© 2003 Robert A. Freitas Jr. All Rights Reserved.

Robert A. Freitas Jr., Nanomedicine, Volume IIA: Biocompatibility, Landes Bioscience, Georgetown, TX, 2003

15.3.1.1 Protein Adsorption on Diamond Surfaces

The first direct study of protein adsorption on diamond, done by Tang et al [521] in 1995, focused on fibrinogen (Section 15.2.2). Fibrinogen, a 340-kilodalton soluble plasma glycoprotein ~47.5 nm in length [526], is the major surface protein to initiate coagulation [518, 527-529] via platelet adhesion to fibrinogen, and inflammation including fibrosis [523-525] around implanted biomaterials. The adsorption and conformational state (“denaturation”) of fibrinogen is commonly used as a biocompatibility indicator [530]. The amounts of “denatured” fibrinogen accumulated on surfaces correlates closely with the extent of biomaterial-mediated inflammation [531].

Tang and colleagues [521] measured ~3.7 mg/m² (~6600 molecules/micron², or ~50% surface coverage) gross surface adsorption of human fibrinogen on chemical-vapor-deposited (CVD) diamond surfaces, after incubation of the plasma-coated diamond surface in a 20 µg/cm³ fibrinogen solution (~0.1% of blood concentration; Appendix B) for 8 hours at room temperature. Much of this adsorbed fibrinogen was only loosely bound, however. A solution of sodium dodecyl sulfate was rinsed over the incubated CVD surface to remove the loosely-bound or elutable (non-denatured) fibrinogen. (Sodium dodecyl sulfate is an anionic detergent commonly used to solubilize proteins, e.g., a surfactant creating negative surface adhesion energy; Section 9.2.3. Although SDS is of course unavailable to wash biomaterials once implanted, the wash results nevertheless indicate the extent to which loosely bound proteins will eventually detach.) After the rinse, ~44% of the bound fibrinogen molecules detached, leaving ~2.1 mg/m² (~3700 molecules/micron²) of spontaneously denatured fibrinogen still present on the CVD diamond surface.

CVD diamond [532-535] might not accurately represent the atomically-smooth flawless diamond surfaces which may characterize the typical MNT-manufactured medical nanorobot exterior. Far from being atomically smooth, CVD diamond films are amorphous and polycrystalline [537], often with grain sizes up to 1-10 microns [535, 536]. In Tang’s experiment [521], diamond wafers with two distinct sides were tested, as follows.

The nucleation side of the diamond wafers was grown in contact with a flat silicon substrate, which was then dissolved away by acid. The formation of SiC on such a substrate allows silicon to bond well with carbon during the growth process [537]. However, the presence of small amounts of surviving carbide in the nucleation diamond surface, or of concave nanoscale surface features recording the removal of SiC by etchant, could markedly alter the protein adsorbent characteristics of the diamond surface at the molecular level. Also, SiC is tolerated by cells up to 0.1 mg/cm³ concentration but is cytotoxic at 1 mg/cm³ [538]. Furthermore, a contact profilometer measured the nucleation surface as having a rugosity of up to 250 nm, a roughness 100-1000 times greater than that which may be expected at the surfaces of the typical diamondoid medical nanodevice.

The growth side of the diamond wafers used in Tang’s experiment was even rougher than the nucleation surfaces, so this surface was ground and polished but only to a rugosity of ~1 micron peak-to-valley. This is approximately the diameter of an entire bloodborne medical nanorobot and clearly not representative of an atomically-precise engineered medical nanodevice surface. There is no indication whether the grinding and polishing of the growth surface was done under oil (thus preserving a predominantly hydrogen-terminated, hence strongly hydrophobic, surface [539]), nor was there any evaluation of whether subsequent etching with H₂SO₄ and H₂O₂ might have produced carbonyl and hydroxyl conversions at the surface (thus possibly creating regions of hydrophilicity). Furthermore, diamond crystals are believed to polish by successive repeated microcleavage along (111) cystallographic planes, which is why polishing is much easier in some directions than others [539]. Non-(111) surfaces, when mechanically polished, will always be rough and will consist of small domains of (111) surface canted at appropriate angles to the macroscopic orientation [539]. Residual asperities of ~5 nm have been reported even for exceptionally carefully polished surfaces [540]. The general conclusion is that the chemical and mechanical processes used in Tang’s experiment seem unlikely to have produced a surface that is well characterized at the molecular level. Protein adhesion to near-atomically smooth diamond surfaces remains to be investigated experimentally.

