging, and the junction pressure is self-limited by the compliance of the Ag-paint film because no external spring-loaded pressure head is used (Zhao et al., 7 Sep 2025). This geometry is particularly suited to fragile or air-sensitive materials and, as emphasized for Fe-based superconductors, soft contacts are often preferred because they avoid tip pressure and the associated spurious zero-bias anomalies (Daghero et al., 2011).

Needle-anvil implementations prioritize controlled positioning and retraceable motion. The cryogenic rig of Biswas et al. employs a sharpened Au or PtIr tip mounted on orthogonal Attocube nano-positioners, ANPz51/LT for the $R_J$ 0 axis and ANPx51/LT for the $R_J$ 1 axis, all housed in a detachable head that fits the 28 mm PPMS vacuum can. The reported stage specifications are $R_J$ 2 mm with 50 nm step resolution per 30 V pulse at 1.8 K and $R_J$ 3 with $R_J$ 4 nm/V, and $R_J$ 5 mm with 48 nm step resolution per 30 V pulse and $R_J$ 6 with $R_J$ 7 nm/V (Tortello et al., 2016). The same head incorporates a Cernox thermometer pocket, a 16 pin electrical connector, a Helmholtz coil, and a cavity for a semi-rigid coaxial cable, making the platform not only directional but multifunctional.

Shear-adjusted contacts are mechanically simpler yet highly effective for layered crystals. In 2H-NbSe $R_J$ 8, single-crystal plates were glued with Ag paste to the arms of a $R_J$ 9-shaped beryllium-bronze damper. Twisting one arm by means of a torsion screw outside the cryostat causes the NbSe$10$0 and Cu electrodes to slide against each other until a metallic microcontact is formed. The same mechanism permits in situ control of both contact force and orientation, and different Cu geometries were used for c-axis and in-plane contacts: a three-facet prism with tip radius $10$1–0.2 mm for c-axis spectra and a $10$2 mm$10$3 parallelepiped for in-plane spectra (Bobrov et al., 2016).

Cryogenic conditions are dictated by the targeted energy scale. The CsV$10$4Sb$10$5 soft-PCAR experiment used a He$10$6 cryostat with base temperature $10$7 K and stable control up to $10$8 K, with a RuO$10$9 thermometer near the crystal and stability better than $100\,\Omega$ 0 mK over each sweep (Zhao et al., 7 Sep 2025). The lateral-scanning rig operated in a PPMS system down to 1.8 K, with all structural parts non-magnetic and thermally anchored via braided copper flexors and miniature Cu blocks (Tortello et al., 2016). The NbSe $100\,\Omega$ 1 study employed two temperature ranges, 1.6–4.2 K in a He bath with pumped vapor and 4.4–10 K in a separate gas-exchange cryostat, with $100\,\Omega$ 2 throughout (Bobrov et al., 2016).

3. Measurement channels and operating protocols

The primary observable in superconducting point-contact spectroscopy is the differential conductance $100\,\Omega$ 3. In Fe-based superconductors, a standard four-probe scheme applies a small dc bias $100\,\Omega$ 4 over $100\,\Omega$ 5 mV together with a low-frequency 1–2 kHz ac excitation, typically 50–100 $100\,\Omega$ 6V, and records $100\,\Omega$ 7 by lock-in amplification from base temperature $100\,\Omega$ 8–2 K up to above $100\,\Omega$ 9, sometimes in magnetic fields up to 6–9 T (Daghero et al., 2011). In CsV $d^2V/dI^2(V)$ 0Sb $d^2V/dI^2(V)$ 1, $d^2V/dI^2(V)$ 2 was recorded in a quasi four-terminal geometry using a small ac modulation of approximately 20 $d^2V/dI^2(V)$ 3V superimposed on a dc bias current, with the lock-in referenced to the ac frequency (Zhao et al., 7 Sep 2025). In the PPMS-based rig, the junction resistance during approach was monitored in a four-probe lock-in scheme with $d^2V/dI^2(V)$ 4A rms at 877 Hz (Tortello et al., 2016).

For electron-phonon interaction spectroscopy, the relevant observables are the first and second harmonics of the voltage response. In 2H-NbSe $d^2V/dI^2(V)$ 5, a small ac current $d^2V/dI^2(V)$ 6 at approximately kHz frequency was superimposed on the dc bias, and the harmonics $d^2V/dI^2(V)$ 7 and $d^2V/dI^2(V)$ 8 were detected with lock-in amplifiers. In the ballistic limit,

$d^2V/dI^2(V)$ 9

so peaks in $\alpha^2F(eV)$ 0 at bias $\alpha^2F(eV)$ 1 correspond to phonon modes of energy $\alpha^2F(eV)$ 2 (Bobrov et al., 2016). This provides a spectroscopic route complementary to Andreev-reflection analysis: instead of fitting subgap conductance, one maps the inelastic bosonic spectrum.

