Frozen Outside, Mobile Within: Boundary-Window Synthesis Through Phase Asymmetry and Timescale Separation

DOI: To be assigned

John Swygert

July 16, 2026

Abstract

A recent graphene oxide membrane experiment demonstrates a broader engineering principle whose significance extends beyond membrane science. By exploiting the lower freezing temperature of nanoconfined water, researchers created a temporary condition in which bulk water surrounding a graphene oxide membrane was frozen while water confined between graphene oxide sheets remained liquid. Within this selectively mobile region, dopamine molecules could rapidly assemble into nanopillars that adjusted the interlayer spacing of the membrane. Further cooling then froze the confined water, halting molecular assembly, while slower polymerization and covalent bonding continued and permanently stabilized the selected structure.

This paper identifies the more general strategy as Boundary-Window Synthesis: the deliberate creation of a temporary spatial and temporal window in which mobility is permitted only where structural reorganization remains desirable, followed by termination of that mobility before irreversible reactions can displace the system from its target configuration. The strategy combines phase asymmetry, confinement, controlled latency, timescale separation and selective structural commitment. It offers a general route for programming matter without attempting to force the final structure in a single step.

The graphene oxide result does not establish a universal physical theory, but it provides a powerful experimental example of a recurring engineering grammar: alter a boundary, open a restricted route, allow a fast reversible process to approach a chosen state, close the route and then permit slower irreversible processes to secure the result. This paper develops that grammar, defines its operational conditions, identifies its failure modes and proposes wider applications in membrane fabrication, energy storage, catalysis, crystal engineering, biomaterials and nanoscale manufacturing.

1. Introduction

Materials engineering frequently confronts a conflict between mobility and stability.

Mobility is necessary while molecules, particles, layers or domains are being arranged. Stability is necessary once the desired arrangement has been achieved. The same freedom that permits a system to form a useful structure can later permit that structure to swell, drift, collapse, aggregate or reorganize into an undesirable state.

Traditional approaches often attempt to solve this conflict by applying a stabilizing chemical reaction while the material is still assembling. This can produce premature crosslinking, blocked transport pathways, nonuniform structures or a final geometry determined more by reaction kinetics than by design. Alternatively, the structure may first be assembled and later stabilized, but it can change during the interval between those two operations.

A recently reported graphene oxide membrane fabrication method offers a particularly elegant solution to this problem. The method does not merely select a chemical crosslinker or apply a lower temperature. It creates a controlled difference between the physical state of water outside a nanoscale boundary and the physical state of water inside it.

Bulk water surrounding the graphene oxide membrane is converted into ice. Water confined between the graphene oxide sheets remains liquid because nanoconfinement lowers its freezing temperature. The surrounding environment is therefore immobilized while the interior remains sufficiently mobile for dopamine molecules to assemble into spacing-controlling nanopillars.

When the desired interlayer spacing has been reached, the temperature is lowered again. The confined water freezes, molecular diffusion is arrested and further assembly is halted. Slower chemical reactions then continue within the immobilized environment, converting the dopamine assemblies into polydopamine structures and bonding them to oxygen-containing sites on the graphene oxide sheets.

The procedure can be summarized in four statements:

The exterior is frozen.

The interior remains mobile.

The structure is allowed to form.

The remaining mobility is then removed before the structure is chemically secured.

This is more than a specialized membrane-manufacturing technique. It is an example of a general method for using boundaries to separate processes that would otherwise interfere with one another.

The Nature study was received on April 10, 2025, accepted on June 4, 2026, and published online on July 15, 2026. Its experimental development predates the author’s relevant TSTOEAO publications. However, the broader boundary, confinement, route-control and timescale-separation framework was developed independently before the author encountered the Nature article. The relationship is therefore treated as independent convergence rather than direct influence in either direction.

2. The Experimental Demonstration

Stacked graphene oxide membranes contain narrow channels between adjacent graphene oxide sheets. These channels can allow rapid water transport while rejecting or selectively transporting ions. Their usefulness, however, depends on maintaining extremely precise interlayer spacing.

