Spin-Assisted LbL-LPE for MOF Thin Films

• Spin-assisted LbL-LPE is a cyclic thin-film deposition method that uses spin-coating dynamics to assemble MOF layers with precise, cycle-defined thickness.
• The process employs an automated direct-dispense system with controlled reagent dosing, significantly reducing chemical consumption and processing time.
• Optimized parameters such as spin rate, cycle count, and ligand stoichiometry are critical for achieving uniformity, high out-of-plane orientation, and reproducible film quality.

Searching arXiv for the cited papers to ground the article in published work. Spin-assisted layer-by-layer liquid-phase epitaxy (LbL-LPE) denotes a cyclic thin-film growth strategy in which precursor solutions are dispensed directly onto a rotating substrate, so that liquid-phase interfacial assembly, rinsing, and thickness accumulation are coupled to spin-coating hydrodynamics. In the literature represented here, the most explicit realization is an automated spin-assisted LbL-LPE workflow for highly oriented mixed-linker MOF thin films, where spin coating reduces chemical consumption, shortens processing time, permits ambient operation, and makes thickness discretely tunable through the number of deposition cycles (Afanasenko et al., 25 Mar 2026). Closely related but distinct bodies of work define the method’s boundaries: conventional dipping-LPE establishes ultrathin oxide-film performance benchmarks without true spin-assisted or cycle-resolved growth (Dubs et al., 2016, Dubs et al., 2019), while a vapor-phase GaN study shows that spin-deposited 2D nanosheet networks can regulate epitaxial nucleation through local transport-percolation effects even though the growth itself is not liquid-phase epitaxy (Beak et al., 7 May 2025).

1. Process architecture and operational definition

In its most explicit form, spin-assisted LbL-LPE is implemented as an automated direct-dispense process on a continuously rotating substrate. The demonstrated platform comprised a Laurell spin coater (WS-650MZ-23NPPB), three Harvard PHD 4400 syringe pumps, and a motorized dispensing arm carrying three PTFE tubes connected to solvent, metal solution, and linker solution reservoirs. An in-house LabVIEW program controlled tube positioning, pumping, dispense volume, and sequence timing (Afanasenko et al., 25 Mar 2026).

One deposition cycle was defined as

$\mathrm{S-M-S-L-S}$

where $S$ is ethanol solvent/rinse, $M$ is zinc acetate solution, and $L$ is the mixed-linker BDC/DABCO solution. Each dispense used $33~\mu\mathrm{L}$, the pumping rate was reported as $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$, the dispensing tube was positioned $10~\mathrm{mm}$ above the substrate, and the substrate rotated continuously at $550~\mathrm{rpm}$ throughout the cycle. The active deposition time per cycle was stated to be only seconds. After the programmed cycles, the sample was removed from the coater and dried gently under nitrogen flow; drying while still spinning preserved the MOF chemical fingerprint in IR but destroyed crystallinity in GIWAXS (Afanasenko et al., 25 Mar 2026).

The operating window was narrow. At $800{-}1000~\mathrm{rpm}$, the liquid dried too quickly during spinning and no film growth occurred. At $450{-}500~\mathrm{rpm}$, rotation was unstable and film homogeneity deteriorated. This identifies spin assistance not as a generic substrate-rotation label, but as a hydrodynamically constrained interfacial-reaction regime in which the liquid layer must persist long enough for coordination assembly while remaining thin enough for efficient rinsing and controlled reagent exchange (Afanasenko et al., 25 Mar 2026).

Thickness control in this implementation was discrete rather than continuous: 60 and 120 deposition cycles were examined. Representative successful films had a total thickness of $S$0, including about $S$1 Au, about $S$2 SAM, and about $S$3 MOF, while cross-sectional SEM gave MOF crystallite thicknesses of about $S$4. This establishes cycle count as the primary thickness-control variable in the reported spin-assisted LbL-LPE protocol (Afanasenko et al., 25 Mar 2026).