In situ operation is not limited to initial contact formation. In the lateral-scanning rig, after cooling to about 1.8 K the tip is first retracted by $\alpha^2F(eV)$ 3m in $\alpha^2F(eV)$ 4, then stepped toward the sample in increments of $\alpha^2F(eV)$ 5 nm until the desired junction resistance is obtained. To probe a new site, the tip is retracted by 500 nm in $\alpha^2F(eV)$ 6, translated by $\alpha^2F(eV)$ 7m, and re-approached, and a full $\alpha^2F(eV)$ 8 curve is recorded at each position (Tortello et al., 2016). Because the system is open-loop, stage capacitance is monitored by software to detect anomalies, and the reported mechanical noise floor is $\alpha^2F(eV)$ 9 nm rms with thermal drift $ab$ 0 nm min $ab$ 1 over 12 h at 1.8 K (Tortello et al., 2016).

Normalization is a critical part of the protocol. In Fe-based superconductors, each raw $ab$ 2 curve at $ab$ 3 is divided by the normal-state conductance measured either at $ab$ 4 or, when accessible, at $ab$ 5; the review explicitly recommends normalization by the actual normal state rather than by straight-line fits so as to preserve high-energy structures (Daghero et al., 2011). In the CsV $ab$ 6Sb $ab$ 7 analysis, the raw $ab$ 8 was divided by a high-bias normal conductance $ab$ 9 before fitting (Zhao et al., 7 Sep 2025). The operational significance is that both gap-edge features and higher-bias bosonic structures are sensitive to how the background is removed.

4. Theoretical descriptions: BTK, anisotropy, multiband effects, and strong coupling

The standard starting point for PCAR analysis is the Blonder-Tinkham-Klapwijk framework. In the 1D isotropic single-gap limit, the dimensionless barrier parameter is $_5$ 00 and the normal transparency is $_5$ 01 (Daghero et al., 2011). Directional spectroscopy requires the angularly resolved extensions. In the 2D Tanaka-Kashiwaya formulation for a cylindrical Fermi surface and arbitrary in-plane injection angle $_5$ 02,

$_5$ 03

and the normalized conductance at $_5$ 04 is

$_5$ 05

For realistic multiband systems with nontrivial Fermi-surface geometry, the 3D generalization integrates over each Fermi-surface sheet $_5$ 06 and includes the projection of the Fermi velocity onto the current direction through a weight factor $_5$ 07 (Daghero et al., 2011).

A compact directional multiband form is

$_5$ 08

with $_5$ 09 (Daghero et al., 2011). Here each band $_5$ 10 can carry its own barrier parameter $_5$ 11, gap function $_5$ 12, and broadening $_5$ 13. In simpler two-band 2D fits, a phenomenological weight factor $_5$ 14 is often used, whereas in full 3D fits the weights arise automatically from the Fermi-surface geometry rather than being treated as free parameters (Daghero et al., 2011).

Anisotropic order parameters are introduced by writing

$_5$ 15

where $_5$ 16 may represent isotropic $_5$ 17, fully anisotropic $_5$ 18 with zeros, $_5$ 19, or related forms (Daghero et al., 2011). For the kagome superconductor CsV $_5$ 20Sb $_5$ 21, the sixfold symmetry of the $_5$ 22 plane was modeled with an anisotropic $_5$ 23-wave gap

$_5$ 24

or equivalently

$_5$ 25

where $_5$ 26 is the average gap amplitude and $_5$ 27 is the anisotropy ratio (Zhao et al., 7 Sep 2025). Quasiparticle lifetime broadening is incorporated by the substitution $_5$ 28, equivalent to a Dynes parameter.

The model used for CsV $_5$ 29Sb $_5$ 30 distinguishes the two principal current directions. For c-axis junctions, the conductance was fitted by a two-isotropic-gap BTK expression,

$_5$ 31

where $_5$ 32 is the spectral weight of the small-gap channel. For in-plane junctions,

$_5$ 33

In both geometries, the calculation proceeds by computing Andreev and normal reflection probabilities with the chosen $_5$ 34, $_5$ 35, and gap structure, convolving with the derivative of the Fermi function, and then either integrating over angle or mixing two channels (Zhao et al., 7 Sep 2025).

