Seed-driven Stepwise Crystallization (SDSC) in r-GeO2
- Seed-driven stepwise crystallization (SDSC) is a segmented MOCVD strategy that overcomes the trade-off between crystal quality and surface coverage by using repeated seeding and controlled thermal cycling.
- The process enhances film epitaxy by promoting lateral coalescence on preexisting rutile seeds, leading to lower defect density and improved crystallinity.
- Quantitative results show coverage improvement from 57.4% to nearly 100% and a 30% reduction in rocking-curve FWHM, demonstrating its practical impact in achieving phase-pure r-GeO2 films.
Seed-driven stepwise crystallization (SDSC) is a segmented metal-organic chemical vapor deposition (MOCVD) strategy for rutile germanium dioxide, -GeO, in which growth proceeds through multiple sequential deposition steps on a pre-templated substrate enriched with rutile seeds. In the GeO literature, SDSC is presented as a response to a persistent trade-off: short single-run depositions can yield good crystal quality but incomplete coverage, whereas longer continuous runs improve coverage while promoting phase segregation into amorphous, -quartz, or cubic GeO. By repeatedly re-seeding the surface and biasing subsequent incorporation toward preexisting rutile facets, SDSC is used to overcome phase segregation and obtain continuous, high-quality -GeO films on -TiO(001) (Rahaman et al., 2024).
1. Materials setting and growth problem
Rutile GeO has been described as an emerging ultrawide bandgap semiconductor with significant potential for power electronics, owing to large-size substrate compatibility and ambipolar doping capability; related work also emphasizes a wide bandgap of 0–1 and potential relevance to optoelectronics (Rahaman et al., 2024, Rahaman et al., 30 Jul 2025). The principal obstacle addressed by SDSC is polymorphic competition during vapor-phase growth. On 2-TiO3, conventional single-step MOCVD tends to produce a compromise between crystalline quality and areal continuity: a 90 min run can give good crystal quality but incomplete coverage, while a 180 min run increases coverage at the expense of quality because phase segregation yields amorphous, 4-quartz, or cubic GeO5 regions (Rahaman et al., 2024).
A later mechanistic study places this problem in a broader phase-selection framework. It states that, although rutile GeO6 is the ground state below the rutile–quartz crossover, free-energy differences under MOCVD are small enough that interfacial and kinetic terms become decisive. In that setting, SDSC is not merely a timing modification but a means of repeatedly re-establishing conditions favorable to rutile nucleation while disfavoring competitive polymorphs (Rahaman et al., 30 Jul 2025).
2. Process architecture and reactor implementation
In its explicit seven-step form, SDSC begins with an initial 180-minute deposition to establish 7-GeO8 nucleation seeds, followed by a 90-minute deposition and then five 60-minute depositions. The reported baseline reactor conditions are an Agilis MOCVD reactor, 9, 80 Torr, TEGe flow of 160 sccm, O0 flow of 2000 sccm, Ar carrier/shroud gas, and substrate rotation of 300 rpm. Substrates are 1-TiO2(001) wafers cleaned by piranha 3 for 10 min, rinsed in acetone, isopropanol, and deionized water, dried under 4, and loaded immediately into the chamber (Rahaman et al., 2024).
Subsequent papers describe the same approach in a more explicitly segmented thermal-cycling form. One study defines SDSC schemes such as “180 min 5 1”, “90 min 6 2”, “60 min 7 3”, “30 min 8 6”, and “30 min 9 12”. In the “30 min 0 6” and “30 min 1 12” variants, each cycle consists of deposition at 2 for 3 min with TEGe and O4 on, followed by precursor shutoff, cooling to 5 at 6, immediate reheating to 7, and resumption of TEGe/O8 flow only after returning to 9. The total per cycle is 0 min (Rahaman et al., 30 Jul 2025).
The later thermal-conductivity report states that the granular recipe for the seed layer and each crystallization step is not fully restated there, and summarizes SDSC only in broad strokes: a seed-layer deposition step under low-coverage/high-nucleation-density conditions, a ramp to a final rutile-growth temperature of 1, and a hold at 2 until the film converts to 3 rutile. That paper again places the process in an Agnitron Agilis reactor at 80 Torr with Ar carrier/shroud gas and 300 rpm rotation (Rahaman et al., 31 Oct 2025).
3. Kinetic and thermodynamic rationale
The theoretical framing of SDSC is given in terms of classical nucleation, island growth, and coverage kinetics. The areal nucleation rate is written as
4
with
5
while a simple expression for lateral growth velocity is
6
The overall crystalline coverage is described by a Johnson–Mehl–Avrami–Kolmogorov law,
7
Within this picture, SDSC exploits the fact that once a rutile seed has formed, subsequent rutile nucleation on that seed face has a much lower effective barrier than fresh nucleation on a clean surface, and the kinetics shift toward lateral coalescence rather than the appearance of unwanted phases (Rahaman et al., 2024).
The same literature gives a cumulative-time form for the stepwise process,
8
and reports an effective 9–3, interpreted as consistent with combined continuous nucleation and lateral coalescence on a seeded template. This suggests that the practical function of segmentation is not simply to prolong deposition time, but to alter the mode by which surface coverage approaches 0 (Rahaman et al., 2024).
In the thermal-cycling formulation, phase selection is further interpreted through repeated perturbation of the competing polymorph landscape. During cooling to 1, metastable 2-quartz and amorphous regions are described as experiencing increased volatility and reduced thermal stability; upon reheating to 3, supersaturation spikes at pre-existing rutile seeds, reducing the critical nucleus size for rutile and enhancing 4 relative to 5 or 6. The net effect is described as suppression of new quartz/amorphous nucleation and preferential lateral expansion of rutile seeds (Rahaman et al., 30 Jul 2025).
