Composite Salt Sorbents (Salt-in-Matrix / SWS)

Impregnate a hygroscopic salt (CaCl₂, LiCl, LiBr) into the pores of a host matrix (silica gel, vermiculite, expanded graphite, templated silica, zeolite, MOF) and you get a Selective Water Sorbent (SWS) / composite “salt in porous matrix” (CSPM) — the Aristov/Gordeeva material class that raises water uptake 3–5× over bare silica gel while regenerating at low temperature. The gains are real but bounded: this article catalogs the uptake numbers, the tri-modal mechanism, and the failure modes (deliquescence, leakage, corrosion, agglomeration, degradation) that keep these materials mostly pre-commercial. It is the sorbent-side companion to Adsorbent Bed Engineering.

The design idea: tune salt + matrix to the cycle

Aristov’s founding “material-scientist approach” is that a sorbent should be designed to a target isotherm for a given cycle (evaporator, condenser, and drive temperatures), and that a salt-in-matrix composite gives two independent knobs — the salt (which sets the hydrate steps and the deliquescence point) and the matrix (which sets pore size, dispersion, and thermal path) — that a single-phase adsorbent lacks. The canonical result is SWS-1L, CaCl₂ confined in mesoporous silica gel: uptake up to ~0.7 g·g⁻¹, 3–5× bare silica’s ~0.1–0.2 g·g⁻¹ working swing, desorption as low as 90–100 °C, and a theoretical cooling COP ~0.7 at 10 °C evaporator / 35 °C condenser. The gap between theory and hardware is the recurring story of the field: the early prototype achieved COP 0.6 but only SCP 20 W·kg⁻¹ against an ideal 378 W·kg⁻¹ — the material was fine, the bed and kinetics were not.

What makes composites outperform their host is that water is taken up three ways at once, a hybrid the CaCl₂-PHTS study states explicitly and the LiCl/vermiculite work traces step by step:

Physical adsorption on the host pore surface (the silica or vermiculite’s own capacity),
Chemisorption — water reacts with the salt to form crystalline hydrates (e.g. LiCl·nH₂O, n = 1, 2), giving a stepped isotherm,
Absorption — at higher humidity the hydrate deliquesces into a confined aqueous salt solution inside the pores.

This is why uptake scales with salt loading — and also why every failure mode below originates in steps 2–3 (leakage, corrosion, and swelling all come from the salt phase, not the host). It also makes composites the textbook combined case of Adsorption vs Absorption.

Uptake & performance numbers

Composite	Water uptake	vs bare host	Cooling performance
SWS-1L (CaCl₂ in mesoporous silica)	~0.7 g·g⁻¹	3–5× bare silica (~0.1–0.2 working)	theoretical COP ~0.7; early prototype COP 0.6, SCP 20 W·kg⁻¹ (ideal 378)
CaCl₂-PHTS (20 wt %, templated silica)	2.44 g·g⁻¹ (40 °C)	bare 0.65 → 2.44	storage energy density 71 → 193 Wh·kg⁻¹ (~2.7×) at 40 ads / 120 des / 10 °C dew
LiCl + CaCl₂ (binary) on zeolite 13X / SAPO-34	—	5.3× / 4× bare matrix	—
CaCl₂ or LiBr on silica	up to 0.75 g·g⁻¹	—	adding LiBr to LiCl/silica raises capacity ~5.5 %
LiCl/silica-gel chiller	—	—	COP 0.41, SCP 250 W·kg⁻¹ at a ~680 s half-cycle
Consolidated CaCl₂/expanded graphite (NH₃)	—	volumetric capacity +52 % vs pure CaCl₂	SCP +353 % vs CaCl₂ powder; prototype 11.4 kW, COP 0.27, SCP 422 W·kg⁻¹
MIL-101(Cr)/CaCl₂ (1:8) (salt-in-MOF)	—	—	+68 % SCP, desorption temperature 110 → 90 °C
LiCl in COF gel (55 wt %)	3.7 g·g⁻¹, 1.2 g·g⁻¹·h⁻¹	—	water-harvesting context; six solar cycles/day

Two distinct jobs for the matrix emerge. Silica, templated silica, and vermiculite hosts add capacity and dispersion; expanded graphite adds thermal conductivity — consolidated CaCl₂/EG reaches 7.05–9.2 W·m⁻¹K⁻¹ against 0.3–0.4 for a powder bed (a 20–30× lift) while contributing essentially nothing to uptake, which is why the SCP jumps ~4.5× (+353 %). The best designs need both, and the graphite result — obtained with ammonia chemisorption, not water — shows the conductivity role is independent of working fluid.

The salt-loading tradeoff is sharp and the CaCl₂-PHTS dataset shows it cleanly: raising CaCl₂ from 0 → 4 → 10 → 20 wt % lifted uptake 0.65 → 0.78 → 1.20 → 2.44 g·g⁻¹ while surface area collapsed 810 → 461 → 322 → 163 m²·g⁻¹ and pore volume fell 0.705 → 0.189 cm³·g⁻¹ as salt filled the pores. Past a salt-specific ceiling, aggregates block the pore network entirely — ZnCl₂ in activated carbon plateaus at 32 wt % while CaCl₂ in activated carbon keeps improving to 70 wt %. Uptake and surface area move in opposite directions: composites are the clearest demonstration that BET area is not what sets capacity here.

