Crystallization or adsorption?
Both routes produce fibre-grade para-xylene. They exploit entirely different physical properties, they fail in different ways, and the right answer depends far more on your feed composition than on which is the newer technology.
Updated
Crystallization: exploiting the freezing point
Para-xylene freezes at +13.3 °C. Meta-xylene stays liquid all the way down to −47.9 °C, ortho-xylene to −25.2 °C, ethylbenzene to −94.9 °C. Chill a mixed xylene stream and para-xylene is the first thing to come out as a solid — and it comes out essentially pure, because a growing crystal lattice rejects molecules of the wrong shape.
- 1Chill
The feed is cooled in scraped-surface or internally-cooled crystallizers against a refrigerant cascade, usually propylene and ethylene stages, down to the crystallization temperature.
- 2Separate
The resulting slurry is filtered or centrifuged. The crystals carry adhering mother liquor with them, so this stage sets the purity ceiling, not the crystallization itself.
- 3Wash and melt
Crystals are washed — typically with molten product — to displace the mother liquor, then melted. This is why plants run several stages: each one takes a purer feed and produces a purer product.
- 4Recycle the filtrate
The para-depleted filtrate goes to isomerization. Intermediate-stage filtrates are recycled back to earlier stages rather than discarded.
Industrial units are built as a cascade — commonly three or four stages in series, with filtrate from each stage returned upstream. The first stage does the bulk removal from a dilute feed; the last one polishes a nearly pure stream to specification.
The eutectic: the hard limit
As para-xylene crystallizes out, the liquid left behind becomes progressively richer in meta-xylene. Follow that path far enough and you reach the eutectic — around 13 % para-xylene — where the whole mixture solidifies at once instead of depositing pure crystals. Past that point you are not separating anything, you are freezing the mixture.
Adsorption: exploiting molecular shape
Para-xylene is the slimmest of the three isomers. A faujasite zeolite, ion-exchanged with barium or potassium, has pore openings that admit para-xylene readily while discriminating against the bulkier meta- and ortho- forms. Pass mixed xylenes over that adsorbent and para-xylene is preferentially held; flush with a desorbent — para-diethylbenzene or toluene — and it is released.
Run as a fixed bed, this would be a batch process with poor adsorbent utilisation. Industrial units instead use a simulated moving bed: a single vessel divided into many beds, with a rotary valve that continuously shifts the feed, desorbent, extract and raffinate ports from one bed to the next. The ports chase the concentration profile around the vessel, so the bed behaves as though it were flowing counter-currently to the liquid, without anything actually moving except the valve.
The result is a continuous, near-isothermal separation reaching 95–97 % recovery per pass at 99.7 %+ purity — a much higher single-pass recovery than crystallization can reach from the same feed. The extract and raffinate then go to fractionation to recover the desorbent for reuse.
Head to head
| Crystallization | Adsorption (SMB) | |
|---|---|---|
| Physical basis | Freezing point difference (~60 °C) | Molecular shape vs pore aperture |
| Per-pass recovery, equilibrium feed | 60–70 % | 95–97 % |
| Per-pass recovery, PX-rich feed | > 90 % | 95–97 % |
| Product purity | 99.5–99.9 % | 99.7–99.9 % |
| Dominant energy cost | Refrigeration to well below 0 °C | Desorbent recovery by distillation |
| Consumables | None | Adsorbent (periodic replacement), desorbent make-up |
| Feed impurity tolerance | High | Lower — the adsorbent must be protected |
| Turndown | Good | Moderate — the profile must stay established |
| Equipment at risk | Scraped-surface crystallizers, filters/centrifuges | Rotary valve, adsorbent bed |
Which one is right
The honest rule is that feed composition decides, not vintage:
- Equilibrium feed (~24 % PX), grassroots world-scale plant — adsorption. The recovery difference dominates: a crystallizer would need a much larger isomerization loop around it to reach the same production, and that loop is expensive.
- Feed already rich in para-xylene (selective toluene methylation, selective disproportionation) — crystallization. Above the eutectic constraint it recovers nearly everything, and it does so with no adsorbent inventory, no desorbent make-up and no rotary valve.
- Debottlenecking an existing adsorption unit — a crystallizer placed on the raffinate or as a second stage can add capacity without touching the SMB.
- Difficult or variable feed — crystallization tolerates impurities that would poison or foul an adsorbent bed.
Hybrid schemes exist precisely because of this: adsorption for the bulk separation from a dilute feed, crystallization to polish an already-rich stream. Each technology is used where its own limitation does not bind.
Frequently asked questions
Which gives higher para-xylene purity?
Both reach fibre grade. Adsorption typically delivers 99.7–99.9 % directly. Multi-stage crystallization with proper washing reaches the same range; purity there is limited by how well mother liquor is displaced from the crystals, not by the crystallization itself.
Why is crystallization still built today?
Because feeds rich in para-xylene changed the arithmetic. The eutectic only limits recovery when the feed is near equilibrium composition. With an 80–90 % para-xylene feed, crystallization recovers more than 90 % per pass with no consumables at all.
What is a simulated moving bed?
A fixed adsorbent bed divided into sections, with a valve that continuously moves the feed, desorbent, extract and raffinate injection and withdrawal points from bed to bed. This makes the solid appear to move counter-currently to the liquid, giving the efficiency of a counter-current process without physically circulating solids.
How cold does a para-xylene crystallizer run?
Well below zero — the exact temperature depends on the feed composition and the stage, since each stage works on a progressively different liquid. The refrigeration cascade needed to reach it is the dominant variable cost of the crystallization route.
Our calculator models the crystallization route in detail: the stage cascade, the filtrate recycles, the refrigeration duty and the isomerization loop wrapped around it.
Related reading
How para-xylene is made from mixed xylenes: crystallization or adsorption, the isomerization recycle loop, yields, losses and the economics.
Where the money goes in a para-xylene plant: the PX–naphtha spread, refrigeration and reboiler duty, the recycle multiplier, ring losses and by-product credits.
Why para-xylene plants recycle: how isomerization re-equilibrates xylenes, the ethylbenzene fork, approach to equilibrium, and how losses multiply.