Sorbents: From Basic Structure to Clinical Application

Basic Principles

The combination of chemicals to form a mixture is a spontaneous and natural process, accompanied by an increase in entropy or randomness. The reciprocal process of separation of a mixture into its constituent species is not a spontaneous process and requires an expenditure of energy. If the mixture comes as two or more immiscible phases, gravity, pressure, or electrical fields can be applied to obtain separation. On the contrary, if it exists in a single homogeneous phase, different processes must be applied, such as the following:

• Separation by phase addition or creation (distillation, crystallization, de-sublimation)

• Separation by barrier (reverse osmosis, dialysis, microfiltration, ultrafiltration)

• Separation by solid agent (adsorption, chromatography, ion exchange)

• Separation by external field or gradient (electrodialysis, electrophoresis)

In clinical settings, blood purification techniques rely mostly on the second and third processes. Membrane separation processes such as hemofiltration or hemodialysis predominantly use diffusion and convection. Diffusion may be limited by the diffusion coefficients of the molecules or by other factors such as temperature, surface area, and thickness of the membrane. On the other hand, convection is limited primarily by the sieving properties of the membrane and the flux of solvent obtained in response to a positive pressure gradient (ultrafiltration). When diffusion and convection are inadequate to remove the target molecules from the patient's blood, the use of sorbents and hemoperfusion may become an additional option for blood purification. In hemoperfusion, blood is circulated through a unit (cartridge) containing the solid sorbent material. Solute removal and blood purification are obtained by absorption (binding) of molecules onto the sorbent particles.

Sorbents can be composed of synthetic or natural materials. In the past, the application of hemoperfusion was limited by the relative bioincompatibility of the sorbent material and the significant side effects derived from its contact with blood. Hemoperfusion sessions were often accompanied by chills, fever, cutaneous rush, thrombocytopenia, leukopenia, and aluminum leaching.

Today, these reactions have become rare and are prevented in two ways:

• In some techniques, plasma is separated from cells before being circulated through the sorbent bed. After the sorbent cartridge, blood is reconstituted so that red cells, white cells, and platelets never come in contact with the sorbent surface, and bioincompatibility reactions are avoided.

• The sorbent material is made bio- or hemocompatible by a specific coating process that covers the particles with biolayers that are well tolerated by blood cells.

There is little debate that the use of sorbents is justified in poisoning or acute intoxications, for which hemoperfusion is the treatment of choice in many instances because of the high affinity of the sorbent for the specific toxic molecule. (This aspect of hemoperfusion is not the focus of this chapter.) However, the use of sorbents in chronic or acute blood purification techniques is still a matter of discussion. In particular, the additional value offered by adsorption must be counterbalanced by the increase in costs that is involved when sorbents are used.

The efficiency of membrane separation processes in hemodialysis is limited by membrane permeability. To overcome this problem, high-flux membranes have been introduced. Today, high cutoff membranes are also available, but their efficiency when used in the diffusive mode is limited by the low diffusion coefficients of high-molecular-weight solutes. Adaptations increasing the degree of convection have been made in chronic treatments (online hemodiafiltration) and in continuous therapies (high-volume hemofiltration). In such circumstances, the high rate of ultrafiltration increases significantly the clearance of solutes in the middle-high-molecular-weight spectrum. Nevertheless, the relative selectivity of adsorptive processes and the possibility of placing the sorbent in direct contact with blood may be seen as a further step toward increasing the efficiency and specificity of the blood purification process for certain types of solutes. In particular, specific molecules can be targeted for removal by selective adsorption mechanisms. Furthermore, solutes with molecular size larger than the pore dimensions of membranes can be removed by direct adsorption onto the surface of the sorbent particles.

However, the process of size-dependent, nonselective adsorption may cause unwanted losses with unexpected removal of antibiotics or other drugs and hormones. The removal kinetics for middle-large molecules during hemoperfusion are distinct in comparison with hemodialysis, and several clinical parameters should be monitored carefully. The basis for safe and efficient application of sorbents in clinical practice resides in a deep knowledge of the materials used and the mechanisms involved in the production and the design of the hemoperfusion device.

Sorbent Materials and Structure

To deliver an adequate adsorbent-based therapy, some important requirements must be fulfilled: (1) an effective, biocompatible, and safe sorbent material; (2) a sorbent cartridge with adequate design and structure; and (3) operating conditions allowing for optimal utilization of the available surface of the sorbent.