Strong-coupling extensions replace the constant gap with an energy-dependent complex Eliashberg gap $_5$ 36 obtained from a chosen bosonic spectral function $_5$ 37, such as a Lorentzian or an antiferromagnetic spin-fluctuation form (Daghero et al., 2011). Substituting $_5$ 38 into the BTK expressions produces high-energy kinks or humps in $_5$ 39 at energies approximately $_5$ 40 and above. This is central to the use of directional PCAR not only as a gap probe but also as a spectroscopic method for the pairing boson.

5. Extended in situ modalities

Directional point-contact spectroscopy can be extended beyond single-junction conductance measurements. In the piezo-based rig, two point contacts on adjacent facets of a single crystal can be used to form a DC SQUID for phase-sensitive Josephson interferometry. With transverse magnetic flux $_5$ 41, the critical current follows

$_5$ 42

where $_5$ 43 (Tortello et al., 2016). Real-time adjustment of the contact areas via the piezos allows the two junctions to be balanced so that $_5$ 44, maximizing the modulation contrast. In the reported Nb thin-film demonstration, a 75% modulation depth was achieved by fine tuning the contacts (Tortello et al., 2016).

Microwave irradiation adds another in situ spectroscopic channel. The same probe can host a semi-rigid coaxial line, 3.5 mm in diameter, routed along the probe and thermally anchored at each stage for microwave delivery over 0.1–20 GHz (Tortello et al., 2016). Under irradiation, Josephson point contacts exhibit Shapiro steps at

$_5$ 45

with step heights governed by Bessel functions $_5$ 46 (Tortello et al., 2016). This mode is conceptually adjacent to PCAR because the same mechanically defined point contacts can be reconfigured from quasiparticle spectroscopy to Josephson dynamics without changing the cryogenic platform.

Lateral scanning is another distinctive extension of the in situ concept. The horizontal piezo stage permits spectroscopy at multiple sites on the same sample, which is useful for assessing spatial reproducibility, identifying damaged regions, or studying heterogeneity. On a Nb thin film, conductance spectra taken at five lateral sites with 100 $_5$ 47m spacing showed junction resistances varying from 78 $_5$ 48 to 12 $_5$ 49 as contact pressure increased, while the fitted $_5$ 50 parameter decreased from 1.8 to 0.9 (Tortello et al., 2016). A plausible implication is that in situ directional point-contact spectroscopy is not restricted to a single fixed junction geometry; it can also function as a spatially resolved probe of local interface transparency and superconducting response.

6. Representative results in superconductors and CDW materials

In Fe-based superconductors, directional PCAR established several broad results. The 1111 and 122 families near optimal doping exhibit at least two nodeless gaps with ratios $_5$ 51–3 and $_5$ 52–9, interpreted as consistent with an $_5$ 53 state and interband pairing (Daghero et al., 2011). In high- $_5$ 54 1111 and 122 materials, no zero-bias peaks appear in ab-plane soft-PCAR spectra, which was taken to rule out line nodes on any Fermi-surface sheet (Daghero et al., 2011). By contrast, in some low- $_5$ 55 or 3D-layered materials, including Ca(Fe,Co) $_5$ 56As $_5$ 57 and Sr $_5$ 58ScFePO $_5$ 59, systematic zero-bias conductance peaks required fits with $_5$ 60-wave or anisotropic $_5$ 61-wave gaps and identified these systems as candidates for nodal pairing (Daghero et al., 2011). The same body of work showed that high-energy conductance structures track a boson energy $_5$ 62– $_5$ 63 that decreases with temperature and matches neutron spin-resonance data, which was interpreted as evidence for spin-fluctuation-mediated pairing (Daghero et al., 2011).

In 2H-NbSe $_5$ 64, directional point-contact spectroscopy was used not primarily for Andreev gap fitting but for anisotropic electron-phonon spectroscopy. For c-axis contacts, the main reproducible peaks averaged over approximately 14 spectra appeared at 5, 10, 13, 16, 19, 25, 31, 39, 44, 48, 52, and 58 meV. For basal-plane contacts, the corresponding main peaks were at 5, 10, 14, 16, 20, 25, 29, 36, 42, 46–48, 51–52, and 56–58 meV (Bobrov et al., 2016). The anisotropy is especially pronounced in the high-energy cutoff, approximately 40 meV along $_5$ 65 versus approximately 60 meV in plane. Comparison with the computed dispersions of Feldman and Motizuki gave agreement within approximately $_5$ 66 meV and supported the phononic assignment of the spectral features (Bobrov et al., 2016). Additional peaks at 5 meV, and occasionally near 2–4 meV, were observed only in basal-plane contacts below 33 K; these vanished above the CDW transition and were assigned to collective amplitude and phase modes of the incommensurate charge-density wave (Bobrov et al., 2016).

