Bearing and magnetic coupling design formagnetically driven agitators in bioprocessMilena McFeeters Chem.Eng. MBA∗ and Bart Duijvelaar M.Sc. MBA†

∗Steridose Inc, 5020 World Dairy Drive, Madison WI, USA, †Steridose AB, Himmelsbodavägen 7, Tumba, Sweden

ABSTRACT The most common purpose of agitation in biopharmaceutical processing is liquid/liquid andsolid/liquid blend (solutions and suspensions). Agitators are performing their duty while submerged in a fluidphase, and consequently designed for this intended use, considering that running in air would not fit anypurpose. However, when dealing with partly manually controlled standard operating procedures (SOPs),or due to different sequences alternating process fluids with clean-in-place (CIP) and steam-in-place (SIP),combined with a relatively basic automation, the risk of having an agitator running without liquid is present.In the case of top entry agitators, this may not constitute a condition that could damage the equipment. Forbottom entry agitators, the situation may be more complex and may result in damage to mechanical seals, orbearings (in the case of magnetically driven agitators) that rely on the process fluids for lubrication. This articlediscusses different bearing designs and materials of construction typically available and dry-running.

CONTENTS

1 Bearing and magnetic coupling design for mag-netically driven agitators in bioprocess 11.1 Introduction . . . . . . . . . . . . . . . . . . 11.2 Bearing design . . . . . . . . . . . . . . . . . 11.3 Magnetic coupling design . . . . . . . . . . 21.4 Materials of construction . . . . . . . . . . . 31.5 Bearing material requirements for use in

bioprocess . . . . . . . . . . . . . . . . . . . 41.6 Conclusion . . . . . . . . . . . . . . . . . . . 4

2 Effect of dry-running conditions on magneticallydriven agitators in bioprocessing 42.1 Introduction . . . . . . . . . . . . . . . . . . 42.2 What constitutes a dry-running condition . 42.3 Bearing damage expected from dry-

running conditions . . . . . . . . . . . . . . 52.4 Design characteristics related to dry-

running capability . . . . . . . . . . . . . . . 52.5 Bearing material requirements for use in

bioprocess . . . . . . . . . . . . . . . . . . . 52.6 Conclusion . . . . . . . . . . . . . . . . . . . 6

Copyright © 2019 Milena McFeetersGenerated by Steridocs: January 21, 2019

1. BEARING AND MAGNETIC COUPLING DE-SIGN FOR MAGNETICALLY DRIVEN AGITA-TORS IN BIOPROCESS

1.1. IntroductionBottom entry magnetically driven agitators feature an im-peller that is driven by a magnetic coupling, rather thanphysically connected to a shaft. This means there is noneed to penetrate a process vessel, eliminating mechan-ical seals and reducing the risk of leaks. By not havinga shaft, it is easy to design the process contact surfacesfor effective CIP and SIP. The impeller typically runs ona bearing that will support it along all axes while rotat-ing. A magnetic coupling, using only permanent magnets,cannot suspend objects in space without radial or axialsupport (Earnshaw’s theorem). In general, bearings aredesigned to work with a lubricant. In the case of mag-netically driven agitators in bioprocess, the lubricant isthe process fluid. This could vary from fluids, such aswater for injection (WFI) to final product, as well as CIPdetergents and clean steam. The lubricating properties ofall these vary greatly.

1.2. Bearing designMost magnetically driven agitators for biopharmaceuticaluse a journal type bearing. See figure 1. Other designshave been used, such as ball bearing, but are not preferreddue to their inherited complex geometry that would neg-

www.steridose.com White paper collection | 1

WHITE PAPER

n Figure 1 Axially supported journal bearing arrangementversus a non-supported design and consequential particlegeneration.

atively impact cleanability. This discussion will focus onjournal bearings.