Sorbents are present in nature as raw materials, or they can be produced synthetically in the laboratory. Natural sorbents such as zeolites (aluminum silicates) are inorganic polymers with remarkable porosity, deriving from their crystal structure, and can be modified synthetically to control the structure of the internal pore system. Other typical sorbents such as porous carbons are cellulose-derived organic polymers prepared by controlled thermal oxidation ( Fig. 189.1 ).

Different polymers of synthetic origin constitute the other class of sorbents. Almost all monomers susceptible to cross-linking can be transformed into large polymeric molecules via a multitude of reactions. Bifunctional monomers tend to aggregate in linear polymeric structures, whereas highly functional monomers tend to polymerize in cross-linked structures. Divinyl-benzene is a potent cross-linker frequently used to build polymeric sorbent molecules. Sorbent polymers also can be functionalized with chemical compounds to target specific molecules for adsorption ( Fig. 189.2 ).

Sorbents exist in granules, spheres, fibers, cylindric pellets, flakes, and powder. They are solid particles with single-particle diameters generally ranging from 50 µm to 1.2 cm. Surface area to volume ratio (S/V) is extremely high in sorbent particles with an effective surface area varying from 300 to 1200 m ² /g. They also are classified according to the size of the pores of the inner structure: (1) macroporous: pore size >500 Å (50 nm); (2) mesoporous: pore size 20-500 Å; and (3) microporous: pore size <20 Å.

The S/V generally is described by the following equation:

where d _p = pore diameter and L = pore length. Considering fractional particle porosity (ɛ _p ) and particle density (ρ _p ), the specific surface area per unit of mass (S _g ) is:

As an example of a clinically realistic application, if ɛ _p = 0.5, ρ _p = 1 g/cm ³ (1 × 106 g/m ³ ), and d _p = 20 Å (20 × 10 ⁻¹⁰ m), S _g = 1000 m ² /g. In other words, 1 g of sorbent material provides a potential surface for adsorption of 1000 m ² . Frequently, however, the available surface is not used fully because many factors contribute to limit the fraction of surface truly available for adsorption.

Requirements for a Sorbent

Sorbent material must have high selectivity/affinity with the capacity to enable sharp separation and minimize the amount of sorbent required to make a suitable commercial product. The sorbent should have favorable kinetics and transport properties for rapid adsorption of target solutes, chemical and thermal stability, low solubility in the contacting fluid, and high mechanical strength to prevent crushing or erosion.

In a sorbent cartridge used for clinical purposes, the material must allow free flow of blood or plasma (fluid phase) and easy filling and emptying of the packed bed. Other requirements are high resistance to fouling to permit long cartridge life span and maximal biocompatibility with no tendency to promote undesirable chemical reactions or side effects. In addition, the sorbent must be cost effective. The possibility of regeneration should be explored to allow possible reuse. Unwanted losses resulting from adsorption of hormones, proteins, and drugs must be characterized and addressed as potential side effects. An adequate regime of anticoagulation also should be defined to prevent clotting or platelet loss in case of direct contact with blood. All these characteristics are tested carefully in vitro and in animals before approval of use in humans to ensure maximal safety of application in clinical settings.

Mechanism of Solute Adsorption in Porous Media

Different steps and mechanisms can be identified in the process of solute adsorption onto a porous material:

• 1.

  External (interphase) mass transfer of the solute by convection from the bulk fluid and then by diffusion through a thin film or boundary layer, to the outer surface of the sorbent

• 2.

  Internal (intraphase) mass transfer of the solute primarily by diffusion from the outer surface of the sorbent into the internal porous structure

• 3.

  Surface diffusion along the surface of the internal pores

• 4.

  Adsorption of the solute onto the porous surface ( Fig. 189.3 )

  

The adsorption mechanism involves physiochemical forces of different nature. The overall rate of solute removal is usually controlled by step 2 or 4.

The interphase mass transfer is a crucial step because it brings the solution (fluid phase) and the molecules to be removed in contact with the sorbent. The cartridge in which the sorbent is contained must promote uniform distribution of internal flow of the fluid phase (plasma or whole blood). Uniform flow distribution profiles are obtained generally using granules or spherical beads of equal size. Packing density between 40% and 60% is considered optimal to prevent preferential channeling of the flow with undesired loss of performance. Any type of channeling phenomenon may affect the quantity of solute adsorbed per unit of sorbent and influence the saturation process of the unit.

