892 IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, VOL. 24, NO. 6, JUNE 2005 New Data-Background Sequences and Their Industrial Evaluation for Word-Oriented Random-Access Memories Said Hamdioui, Member, IEEE, and John Eleazar Q. Delos Reyes Abstract—This paper improves upon the state of the art in the testing of intraword coupling faults (CFs) in word-oriented mem- ories. It first presents a complete set of fault models for intraword CFs. Then, it establishes the data background sequence and tests for each intraword CF, as well as a test with complete fault coverage of the targeted faults. All introduced tests will be evaluated indus- trially, together with the most well-known memory tests. The tests will be applied to big arrays with an interleaved bit-organization as well as to small arrays with an adjacent bit-organization in order to investigate the influence of the memory organization on the intra- word CFs. The test results show that the intraword CFs are also significantly important for interleaved memories, even when the cells within a single cell are not physically adjacent. This is due to coupling between the adjacent bit lines and word lines running across the memory array. The paper concludes that intraword CFs should be considered for any serious test purpose or leave substan- tial defects undetected, especially when considering a high-volume production and a very low defect-per-million (DPM) level. Index Terms—Bit-oriented memories (BOMs), data back- grounds (DBs), fault models (FMs), memory tests, word-oriented memories (WOMs). I. INTRODUCTION R ANDOM-ACCESS memories can be organized as bit-ori- ented memories (BOMs) or as word-oriented memories (WOMs). WOMs contain more than one bit per addressable word, i.e., , whereby represents the number of bits per word, and is usually a power of 2. Read operations read bits simultaneously, while write operations write data bits simultaneously; where the data to be written in each cell can be specified independently from the data for the other cells. Traditionally, WOMs have been tested by repeated applica- tion of BOM tests, where different data backgrounds (DBs) are used during each iteration [5], [15]–[17], [19]. The disadvan- tages of this methodology are test time inefficiency and limited fault coverage of coupling faults (CFs) between cells within the same word, which are called intraword CFs. In [5], a system- atic method to solve the problem of detecting intraword CFs Manuscript received January 7, 2004; revised June 25, 2004. This paper was recommended by Associate Editor S. Hellebrand. S. Hamdioui is with the Computer Engineering Laboratory, Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, 2628 CD Delft, The Netherlands (e-mail: Said.Hamdioui@ philipscrolles.st.com, S.Hamdioui@ewi.tudelft.nl). J. E. Q. D. Reyes is with the Intel Corporation, Hillsboro, OR 97124 USA. Digital Object Identifier 10.1109/TCAD.2005.847904 was designed. The solution was based on observing that most march tests contain read and write operations with some data values as well as the complementary data values. In [16], a new method based on -out-of- codes has been presented to derive an optimal set of DBs. However, the solutions reported in [5] and [16] were restricted only to state CFs. Currently published work shows, based on defect injection and circuit simulation, the exis- tence of many new CFs [3], [7], [10]. In [15], the transformation of BOM tests into WOM tests has been established, based on re- placing the data values used by the march tests with march DBs (MDBs) and walking DBs (WDBs). The MDBs replace the data value in the “state changing” march elements, which are ele- ments that consist, in total, of an odd number of transition write operations [e.g., ], while the WDBs replace the data value in the “state remaining” march elements, which are elements that consist, in total, of an even number of transition write operations [e.g., ]. However, the solu- tion proposed in [15] increases the test time by a factor , where is the number of DBs. In addition, it does not guarantee the detection of all intraword CFs. In [17] and [19], DB sequences (DBSs) for intraword idempotent and disturb CFs have been presented. In addition, a systematic way for converting BOM tests into WOM tests has been introduced, based on concate- nating the BOM test (which detects faults between words) and the WOM test (which detects the faults within a word). How- ever, only idempotent and disturb CFs have been considered. As mentioned previously, currently published work shows the existence of many new CFs. On the other hand, most used in- dustrial tests repeat the BOM’s tests with different DBs, which does not guarantee the detection of all intraword faults. Solid, checkerboard, column stripe, and row stripe are the usual DBs in that case [13], [18]. This paper considers all possible intraword CFs. The DBSs required for the detection of each of these faults will be es- tablished, and compiled into tests detecting the targeted faults. The tests will then be evaluated industrially. This paper is or- ganized as follows. Section II describes the fault models (FMs) for WOMs. Section III establishes the DBS and the required op- eration sequences for each intraword CF. Section IV introduces tests covering the targeted faults. Section V presents the indus- trial evaluation of the introduced tests and the most well-known memory tests for big arrays with an interleaved bit-organiza- tion and for small arrays with an adjacent bit-organization. Sec- tion VI ends with the conclusion. 0278-0070/$20.00 © 2005 IEEE