2. Interfacial chemistry, substrate preparation, and growth window

The proof-of-concept framework was the flexible pillar-layered MOF $S$5, with zinc acetate dihydrate as the metal source, benzene-1,4-dicarboxylate as the sheet-forming linker, and DABCO as the pillar linker. Substrates were gold-coated silicon wafers consisting of polycrystalline Au films of about $S$6 thickness on Ti-precoated Si, with roughness below $S$7. Before functionalization, wafers were sonicated in absolute ethanol for 30 min, then modified with either 16-mercaptohexadecanoic acid (MHDA) or the pyridine-terminated thiol PP1 by 24 h immersion in $S$8 ethanolic solution; the MHDA solvent contained 10 vol% acetic acid (Afanasenko et al., 25 Mar 2026).

Surface state was treated as a critical control variable. Fresh ordered MHDA surfaces showed $S$9, whereas disordered or degraded MHDA surfaces showed $M$0. For PP1, $M$1 indicated well-ordered films and $M$2 indicated loss of order. Only freshly functionalized substrates less than 48 h old were used for deposition (Afanasenko et al., 25 Mar 2026).

Precursor solutions were prepared in ethanol. BDC was first dissolved at $M$3 and its concentration verified by UV-Vis at $M$4 before DABCO addition; this verification step was necessary because the absorbance of the two linkers overlapped in the mixed solution. The optimized working stoichiometry was

$M$5

equivalently $M$6. Working solutions were used for no more than 3 d to avoid ligand crystallization or $M$7 formation (Afanasenko et al., 25 Mar 2026).

Stoichiometric control was decisive. At BDC:DABCO $M$8, framework formation was strongly hindered; GIWAXS showed weak diffuse reflections on PP1 and none on MHDA, while IR showed Zn coordination to BDC but no DABCO coordination signature, specifically lacking the C–N–C bending vibration at $M$9. The best regime was BDC:DABCO $L$0 to $L$1, with $L$2 optimal. At higher DABCO content, $L$3 and $L$4, orientation degraded and IR revealed a band at $L$5 assigned to unreacted DABCO (Afanasenko et al., 25 Mar 2026).

The metal-to-ligand ratio imposed a second window. At the optimal ligand ratio of $L$6, reducing Zn from $L$7 to $L$8 preserved orientation but produced poor coverage, especially at wafer edges. Increasing Zn to $L$9 still yielded well-oriented, uniform films, whereas Zn concentrations above $33~\mu\mathrm{L}$0 destroyed preferential orientation. The practical process window was therefore centered around $33~\mu\mathrm{L}$1 Zn, $33~\mu\mathrm{L}$2 BDC, $33~\mu\mathrm{L}$3 DABCO, $33~\mu\mathrm{L}$4, and 60 cycles on fresh SAMs (Afanasenko et al., 25 Mar 2026).

3. Orientation control, texture metrics, and growth mechanism

The most distinctive outcome of the reported spin-assisted LbL-LPE protocol was strong out-of-plane orientation rather than full epitaxy. GIWAXS confirmed that the films adopt the tetragonal $33~\mu\mathrm{L}$5 structure of bulk $33~\mu\mathrm{L}$6. The dominant diffraction feature was the (001) reflection, indicating that the (001) lattice planes were parallel to the substrate surface and the crystallographic $33~\mu\mathrm{L}$7-axis was perpendicular to the surface. The films were explicitly described as textured rather than fully epitaxial, because no preferred in-plane azimuthal alignment was observed and crystallites were isotropically distributed in-plane (Afanasenko et al., 25 Mar 2026).

Orientation quality was quantified by the degree of orientation (DO), the Hermans orientation parameter (HOP), and the azimuthal full width at half maximum (FWHM). The paper defined

$33~\mu\mathrm{L}$8

$33~\mu\mathrm{L}$9

and

$0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$0

Here, DO estimates the fraction of preferentially oriented crystallites, while HOP is intensity-weighted and reports how tightly they are aligned. The use of all three metrics was important because DO alone could remain high even when the orientation distribution broadened (Afanasenko et al., 25 Mar 2026).

