Doping is a ubiquitous and powerful tool used to alter a material’s electronic properties. Conventional Si solar cells would not exist without it, and it has been used to increase the efficiency of organic and inorganic photovoltaics alike. While doped carbon nanotubes (CNTs) have been the subject of numerous spectroscopic studies, few investigations have examined their performance in photovoltaic devices. Here, we incorporate chemically p-doped CNTs into photovoltaic heterojunctions with C60. In a conventional CNT–C60 heterojunction device, photocurrent can be produced when photogenerated excitons in the CNT layer diffuse to the CNT–C60 heterointerface. The offset in conduction band energy drives dissociation of these excitons via photoelectron transfer from CNT to C60, which yields separated charges that can be collected. Our objective here was to determine the following: 1) does p-doping increase or decrease the efficiency with which excitons dissociate into electrons and holes? 2) Do p-dopants trap excitons and inhibit their diffusion to the heterointerface? 3) How efficiently can a photogenerated trion (exciton coupled to an injected hole) be dissociated to produce a photocurrent?To answer these questions, we fabricated polymer wrapped (6,5) CNT–C60 heterojunctions over a range of CNT thickness with various levels of p-doping introduced by triethyloxonium hexachloroantimonate treatment. We measured photocurrent to quantify relative photoelectron transfer efficiency for both excitons and trions and compared this to photoluminescence and transient absorbance measurements on doped CNT films. We found that p-doping decreases photocurrent because injected holes bleach the CNTs’ S11 transition and decrease the amount of light absorbed. Meanwhile, the photogenerated exciton to collected electron transfer efficiency remains unchanged for doping densities up to 50μm-1. In contrast, photoluminescence quantum yield drops by over 90% at an equivalent doping density. We attribute this disparity to the fast electron transfer at the CNT–C60 interface, which occurs on the order of femtoseconds. We also found that moderate doping can promote photocurrent at the trion peak, with a photogenerated trion to collected electron transfer efficiency about 1/4 that of the photogenerated exciton to collected electron transfer efficiency. Photocurrent at the trion absorption peak is invariant with reverse bias up to 0.3 V, despite the charged nature of the trion, supporting evidence for the immobile nature of trions. These results indicate that p-CNTs are a viable material for use in photovoltaics, albeit with weaker absorption, and that photocurrent from trions is possible, although with lower quantum efficiency. Figure 1
Read full abstract