Because blood is a non-Newtonian fluid, accurate analysis of the flow distribution in different conditions of flow and viscosity should be made. Flow distribution in packed beds can be modeled theoretically using equations of physical chemistry and transport. The packing structure is usually complex, and the resulting flow pattern is complicated. There are tortuous paths through the interstitial space of the bed, which consists of channels (pores) of various diameters (interparticle porosity). The packed bed can simulate a bundle of tortuous capillary tubes. In well-packed beds with relatively constant interparticle porosity, the variation of flow velocity among individual channels is relatively small. However, if packing is not homogeneous, channels of different size can be present with significant variation of fluid phase velocity, leading potentially to blood stagnation as a result of high resistance in areas with small-diameter channels and consequent clotting. On the other hand, areas having large diameter channels offer relatively little resistance to flow and the undesirable phenomena of preferential flow channeling may result, with poor utilization of the sorbent potential, reduction in adsorption performance, and rapid saturation of the unit.

Physical laws and equations governing flow distribution in packed beds go beyond the scope of this chapter. Nevertheless, for the benefit of the reader, a quick summary of governing laws is included in the following section.

Flow Distribution in Packed Beds

The fundamental principle governing the flow of fluids through packed beds is Darcy's law , which states that the flow velocity is directly proportional to the pressure gradient and the specific permeability coefficient of a unit whereas it is inversely proportional to the viscosity of the fluid phase and the length of the conduit:

In this expression, P _i and P _o are the pressures at the inlet and at the outlet of the cartridge, η is the viscosity, L is the length of the conduit, B ^o is the specific permeability coefficient, and v _o is the superficial velocity (the average linear velocity the fluid would have in the cartridge if no packing were present). It is calculated by dividing the flow rate by the cross-sectional area of the empty cartridge (specific permeability coefficient for open tubes is equal to r ² /8).

The component of the cross section of the bed available for flow is expressed by the interparticle porosity (ɛ). Random packing of equal-size particles usually results in ɛ = 0.4 ± 0.03. The total porosity of beds packed with porous particles is of course larger because of the intraparticle porosity that allows some flow through the particles. The true average fluid velocity (v) is obtained, from Eq. 3 as

The dimension of the specific permeability B ^o is cm ² , but it is also given in Darcy units (1 Darcy = 10 ^–8 cm ² ).

The hydraulic radius concept is frequently used to calculate flow through channels of different geometry. The hydraulic radius r _h is defined in the following way:

and the average flow velocity (v) is expressed as

Several equations have been derived to relate the specific permeability to the particle diameter and the bed porosity. The best-known expression is the Kozeny-Carman equation, which gives the specific permeability as

where d _p is the particle diameter. The average fluid velocity is then given by

This equation is valid for laminar flow and for beds having porosity less than 0.5.

For packed beds the Reynolds number (Re) is calculated with particle diameter substituted for tube diameter:

where Re is the dimensionless Reynolds number, v is the fluid velocity (cm/sec), ρ is the fluid density (g/cm ³ ), d _p is the particle diameter (cm), and η is the fluid viscosity (poise).

Turbulence and transition from laminar to turbulent flow are not nearly as well defined in packed beds as in open tubes. It is assumed that turbulence in packed beds develops gradually as Re increases from 1 to 100. Actually, even at low Reynolds numbers, in packed tubes there is a lateral movement of the fluid elements because of stream splitting. At high flow velocities, this leads to a substantial “convective diffusivity,” analogous to the eddy diffusivity in turbulent flow. The flow profile then can be approximated as plug flow.

The most uniform flow profile can be obtained when beds are packed carefully with spheric particles of equal size. If the ratio of the tube diameter to the particle diameter is less than 100, this ratio may have a significant effect on the flow profile.

In commercial cartridges, the tube diameter to particle diameter ratio is far from the above-mentioned ranges; cartridge and particle diameters around 5 cm and 1000 microns, respectively, are common. In some experimental analysis, the flow observed is close to optimal and easily can be assimilated to a plug flow with absence of channeling phenomena ( Fig. 189.4 ). This results in straightforward calculation of the saturation time and the maximal solute removal per unit of sorbent. From these data, the optimal amount of sorbent used in one unit can be calculated according to the treatment duration, the average concentration of the solute at the beginning of the session, and the volume of distribution of the solute in the body.

The internal mass transfer (intraphase) can be seen as a primarily convective transport of the solute through the structure of the sorbent resulting from flow of the fluid phase inside the sorbent particle. This once again depends on the packing density, the pressure gradient, and the permeability coefficient of the particle. Often this mechanism is far from being optimized and the sorbent is generally used only in minimal part because of insufficient permeation of the bulk solution into the structure of the particle.

