Carbon nanotubes (CNTs) are the strongest materials ever discovered by mankind. With measured tensile strengths as high as 150 GPa and a high modulus of 1 TPa, CNTs are very promising for many high-strength, lightweight applications. Several approaches have been reported to process micrometer-long CNTs into fibers, including dispersing CNTs into polymer or acid solutions and then spinning them into composite or pure CNT fibers. Unfortunately, difficulties in CNT dispersion and alignment have hindered the full utilization of CNT strength. Recently, CNT yarns have been directly spun from CNT arrays. This new approach has the advantage of good CNT alignment and high CNT volume fraction, and therefore is very promising for producing higher strength CNTyarns and fibers. So far, CNT fibers can only be spun from CNTarrays that are less than 300 lm long, and this has limited their tensile strengths to only a few hundred MPa. Therefore, it is critical to synthesize spinnable long CNT arrays so that higher strength can be obtained in the spun CNT fibers. Long active growth is the key to growing long CNT arrays. However, during CVD processes using thin Fe catalyst films, the catalysts usually become inactive within several minutes due to either interdiffusion with the substrate or the accumulation of amorphous carbon. In order to achieve sustained growth for long CNT arrays, catalysts like ferrocene have been added continuously or intermittently into the system to nucleate new catalyst particles during the growth. This makes individual CNTs much shorter than the array height. In addition, CNTs synthesized in this way are often thick and dirty with undesirable excessive catalytic particles and amorphous carbon. Alternatively, sustained growth was realized by introducing a thin buffer layer, either Al2O3 [18,19]