Failure modes (and why they’re not commercial yet)

Failure	Mechanism / threshold	Mitigation
Deliquescence & leakage	Salt solution leaks from the pores when the operating relative pressure exceeds the equilibrium pressure of the impregnating solution. Saturated CaCl₂ at 20–30 °C sits at relative pressure 0.22 — adsorption must stay below it.	Choose the impregnating concentration 1.5–2× below the critical value; hydrophobic breathable coatings
Corrosion + non-condensables	Leaked salt solution corrodes metal heat-exchanger parts and emits non-condensable gas, which degrades the vacuum a water chiller depends on	Confinement; anchoring the cation to the matrix
Agglomeration	Bulk salt aggregates on hydration, blocking pores and slowing mass transfer — the key barrier for salt-only sorbents	Macroporous host (expanded vermiculite); expanded-graphite consolidation; binary-salt tuning
Slow kinetics / long cycles	Bulk salt sorbs slowly; the LiCl/silica chiller needed a ~680 s (11 min) half-cycle for COP 0.41 / SCP 250 W·kg⁻¹	Matrix dispersion — the vermiculite host gives a “giant acceleration” of sorption dynamics vs bulk salt
Hydrothermal degradation	Front-loaded: in Fraunhofer ISE cycling, the Cu-BTC MOF lost 40 % of capacity in the first 15 cycles. Silica composites show only “slight destruction” to 100 cycles, but long-cycle stability is unverified for commercial life	Early-cycle screening (most loss is initial); stable host selection (silica, AlPO/SAPO over Cu-MOFs)

The pattern mirrors Limitations & Mitigations: the material that gives the biggest capacity jump (high salt loading) is exactly the one most prone to leakage, corrosion, and pore blockage — so practical composites run at conservative loadings with a confinement strategy, trading peak uptake for durability. The Fraunhofer method is worth noting in itself: short-cycle thermogravimetric screening of powders plus a custom life-cycle rig for composites, on the finding that water-vapour cycling imposes hydrothermal stress that some structured hosts cannot survive and that most of the loss appears in the first cycles.

2025–2026: the confinement-engineering counterattack

The failure modes above are being attacked head-on, and the 2025–2026 materials trend is rigid salt confinement, not new bare frameworks:

COF-gel confinement. The first covalent-organic-framework gel salt composites (Zheng et al., Water Research 2025) nano-confine LiCl in a rigid, non-swelling micro/macroporous COF network made by one-step gelation, without ionic interaction between salt and host. The result is stable at 55 wt % salt with zero observed leakage, 3.7 g·g⁻¹ maximum uptake, 1.2 g·g⁻¹·h⁻¹ sorption rate, and six solar-driven adsorption–desorption cycles per day outdoors at 60 % RH (3.2 L·kg⁻¹·day⁻¹) — fast cycling that implies a low regeneration temperature. It is published for atmospheric water harvesting, but the composite concept and the failure modes are identical to cooling working pairs; the transfer is inferred, not demonstrated.
Encapsulation on the chiller side. The same raw source records SJTU’s encapsulated-composite chiller (Cui/Wang, Energy 2024), which seals deliquescent salt in permeable shells on the adsorber heat exchanger and runs on 50 °C hot water — data-centre waste-heat grade, below anything silica gel or zeolite can use — at SCP 102 W·kg⁻¹, about 2× silica gel at that drive temperature.
Salt-MOF composites are quantified. The ORNL review (Shamim et al., Energy 2025) reports MIL-101(Cr)/CaCl₂ at a 1:8 ratio delivering +68 % SCP while cutting the desorption temperature 110 → 90 °C, alongside MIL-101/graphene-oxide/CaCl₂ variants. The same review’s four commercialization barriers — hydrothermal cycling stability, scalable synthesis, coating/integration onto heat-exchanger surfaces, and cost — apply to salt composites as much as to bare MOFs, and it places the bottleneck at manufacturing and coated-bed engineering rather than discovery.
Degradation remains the open question. Confinement mitigates leakage and swelling, but none of these routes has yet published chiller-duty cycle counts of the kind the Fraunhofer rig produces; the durability discussion continues in Limitations & Mitigations.

Sources

Aristov — New family of solid sorbents (SWS) — the salt+matrix design approach, SWS-1L uptake, 90–100 °C desorption, theoretical COP 0.7 vs prototype COP 0.6 / SCP 20
CaCl₂-PHTS salt-loading & uptake — 4/10/20 wt % uptake vs surface area and pore volume; explicit tri-modal mechanism; energy density 71 → 193 Wh·kg⁻¹
Aristov — CSPM salt-in-matrix review — binary-salt uptake multiples, leakage/corrosion/NCG failure modes, deliquescence 0.22, 1.5–2× concentration margin, loading ceilings, mitigations
LiCl/vermiculite composite — hydrate steps → solution sequence, anti-agglomeration macroporous host, kinetics acceleration, 680 s half-cycle chiller
Consolidated CaCl₂/expanded-graphite — conductivity 7–9 vs 0.3–0.4 W·m⁻¹K⁻¹, SCP +353 %, NH₃ prototype; the matrix’s capacity-vs-conductivity roles
Fraunhofer ISE — cycle stability — front-loaded hydrothermal degradation, Cu-BTC −40 % in 15 cycles, screening method
Zheng et al. — COF-gel LiCl composite (Water Research 2025) — 55 wt % LiCl with zero leakage, 3.7 g·g⁻¹, six cycles/day; SJTU 50 °C encapsulated-composite chiller
ORNL — MOF adsorption heat pump review (Energy 2025) — MIL-101(Cr)/CaCl₂ +68 % SCP and 110 → 90 °C; the four commercialization barriers