The physical-chemical mechanisms regulating surface adsorption are multiple. Once the molecule is brought to the surface of the sorbent, different chemical and physical mechanisms are involved:

• van der Waals forces are generated by the interaction between electrons of one molecule and the nucleus of another molecule; these are weak and generally reversible.

• Ionic bonds are generated by electrostatic attraction between positively charged and negatively charged ions; these are typical of exchange ion resins.

• Hydrophobic bonds represent strong binding forces generated by the hydrophobic affinity of the sorbent and the solute molecules ( Fig. 189.5 ).

  

Efficiency of Adsorption

Porous polymers can be designed and constructed with different internal surface selectivity and various pore sizes. As a consequence, mass separation can be based upon size, geometry, and individual binding properties. To achieve a selective or partially selective adsorption process, the clinician needs to know the properties of the molecules to be separated or removed. If the information is lacking, the properties of the molecules under analysis can be ascertained by combining a number of available analytic measurements to develop a better understanding of the distinctive molecular pattern. A more empiric attempt can be made by trial and error through adsorption isotherms ( Fig. 189.6 ).

When a liquid mixture is brought into contact with a microporous solid, adsorption of certain components in the mixture takes place on the internal surface of the solid. The maximum extent of adsorption occurs when equilibrium is reached and no further net adsorption occurs. No theory for predicting adsorption curves is embraced universally. Instead, laboratory experiments must be performed at fixed temperature for each liquid mixture and adsorbent, to provide data for adsorption isotherms curves. (Separation processes are energy intensive and affect entropy.) Adsorption isotherms can be used to determine the amount of adsorbent required to remove a given amount of solute from the solvent (within a specific unit) at equilibrium.

Another measure of the efficiency of the unit is obtained by using marker molecules to determine the “mass transfer zones” at different times inside the unit. The mass transfer zone is the portion of the cartridge length that extends from the point at which the sorbent is fully saturated to the point at which the sorbent is completely unsaturated (no solute on the particles). Figs. 189.7 and 189.8 describe graphically the concept of mass transfer zone and different possible profiles occurring in a sorbent unit. Mass transfer zone determination also helps to define the quality of design and performance of the unit along with the expected life span before saturation. The mass transfer zone is a function of packing density and unit design. Poor design and inadequate packing density result in mass transfer zones exceeding the length of the unit and are characterized by a flow-through condition (“breakthrough” of the solute in the fluid phase leaving the unit) even at the beginning of the treatment.

Biocompatibility of Sorbents

The biocompatibility of a system using sorbents for extracorporeal therapies should be studied considering different aspects. First, the sorbent must be resistant and have sufficient mechanical strength to prevent cracking of the solid component, with release of microparticles and fragments to the systemic circulation. To further prevent this unwanted effect, cartridges are provided with a screen that allows free passage of blood but retains particles or their fragments ( Fig. 189.9 ). A derivative measure of biocompatibility is given in clinical practice by continuous measure of end-to-end pressure drop in the unit throughout the treatment. Fouling of the screens resulting from cell or albumin adhesion may result in increased resistance to flow and thus in increased pressure drop inside the cartridge. Accelerated clotting of the unit will also cause a sudden increase in end-to-end pressure drop.

The second aspect is the intrinsic structure of the sorbent material. The inner surface of the sorbent should be compatible with blood to avoid cell and protein deposition that may occupy the adsorption sites and impair the sorbent capacity ( Fig. 189.10 ). When the material is intended for direct contact with blood, biocompatibility should be directed further toward preventing unwanted reactions in circulating blood (from complement activation to cytokine release), leukopenia, thrombocytopenia, development of antibodies, and significant adsorption of albumin. All these effects can be mitigated by coating the surface of the granules or the fibers with a biocompatible material such as polysulfone. In this case, however, the coating may render the sorbent less efficient because the intraphase component of the transport may be negatively affected ( Fig. 189.11 ). The coating acts as a size exclusion barrier and prevents larger solutes from reaching the intraparticle site of adsorption. The mass transfer considerations for the removal of cytochrome C by a hypothetical coated sorbent is shown in Fig. 189.12 .

To obviate the need for coating the sorbent, some techniques separate plasma from cells and circulate cell-free plasma through the sorbent bed, avoiding direct contact with cells. Downstream in the circuit, blood is reconstituted by mixing purified plasma with cells. In some cases, only plasma water ultrafiltrate is regenerated by exposing it to the sorbent bed and subsequently reinfusing it downstream into the circuit.