Untreated graphene oxide membranes tend to swell in water. Chemical crosslinking can improve stability, but conventional crosslinkers often establish only one fixed spacing or obstruct portions of the transport channels. The challenge is therefore not simply to stabilize the membrane. It is to stabilize the membrane at a deliberately selected spacing while retaining sufficient open channel area for rapid transport.

Lu and colleagues addressed this challenge by separating dopamine assembly from dopamine reaction.

Dopamine molecules were introduced into the graphene oxide membrane. The surrounding water was frozen while nanoconfined water between the graphene oxide sheets remained liquid. In this condition, the dopamine molecules retained sufficient mobility to aggregate near oxygen-containing regions of the graphene oxide surface.

The size of these molecular assemblies increased with assembly time and dopamine concentration. Because their growth affected the separation between adjacent graphene oxide sheets, interlayer spacing could be tuned by controlling macroscopic variables such as concentration, temperature and exposure time.

Once the selected spacing had been achieved, additional cooling froze the confined water and arrested further assembly. Chemical polymerization and covalent bonding, which occurred on a slower timescale, were then allowed to continue. These slower reactions converted the temporary molecular assemblies into stable polydopamine nanopillars bonded to the graphene oxide structure.

The resulting membranes demonstrated dry-state interlayer spacing tunable from approximately 5.9 to 10.2 angstroms, with reported batch variation below 0.1 angstrom under selected fabrication conditions. The stabilized membranes resisted swelling, remained intact during water flushing and maintained approximately 99 percent sodium chloride rejection after 30 days of continuous operation.

The membranes were also used for highly selective ion separation. The reported rubidium-to-potassium separation factor reached 5,320 even though the hydrated sizes of the two ions differed by less than 0.1 angstrom.

These performance results are important, but the fabrication logic behind them may be even more broadly significant.

3. Boundary-Window Synthesis

The method can be generalized as Boundary-Window Synthesis, or BWS.

Boundary-Window Synthesis is a process in which a physical boundary creates a temporary region whose mobility, phase, transport properties or reaction accessibility differ from those of the surrounding environment. That temporary region is used to permit controlled structural reorganization. The window is then closed, and a slower process secures the resulting state.

A complete Boundary-Window Synthesis process contains five elements:

1. A boundary-sensitive difference in physical state or mobility.
   
   

2. A fast process capable of reorganizing the material.
   
   

3. A slower process capable of permanently stabilizing it.
   
   

4. A controllable mechanism for stopping the fast process.
   
   

5. A target state that can be reached before irreversible commitment dominates.
   
   

In the graphene oxide system, the boundary is formed by the nanoscale separation between adjacent graphene oxide sheets. The fast process is dopamine diffusion and assembly. The slow process is polymerization and covalent bonding. Cooling the confined water below its freezing point stops the fast process. The selected interlayer spacing is the target state.

The critical feature is not simply that different events happen at different rates. Many systems contain fast and slow processes. The critical feature is that the experimenter deliberately controls the interval during which the fast process remains available.

That interval is the boundary window.

4. Phase Asymmetry

The graphene oxide procedure relies on a state in which the same substance—water—occupies different phases across closely adjacent regions.

Let the freezing temperature of bulk water be represented by:

Tfb

Let the freezing temperature of water under nanoconfinement be represented by:

Tfc

When confinement depresses the freezing temperature:

Tfc < Tfb

An assembly temperature can then be selected so that:

Tfc < TA < Tfb

At this temperature, bulk water is frozen while confined water remains liquid.

The surrounding environment is immobilized, but the interior pathway remains active.

After the desired molecular assembly has formed, the system is moved to a lower temperature:

TR < Tfc

The confined water then freezes as well, terminating diffusion-dependent assembly.

The significant engineering resource is therefore not ice alone, water alone or low temperature alone. It is the controllable interval between two boundary-dependent transition temperatures.

A phase transition that would normally be treated as one bulk event becomes two separately accessible events:

First, the exterior freezes.

Later, the interior freezes.

The separation between these events produces an operational window.