4. Quantitative evolution of coverage and crystalline quality
For the seven-step implementation, SEM image analysis with ImageJ and HR-XRD rocking curves were used to track both rutile coverage and crystalline quality. The reported progression is as follows (Rahaman et al., 2024):
| Step (duration) | Rutile coverage 7 | FWHM 8 |
|---|---|---|
| 1 (180 min) | 57.4% | 9 |
| 2 (90 min) | 77.49% | 0 |
| 3 (60 min) | 79.73% | 1 |
| 4 (60 min) | 93.27% | 2 |
| 5 (60 min) | 99.17% | 3 |
| 6 (60 min) | 4 | 5 |
| 7 (60 min) | no change | not specified |
Over steps 1 to 6, the rocking-curve width decreases by 6, from 7 to 8, which the source interprets as a substantial reduction in mosaicity and defect density. The key point is that coverage and crystallinity improve concurrently, rather than exhibiting the usual single-run trade-off (Rahaman et al., 2024).
A separate segmentation study shows that the specific schedule matters. The reported rocking-curve FWHM values are 9 for “180 min 0 1”, 1 for “90 min 2 2”, 3 for “60 min 4 3”, 5 for “30 min 6 6”, and 7 arcsec for “30 min 8 12”. The same work characterizes the “30 min 9 12” film as phase-pure, nearly full coverage, and high crystalline quality (Rahaman et al., 30 Jul 2025). A plausible implication is that SDSC is better understood as a family of segmented growth schedules whose efficacy depends on how deposition time, interruption frequency, and thermal cycling are balanced.
5. Epitaxy, morphology, and functional consequences
The crystallographic registry reported for SDSC-grown films is
0
with
1
plus equivalent 2 variants from tetragonal symmetry. The lattice mismatch is given as
3
For the “30 min 4 12” sample, HRTEM and SAED are reported to confirm coherent epitaxy, with no amorphous interlayer and only a zig-zag atomic boundary (Rahaman et al., 30 Jul 2025).
The downstream consequence of this phase control is illustrated by the thermal-transport study on SDSC-grown films. There, the as-grown film is 5 thick, has AFM RMS roughness 6, shows dominant 7-GeO8(002) with minor quartz peaks, and has a rocking-curve FWHM 9. After chemical mechanical polishing, the thickness is reduced to 0, the roughness to 1, with local roughness 2, quartz-GeO3 peaks vanish, and the rocking-curve FWHM remains 4. Additional characterization reports four-fold symmetry in 5-scans, coherent epitaxy and partial relaxation in reciprocal-space maps, suppression of TiO6 cathodoluminescence bands at 2.80, 2.56, and 2.10 eV, and enhancement of GeO7 intrinsic 8 and 9 emissions at 2.64 and 2.38 eV, respectively (Rahaman et al., 31 Oct 2025).
Using TDTR with an 00 Al transducer, 01, and pump–probe spot 02, that study reports a cross-plane thermal conductivity of 03 at 300 K for high-quality 04 05-GeO06 films grown by MOCVD. In the paper’s own framing, phase control was achieved through SDSC (Rahaman et al., 31 Oct 2025).
6. Generalization, interpretation, and related uses of seed-driven stepwise crystallization
The GeO07 literature reduces SDSC to three operational pillars: an initial step to generate uniform nucleation seeds of the desired phase; subsequent steps timed to reinforce lateral growth and coalescence while avoiding overgrowth that promotes secondary phases; and preservation of a high-mobility, near-equilibrium surface between steps. The same source argues that any material system in which heterogeneous nucleation on a seed is much easier than homogeneous nucleation of competing phases, and in which adatom diffusion is sufficient at growth temperature, stands to benefit (Rahaman et al., 2024).
Possible extensions explicitly proposed include rutile Al08O09 or Ga10O11 films on lattice-matched substrates, III-nitrides such as AlN or AlGaN, diamond or SiC growth by HFCVD or MBE, and 2D transition-metal dichalcogenides such as MoS12 or WS13 in CVD (Rahaman et al., 2024). A later GeO14 study generalizes the same idea more narrowly as a pathway for selective vapor-phase growth of metastable or unstable phases, with tunable cycle duration, cooling/heating amplitude, and seed density (Rahaman et al., 30 Jul 2025).
A terminological boundary is also useful. Separate literature uses a seed-driven, stepwise crystallization logic in a computational context under the name HSEED, a heterogeneous seeded molecular dynamics framework for ice nucleation on crystalline surfaces. That method combines random structure search with seeded MD and is aimed at heterogeneous ice nucleation rather than MOCVD thin-film growth (Pedevilla et al., 2018). The shared language underscores a common emphasis on seeded heterogeneous nucleation, but the GeO15 SDSC literature refers specifically to segmented vapor-phase film growth on 16-TiO17.
Overall, SDSC in the 18-GeO19 context denotes a seed-mediated, segmented crystallization strategy that converts the usual coverage-quality trade-off into a cumulative process of rutile seed formation, lateral coalescence, and suppression of amorphous and quartz competitors. The reported outcomes—coverage rising from 57.4% to 20, rocking-curve narrowing from 21 to 22, phase-pure films with FWHM as low as 597 arcsec in segmented variants, and a cross-plane thermal conductivity of 23—have made SDSC the central process concept in current MOCVD studies of rutile GeO24 (Rahaman et al., 2024, Rahaman et al., 30 Jul 2025, Rahaman et al., 31 Oct 2025).