Advanced Piezoelectric Semiconductors
Aluminum Scandium Nitride (AlScN) RF Filter Lab
Explore reactive magnetron co-sputtering of $\text{Al}_{1-x}\text{Sc}_x\text{N}$ thin films, wurtzite $c$-axis lattice softening, piezoelectric $d_{33}$ scaling, electromechanical coupling ($k_t^2$), and 20-year DCF cleanroom fab economics for next-generation 5G/6G Sub-6 GHz and Wi-Fi 7 Bulk Acoustic Wave (BAW/FBAR) filters.
Peak Resonator Coupling ($k_t^2$)
15.8%
d33: 22.4 pC/N
Center Freq ($f_0$)
6.38 GHz
Bandwidth
415 MHz (6.5%)
Annual Revenue
$105.5M
38.2M chips/yr
Annual EBITDA
$58.4M
55.4% margin
20-Yr NPV (8.5%)
$248.6M
Discounted FCF
Project IRR
28.4% IRR
3.2 Years
Levelized Wafer Cost
$1,120 / wafer
$0.048 / chip
Sc Target Cost
$42.8 / wafer
4N5 Scandium metalAlScN BAW/FBAR Cleanroom Fabrication Flowsheet
Sputter PVD & Cleanroom Fab Parameters
Piezo-Acoustic Physics & Formulas
1. Piezoelectric Coefficient Scaling ($d_{33}$)
Scandium substitution induces elastic lattice softening along the $c$-axis, flattening ionic potential wells and boosting $d_{33}$:
$d_{33}(x) = 5.5 \times (1 + 7.2x + 4.8x^2)\text{ [pC/N]}$
Pure AlN = $5.5\text{ pC/N}$; at $x=0.35$, $d_{33} \approx 29.5\text{ pC/N}$ ($\sim 5.4\times$ increase).
2. Acoustic Velocity & Resonator Center Frequency
Longitudinal sound velocity $v_a(x)$ decreases with scandium mass and softness, governing fundamental resonance $f_0$:
$f_0 = \frac{v_a(x)}{2 \cdot t_{\text{piezo}}} = \frac{10,400 \times (1 - 0.70x)}{2 \cdot t_{\text{piezo}}}\text{ [GHz]}$
Allows $650\text{ nm}$ films to resonate precisely at $6.38\text{ GHz}$ for Wi-Fi 7 / 5G n77.
3. Electromechanical Coupling & Fractional Bandwidth
Coupling coefficient $k_t^2$ sets maximum attainable filter fractional bandwidth (FBW):
$\text{FBW} \approx \frac{4}{\pi^2} k_t^2(x) \quad \text{where } k_t^2(x) = 6.5\% \cdot (1 + 6.3x - 3.9x^2)$
Surpasses the $3\%$ AlN barrier, achieving $6.5-9.0\%$ FBW for wideband carrier aggregation.
4. Acoustic Figure of Merit (FoM)
Measures overall resonator insertion loss and steep roll-off skirts:
$\text{FoM} = k_t^2 \times Q_m \approx 15\% \times 800 \approx 120$
20-Year DCF Cash Flow Schedule ($M USD)
Cleanroom Fab Model| Period | Revenue | Operating Costs | EBITDA | Free Cash Flow | Discounted FCF | Cumulative Cash |
|---|
AlScN Piezoelectric & RF Engineering Knowledge Check
Frequently Asked Questions (FAQ)
When scandium (Sc³⁺) replaces aluminum (Al³⁺) in the hexagonal wurtzite lattice, its larger ionic radius and chemical bonding preference for octahedral coordination soften the elastic stiffness constant c33 along the c-axis. This microscopic elastic flattening dramatically increases polarizability under stress, raising the longitudinal piezoelectric coefficient d33 from ~5.5 pC/N in pure AlN up to ~28-30 pC/N near 35-40 atomic percent Sc.
The electromechanical coupling coefficient kt2 measures the efficiency with which the piezoelectric material converts electrical energy into acoustic energy and vice versa. Filter fractional bandwidth (FBW) is fundamentally limited by coupling (FBW ≈ 4/π² · kt²). Pure AlN offers kt2 ≈ 6.5% (yielding 2.5-3.0% FBW), which cannot cover the ultra-wide 160MHz/320MHz channels of Wi-Fi 7 or the 900MHz bandwidth of 5G n77 (3.3-4.2 GHz). Doping with 28% Sc boosts kt2 to ~16%, enabling 6.5-8.5% FBW in a single compact acoustic filter without complex multiplexing.
Both devices trap longitudinal acoustic waves inside an active piezoelectric thin film. Solidly Mounted Resonators (SMR) achieve acoustic reflection on the substrate side by fabricating an acoustic Bragg reflector (alternating quarter-wavelength layers of high and low acoustic impedance materials such as Tungsten and SiO2) on top of a standard silicon substrate. Film Bulk Acoustic Resonators (FBAR) instead create an air-cavity beneath the active membrane using backside deep reactive ion etching (DRIE) or sacrificial layer release. SMR offers superior mechanical ruggedness and thermal dissipation, whereas FBAR delivers slightly higher unloaded quality factors (Q) and lower parasitic capacitance at extreme high frequencies.
Key industrial challenges include high-purity scandium metal target metallurgy (suppressing oxide/carbide micro-inclusions that cause pinholes and electrical breakdown), film stress management across large wafers (avoiding film peeling or wafer bow during subsequent photolithography), controlling abnormal grain orientation (AGO) that degrades c-axis texture (FWHM rocking curves must stay below 1.5°), and preventing cubic rock-salt phase precipitation at scandium concentrations approaching 40 atomic percent.
A 200mm (8-inch) AlScN filter fab requires $100M to $250M in CapEx for dual-cathode reactive pulsed-DC PVD cluster tools, stepper lithography, DRIE cavity etchers, and wafer-level packaging. High-purity 4N5 scandium metal sputter targets represent a significant variable cost ($3,500 to $4,500/kg). However, because each 200mm wafer yields over 20,000 good filter dies and sells for $2,000 to $3,500 per wafer, unit filter chip production costs remain below $0.15/chip, delivering EBITDA margins above 50% and rapid capital payback.