Engineering Guide · Vendor-Neutral

Depth Filter vs Centrifugation for Harvest Clarification: Which Should You Pick?

Published June 16, 2026 · 13 min read
Depth filter vs centrifugation side-by-side comparison for bioprocess harvest clarification harvest broth clarified centrate 95-99% recovery 10 um 5 um 2 um 0.5 um + charge Depth filter Graded fibrous bed Single-use · gentle · low capex VS harvest broth 8000-15000 g solids to discharge centrate 92-97% recovery desludge / solids Disc-stack centrifuge Continuous mechanical separation Stainless · high capex · CIP/SIP
Figure 1: Depth filtration (left) drives harvest broth through a graded fibrous bed of cellulose plus diatomaceous earth, retaining cells and debris by size while a positively charged binder also adsorbs DNA and acidic host-cell protein. A disc-stack centrifuge (right) spins broth at 8,000-15,000 g across stacked conical discs, sedimenting cells outward to the bowl wall while clarified centrate exits axially through the centripetal port.
Quick Verdict

For CHO mAb harvest up to 2,000 L at moderate cell density, single-use depth filtration alone is the default. Above that scale, or at 30+ million cells/mL with 70-80% viability, switch to a disc-stack centrifuge (Alfa Laval BTPX / Culturefuge or GEA biopharma separators) followed by depth filter polish. The break-point is the depth filter throughput ceiling of about 60-150 L/m2; once the area needed to absorb 10,000+ L of high-density broth exceeds the floor space and cartridge cost of a centrifuge train, the centrifuge wins on COGS even with the higher capex.

Key differences at a glance

Side-by-side comparison

FactorDepth filtrationCentrifugation (disc-stack)
Separation principle Size exclusion + adsorption in graded fibrous bed Centrifugal sedimentation at 8,000-15,000 g
Throughput per unit 60-150 L/m2 (CHO); 30-80 L/m2 (E. coli homogenate) Continuous 500-2,000 L/h
Product recovery 95-99% (gentle, no shear) 92-97% (cell shear releases HCP/DNA into centrate)
HCP / DNA reduction 30-60% HCP, 60-90% DNA (charged depth filters) Negligible (cells only)
Sterilisation / cleaning Single-use, no CIP/SIP CIP + SIP cycles between batches (4-8 h)
Capital cost 50-150k USD (production-scale holders) 500k-1.2M USD (BTPX / Culturefuge class)
Consumables per batch 20-60k USD (cartridge stacks) 0 (reusable bowl) or 5-15k USD (single-use bowl)
Best-fit scale 50-5,000 L (alone); polish for any scale 2,000-25,000 L (alone needs depth polish)
Modality fit CHO mAb, AAV (HEK293), microbial after pre-clarification CHO mAb high-titre, E. coli homogenate, yeast

Values reflect typical published specifications. Your vendor's current datasheet takes precedence.

Depth filtration in detail

Depth filters are single-use cassettes or pods containing a graded fibrous bed of cellulose fibres and diatomaceous earth, often impregnated with a positively charged binder resin. The bed is structured so that broth flows through progressively tighter layers: coarse fibres at the inlet face capture intact cells and large debris, finer layers trap sub-micron debris, and the charged surfaces adsorb negatively charged DNA and a fraction of acidic host-cell protein on the way through. Modern grades from Merck Millipore (Millistak+ HC Pro), Sartorius (Sartoclear DL), Pall (Stax), 3M (Zeta Plus Encapsulated), and Eaton (BECO Integra) span 10 µm down to 0.5 µm nominal cut-offs and are typically used in two-stage trains (a primary coarse grade plus a tighter secondary grade) before a sterile-grade membrane.

