Half-Cone Transducer for YIG MSW Filters

• Half-Cone Transducer is a single-side-tapered electrode design that spatially tailors spin-wave excitation in YIG MSW filters to suppress undesired edge-localized modes.
• It maintains constant inter-transducer spacing, preserving phase coherence while reducing current density at the film edges to mitigate spurious responses.
• Experimental and simulated results demonstrate low insertion loss, high out-of-band rejection, and ultra-wide frequency tuning across single- and dual-cavity implementations.

Searching arXiv for papers on "half-cone transducer" and related terms to ground the article in current literature. I’ll query arXiv for the exact phrase and closely related magnetostatic-wave/YIG filter work. The Half-Cone Transducer (HCT) is an apodized radio-frequency transducer for yttrium iron garnet (YIG) magnetostatic-wave (MSW) cavity filters that spatially tailors spin-wave excitation to maximize coupling into the desired primary cavity modes while suppressing width-mode spurs generated by finite cavity dimensions. In the reported implementation, the HCT is used in Damon–Eshbach magnetostatic surface-wave (MSSW) cavities under an in-plane bias field $H_0$, where conventional uniform transducers excite both the intended longitudinal cavity modes and undesired edge-localized transverse modes. The HCT addresses this by tapering only one side of each electrode, thereby reducing current density and RF magnetic field near the film edges while preserving constant spacing between input and output transducers. This geometry enables spurious-free, low-loss, ultra-wide frequency-tunable MSW cavity filters in both single-cavity and dual-cavity realizations (Wu et al., 19 Jul 2025).

1. Definition, motivation, and operating principle

In the YIG MSW-filter context, the HCT is a single-side-tapered electrode geometry that shapes the amplitude and phase of the excitation field $H_{\mathrm{rf}}(\mathbf{r})$ across the cavity aperture. Conventional straight strips, loops, and microstrips radiate an RF magnetic field that is spatially broad and spectrally indiscriminate. In a finite YIG film, that excitation couples not only to the desired primary longitudinal cavity modes but also to quantized width modes across the finite film width. These width modes are edge-localized, appear just below the passband due to MSSW dispersion, and degrade selectivity, ripple, and insertion loss (Wu et al., 19 Jul 2025).

The HCT imposes a spatial weighting $w(x,y)$ on $H_{\mathrm{rf}}(x,y)$. Wider electrode sections carry lower current density, while narrower sections carry higher current density; the resulting field distribution suppresses excitation near the film edges, where width modes reside, and maintains strong excitation of the central MSSW paths associated with the primary modes. The defining geometric distinction is that only one edge of each electrode is widened, while the opposite edge remains straight, so the spacing between the two transducers across the cavity remains constant (Wu et al., 19 Jul 2025).

This constant-spacing property is central to the device concept. A full-cone taper, in which both sides are tapered, changes the inter-transducer spacing across the aperture and perturbs the MSSW phase at the receiver. The half-cone geometry instead preserves longitudinal modal phase coherence while suppressing edge-weighted modes. This suggests that the HCT is not merely an impedance- or aperture-shaping element, but a mode-selective excitation structure whose effectiveness depends on simultaneous control of spatial overlap and phase integrity (Wu et al., 19 Jul 2025).

2. Device architecture and physical realization

The reported HCT is implemented on a single-crystal YIG MSW cavity. The film thickness is approximately $t \approx 15~\mu\mathrm{m}$ on a $500~\mu\mathrm{m}$ gadolinium gallium garnet (GGG) substrate. Thick YIG is used because it minimizes MSW loss and improves coupling and power handling. Typical simulated cavity dimensions are $L \approx 280~\mu\mathrm{m}$ along the propagation direction $x$ and $W \approx 400~\mu\mathrm{m}$ along the transverse direction $y$, with fabricated cavities of comparable size (Wu et al., 19 Jul 2025).

The bias field $H_{\mathrm{rf}}(\mathbf{r})$0 lies in-plane parallel to the transducers, along $H_{\mathrm{rf}}(\mathbf{r})$1, जबकि waves propagate along $H_{\mathrm{rf}}(\mathbf{r})$2, exciting surface Damon–Eshbach MSSW in a nonreciprocal configuration. In the model, the film edges act as reflective boundaries, and the transverse component satisfies $H_{\mathrm{rf}}(\mathbf{r})$3 at $H_{\mathrm{rf}}(\mathbf{r})$4. Aluminum electrodes of thickness approximately $H_{\mathrm{rf}}(\mathbf{r})$5 and central width approximately $H_{\mathrm{rf}}(\mathbf{r})$6 are patterned on the YIG. The device uses benzocyclobutene planarization around the cavity, ground-signal-ground pads, and wave ports, with no ground plane under the cavity. For simulations, the transducer is placed at approximately $H_{\mathrm{rf}}(\mathbf{r})$7 to achieve strong coupling and good out-of-band rejection (Wu et al., 19 Jul 2025).