In the kagome superconductor CsV $_5$ 67Sb $_5$ 68, directional soft PCAR provided explicit evidence for a strongly anisotropic superconducting gap pairing state (Zhao et al., 7 Sep 2025). The spectra measured with current injected parallel to the $_5$ 69 plane exhibited an in-gap single conductance peak, whereas spectra in the perpendicular direction showed a double-peak structure. The c-axis spectra at 0.28 K were fitted by a two-isotropic-gap BTK model, yielding a small gap $_5$ 70 meV and a large gap $_5$ 71 meV in sample #1, and $_5$ 72 meV and $_5$ 73 meV in sample #2, with barrier strengths $_5$ 74 and 0.78, broadening $_5$ 75–0.15 meV and $_5$ 76–0.12 meV, and weight $_5$ 77 (Zhao et al., 7 Sep 2025). The ab-plane spectra at the same base temperature were fitted by a single anisotropic-gap BTK model with average gap $_5$ 78 meV in sample #1 and 0.49 meV in sample #2, anisotropy ratio $_5$ 79, broadening $_5$ 80 meV, and barrier $_5$ 81 (Zhao et al., 7 Sep 2025). Using

$_5$ 82

the reported values give $_5$ 83 meV and an anisotropy

$_5$ 84

The extracted picture was that the superconducting gaps comprise an isotropic large gap and a strongly anisotropic gap originating from different Fermi-surface sheets (Zhao et al., 7 Sep 2025).

7. Methodological limits, interpretive cautions, and best practices

The most important limitation of soft contacts is angular averaging. In CsV $_5$ 85Sb $_5$ 86, the authors explicitly note that soft contacts average over a small solid angle around the nominal injection direction rather than over a single $_5$ 87 vector, and that averaging over many nano-contacts can mask very sharp features (Zhao et al., 7 Sep 2025). The same study states that high-purity samples with high residual-resistivity ratio and very thin silver layers improve $_5$ 88-space resolution, while temperature stability below 0.3 K and accurate thermometry are required for resolving sub-meV gaps (Zhao et al., 7 Sep 2025). Mechanical and chemical stability of the Ag-paint interface may also degrade over repeated thermal cycles (Zhao et al., 7 Sep 2025). These constraints do not invalidate soft-PCAR results, but they delimit the angular sharpness and long-term reproducibility of the method.

A common interpretive issue concerns zero-bias structures. In the Fe-based literature, zero-bias conductance peaks are used as evidence that anisotropy or sign change must be included in the gap function, but the same review recommends soft contacts when possible specifically to avoid tip pressure and spurious zero-bias anomalies (Daghero et al., 2011). This suggests that a zero-bias feature is not self-interpreting: its evidentiary value depends on junction formation method, contact pressure, current direction, and whether the conductance can be fitted without invoking gap zeros. In the same vein, relative band weights in simple multiband fits should not automatically be read as direct Fermi-surface measures, because in full 3D BTK the weights arise from the Fermi-surface geometry and the projection of the Fermi velocity along the current direction (Daghero et al., 2011).

Best practice in directional point-contact spectroscopy is correspondingly procedural. The Fe-based review recommends exposing well-defined $_5$ 89 or $_5$ 90 faces, using a rotatable stage if angular dependence is to be studied in detail, normalizing the spectra by the actual normal state rather than by straight lines, and matching the model complexity to the data: 1D or 2D BTK for isotropic gaps and simple geometries, full 3D formulations for anisotropic gaps and realistic Fermi surfaces, and Eliashberg-based extensions when strong-coupling features are present (Daghero et al., 2011). For layered or CDW-bearing materials, the NbSe $_5$ 91 work recommends clean, oxide-controlled electrodes, in situ shear or torsion adjustment, S–c–N contacts such as Cu counterelectrodes to reduce self-heating and enhance EPI features, and comparison with computed or neutron/Raman density of states to assign phonon branches and identify low-frequency collective modes (Bobrov et al., 2016). Taken together, these practices define in situ directional point-contact spectroscopy not as a single instrument or contact geometry, but as a family of orientation-resolved microjunction techniques whose reliability depends on simultaneous control of crystallographic alignment, interface transparency, thermal environment, and model selection.