Cost structure in semiconductor manufacturing - A transparent view for cost engineers

Cost structure in semiconductor manufacturing - A transparent view for cost engineers

The increasing complexity and miniaturization of semiconductor components has not only increased the technological requirements, but also the need for precise cost calculation. It is becoming increasingly important for cost engineers to transparently understand and evaluate the key cost drivers along the entire value chain of IC manufacturing ("semiconductor manufacturing cost").

The manufacturing process for integrated circuits (ICs) is divided into three main phases: Wafer production, front-end and back-end processes. While the front-end processes - especially lithography - account for the majority of production costs, the correct allocation of overheads also plays a key role in realistic overall costing.

Wafer production begins with the Czochralski process for manufacturing monocrystalline silicon ingots. After cutting, grinding and cleaning, these are turned into usable wafers, which form the basis for theIC. Even here, the costs vary considerably - depending on the diameter and purity of the wafer as well as the desired technology node.

In the front-end process, the circuits are applied to the wafer. This takes place in several steps: oxidation, coating, exposure (lithography), etching, doping and polishing. Lithography is the most cost-intensive single process, especially when EUV technologies for advanced nodes (below 7 nm) are used. A singleEUV exposure device can cost up to 300 million US dollars and therefore represents a significant investment. Accordingly, machine depreciation is a dominant cost item that cost engineers must carefully consider in their calculations.

Another key aspect in the cost analysis is the inclusion of material costs. These include not only the silicon wafers, but also chemical process materials, gases, photoresists and mask sets, which become more complex and expensive with increasing structure depth - in the case of state-of-the-art nodes, they can cost several million dollars per set.

The back end includes separating the chips, applying them to a substrate, the electrical connection via wire bonding and the final encapsulation. Even if this section only accounts for around 20 % of the total costs, it should not be underestimated - especially in terms of material consumption and labor costs.

The cost approach developed is based on a bottom-up calculation of a 32-bit microcontroller with BGA-416 housing with an annual production of 10 million units. The analysis shows that front-end processes, including wafer costs, account for around 80% of total production costs. Within this 80 %, lithography alone accounts for around 25 %, while a further 35 % of total costs are classified as overheads - including IT, infrastructure, energy, qualified personnel and cleaning processes. These expenses are heavily dependent on the location and process, which is why benchmarks from comparable high-tech industries such as aviation or chemicals are often used as an approximation.

For a valid cost estimate, capital and operating costs, material consumption, cycle times and throughput must therefore be analyzed in a differentiated manner. Cost engineers should always take into account the specific framework conditions such as production location, technology nodes and volumes in order to obtain robust and reliable results. This is the only way to make well-founded decisions in purchasing, price negotiations and strategic procurement.

📄 C lick here for the full white paper: White Paper: Cost Calculation of Integrated Circuits (PDF)

‍