This principle need not be limited to freezing. Comparable windows could potentially be produced through boundary-dependent changes in:

Glass-transition temperature

Solubility

Viscosity

Electrical conductivity

Magnetic ordering

Crystallization

Vapour pressure

Ion mobility

Reaction accessibility

Mechanical rigidity

The general requirement is that the boundary shift a relevant transition or transport threshold enough to allow one region to remain active while another becomes inactive.

5. Timescale Separation

Phase asymmetry alone does not explain the result. The fabrication process also depends on a useful difference between molecular assembly time and chemical reaction time.

Let:

τA represent the characteristic time of structural assembly.

τR represent the characteristic time of the stabilizing reaction.

The useful condition is:

τA << τR

In plain language, the desired structure must form substantially faster than the chemistry that permanently commits it.

A second timescale concerns stopping the assembly process. Let:

τS represent the time required to arrest mobility.

For precise control:

τS << τR

Ideally, τS should also be short compared with the time required for the assembly to change significantly after reaching its target.

The boundary window must remain open long enough for the desired assembly to form but not so long that slow reactions or continued aggregation dominate:

τA < ΔtW < τR

Here, ΔtW is the duration of the useful boundary window.

This relationship reveals why simply allowing assembly and polymerization to occur simultaneously is less controllable. When irreversible bonding becomes significant before the temporary structure has reached its target, the material can be trapped in a distribution of incomplete or obstructive states.

Boundary-Window Synthesis instead establishes an ordered sequence:

Mobility first.

Arrangement second.

Arrest third.

Commitment fourth.

This order is fundamental.

6. Selective Mobility Before Irreversible Commitment

A general design law can be extracted from the experiment:

Permit mobility only where structure must still form, and remove that mobility before irreversible commitment can move the system away from its target.

This may be described as the law of Selective Mobility Before Irreversible Commitment.

Many manufacturing problems arise because mobility and commitment are permitted in the same region at the same time. Molecules move while they are crosslinking. Particles aggregate while a matrix is hardening. Crystals continue growing while defects are being locked into place. Solvents escape while a structure is still attempting to organize.

The graphene oxide process spatially and temporally separates those functions.

The confined region performs the assembly.

The frozen surrounding region suppresses unwanted bulk movement and swelling.

The second freezing step halts further assembly.

The slower chemistry performs the commitment.

The process does not eliminate motion indiscriminately. It assigns mobility to a particular location and a particular interval.

It also does not accelerate every process. It takes advantage of the fact that one process is already fast and another is already slow. Control is achieved by preserving that difference rather than attempting to make the system uniformly rapid.

7. TSTOEAO Interpretation

Within the substrate grammar of TSTOEAO, the membrane experiment can be described through the sequence:

Gradient → boundary condition → available route → correction → cost-location → equilibrium target.

7.1 Gradient

The membrane must satisfy two competing requirements.

It must contain channels narrow enough to exclude unwanted ions.

It must retain sufficient open transport area to permit rapid water movement.

A membrane that is too open loses selectivity. A membrane that is too compressed, crosslinked or obstructed loses flux.

7.2 Boundary condition

Graphene oxide sheets establish a nanoscale confinement boundary. That boundary alters the freezing behaviour and molecular environment of the enclosed water.

The confined water is not an entirely separate substance. Its behaviour changes because the available physical routes and molecular arrangements have changed.

7.3 Available route

At the assembly temperature, the bulk route is immobilized by freezing, while the confined route remains available.

Dopamine molecules can continue moving and aggregating inside the interlayer space, but uncontrolled bulk swelling is suppressed.

7.4 Correction

Dopamine assembly forms nanopillars that alter the separation between adjacent graphene oxide sheets. The structure moves toward a selected interlayer spacing.

7.5 Cost-location

The dopamine assemblies preferentially develop near oxygen-containing regions of the graphene oxide sheets. Their placement allows bonding and stabilization while leaving substantial carbon-rich regions available for water and ion transport.

The structural cost of stabilization is therefore not distributed uniformly. It is preferentially localized where support and bonding are most useful.