How it works

Harvest broth is pushed by a transfer pump through the cassette stack at constant flow (typically 100 LMH, or litres per square metre per hour). Solids accumulate in the upstream layers; trans-membrane pressure climbs from 0.1 bar at start of run to 1-2 bar at terminal capacity. Throughput is sized as litres of broth processed per square metre of filter area (L/m2). Once terminal pressure is reached, the cassette stack is flushed with buffer to recover residual product, then discarded. Total capacity for CHO mAb harvest typically lands at 60-150 L/m2; E. coli or yeast homogenate is much harsher and fouls cassettes at 30-80 L/m2. Nejatishahidein and Zydney (2021) describe how the cellulose fibres provide depth retention while diatomaceous earth provides surface area, and how the charged binder layer offers an impurity-clearance bonus that the centrifuge cannot match.

When depth filtration wins

Depth filtration is the right answer for CHO mAb harvest up to about 2,000 L if the broth is reasonably clean (cell density <20 million cells/mL, viability >85%). At those conditions, the solids load is manageable, depth filter capacity is generous, and the elimination of CIP/SIP turnaround time accelerates batch cycle by 4-8 hours per run. Depth filtration is also the right answer for any modality where centrifugal shear is unacceptable: AAV harvest from HEK293, lentivirus from suspension HEK, and exosome production all default to depth filter trains because the centrifuge bowl mouth would shear the enveloped or capsid-fragile particles. Single-use depth filtration is also the dominant choice in clinical-phase manufacturing where the cost of CIP/SIP validation, cleaning chemistry, and equipment ownership cannot be amortised across many batches.

Centrifugation in detail

A disc-stack centrifuge for biopharma is a continuous, self-cleaning, hermetically-sealed mechanical separator: harvest broth is fed into the centre of a spinning bowl, which contains a stack of conical discs that subdivide the flow into thin sheets. The high centrifugal field (8,000-15,000 g) drives cells outward to the bowl wall, where they accumulate as sludge, while clarified centrate exits axially through a centripetal port at the top. Bowls are periodically desludged through hydraulic ports that open at the equator, ejecting the cell paste into a collection chamber. The dominant biopharma platforms are the Alfa Laval BTPX and Culturefuge series, GEA disc-stack separators, and the Andritz biopharma centrifuge line. Alfa Laval's single-use CultureOne is the emerging single-use disc-stack alternative for facilities that want to avoid CIP/SIP entirely.

How it works

Centrifuges are sized by equivalent settling area (the Sigma factor), which captures bowl geometry, disc count, disc spacing, and rotational speed. A BTPX-305 or Culturefuge 200-class machine handles about 500 L/h at hydraulic capacity; the larger BTPX-510 or Culturefuge 400-class reaches 1,000-2,000 L/h. Centrate exits at 50-200 NTU turbidity (compared to 1,000-5,000 NTU broth), which is too cloudy for a sterile filter, so a polishing depth filter is always installed downstream. Joseph et al. (2016) showed that capillary-shear devices and disc-stack centrifuges generate equivalent levels of cell breakage, measured by LDH release, and that this released DNA and host-cell protein then load the polishing depth filter — so over-shearing in the centrifuge directly raises depth filter consumption downstream.

When centrifugation wins

Centrifugation wins at scale and on solids load. Above 5,000 L of broth per batch, or when cell density and viability at harvest push packed cell volume above 10%, the depth filter area needed to absorb the solids becomes impractical (you would need 50-100 m2 of primary filter, occupying a full equipment skid). A disc-stack centrifuge handles the same broth continuously in 6-12 hours and discharges concentrated cell paste for biomass disposal. Centrifugation is also the right answer for E. coli, yeast, and Pichia harvest after high-pressure homogenisation, because the homogenate solids load is too high for direct depth filtration. Hermetic-seal centrifuges from Alfa Laval and GEA have closed the historical gap on bioburden and shear that gave depth filtration its original advantage in cell therapy and viral vector manufacturing.

Pros and cons

Depth filtration

Advantages

  • Low capex (50-150k USD production-scale holders) and no CIP/SIP validation overhead.
  • Gentle: 95-99% product recovery with negligible shear damage.
  • Charged grades adsorb 60-90% of DNA and 30-60% of host-cell protein in the same pass.
  • Compatible with shear-sensitive modalities (AAV, lentivirus, exosomes).
  • Single-use eliminates cleaning, changeover, and cross-contamination concerns.