The HCT itself is defined by a three-point arc that produces a half-cone taper along both $H_{\mathrm{rf}}(\mathbf{r})$8 and $H_{\mathrm{rf}}(\mathbf{r})$9. The optimized dimensions in HFSS are approximately $w(x,y)$0 for the half-cone length along $w(x,y)$1 and $w(x,y)$2 for the aperture along $w(x,y)$3. Each port comprises a pair of apodized aluminum electrodes contacting the BCB-planarized area, and the vector network analyzer drives $w(x,y)$4 excitation into $w(x,y)$5 ports (Wu et al., 19 Jul 2025).

The YIG material parameters are treated within a standard MSW framework. The saturation magnetization is typically about $w(x,y)$6 (approximately $w(x,y)$7); the gyromagnetic ratio is $w(x,y)$8 (approximately $w(x,y)$9); the exchange stiffness used in the Kalinikos formalism is $H_{\mathrm{rf}}(x,y)$0; and the Gilbert damping is small in liquid-phase-epitaxy-grown YIG (Wu et al., 19 Jul 2025).

3. Magnetostatic-wave theory and mode selectivity

The HCT operates in a finite MSSW cavity whose in-plane spectrum is quantized along both the propagation and width directions. In the magnetostatic limit, with $H_{\mathrm{rf}}(x,y)$1 and $H_{\mathrm{rf}}(x,y)$2, the Damon–Eshbach surface-wave dispersion is

$H_{\mathrm{rf}}(x,y)$3

while the backward-volume MSW dispersion is

$H_{\mathrm{rf}}(x,y)$4

The paper further uses the dipole–exchange Kalinikos–Slavin formalism,

$H_{\mathrm{rf}}(x,y)$5

with

$H_{\mathrm{rf}}(x,y)$6

to account for both magnetostatic and exchange effects (Wu et al., 19 Jul 2025).

Finite cavity dimensions impose the quantization conditions

$H_{\mathrm{rf}}(x,y)$7

for longitudinal primary modes and

$H_{\mathrm{rf}}(x,y)$8

for width modes. Because MSSW dispersion shifts to lower frequency for larger transverse $H_{\mathrm{rf}}(x,y)$9, higher-order width modes appear below the desired passband and manifest as spurious peaks. Mode profiles show that primary modes with $t \approx 15~\mu\mathrm{m}$0 are concentrated near the cavity center along $t \approx 15~\mu\mathrm{m}$1, whereas width modes with $t \approx 15~\mu\mathrm{m}$2 become increasingly edge-localized (Wu et al., 19 Jul 2025).

The coupling mechanism is expressed through the RF-field contribution to the effective field in the Landau–Lifshitz–Gilbert equation,

$t \approx 15~\mu\mathrm{m}$3

with $t \approx 15~\mu\mathrm{m}$4 generated by the transducer current density through Ampère’s law,

$t \approx 15~\mu\mathrm{m}$5

Coupling to a cavity eigenmode $t \approx 15~\mu\mathrm{m}$6 is governed by the overlap integral

$t \approx 15~\mu\mathrm{m}$7

In a reciprocal Fourier-space description, if $t \approx 15~\mu\mathrm{m}$8 denotes the spectral content of the excitation field and $t \approx 15~\mu\mathrm{m}$9 the modal spectral weight, then

$500~\mu\mathrm{m}$0

The HCT apodization reduces $500~\mu\mathrm{m}$1, thereby suppressing coupling to higher-order width modes while maintaining strong overlap with the primary modes (Wu et al., 19 Jul 2025).

4. Half-cone versus straight and full-cone transducers

The comparison among straight, full-cone, and half-cone transducers is central to the interpretation of the HCT. Straight transducers excite both primary and width modes because their $500~\mu\mathrm{m}$2 distribution extends strongly into the edge regions of the finite-width cavity. Full-cone tapers suppress some undesired modes, but they narrow both electrodes toward the center and thereby alter the inter-transducer spacing across the aperture. This changes the longitudinal phase relationship seen by the receiver and degrades coupling to higher-order primary modes, especially near the upper passband (Wu et al., 19 Jul 2025).