For MHDA, 60 cycles, and BDC:DABCO $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$1, the reported values were

$0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$2

In the main text, this condition was summarized as about $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$3 DO and HOP $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$4, and described as approaching single-crystal-like alignment. For MHDA, 60 cycles, $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$5, the values were $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$6, $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$7, and $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$8. For $0.2~\mu\mathrm{L}\,\mathrm{min}^{-1}$9, the values were $10~\mathrm{mm}$0, $10~\mathrm{mm}$1, and $10~\mathrm{mm}$2. For $10~\mathrm{mm}$3, they were $10~\mathrm{mm}$4, $10~\mathrm{mm}$5, and $10~\mathrm{mm}$6. These data show that high DO can coexist with degraded orientational sharpness, making HOP and FWHM necessary discriminants (Afanasenko et al., 25 Mar 2026).

Cycle number altered structure as well as thickness. For MHDA at $10~\mathrm{mm}$7, increasing to 120 cycles preserved high DO,

$10~\mathrm{mm}$8

but reduced orientational sharpness to

$10~\mathrm{mm}$9

The film therefore remained preferentially oriented while developing a much broader orientation distribution because of tilt and twinning. The mechanistic interpretation in the paper was that under insufficient DABCO coordination, defects in upper layers permitted rotation of some Zn$550~\mathrm{rpm}$0 paddle-wheel units around the $550~\mathrm{rpm}$1-bonds of the BDC linker, generating twinned crystallites; at excessive DABCO concentrations, competition for Zn$550~\mathrm{rpm}$2 binding suppressed orderly formation of the Zn-BDC 2D sheets and promoted more isotropic crystal orientations (Afanasenko et al., 25 Mar 2026).

Morphology and compositional mapping indicated island-type growth. Top-view SEM showed densely packed plate-like crystallites lying parallel to the substrate under optimal $550~\mathrm{rpm}$3 and $550~\mathrm{rpm}$4 conditions, while higher DABCO ratios yielded broader tilt-angle distributions and additional (100)-oriented domains. Cross-sectional FIB-SEM showed that both (001)- and (100)-oriented crystallites had similar thicknesses, around $550~\mathrm{rpm}$5, and EDS suggested that regions between crystallites contained little or no metal. The authors interpreted this as a Volmer-Weber growth mechanism rather than immediate formation of a dense continuous layer (Afanasenko et al., 25 Mar 2026).

4. Characterization logic and reproducibility criteria

The reported spin-assisted LbL-LPE workflow integrated correlative characterization as part of process control rather than solely as ex post validation. GIWAXS was used for phase identification and orientation analysis; measurements were performed at the Austrian SAXS beamline at Elettra with $550~\mathrm{rpm}$6 (8 keV), a sample-to-detector distance of 260 mm, an incident angle of $550~\mathrm{rpm}$7, a q-range of $550~\mathrm{rpm}$8, a Pilatus3 1M detector, and triplicate 30 s exposures. The reciprocal-space patterns distinguished sharp spot-like (001) texture from broader, azimuthally spread scattering in less aligned films (Afanasenko et al., 25 Mar 2026).

IR spectromicroscopy served simultaneously as a framework-chemistry probe and as a spatial uniformity metric. Grazing-angle spectra on reflective substrates showed asymmetric and symmetric carboxylate stretches at 1624 and $550~\mathrm{rpm}$9, a DABCO-associated C–N–C bending vibration at $800{-}1000~\mathrm{rpm}$0, and, under DABCO-rich conditions, the $800{-}1000~\mathrm{rpm}$1 band of unreacted DABCO. The peak assignments were supported by DFT-calculated spectra, with the calculated spectrum of model V-exp. scaled by 0.978 or about 0.98 to account for harmonic overestimation of vibrational frequencies (Afanasenko et al., 25 Mar 2026).

ToF-SIMS provided depth-resolved validation of layering and composition. Depth profiles and 3D reconstructions tracked $800{-}1000~\mathrm{rpm}$2, $800{-}1000~\mathrm{rpm}$3, $800{-}1000~\mathrm{rpm}$4, $800{-}1000~\mathrm{rpm}$5, and $800{-}1000~\mathrm{rpm}$6 signals. Sulfur localized at the Au/MOF interface as expected for the SAM; Zn-related fragments co-localized with aromatic fragments throughout the MOF layer; comparison of $800{-}1000~\mathrm{rpm}$7 and $800{-}1000~\mathrm{rpm}$8 films showed increased $800{-}1000~\mathrm{rpm}$9 in the DABCO-rich film; and two-dimensional $450{-}500~\mathrm{rpm}$0 imaging showed homogeneous Zn-containing signal distribution over the surface (Afanasenko et al., 25 Mar 2026).