A journal bearing will have a stationary component(male bearing) and a rotating component (female bear-ing) that is affixed to the impeller. While most designsuse the same material for both male and female compo-nents, some use dissimilar materials. Unlike mechanicalseal stationary and rotating seal faces, the male and fe-male components in a journal bearing are not intendedto seal and a small gap between them exists so that thecomponents can be lubricated by the process fluid. Inbiopharmaceutical applications, this gap is also essentialto ensure cleanability and drainability. Therefore, bearingexposure to process fluids during formulation and CIP isimportant to ensure cleanability and avoid dry-running.In mechanical seal’s design, the rotating and stationaryseal rings could be made of dissimilar materials, wherethe softer material would wear out faster, but will be heldagainst the opposite face by spring force. With journalbearings, there is no spring that keeps the components to-gether, so if one component wears faster than the other, thegap between them will increase. While a gap is necessaryto provide lubrication, a big gap can result in wobblingand instability of the assembly that can cause prematuredamage.

1.3. Magnetic coupling designBearing design and magnetic coupling design are stronglyrelated. The bearing design needs to withstand the forcesimposed by the weight and rotation of the impeller andby the magnetic field that creates the magnetic coupling.Magnetic couplings can be axial or radial. Radial cou-plings are preferred by most manufacturers since they donot add additional axial loads to the bearing. Journal bear-ings are typically able to withstand the forces imposedby a radial coupling. They also provide more space toaccommodate more magnets and therefore can providehigher maximum torque ratings. See figure 2.

Axial couplings would add forces acting upon the bear-ing that would require increased robustness. In thosecases, ball or roller bearings may be more adequate towithstand the additional load, but, as discussed, are notconsidered to be hygienic. This discussion will focus onjournal bearings used in combination with radial magneticcouplings.

Magnetic couplings can be of non-floating or floating

n Figure 2 Principle of a radial magnet coupling.

−2 −1 0 1 2 −2

0

2−4

−2

0

2

4

n Figure 3 Illustration of the stationary points that canbe achieved by permanent magnet configurations. Themagnetic field’s stationary points are mathematically limitedto saddle points, providing stability in one direction, but notin the other.

2 | Milena McFeeters et al. www.steridose.com

type. A non-floating coupling is aligned in a way thatensures the female bearing in the impeller is resting onthe male bearing, thus providing mechanical support bothalong the radial, as well as axial axes. A floating couplingis aligned in a way that when engaging the coupling, theimpeller slightly lifts up avoiding contact on the axial sur-faces of the male and female bearing. While in operation,with a liquid film between the bearings, the designs areroughly equally non-contacting.

1.3.1. Earnshaw’s theorem

From the discovery of magnetism by the ancient Greeks, totoday’s children playing with magnets, mankind has triedto make objects float in mid-air by using the attracting andrepelling forces between magnets. For many centuries,and in some cases even today, the prospect of floatingobjects has triggered many attempts along the lines ofif I could only arrange these magnets in a sufficiently clevergeometry, I can make things float. The internet is full withexamples of these attempts.

In 1842 the British mathematician Samuel Earnshawproved mathematically that permanent magnet configura-tions cannot lead to magnetic fields with stationary pointsthat are stable in all directions [5], see figure 3. With theexception of a small number of special cases, magnetically’floating’ objects are always supported along at least oneaxis. Real levitation by electromagnetic forces is usuallyachieved by a feedback loop that adjusts (electro)magneticfields in fractions of seconds depending on the exact posi-tion of the object, in practice ’broom-stick-balancing’ theobject on the saddle-point.

1.3.2. Implications of Earnshaw’s theorem on floatingbearing arrangements

As a result of the limitations of positioning objects bymagnetic forces (see section 1.3.1) even the floating bear-ing arrangement will have contacting surfaces betweenthe male and female bearing, see figure 1. As mentionedbefore (section 1.3), with a liquid film between the bear-ings, this effect might be minimized (to about the samemagnitude for both floating as well as non-floating ar-rangements). Under start-up-conditions or dry-runningconditions, this will not hold true.

Moreover, in most applications, the flow pattern is inturbulent regime. Also, the impeller is located offset fromthe center of the vessel, and therefore the eddies aroundthe impeller are of different magnitudes. As the impeller ispushed side to side by the eddies, the radial surfaces of thejournal bearing would become in contact, avoiding the im-peller from moving outside of its intended position. Thismeans that the non-contacting bearing now has become acontacting one, with comparable particle-generation andwear to non-floating bearing arrangements.

