The Research And Implementation Of TSV Open Test Algorithm In 3D SRAM Based On Fault Primitive

In 3D-IC technology based 3D SRAM, the large number of TSVs implemented in circuits. The manufacturing process of TSV is not mature yet which makes TSVs becoming more prone to open defects, brings new challenges to the 3D SRAM test. On the one hand, the existing 2D MBIST method has the ability to detect the faults in 3D SRAM. However, this method cannot tell whether it's derived from TSV or SRAM itself. On the other hand, although the specific circuit of TSV can find out where the faults are, it need extra specific circuit to implement which increases both the area overhead and its design complexity. In consideration of what discussed above, this paper proposes a new idea to detect the open defects of TSV based on a test algorithm. It is an effective method to detect the open defects of TSV in 3D SRAM without any additional testing circuit and extra overhead. The main content of this dissertation includes the following items:1, Built a Electrical model of 3D SRAM. The existing 3D SRAM architecture are analyzed and summarized, we built a Electrical model of 3D SRAM based on the memory array is divided into many small pieces and stacked in 3D fashion, TSVs are implemented as the bitlines and worldliness and used to link these memory arrays from different layer to peripheral logic in bottom layer in this model.2, Failure probability of memory caused by TSV open fault. According to the fabrication process, characteristic and failure mechanism of TSV, we built a malfunction coupling model of TSV to analyze the impact of memory from TSV open fault. Then, use Hspice to assess the failure probability of memory caused by TSV open fault. Results show that circuit malfunction will be happen when the TSV exist open fault, leading to a read/write error; When considering PVT influence, the max failure probability of memory will be reach to 100%.3, The model extraction of TSV open malfunction. We does a simulation experiment using Hspice under different type of memory operation, get a faulty behavior during TSV of word-lines and bit-lines exist open fault. Then according to the fault behavior and the characteristics of memory malfunction, the physical faults caused by TSV open defects are map into SRAM malfunction, obtaining six different functional fault models. We can obtain the corresponding test sequences through these malfunction model. It is an effective method of test and solve the open defects of TSVs.4, Propose a new test algorithm-- TSV open test algorithms. In this paper, we use the fault primitives and the existing 2D Memory test algorithm based on theoretical analysis to obtain the simplest March test sequences corresponding to these six malfunction models. Then get a TSV open test algorithms after merge and optimization of the algorithm on the basis of the simplest March elements, used to cover the TSV open function failure. The complexity of the optimized algorithm decreased from 28 N to 9N, reduced to 40%, the complexity of the algorithm(9N) compared with traditional 2D March algotithm(14N) greatly reduced the test time. It is an effective method of test and solves the problem for open defects of TSVs.5. In order to verify the functional correctness and effectiveness of the TSV test algotithm, the simulation also done using NC-verilog and Nanosim. The results show that there is no functional problem with the algorithm and realized the purpose of TSV open fault detection.Compared with the TSV test method using the test circuit, the proposed method can use the traditional 2D BIST structure and without any additional test circuit.The results show that, the method proposed in this paper is simple and realized the purpose of TSV open fault detection by using the traditional BIST structure, so this method can be used in project.