The half-cone geometry preserves constant spacing across the aperture and therefore maintains longitudinal modal phase coherence. This allows suppression of width modes without sacrificing coupling to high-order primary modes. In the reported simulations, only the half-cone design simultaneously preserves high-order primary-mode coupling and suppresses spurious responses. The distinction is not merely geometric; it determines whether mode selectivity can be improved without introducing additional insertion-loss penalties at high frequency (Wu et al., 19 Jul 2025).

The reported mechanism for insertion-loss improvement includes impedance effects as well. The half-cone lowers the effective electrode impedance $500~\mu\mathrm{m}$3, improving matching to $500~\mu\mathrm{m}$4 ports. In the dual-cavity case, this was corroborated by Smith-chart analysis showing HCT impedances closer to $500~\mu\mathrm{m}$5 than the straight-line reference over $500~\mu\mathrm{m}$6–$500~\mu\mathrm{m}$7; for example, at $500~\mu\mathrm{m}$8, $500~\mu\mathrm{m}$9 for the HCT was approximately $L \approx 280~\mu\mathrm{m}$0, versus $L \approx 280~\mu\mathrm{m}$1 for the reference (Wu et al., 19 Jul 2025).

A plausible implication is that the HCT should be understood as a joint field-shaping and matching strategy. Its coarse apodization of current density avoids the delicate phase-cancellation schemes associated with more complicated prior-art solutions, while retaining the modal coherence needed for wideband tunability (Wu et al., 19 Jul 2025).

5. Experimental validation and quantitative performance

The HCT was validated through theoretical analysis, 3D finite-element simulation in Ansys HFSS, and experiment. The simulations used the magnetostatic approximation and an anisotropic Polder tensor derived from the LLG equation, with $L \approx 280~\mu\mathrm{m}$2 boundary conditions at $L \approx 280~\mu\mathrm{m}$3 and wave ports for excitation. The Kalinikos dipole–exchange dispersion guided mode identification. Experimentally, the devices were measured on a magnetic probe station with a calibrated in-plane bias field; a Keysight P5026B vector network analyzer drove $L \approx 280~\mu\mathrm{m}$4 into $L \approx 280~\mu\mathrm{m}$5 ports after SOLT calibration, using $L \approx 280~\mu\mathrm{m}$6-pitch GSG probes, and the measurements were performed without de-embedding (Wu et al., 19 Jul 2025).

For the single-cavity half-cone MSW filter, the tunable center frequency spans approximately $L \approx 280~\mu\mathrm{m}$7–$L \approx 280~\mu\mathrm{m}$8 under a bias field of approximately $L \approx 280~\mu\mathrm{m}$9–$x$0. The insertion loss remains approximately $x$1–$x$2 across the range, compared with $x$3–$x$4 for the straight-line reference. Representative $x$5 bandwidth values are approximately $x$6–$x$7. All width-mode spurs are suppressed with levels up to approximately $x$8 below the passband, and measured spurious suppression ratios are approximately $x$9–$W \approx 400~\mu\mathrm{m}$0 across bias points. The spurious suppression ratio is defined as

$W \approx 400~\mu\mathrm{m}$1

and insertion loss is taken from the transmission coefficient through

$W \approx 400~\mu\mathrm{m}$2

The passband is flatter than that of straight transducers, and the skirts are spur-free (Wu et al., 19 Jul 2025).

For the dual-cavity HCT-based filter, consisting of two identical cavities in parallel, the tunable center frequency spans approximately $W \approx 400~\mu\mathrm{m}$3–$W \approx 400~\mu\mathrm{m}$4, corresponding to an approximately $W \approx 400~\mu\mathrm{m}$5 tuning range. The insertion loss is approximately $W \approx 400~\mu\mathrm{m}$6–$W \approx 400~\mu\mathrm{m}$7 across the span, compared with approximately $W \approx 400~\mu\mathrm{m}$8–$W \approx 400~\mu\mathrm{m}$9 for the reference, with a difference up to $y$0 at high bands. The out-of-band rejection reaches up to approximately $y$1 at $y$2, and the passband remains spurious-free across all measured bias points (Wu et al., 19 Jul 2025).

Configuration	Frequency range	Insertion loss
Single-cavity HC-MSWF	$y$3–$y$4	$y$5–$y$6
Dual-cavity HC-MSWF	$y$7–$y$8	$y$9–$H_{\mathrm{rf}}(\mathbf{r})$00

The reported device footprints are approximately $H_{\mathrm{rf}}(\mathbf{r})$01 for the single-cavity implementation and approximately $H_{\mathrm{rf}}(\mathbf{r})$02 for the dual-cavity implementation. These figures are presented alongside spurious-free operation, low insertion loss, high out-of-band rejection, and ultra-wide tuning span (Wu et al., 19 Jul 2025).

