Currently, the electricity to run cloud computers is usually generated from fossil fuels (e.g., petroleum, natural gas), which will cause carbon pollution. Therefore, data centers, as places to accommodate cloud computers, are now facing a serious problem of high carbon pollution. In this letter, an operational method is proposed to achieve zero-carbon data centers by carefully matching delay tolerant tasks to computing resources (e.g., CPU) when zero-carbon electricity (wind and solar energy) is available. We designed a unified measurement called CPU×Time, by which the complex matching problem involving tasks, computing resources, and zero-carbon electricity is simplified into a bin-packing scheduling. Simulations show that the proposed bin-packing scheduling method can achieve high resource utilization without carbon pollution.