7.6 Equilibrium target

The final target is not maximum rigidity, minimum spacing or maximum permeability considered separately.

The target is a stable, non-swelling membrane whose channel geometry produces a useful balance among:

Mechanical integrity

Water permeance

Salt rejection

Target-ion selectivity

Manufacturing reproducibility

The experiment therefore illustrates equilibrium as an engineered relationship among competing requirements rather than a single maximized variable.

8. Controlled Latency as a Manufacturing Variable

Latency is generally treated as delay, inefficiency or unwanted waiting. In Boundary-Window Synthesis, delay becomes a control resource.

The slower polymerization reaction gives the system time to establish a physical arrangement before that arrangement is permanently committed. The slow reaction is useful precisely because it does not immediately erase the experimenter’s ability to tune the structure.

The process creates a controlled interval between:

The appearance of the desired geometry

And the irreversible fixation of that geometry

This interval is a form of productive latency.

Too little latency would cause premature fixation.

Too much latency could permit drift, collapse or uncontrolled aggregation.

The correct latency allows the target structure to become measurable and selectable before it becomes permanent.

This suggests a broader principle for manufacturing:

A process should not always minimize delay. It should place delay between events that must remain separately controllable.

Latency has value when it preserves optionality.

9. An Operational Boundary-Window Protocol

A generalized Boundary-Window Synthesis programme could proceed through the following stages.

9.1 Define the target variable

The desired variable may be channel spacing, porosity, particle separation, crystal orientation, layer thickness, molecular loading or domain geometry.

The variable must be measurable during development, even if real-time measurement is not ultimately used during production.

9.2 Identify the reversible organizing process

Determine which process can move the system toward the target without immediately making the change permanent.

Examples include diffusion, self-assembly, adsorption, molecular alignment, solvent redistribution, particle migration or reversible crystallization.

9.3 Identify the irreversible stabilizing process

Determine which slower process can secure the selected arrangement.

Examples include polymerization, crosslinking, covalent bonding, sintering, vitrification, curing, mineralization or phase conversion.

9.4 Measure the competing timescales

The organizing process and stabilizing process must be characterized independently whenever possible.

If their timescales substantially overlap, control may require altered temperature, catalyst concentration, pH, pressure, illumination or reactant accessibility.

9.5 Create the boundary window

Establish a condition in which mobility remains available only in the region where organization is required.

This can be achieved through confinement-dependent phase behaviour or another boundary-sensitive property.

9.6 Stop organization at the target

Apply a rapid, controllable change that arrests diffusion or structural reorganization.

The stopping operation must be faster than the unwanted continuation of the organizing process.

9.7 Permit stabilization

Allow the slower commitment process to proceed after the target arrangement has been arrested.

9.8 Verify preservation

The final structure must be tested under the environmental conditions relevant to its intended use. A structure that is precise only when dry, unloaded or chemically isolated has not necessarily retained the programmed state.

10. Potential Applications

The following applications are proposed as research directions rather than established consequences of the graphene oxide result.

10.1 Additional layered membranes

Other layered materials, including MXenes, clays, vermiculites and layered polymers, may contain confined solvents whose transition behaviour differs from bulk solvent behaviour.

Boundary windows could permit controlled pillar formation, ion-channel construction or selective interlayer functionalization.

10.2 Battery and electrochemical interfaces

Battery performance frequently depends on the structure of electrode–electrolyte interfaces. A temporary mobility window might allow ions, polymers or precursor molecules to organize at an interface before a slower electrochemical or chemical reaction forms a stable protective layer.

The objective would not necessarily be to freeze water. It would be to separate interface formation from interface commitment.

10.3 Catalysis

Catalytic selectivity depends on the location, spacing and accessibility of active sites. Boundary-controlled mobility could allow catalyst precursors to occupy preferred nanoscale regions before immobilization.

This may reduce active-site clustering or prevent stabilizing materials from obstructing transport pathways.

10.4 Crystal and polymorph control

Crystal fabrication often suffers from competition among nucleation, growth and phase conversion. A boundary window could permit one stage to proceed while temporarily suppressing another.