Disadvantages

  • High consumable cost per batch (20-60k USD in cartridges).
  • Capacity ceiling: 60-150 L/m2 for CHO, much lower for E. coli homogenate.
  • Footprint scales linearly with batch volume; impractical above ~10,000 L.
  • Sustainability concerns: kilograms of plastic and DE per cartridge.
  • Lot-to-lot variability in cellulose/DE media can shift filter capacity by 20-30%.

Centrifugation

Advantages

  • Continuous flow at 500-2,000 L/h; handles 10,000-25,000 L batches without scaling area.
  • Near-zero per-batch consumable cost (reusable stainless bowl).
  • Handles E. coli, yeast, and Pichia homogenate that would plug depth filters.
  • Hermetic-seal designs (modern Culturefuge) minimise aerosol and bioburden.
  • Lower COGS per gram of mAb above ~5,000 L scale.

Disadvantages

  • High capex (500k-1.2M USD) plus CIP/SIP skid and validation.
  • Cell shear at bowl mouth releases DNA and HCP into centrate, loading downstream filters.
  • Always needs a polishing depth filter to drop centrate turbidity below 2 NTU.
  • Shear unsuitable for AAV, lentivirus, and other enveloped or capsid-fragile particles.
  • 4-8 h CIP/SIP cycle between batches extends overall campaign time.

Which should you choose?

The decision is rarely depth filter or centrifuge in isolation. It is depth filter alone, centrifuge plus depth filter polish, or (rarely) tangential flow microfiltration plus depth filter. Pick by the dominant constraint in your process: scale, modality, COGS target.

CHO mAb, <2,000 L, clinical phase

Cell density 5-15 million cells/mL, viability >85%. Single-use simplicity matters more than per-batch consumable cost. Charged depth filter primary plus secondary plus sterile filter.

Choose depth filtration alone

CHO mAb, >10,000 L, commercial

High-titre fed-batch at 30-40 million cells/mL, harvest viability 70-80%, 10-20% packed cell volume. Depth filter area would exceed 100 m2; centrifuge is the only economic option.

Choose centrifuge + depth filter

E. coli or yeast homogenate

Post-homogenisation cell debris is too fine and dense for direct depth filtration. Centrifuge removes the bulk of solids; depth filter then polishes for sub-2-NTU clarity before chromatography.

Choose centrifuge + depth filter

AAV / lentivirus / exosome harvest

Centrifugal shear damages capsid integrity and lyses producer cells. Depth filtration plus tangential flow microfiltration replaces the centrifuge step entirely.

Choose depth filtration alone

Real-world use cases

Typical setups where bioprocess teams have converged on one choice or the other.

CHO mAb, 2,000 L clinical
Depth filter primary + secondary + sterile

3 g/L mAb at 15 million cells/mL, 88% viability. Primary Millistak+ HC Pro D0HC at 100 L/m2 (20 m2 needed), secondary X0HC at 200 L/m2 (10 m2), 0.22 µm sterile. Total batch turnaround 4 h. No CIP.

CHO mAb, 12,000 L commercial
Disc-stack centrifuge + depth filter polish

5 g/L mAb at 35 million cells/mL, 75% viability, 18% packed cell volume. Alfa Laval BTPX-510 at 1,500 L/h with hermetic seal. Centrate at 80 NTU. Polishing Sartoclear DL30 at 250 L/m2 (48 m2) drops to 1.2 NTU.

E. coli inclusion body harvest
Centrifuge after homogenisation

BL21(DE3) fed-batch fermentation, harvested at OD600 90, homogenised at 800 bar. GEA HSE 30 disc-stack at 600 L/h separates inclusion bodies as paste; depth filter polish of supernatant only for trace IB removal. Centrifuge is the only viable option at this solids load.