Still, we can hypothesize that completely fibrinogenophobic surfaces might be engineered using atomically-smooth diamondoid materials, keeping in mind the important role of hydrophobic forces in surface denaturation (e.g., see Section 15.3.4.1). To do this will require a thorough molecular-level understanding (by 2002, not yet complete) of the adhesion and conformational properties of fibrinogen, as summarized below and in Section 15.2.2.

It has long been known that fibrinogen preferentially adsorbs on a hydrophobic surface, and albumin on a hydrophilic surface, during competitive binding [541]. One experiment [542] found 10,800 molecules/micron² of fibrinogen (~complete monolayer coverage) adsorbed on a hydrophobicized quartz surface (contact angle ~70^o) after 30 seconds incubation with a 2.9-µM fibrinogen solution (~30% of physiological in human serum; Appendix B), but only 8400 molecules/micron² after a 60-second exposure of a hydrophilicized silicon surface (contact angle ~28^o) to the same solution. Hydrophobic surfaces generally have higher adsorbence of adhesion proteins such as complement C3, fibronectin, and vitronectin, while hydrophilic surfaces have higher adsorption of albumin and immunoglobulin IgG [543]. (The fibrinogen molecule’s own surface properties are very hydrophilic, changing to moderately hydrophobic as it converts to fibrin during coagulation [544].) Fluorocarbon films (very hydrophobic) generally show high protein retention [1113-1114] (Section 15.3.4.1).

Surface functionality has been shown to influence protein-surface interactions [1111-1113]. Tang et al [531] found that surfaces having high concentrations of specific surface functionalities including -OH (hydrophilic), -NH₂ (hydrophilic), -CF₃ (hydrophobic), and siloxyl groups (hydrophobic) exhibited significant differences in both the adsorption and “denaturation” of adsorbed fibrinogen. But hydrophobicity alone did not dictate fibrinogen-surface interactions on these surfaces. Soaking in saline solution for 15 days increased oxygen incorporation in the -NH₂, siloxyl, and CF₃ rich films, and slightly decreased the oxygen content in the -OH rich films. After this soaking, the two hydrophilic films became somewhat less hydrophilic whereas the two hydrophobic films became somewhat more wettable [531]. These kinds of changes may be important for medical nanorobots expecting to remain in vivo for extended times. Rapoza and Horbett [545] observed rapid denaturation transitions, requiring <2 hours after adsorption on hydrophobic polymers containing no oxygen. More gradual conformation changes, occurring only after a time lag of 1-4 hours, were seen on hydrophobic polymers containing oxygen. Note that the existing literature discusses surface changes in response to short-term exposures. Little is known of the effects of long-term exposures lasting months or even years. Such exposures and their consequences should also be investigated because permanently implanted nanorobotic organs and nanorobots used for surveillance or early detection of disease could have very long mission durations.

