Abstract
Many IoT and wearable devices require an on-chip small-to-mid-capacity nonvolatile memory (NVM) with a fast read-access time (T <sub>AC</sub> ) and reliable read operations: for applications including data-logging, configurable look-up tables (LUT), eFuse, and physically unclonable functions (PUF). STT-MRAM [1-4] is a good candidate for these applications due to its fast write speed, low-voltage write, and high endurance. However, STT-MRAM suffers from a small read-signal margin (RSM) due to a small tunnel magnetoresistance ratio (TMR: (R <sub>AP</sub> -R <sub>P</sub> )/R <sub>P</sub> ) between the cell resistance of parallel (P, R <sub>P</sub> ) and anti-parallel (AP, R <sub>AP</sub> ) states [1-6]. Moreover, the read-disturb behavior of STT-MRAM cells is sensitive to the BL read voltage (V <sub>BL-RD</sub> ) and the stress/read time. Compact 1T1MTJ arrays are suitable for high-density applications [5-6]; however, they use a power-hungry current-mode read scheme with a slow read speed due to the small RSM. Researchers have proposed 2T2MTJ (Fig. 30.3.1) arrays [1-4] with differential bitlines (BL and BLB) and a voltage-mode read scheme, with an enlarged RSM (V <sub>RSM</sub> ), for fast, low-power read operations. V <sub>RSM</sub> refers to the voltage difference between BL (V <sub>BL</sub> ) and BLB (V <sub>BLB</sub> ). 2T2MTJ STT-MRAM read operations still face the following challenges: (1) V <sub>BL</sub> and V <sub>BLB</sub> both drop from V <sub>BL-RD</sub> to 0V quite quickly due to the large cell read current (V <sub>P</sub> and V <sub>AP</sub> ) or low R-value in both R <sub>P</sub> and R <sub>AP</sub> , resulting in small sensing window (T <sub>SMW</sub> ), which is the period when V <sub>RSM</sub> >offset; (2) the maximum V <sub>RSM</sub> (V <sub>RSM-MAX</sub> ) occurs at different times (t <sub>RSM-MAX</sub> ) for different cells due to TMR (R <sub>AP</sub> /R <sub>P</sub> ) variation; and (3) a degraded V <sub>RSM</sub> due to the use of a low V <sub>BL-RD</sub> to avoid read disturbs for high data-reliability applications. (1) and (2) lead to a decrease in V <sub>RSM</sub> after reaching its peak (t <sub>RSM-MAX</sub> ), despite an increase in BL development time (t <sub>BL</sub> ). When using a conventional voltage-mode sense amplifier (CNV-VSA) with a common activated (SAEN=1) timing (f <sub>SAEN</sub> ) under the effects of (1)-(3), the signal to be amplified (ΔV <sub>IN</sub> <V <sub>RSM-MAX</sub> ) is subject to degradation at the VSA's differential inputs, resulting in a sensing failure at a low V <sup>BL-RD</sup> .