Redox reactions of CNTs are attributed to metal impurities and the structure of carbon lattice. First, metal impurities are brought by synthesizing catalysts that are transition metals, such as Fe, Ni, Co. Mo, and so on. Among them, iron will be discussed as a typical model. A good method to remove metal impurities from CNTs is a wash in concentrated nitric acid at 350°C . A strong acid wash may alter the surface of carbon nanotubes, modifying surface reactivity. Thus, it is best to avoid strong acid washing that would damage CNTs. Interestingly, Guo et al.  reported that Fe2O3 and FeC encapsulated by carbons cannot be mobilized nor be bioavailable to an acid wash. Furthermore, Liu et al.  discussed that it was not necessary to remove metal impurities from CNTs as long as those metals were enclosed and not bioavailable. Thus, it is possible to manage iron impurities by an appropriate acid wash, and bioactivity will be reduced as long as remaining impurities are not bioavailable.
To various degrees, transition metal impurities are usually oxidative to peroxides while metal oxides are relatively stable. It is known that Fe (II) or Fe2+ ion generates OH radicals (OH•), a form of reactive oxygen species (ROS), by the Fenton reaction, and that ROS induce inflammation of tissues. As Fe3+ generated by the oxidation is again reduced to Fe2+ with peroxide, iron can continue to cause oxidant-induced inflammation in living tissues. In contrast, Fe (III) oxide (Fe2O3) and carbide (FeC) do not generate ROS, because Fe (III) cannot be an electron donor except upon treatment with a strong reduction agent. Fe (II) is supplied not only externally as metal impurities but also internally in a living body. Since Fe (II) essentially catalyzes peroxide generating OH•, redox (reduction) reactions are required to eliminate the radicals. A question is what redox potential do CNTs exhibit.
Redox potential determines the reaction tendency of chemicals in a system. CNTs inevitably have dangling bonds at which unpaired electrons can be easily exchanged with the other species. If those dangling bonds donate electrons to OH•, CNTs become ROS quenchers. According to recent findings [8–11], CNTs scavenge ROS so that those dangling bonds work as electron donors. In other words, CNTs can potentially reduce OH•. If this redox potential hypothesis is true for CNTs, they may decrease oxidant-induced inflammation of tissues. One should be able to stoichiometrically predict oxidant stress once redox potential of CNTs in a reaction system is determined. This hypothesis might apply for the other nanomaterials as well. Furthermore, a systematic protocol based on chemical reaction kinetics can be applied to develop predictive in vitro assays of redox potential for various types of CNTs. Accumulating those redox reaction data, guidance to design safer CNTs can be developed. In conclusion, redox potential might be a useful tool to estimate ROS generation and bioactivity with CNTs. To utilize it, we have to investigate redox potentials of biological systems further and identify the role of oxidant stress in the toxicity of CNTs.