The lack of support in a floating-bearing arrangementmay also allow the impeller to wobble up and down(which is not possible with a non-floating design). Thiscondition may result in damage to the impeller and pre-mature bearing failure, due to increased vibration.

1.4. Materials of constructionMaterials of construction play the most important role inthe ability of a journal bearing to perform satisfactorilyin an application (with regards to process fluid, temper-ature, mixer rpm and bearing design). Some of the mostcommon bearing materials are described below (and com-pared in table 1):

Silicon Carbide: This material is a hard ceramic com-pound of silicon and carbon. There are two types:reaction bonded, which is considered a multi-phasemixture, or sintered, which is considered a crystallinematerial. Sintered silicon carbide is commonly usedin mag mixer’s journal bearings. It is a hard mate-rial, with excellent chemical resistance, but it is brit-tle. This characteristic makes it susceptible to dam-age during assembly/disassembly and system upsets,such as dry-running or other conditions that wouldgenerate excessive vibration.

Tungsten Carbide: This material is a crystalline ce-mented material produced with a binding metallicmatrix. Its chemical compatibility varies dependingon the metallic binder that is used. Some alloy andnickel binder tungsten carbides offer the highest cor-rosion resistance, but its chemical compatibility is notas wide as that of sintered silicon carbide. Straightcobalt-binder tungsten carbide is not recommendedin biopharmaceutical applications due to its low cor-rosion resistance that may cause cobalt to leach outwhen exposed to low pH fluids. Tungsten carbide isimpact resistant, which makes it less prone to dam-age during assembly and disassembly than siliconcarbide. The ductile metal binder also diminishesthe brittle characteristics of ceramic carbides, improv-ing toughness and durability. It is considerably morelikely to survive system upsets and dry-running con-ditions.

Zirconium Dioxide: This material is a crystalline, sin-tered ceramic material. Compared to silicon carbide,it is more impact resistant and less likely to break dur-ing assembly/disassembly. It would survive systemupsets and dry-running better than silicon carbide.However, it is less wear-resistant and therefore proneto higher particle generation.

Diamond Coated Silicon Carbide: A crystalline dia-mond film can be deposited on ceramic substrates,such as silicon carbide, by turning natural gas intodiamond by means of plasma-assisted carbon deposi-tion techniques[8]. The resulting film creates a bondwith the carbon atoms in the silicon carbide structure.The main advantages of this film are related to theproperties of diamond, such as high wear resistance,low coefficient of friction that increases dry-runningtolerance, high thermal conductivity that results inlower running temperatures, and excellent chemicalcompatibility. While it may not change the brittlenature of the silicon carbide substrate (it wouldstill be susceptible to damage during assemblyand disassembly), it greatly improves survivabilityduring system upsets, such as dry running.

www.steridose.com White paper collection | 3

Bearing materiala Tungsten Carbide Silicon Carbide Dri-amond™

Chemical compatibility pH range 2-14b Best - inert Same as silicon carbide

Ease of installation andhandling

Best Harder than tungsten car-bide, but more brittle

Somewhat better than sili-con carbide

Survivability under dry-running conditions

Good Poor Best - lowest coefficient offriction

a The use of dissimilar materials in male and female bearings is not recommended.b Exact value depends on particular fluid properties, temperature and exposure time.

n Table 1 Comparison of bearing materials.

1.5. Bearing material requirements for use in biopro-cess

According to the latest ASME BPE edition [1], per PM-2.1.3 and table PM-2.2.1-1, bearing materials, such as sili-con carbide, tungsten carbide, etc., shall comply with therequirements of USP<87> Biological Reactivity Tests, inVitro [2] or ISO 10993-5 for biocompatibility. Due to theimportance of particle generation in biopharmaceuticalprocesses, most manufacturers also test their bearings inaccordance to USP<788> Particulate Matter in Injections[3]. The USP defines maximum particle counts of 25/mLfor particles ≥10 µm and 3/mL for particles ≥25 µm.