A selected polymorph or orientation might then be arrested before slower conversion processes dominate.

10.5 Biomaterials

Biological structures frequently form through a sequence of self-assembly followed by crosslinking or mineralization. Boundary-Window Synthesis may provide methods for creating scaffolds in which cells, proteins or polymers organize under permissive conditions before a slower process strengthens the structure.

Temperature, pH, ionic concentration or enzymatic accessibility could provide the boundary-sensitive control variable.

10.6 Additive and nanoscale manufacturing

Many printing and coating processes require deposited material to remain mobile long enough to level or self-organize but not so long that it spreads beyond its intended geometry.

A spatially selective mobility window could improve resolution by allowing local reorganization while immobilizing neighbouring material.

11. Testable Predictions

The Boundary-Window Synthesis framework produces several predictions.

First, control precision should depend on the separation between the relevant transition thresholds. A wider difference between the confined and bulk transition conditions should create a more easily controlled processing window.

Second, reproducibility should decline when the timescales of assembly and irreversible reaction become comparable.

Third, premature acceleration of the stabilizing reaction should increase structural heterogeneity or channel obstruction, even when total stabilizer concentration remains unchanged.

Fourth, there should be an optimal boundary-window duration. A window that is too short will produce incomplete organization. A window that is too long will permit excessive aggregation, drift or early chemical commitment.

Fifth, the best performance should not necessarily occur at maximum pillar density, maximum crosslinking or maximum rigidity. Performance should peak where structural support is sufficient without excessive occupation of transport pathways.

Sixth, changes in confinement dimensions should alter both the phase window and the molecular assembly behaviour. The same fabrication temperature may therefore produce different results when channel width, surface chemistry or oxidation level changes.

Seventh, similar control should be achievable with other molecular systems when the organizing process and stabilizing reaction remain sufficiently separable. The original researchers’ demonstration of alternative reactive pathways provides preliminary support for this expectation, but each chemical system will require independent validation.

These predictions are experimentally falsifiable. Failure to observe relationships between phase-window width, timescale separation and structural controllability would limit the generality of the proposed framework.

12. Limitations and Failure Modes

Boundary-Window Synthesis is not automatically beneficial whenever confinement and multiple timescales are present.

A confined region may be too heterogeneous to produce a uniform transition threshold. Defects, variable layer spacing and uneven surface chemistry can create multiple local windows rather than one controllable window.

Large-scale production may also introduce thermal gradients. A membrane that can be uniformly frozen and annealed at laboratory scale may experience different assembly durations across an industrial roll, tube or module.

The arrest process may damage the material. Freeze–thaw cycles can produce mechanical stress, delamination or pore formation in systems less tolerant than graphene oxide.

The slow stabilizing reaction may not remain independent of the arrested structure. Polymerization can release heat, alter local volume or displace molecules even when bulk diffusion has been reduced.

The process may also require substantial energy or specialized temperature control. Superior membrane performance does not by itself establish economic superiority over conventional desalination or ion-separation technologies.

The reported seawater experiments also used pretreated feed solutions rather than untreated raw seawater. Long-term fouling, biological contamination, chemical variation and full industrial-scale operation remain separate questions.

Boundary-Window Synthesis should therefore be understood as a design framework, not a guarantee of commercial success.

13. Discussion

The most important lesson of the graphene oxide experiment is not that freezing can stabilize a membrane.

The deeper lesson is that a boundary can divide what appears to be one physical event into multiple controllable events.

Bulk water freezing and confined water freezing become separate transitions.

Molecular assembly and chemical stabilization become separate operations.

Structural mobility and structural commitment become separate permissions.

The process succeeds because these events are no longer forced to occur simultaneously.

This reflects a more general approach to engineering. Instead of attempting to impose a final equilibrium directly, the experimenter constructs a route through temporary states. Each state removes some possibilities while preserving the possibility needed for the next stage.

The surrounding ice prevents uncontrolled swelling.

The confined liquid preserves assembly.

The second freezing step removes further assembly.

The slow chemistry preserves the selected spacing.