6. Design guidance, limitations, and broader context

The reported design guidance emphasizes apodization strategy, cavity geometry, and bias control. The half-cone taper should be used to reduce current density at film edges while preserving constant inter-transducer spacing; full-cone tapers should be avoided when high-order primary-mode integrity is important. For the reported geometry, $H_{\mathrm{rf}}(\mathbf{r})$03 and $H_{\mathrm{rf}}(\mathbf{r})$04 were optimal. More generally, initial targets of $H_{\mathrm{rf}}(\mathbf{r})$05–$H_{\mathrm{rf}}(\mathbf{r})$06 and $H_{\mathrm{rf}}(\mathbf{r})$07 are suggested, followed by FEM refinement. Placement near $H_{\mathrm{rf}}(\mathbf{r})$08 provides a balance between coupling and skirt rejection (Wu et al., 19 Jul 2025).

The paper also identifies trade-offs. Stronger apodization improves spur suppression but can reduce overall coupling. Larger cavity width $H_{\mathrm{rf}}(\mathbf{r})$09 increases mode density, making HCT-based suppression increasingly important. Thicker films around $H_{\mathrm{rf}}(\mathbf{r})$10 reduce propagation loss and improve coupling, but increase volume and bias-field requirement. Higher $H_{\mathrm{rf}}(\mathbf{r})$11 shifts the passband upward through $H_{\mathrm{rf}}(\mathbf{r})$12, so bias-field uniformity across the cavity is necessary. In high-frequency bands above $H_{\mathrm{rf}}(\mathbf{r})$13, dual-cavity parallel coupling lowers $H_{\mathrm{rf}}(\mathbf{r})$14 and reduces insertion loss, at the cost of increased footprint and more demanding bias uniformity (Wu et al., 19 Jul 2025).

The reported limitations are also specific. The study focuses on MSSW in $H_{\mathrm{rf}}(\mathbf{r})$15 films; thinner films and strong exchange may require refined apodization and dispersion modeling. Bias-field generation remains an integration challenge for handset-scale front-ends, although co-integration with low-power bias circuits is identified as promising. The same spatial-tailoring concept is described as extendable to other MSW devices, including notch filters, delay lines, isolators, and circulators, through custom $H_{\mathrm{rf}}(\mathbf{r})$16 and multi-port apodization designs (Wu et al., 19 Jul 2025).

In broader arXiv usage, the phrase “half-cone” can denote unrelated concepts. In ultrasonics, a conical acoustic lens attached to an unfocused ultrasonic transducer can generate a quasi-nondiffracting Bessel beam with extended depth of field, but that device is an axicon-like acoustic-lens architecture rather than a YIG spin-wave transducer (Song et al., 2020). In inverse problems, “half-cone” or single-nappe acquisition arises in the conical Radon transform, where limited semicircular cone data breaks the symmetry assumptions used in explicit inversion formulas (Nguyen et al., 2019). In the MSW-filter literature, by contrast, the Half-Cone Transducer denotes a single-side-tapered RF excitation structure for selective cavity-mode control (Wu et al., 19 Jul 2025).

7. Significance for tunable RF filtering

Within the reported YIG MSSW platform, the HCT demonstrates that spatially tailored spin-wave excitation can suppress width-mode spurs without degrading primary-mode coupling. The resulting filters combine spurious-free passbands with low insertion loss across unusually wide tuning spans, including a dual-cavity tuning range of approximately $H_{\mathrm{rf}}(\mathbf{r})$17 from $H_{\mathrm{rf}}(\mathbf{r})$18 to $H_{\mathrm{rf}}(\mathbf{r})$19. The approach avoids multi-section couplers or rotated slabs and is described as robust to fabrication tolerances because it relies on coarse apodization of current density rather than delicate phase-cancellation schemes (Wu et al., 19 Jul 2025).

For RF-front-end applications, the relevance of the HCT lies in reconfigurable spectrum control under continuous bias tuning. The reported data support low-loss operation, strong out-of-band rejection, compact footprint, and preservation of passband cleanliness over a broad frequency span. This suggests that the HCT is best understood as a device-level solution to a longstanding finite-cavity problem in tunable MSW filtering: not the extension of tuning range per se, but the elimination of the spurious-mode penalty that previously accompanied that tunability (Wu et al., 19 Jul 2025).