AAV from HEK293, 500 L
Depth filter + TFF microfiltration

HEK293 suspension transient transfection harvested at 2 million cells/mL post-detergent lysis. Pall Stax PDH4 primary depth filter at 60 L/m2 (10 m2) followed by 0.45 µm TFF microfiltration. Centrifuge is avoided to protect capsid integrity.

Sizing a clarification train for your harvest broth?

Run depth filter area, capacity (L/m2), and flux for primary and secondary stages — or calculate disc-stack centrifuge throughput, Sigma factor, and equivalent g-force from RPM and rotor geometry.

Open the Filtration Calculator

Cost and lifecycle considerations

Total cost of ownership has four moving parts

Capital cost of the centrifuge or filter holders, recurring consumables (depth filter cartridges or single-use bowls), CIP/SIP chemicals and labor (centrifuge only), and product loss (lower for depth filtration). Depth filtration wins on capex; centrifugation wins on per-batch consumables at scale.

At small scale (1,000-2,000 L mAb), the centrifuge capex of 500k-1.2M USD overwhelms the depth filter cartridge cost of 5-15k USD per batch. Depth filtration alone is the obvious choice. The break-even point is roughly at 5,000-8,000 L: at 10 batches per year, a depth-filter-only train costs 200-600k USD/year in cartridges, while the centrifuge amortised over 10 years plus 50k USD/year in CIP and 10k USD/batch in polishing depth filter costs 150-300k USD/year.

Above 10,000 L the centrifuge wins decisively. Depth-filter-only at 15,000 L would need 100-200 m2 of primary cartridge area per batch, consuming 60-150k USD in cartridges per run and 1-1.5M USD/year in consumables at 10 batches/year. The centrifuge plus polishing depth filter runs at 200-400k USD/year. Add the floor space of the depth filter skid (which scales linearly with batch volume) and the cycle-time penalty (4-6 h to install and prime 100 m2 of cartridge stacks), and the centrifuge becomes the only realistic option for commercial mAb manufacturing at this scale.

Cost component (10,000 L scale, 10 batches/yr)Depth filtration aloneCentrifuge + depth filter polish
Capital: holders / centrifuge skid~100-150k USD~600k-1.2M USD
Consumables per batch (cartridges)~60-120k USD (100-150 m2)~10-15k USD (polish only, 30-50 m2)
CIP/SIP chemicals + labor per batch0 USD~5-8k USD
Product loss penalty (assume 1 g/L titre)~1-3% (95-99% recovery)~3-5% (92-97% recovery, plus polish loss)
3-year TCO (consumables + capex amortised)~1.9-3.7M USD~1.0-1.5M USD

Vendor landscape

Major vendors in each camp, with one-line positioning notes.