Reproducibility was quantified rather than asserted. Triplicate GIWAXS analysis was performed for most conditions, and eight samples were analyzed for the especially important 60-cycle $450{-}500~\mathrm{rpm}$1 case. Grazing-angle IR spectromicroscopy rastered 400 spectra across the entire substrate for 60-cycle $450{-}500~\mathrm{rpm}$2 films, demonstrating chemically uniform coverage including wafer edges. GIWAXS scans across the wafer showed higher and more uniform integrated out-of-plane intensity for 60-cycle films than for 120-cycle films. The workflow’s reproducibility claim therefore rested on software-controlled dispense sequence, volume, tube position, deposition rate, and spin rate, together with integrated quality checks on SAM order and precursor concentration (Afanasenko et al., 25 Mar 2026).

A concise comparison of the studies most relevant to the topic is useful:

Study	Process type	Relevance to spin-assisted LbL-LPE
(Afanasenko et al., 25 Mar 2026)	Automated direct-dispense spin-assisted LbL-LPE	Explicit cycle-based implementation
(Dubs et al., 2016)	Standard dipping LPE of YIG	Ultrathin conventional LPE benchmark
(Dubs et al., 2019)	Isothermal dipping LPE of YIG with substrate rotation	Ultrathin conventional LPE benchmark; not LbL
(Beak et al., 7 May 2025)	Spin-coated GO mask plus HVPE GaN growth	Analog for spin-assisted nucleation mediation, not LPE

5. Relation to conventional liquid-phase epitaxy benchmarks

Conventional LPE defines the quality baseline that spin-assisted LbL-LPE must match or surpass. In one benchmark study, YIG films of 83–113 nm were grown by standard dipping LPE from a PbO–B$450{-}500~\mathrm{rpm}$3O$450{-}500~\mathrm{rpm}$4-based high-temperature solution in a Pt crucible at about $450{-}500~\mathrm{rpm}$5 on one-inch (111) GGG substrates. The films exhibited RMS roughnesses of 0.25–0.8 nm, a film/substrate interface width below 5 nm by XPS for sample B, nearly perfect crystallinity without significant mosaicity, $450{-}500~\mathrm{rpm}$6, in-plane coercive fields of 0.10–0.20 Oe, and a minimum $450{-}500~\mathrm{rpm}$7 of 1.4 Oe at 6.5 GHz with $450{-}500~\mathrm{rpm}$8 to $450{-}500~\mathrm{rpm}$9. The paper explicitly stated that growth below 100 nm remained a major challenge for the classical thick-film LPE technique (Dubs et al., 2016).

A subsequent study extended conventional LPE YIG down to approximately 10–11 nm. Growth used PbO–B$S$00O$S$01-based high-temperature solutions at about $S$02 on (111) GGG by isothermal dipping, with horizontal substrate rotation at 100 rpm during growth, post-withdrawal spin-off of melt remnants at 1000 rpm, and thickness control through low supercooling $S$03. Films of 10–110 nm were obtained within one minute deposition time. For thicknesses $S$04, the films showed RMS roughness 0.2–0.4 nm, coherent pseudomorphic growth without significant compositional strain or geometric mosaicity, a $S$05 interfacial transition width of $S$06, ideal Fe:Y stoichiometry of 5:3, $S$07, $S$08, and $S$09 at 15 GHz. At about 11 nm, magnetic behavior degraded through increased inhomogeneous broadening, a small two-magnon scattering contribution, reduced $S$10, and reduced $S$11 (Dubs et al., 2019).