1.6. ConclusionBearings in magnetically driven agitators are lubricatedby the process fluid(s). The bearing design, magnetic cou-pling design and selection of materials of constructiongreatly affects magnetic coupled mixer performance. Jour-nal bearings and radial couplings are the most commonchoices for magnetically driven agitators in bioprocess.Non-floating couplings offer more mechanical stabilitythat may prevent premature damage. The material of con-struction can make the most significant difference in theability to run dry. Ductile materials, such as tungsten car-bide, may not be as prone to fracture during dry-runningas ceramic materials, such as silicon carbide. Diamondcoatings have shown to greatly improve dry-running ca-pabilities in ceramic materials. Materials shall complywith regulatory requirements, such as USP<87> [2] andUSP<788> [3].

2. EFFECT OF DRY-RUNNING CONDITIONSON MAGNETICALLY DRIVEN AGITATORS INBIOPROCESSING

2.1. IntroductionBottom entry magnetically driven agitators feature animpeller that is driven by a magnetic coupling, rather thanphysically connected to a shaft. This means there is noneed to penetrate a process vessel, eliminating mechanicalseals and reducing the risk of leaks. By not having a shaft,it is easy to design the process contact surfaces for effectiveCIP and SIP.

In general, bearings are designed to work with a lu-bricant. In the case of magnetically driven agitators inbioprocess, the lubricant is the process fluid. This couldvary from fluids, such as water for injection (WFI) to final

product, as well as CIP detergents and clean steam. Thelubricating properties of all these vary greatly. But first, itis important to define what dry-running conditions meanwith respect to the bearings.

2.2. What constitutes a dry-running conditionThe only component of a magnetically driven agitatorassembly that is susceptible to dry-running is the bear-ing. Therefore, dry-running conditions need to be seenfrom the perspective of the bearing assembly surround-ings. Dry-running is defined as the absence of a liquidfilm between bearings. It is not related to the impellerbeing submerged in fluid or not. A film between bear-ings can still exists when the vessel is empty, and undersome circumstances, may not actually exist in a full vessel.Some example of commonly misunderstood conditionswhere dry running may occur are listed below:

• Running an agitator while emptying the vessel: inthis case, the bearing assembly in the mixer was cov-ered with liquid, and the liquid is drained from thevessel. The bearings would still have a liquid film be-tween them even after the vessel is empty. This liquidfilm would remain for a certain time that varies withthe speed that the agitator is running, the fluid phys-ical properties (specific gravity and viscosity) andtemperature. As long as there is still a liquid film be-tween bearings, this is not considered a dry-runningcondition.

• Running during CIP without submergence: in thiscase, a spray ball is delivering cleaning solutions andrinse water to the vessel. A properly designed sprayball will aim one or more streams to the impeller head,so that it can be wetted. Like all other surfaces insidethe vessel, any surface that cannot be wetted, cannotbe cleaned either. If the mixer speed is kept to themanufacturer’s recommendation in this condition,the flow from the spray ball will replenish the liquidfilm between bearings. This does not constitute adry-running condition. Often, it is possible that thereare a few minutes between the start and stop of eachCIP cycle that may lead to some periods of not havingflow from the spray ball(s). If the mixer is not stoppedin this sequence and runs continuously through theentire cycle, this would be a similar situation as whenrunning the agitator while emptying the vessel. Ifthe lapse in between spray ball flow is too long and,particularly if using hot fluids (e.g. WFI at 80◦C),

4 | Milena McFeeters et al. www.steridose.com

the liquid film may evaporate quicker leading to adry-running condition.

• Running during SIP: in this case, the vessel is empty.It is uncommon to run mixers during SIP, but it maybe necessary to start the mixer for a short intervalevery now and then to remove condensate. Similarto the CIP condition, as long as the mixer is run perthe manufacturer’s recommendation for SIP, this alsodoes not constitute dry-running.