Depth filter vendors

Centrifuge vendors

Frequently asked questions

What is the difference between depth filtration and centrifugation for harvest clarification?
Depth filtration is a static unit operation that drives harvest broth through a graded fibrous bed (typically cellulose plus diatomaceous earth and a positively charged binder) where cells, debris, DNA, and host-cell protein are captured by a combination of size exclusion, adsorption, and depth retention. Centrifugation is a continuous mechanical separation that uses centrifugal force (typically 6,000-15,000 g in a disc-stack bowl) to sediment cells out of the broth. Depth filters are single-use and cleaning-free; centrifuges are stainless-steel CIP/SIP assets with hours-long cleaning cycles. The two are often used in series: centrifuge to remove most of the solid load, depth filter to polish the centrate to under 2 NTU before sterile filtration.
When should you pick depth filtration alone vs centrifugation plus depth filtration?
Depth filtration alone is the default for runs up to about 2,000 L if the broth is reasonably clean (CHO mAb at under 20 million cells/mL with viability above 85%). Above 2,000 L, or for high-titre fed-batch broth at 30-40 million cells/mL with viability dropping to 70-80% at harvest, the solids load (10-30% packed cell volume) overwhelms depth filter capacity and the train becomes centrifuge plus depth filter. The break-point is set by depth filter throughput (typically 60-150 L/m2 for high-density CHO) and the floor space cost of putting in enough depth filter area to absorb 10,000+ L of broth.
How much harvest broth can one depth filter handle?
Typical throughput for primary clarification of CHO mAb harvest is 60-150 L/m2 of depth filter area before terminal pressure (1-2 bar) is reached. A Millistak+ Pod with 1.1 m2 cartridge therefore clarifies 66-165 L of broth. E. coli homogenate after high-pressure homogenisation tends to plug faster, typically 30-80 L/m2. Single-use depth filter cassettes are stacked in holders to reach total areas up to 50-150 m2 at production scale (5,000-15,000 L clarified per cycle). The Depth Filtration Sizing article gives the worked sizing equations and safety factors for CHO and E. coli.
How much does a disc-stack centrifuge cost for biopharma?
A new pilot-scale disc-stack centrifuge (Alfa Laval Culturefuge or BTPX 305 class, about 500 L/h hydraulic capacity) lands at 250-500k USD. Production-scale (BTPX 510 or Culturefuge 400 class, 1,000-2,000 L/h) reaches 500k-1.2M USD. Single-use disc-stack alternatives like the Alfa Laval CultureOne are in the 200-400k USD bracket for the controller, with disposable bowls at 5-15k USD per run. A comparable production-scale depth filtration train is 50-150k USD in stainless-steel holders plus 20-60k USD in cartridges per run.
Does centrifugation damage cells or reduce product recovery?
Yes, modestly. Joseph et al. (2016) showed that capillary-shear devices and disc-stack centrifuges generate equivalent shear stress (measured by LDH release), and that cell breakage at the bowl mouth releases additional DNA and host-cell protein into the centrate. The downstream depth filter then has to absorb this extra burden, so total clarification capacity drops if the centrifuge is over-sheared. Product recovery from centrifugation is typically 92-97%; depth filtration alone runs 95-99% if the filter is properly conditioned. The shear penalty is the main reason newer disc-stack designs (Culturefuge, hermetic seals) market gentle harvest as a key feature.
Is depth filtration single-use or reusable?
Modern primary clarification depth filters are almost exclusively single-use cassettes or pods (Millistak+ HC Pro from Merck Millipore, Sartoclear from Sartorius, Stax from Pall, Zeta Plus from 3M, BECO from Eaton). Cellulose plus diatomaceous earth media are difficult to clean and validate for reuse, and single-use eliminates cleaning chemistry, validation studies, and changeover time. Reusable stainless-steel depth filter housings exist for very-large-scale operations (above 15,000 L) but are uncommon in modern biologics. See the single-use vs stainless steel facility comparison for the broader trade.
Can depth filters remove host-cell protein and DNA, or just cells?
Both. Nejatishahidein and Zydney (2021) describe how positively charged binder resin in modern depth filters (Millistak+ HC Pro, Sartoclear DL) adsorbs negatively charged DNA and a fraction of acidic host-cell protein during the clarification pass. This gives depth filtration an impurity-reduction role beyond pure solid removal: a well-chosen depth filter grade can reduce HCP by 30-60% and DNA by 60-90% before the Protein A capture step. Centrifugation removes only cells and large debris; downstream HCP and DNA reduction then depends entirely on chromatography. See HCP reduction strategies for the full downstream picture.
Which is better for AAV and viral vector harvest, depth filter or centrifuge?
Depth filtration is the default for AAV harvest from HEK293 transient transfection at scales up to a few hundred litres, because centrifugal shear can damage the AAV capsid and because HEK293 lysates are often pre-clarified by mild detergent or freeze-thaw, leaving a manageable solid load. At larger scales (above 1,000 L) and for lentivirus production, depth filter plus tangential flow microfiltration replaces the centrifuge step because the centrifuge would shear-damage the enveloped virus particles. For mammalian-cell-secreted mAbs at very large scale (above 10,000 L), centrifuge plus depth filter remains the dominant train. The AAV downstream processing article covers the full vector-specific train.

Resources and references