The final membrane is therefore not produced by one force, one reaction or one boundary. It is produced by a timed portfolio of boundaries and transitions.

The technique also demonstrates that confinement is not merely a reduction in available space. Confinement can generate new processing opportunities by shifting transition temperatures, altering solvation, changing interaction strengths and restricting the routes through which matter can reorganize.

The useful engineering question is therefore not only:

What does confinement prevent?

It is also:

What temporary state does confinement make possible?

Conclusion

The fabrication of polydopamine-pillared graphene oxide membranes demonstrates a powerful method for controlling material structure through phase asymmetry and timescale separation.

Bulk water is frozen while nanoconfined water remains mobile. Rapid molecular assembly adjusts the interlayer spacing. Further cooling arrests that assembly. Slower polymerization and covalent bonding then secure the selected geometry.

This paper identifies the general strategy as Boundary-Window Synthesis.

Its central law is straightforward:

Permit mobility only where structure must still form, and remove that mobility before irreversible commitment can move the system away from its target.

Boundary-Window Synthesis transforms confinement, phase transitions and latency into coordinated engineering variables. It replaces simultaneous uncontrolled formation and stabilization with an ordered sequence of mobility, organization, arrest and commitment.

The graphene oxide result does not prove TSTOEAO as a universal physical account. It does, however, provide a precise experimental example of its recurring engineering grammar: boundaries alter available routes, route availability determines where correction can occur, the cost of stabilization is selectively located and a useful equilibrium is reached through controlled sequencing rather than direct force alone.

The exterior freezes.

The interior continues working.

The target structure forms.

Then the remaining route is closed.

Matter is not merely constrained. It is given a temporary freedom, in the correct location, for the correct duration, and that freedom is withdrawn at the moment its work is complete.

References

1. Lu, Y., Wei, L., Xie, Z.-a., Wu, X., Xue, H., Su, H., Liu, J., Jing, M., Chen, J., Shi, G., Su, Z., Bian, F., Xu, Z.-K., Zhu, C., Fang, W.-H., Zeng, X. C. & Wang, J. “Graphene Oxide-Polydopamine Membranes with Controlled Interlayer Spacing.” Nature (2026). DOI: 10.1038/s41586-026-10765-4.
   
   

2. Joshi, R. K. et al. “Precise and Ultrafast Molecular Sieving Through Graphene Oxide Membranes.” Science 343, 752–754 (2014).
   
   

3. Abraham, J. et al. “Tunable Sieving of Ions Using Graphene Oxide Membranes.” Nature Nanotechnology 12, 546–550 (2017).
   
   

4. Chen, L. et al. “Ion Sieving in Graphene Oxide Membranes via Cationic Control of Interlayer Spacing.” Nature 550, 380–383 (2017).
   
   

5. Epsztein, R., DuChanois, R. M., Ritt, C. L., Noy, A. & Elimelech, M. “Towards Single-Species Selectivity of Membranes with Subnanometre Pores.” Nature Nanotechnology 15, 426–436 (2020).
   
   

6. Zhang, W.-H. et al. “Graphene Oxide Membranes with Stable Porous Structure for Ultrafast Water Transport.” Nature Nanotechnology 16, 337–343 (2021).
   
   

7. Andreeva, D. V. et al. “Two-Dimensional Adaptive Membranes with Programmable Water and Ionic Channels.” Nature Nanotechnology 16, 174–180 (2021).
   
   

8. Esfandiar, A. et al. “Size Effect in Ion Transport Through Angstrom-Scale Slits.” Science 358, 511–513 (2017).
   
   

9. Trushin, M., Andreeva, D. V., Peeters, F. M. & Novoselov, K. S. “Structure and Flow of Low-Dimensional Water.” Nature Reviews Physics 7, 502–513 (2025).
   
   

10. Swygert, J. Boundary Portfolio Engineering. TSTOEAO (2026).
    
    

11. Swygert, J. Operationalizing Boundary Portfolio Engineering. TSTOEAO (2026).
    
    

12. Swygert, J. Productive Confinement. TSTOEAO (2026).