• Running without liquid in the vessel for an undeter-mined time: this may occur if there are no safeguardsto stop the mixer once all the liquid has been drainedfrom the vessel. For example, if an operator may for-get turning off the mixer and leaves for the day (oreven worse, for the weekend), the mixer may be run-ning with no liquid for multiple hours and possiblydays. Eventually, the liquid film will evaporate. Thisis a common situation where dry-running can occur.

• Running the mixer too fast for the volume containedin the vessel: in this case, and when a strong vortexthat draws air to the impeller is pulled, it may exposethe bearing assembly to air instead of the processliquid and could result in dry-running.

When working with low liquid levels, impeller cav-itation is another consideration, due to insufficient netpositive suction head. This could also damage the bear-ings, but it is a different phenomenon that should not beconfused with dry-running.

2.3. Bearing damage expected from dry-running condi-tions

According to Folger [6], the number one cause for bearingdamage is inadequate lubrication. Therefore, dry-runningis an important consideration in terms of bearing fail-ure. Besides the operating conditions (temperature, mixerspeed, etc.), the bearing type, specific design and mate-rials of construction will determine how long a bearingcan survive dry-running conditions. The range can varyfrom a few minutes to several days. The one thing thatis common in all cases, is that no bearing can survivedry running conditions forever. The most common fail-ure types related to dry-running conditions are describedbelow:

• Bearing fracture: The bearing cracks and breaks intosmall pieces. If the mixer is not stopped right away,the resulting particles can damage other componentsin the assembly, such as the impeller and weld-plate.This extended damage is more prominent with ma-terials with high hardness, such as silicon carbide.Bearing fracture is the result of noise generation whenthe bearings are running dry. The noise will createvibration that will cause the fracture. Brittle materials,such as silicon carbide, are more prompt to this typeof failure.

• O-ring damage: During dry-running conditions, sur-faces can reach high temperatures due to frictionalheating. The amount of heat that is generated will de-pend on the specific material’s coefficient of friction.Materials such as silicon carbide can reach tempera-tures in excess of 500ºF (260ºC) within minutes. Any

O-rings in the bearing assembly will be affected bythe temperature and will expand. This expansioncould also cause damage to the bearing surfaces, de-pending on their design. Most O-ring materials willnot survive such elevated temperatures and will nolonger seal adequately. O-rings are often used to sealthe threads between weld-plate and male bearing.

• Increased particle generation: The lack of lubricationduring dry-running will contribute to increased par-ticle generation. If the particle generation surpassesthe limits imposed by the process, then it is consid-ered that the bearing has failed, even if no fracturehas occurred.

2.4. Design characteristics related to dry-running capa-bility

2.4.1. Bearing designBearing design was discussed in 1.2.

2.4.2. Materials of constructionMaterials of construction play the most important role inthe ability of a journal bearing to withstand dry-runningconditions. Considering all other variables equal (processfluid, temperature, mixer rpm and bearing design), somesignificant changes can be seen by just selecting a differentmaterial of construction. The way these bearings may failwould also be different depending on the material. Someof the most common bearing materials are summarized intable 1.

When exposed to dry-running conditions, ceramic hardmaterials, such as silicon carbide, fail due to fractures onthe surface. More ductile materials, such as tungsten car-bide, do not fracture easily, but may show wear patternsafter long periods of dry-running that would provide ev-idence of an increase in particle generation. Diamondcoated silicon carbide would fail in the same way as sili-con carbide, but over a much longer dry-running exposuretime. The exact time to failure for each material will varydepending on the application conditions. According toDiaccon [9], a diamond coated silicon carbide mechanicalseal face could last hours before failure in a dry-runningcondition, compared to just seconds for uncoated siliconcarbide. Advanced Diamond Technologies reports sur-vivability in dry-running conditions of diamond coatedsilicon carbide mechanical seal faces to be close to 30 timeslonger than uncoated silicon carbide [10].

Steridose testing of their Dri-amond™ option shows im-provement of dry-running capability of journal bearingsby up to 30 times compared to uncoated silicon carbide.See figure 4. The worst-case test refers to no liquid filmpresent between the bearing surfaces (parts were installeddry and liquid was never present), while the acceleratedfailure refers to an initial liquid film present at the onsetof the test, with accelerated evaporation (bearings weresubmerged at first in 90º water and liquid from vessel wasallowed to drain completely before starting test).