These conventional YIG studies are often misread as evidence for spin-assisted LbL-LPE because they involve rotation. That interpretation is not supported by the papers. Neither study reported a cyclic layer-by-layer dosing scheme, monolayer counting, self-limiting adsorption, forced-convection control of a spin-thinned liquid layer during growth, or any explicit spin-assisted LbL mechanism. The 100 rpm motion in (Dubs et al., 2019) was ordinary substrate rotation during LPE growth, and the 1000 rpm step was a post-growth melt-removal operation. Their relevance is therefore comparative: they define what optimized non-LbL liquid-phase epitaxy can already achieve in ultrathin oxide films (Dubs et al., 2016, Dubs et al., 2019).

6. Spin-assisted nucleation mediation beyond strict LPE and current limitations

A neighboring line of work clarifies how spin-assisted solution processing can control epitaxial access even outside true liquid-phase epitaxy. In a GaN study on $S$12-plane sapphire patterned with a $S$13 SiO$S$14 layer and $S$15-diameter circular openings, GO suspensions at $S$16 and $S$17 were spin-coated, thermally annealed at $S$18 in nitrogen to form rGO, and then subjected to GaN HVPE at $S$19 for 1 min without a low-temperature GaN buffer layer. Addition of TBAOH at a final concentration of $S$20 to the $S$21 GO solution yielded a more uniform, moderately stacked rGO morphology than concentration increase alone. Raman G-peak maps showed that higher GO concentration increased average loading but retained a long low-intensity tail, whereas TBA narrowed the coverage distribution and removed the low-intensity tail. Morphological uniformization suppressed both ELOG-like and no-nucleation modes while expanding THE-like nucleation regions (Beak et al., 7 May 2025).

The conceptual contribution of that study is a three-regime nucleation model governed by local rGO thickness and continuity: ELOG-like nucleation on exposed or weakly masked substrate regions, THE-like nucleation through thin, porous, or perforated GO/rGO regions, and complete nucleation suppression in overly thick zones. The authors used “percolatively connected” to describe an rGO nanosieve that was macroscopically continuous yet microscopically permeable to precursor transport. This is not electronic percolation measured by conductivity; it is inferred transport percolation for growth precursors. The paper did not report spin speed, acceleration, spin time, dispense volume, or repeated spin cycles, and it explicitly did not describe repeated layer-by-layer assembly in the classical molecular sense (Beak et al., 7 May 2025).

This neighboring result is relevant because it shows that spin-deposited nanosheet stacking can mediate where epitaxy can and cannot occur. A plausible implication is that some design principles of spin-assisted LbL-LPE extend beyond the MOF case: additive-enabled colloidal control can be more important than average solids loading, local stack-thickness distributions can dominate nucleation outcomes, and the most useful interlayers may be continuous at the device scale yet selectively permeable at the nanoscale. At the same time, the translation to true liquid-phase epitaxy is limited by temperature and chemistry mismatch, the stochastic single-pass assembly character of the GO film, the absence of direct percolation imaging, and the lack of demonstrated stability under melt- or solvent-based LPE conditions (Beak et al., 7 May 2025).

Several general limitations therefore remain. The direct evidence for spin-assisted LbL-LPE is strongest for MOF thin films on SAM-functionalized Au, where the process is cycle-defined, automated, and quantitatively benchmarked (Afanasenko et al., 25 Mar 2026). It is not yet evidence for a universal inorganic LPE method. The current process window is narrow: SAMs must be fresh, precursor solutions must be used within about 3 d, drying during spinning destroys crystallinity, too low or too high spin rate prevents uniform growth, and increasing to 120 cycles broadens the orientation distribution and can introduce byproducts at wafer edges (Afanasenko et al., 25 Mar 2026). Conventional LPE already achieves ultrathin YIG with exceptionally low damping and sharp interfaces, but without true LbL control (Dubs et al., 2016, Dubs et al., 2019). The present state of the field therefore supports a precise formulation: spin-assisted LbL-LPE is a validated, cycle-controlled liquid-phase epitaxy strategy for orientation-engineered MOF films, a benchmark target for ultrathin oxide epitaxy, and a conceptually connected framework for spin-assisted control of epitaxial transport and nucleation through solution-processed interlayers (Afanasenko et al., 25 Mar 2026, Dubs et al., 2016, Dubs et al., 2019, Beak et al., 7 May 2025).