Hydrophobicity and surface functionalities are accessible parameters well within the reach of diamondoid medical nanorobotic design. In general, the high surface energy of natural diamond makes it extremely hydrophobic [547-549, 658]. Yoder [550] reports that ocean barnacles do not attach to diamond. However, a diamond surface may have any of several different crystal planes exposed. These planes may be passivated with any of a number of passivating atoms or molecules, all of which may affect the hydrophobicity of the surface. For example, a hydrogen-terminated (111) crystallographic surface with each H bonded to a single C looks like a hydrocarbon (e.g., like oil) and is not wetted [539]. On the other hand, oxygenation of a diamond surface (e.g., by heating to >250 ^oC in an O₂ atmosphere, or by ion bombardment) promotes formation of hydrophilic surface groups [552] with a complicated surface chemistry [553, 554], including a significant proportion of carbonyl (C=O) groups [554]. The outer faces of natural hydrophobic diamond may be terminated partly by hydrogen and partly by bridging oxygen (C-O-C) [539]. Aging such surfaces in water for several weeks can change them to hydrophilic behavior [554], possibly indicating conversion to hydroxyl (-OH) groups. Hansen [551, 552] suggests that the small amount of oxygen on the atmospherically equilibrated polished surface is present largely as -OH groups because surface wettability appears insensitive to pH values below pH 11. Fe⁺⁺⁺ (and Al⁺⁺⁺ to a much lesser extent) can form surface complexes with these hydroxyl groups, whereas Na⁺, Ca⁺⁺, and Cr⁺⁺⁺ cannot. Complete atomic-scale positional and compositional control of the diamond-passivation layer on the exteriors of medical nanorobots should permit the engineering of adhesion-protein-selective surfaces of appropriate hydrophobicity, hydrophilicity, or adsorptivity (Section 15.2.2). For instance, it is well known that hydrophobic surfaces can be progressively hydrophilized via selective ion implantation [939] or RF plasma treatment [1111-1113]. Hosotani [555] has created an intraocular lens coated with diamond-like carbon film that is more hydrophobic and oleophilic than an uncoated lens.

Under aqueous conditions, surface-dependent structural deformation or spreading of the adsorbed fibrinogen molecule is larger on positively-charged surfaces than on negatively-charged implant surfaces – specifically, the molecular length and the D and E domain widths of fibrinogen are increased, while the corresponding molecular heights are decreased [556]. As an insulator, diamond in water generally does not present a highly charged surface [633, 639, 640]. But surface charge also lies within the purview of diamondoid nanorobot design (Section 15.5.6). For example, a hydrogen-terminated diamond surface has negative electron affinity [557, 558]. Since hydrogen is less electronegative [559] than carbon, the surface externally appears to be a weak array of positive point charges* at molecular contact distances, arising from surface dipoles as polarized covalent bonds would be expected to produce [558]. On the other hand, the oxygenated diamond (100) surface has a positive electron affinity [560] as does the fluorinated (111) surface. (Methods for selectively fluorinating diamond surfaces have also been investigated [561].) Both oxygen and fluorine are more electronegative [559] than carbon, so either surface would externally appear to be an array of negative point charges at molecular contact distances. The controlled coating of a diamond surface with atoms with different electronegativity might provide fine control of the electron affinity while maintaining chemical inertness of the diamond surface. Diamond glucose sensors employed in diabetic blood analysis use heavily boron-doped diamond as one electrode. On the other electrode, glucose oxidase enzyme (a protein) is immobilized on the diamond surface by electrochemical deposition, or is “wired” directly to the diamond electrode by covalent bonding to the electrode surface [655].

* E. Pinkhassik points out that the difference in electronegativity between carbon and hydrogen is very small (2.5 and 2.1, respectively, on the Pauling scale, or 2.746 and 2.592, respectively, on the more modern Sanderson scale [5843]), while the difference between carbon and oxygen, or between carbon and fluorine, is much larger – e.g., 2.5/2.746 for C, 3.5/3.654 for O, and 4/4.0 for F. The C-H dipole is very weak, so a surface coated with C-H groups with carbon being in sp³ hybridization can hardly be considered an array of positive point charges. In fact, the polarizability of the C-H bond may be more significant than the partial point charge. On the other hand, C-O and C-F dipoles are strong, and the partial point negative charges are likely to be larger.

Other blood proteins also must be tested for their adsorptivity to diamond. For example, one study [583, 4723] of protein adsorption on diamond-like carbon (DLC) coatings (Section 15.3.1.2) found that DLC exposed to bovine serum albumin (BSA) at a concentration of 5 mg/cm³ (~10% of physiological for human serum albumin) adsorbs 20 times less BSA than SiO₂ or TiO₂ control surfaces. Phytis L.D.A., the sponsor of this study [583, 4723] and the only manufacturer of a DLC-coated stent, claims that “diamond-like coated surfaces showed only minimal adhesion of proteins at the surface; those adhesions are reversible and do not result in denaturations of protein.”

Last updated on 30 April 2004