2.5. Bearing material requirements for use in biopro-cess

Bearing material requirements for use in bioprocess werediscussed in section 1.5.

www.steridose.com White paper collection | 5

0 10 20 30 40

Silicon Carbide

Dri-amond™

Silicon Carbide

Dri-amond™

TIME TO FAILURE (ARBITRARY UNITS)

WORST CASE TEST

WORST CASE TEST

ACCELERATED FAILURE TEST

ACCELERATED FAILURE TEST

versus

versus

n Figure 4 Steridose Dri-amond™ bearings time-to-failure under adverse running conditions versus standardsilicon carbide. Refer to the text for descriptions of ’worstcase’ and ’accelerated failure test’.

2.6. ConclusionBearings in magnetically driven agitators are lubricatedby the process fluid(s). Therefore, it is important to un-derstand when a dry-running condition may occur. Dry-running is defined as the absence of liquid film betweenbearings. The bearing design, magnetic coupling designand selection of materials of construction can help increas-ing survivability of the bearing during dry-running con-ditions. However, no bearing design can run dry forever.The material of construction can make the most significantdifference in the ability to run dry. Ductile materials, suchas tungsten carbide, may not be as prone to fracture duringdry-running as ceramic materials, such as silicon carbide.Diamond coatings have shown to greatly improve dry-running capabilities in ceramic materials. Materials shallcomply with regulatory requirements, such as USP<87>[2] and USP<788> [3].

REFERENCES[1] American Society of Mechanical Engineers, ASME

BPE 2016 Bioprocessing Equipment, ASME, 2016.

[2] Retrieved from http://www.pharmacopeia.cn/v29240/usp29nf24s0_c87.html , USP <87>, U.S. Pharmacopeia .

[3] Retrieved from http://www.pharmacopeia.cn/v29240/usp29nf24s0_c788.html , USP <788>, U.S. Pharma-copeia .

[4] Milena McFeeters, Bearing and magnetic coupling de-sign, Steridose white paper, June 2018.

[5] Retrieved from https://en.wikipedia.org/wiki/Earnshaw%27s_theorem .

[6] R.Folger et al, Bearing Killers: Preventing Com-mon Causes of Bearing System Damage, Retrievedfrom: https://www.timken.com/wp-content/uploads/2017/04/Bearing-Killers-Technical-White-Paper.pdf .

[7] Fluid Sealing Association, What is the best silicon car-bide wear face material for my mechanical seal?, Pumps& Systems, January 2006.

[8] Bart G. Duijvelaar, Production of and Vickers hardnessmeasurements on DC sputtered TiN, TiC and TiCN coat-ings and nano indentation on a-C:H., Master’s thesis,Eindhoven University of Technology, 1998.

[9] DiaCCon GmbH, Kristalliner Diamant – die Revolutionfur Gleitringdichtungen , Dichtungstechnik, 2006.

[10] Advanced Diamond Technologies , http://www.thindiamond.com/uncd-technology/technology-overview/ .

[11] Ceratec Technical Ceramics BV , http://www.ceratec.nl/materials.html .

[12] Dexter Magnetic Technologies , https://www.dextermag.com/products/permanent-magnets/ .

ABOUT THE AUTHORSMilena McFeeters is the CEO of Steridose. Ms.McFeeters has 20 years of experience workingwith bioprocess equipment. She is a memberof ASME BPE, currently serving as Chair ofthe Sealing Components sub-committee. Ms

McFeeters holds a Licenciate degree in Chemical Engineer-ing from the University of Costa Rica, and an MBA fromNational University.

Bart Duijvelaar is a product manager at Steri-dose, located at the Steridose head office inTumba, Sweden. Mr. Duijvelaar has 17 yearsof experience with sales and product manage-ment of rotating equipment to a wide variety of

industries and in a wide variety of international businesssettings. He holds an MSc in Applied Physics from Eind-hoven Technical University and an MBA from EdinburghBusiness School.

6 | Milena McFeeters et al. www.steridose.com