Rice team flips an altermagnet's Hall signal with strain, not a magnetic field

A group led by Pengcheng Dai at Rice University reports in Physical Review X that stretching hexagonal manganese telluride along one axis both cleans up its magnetic domain structure and reverses the sign of its anomalous Hall voltage ^[1][8][9]. The consequence for anyone building spin-based electronics is a control knob that is mechanical rather than magnetic, in a material that carries almost no net magnetization to begin with ^[3][9].

The setup matters. In ordinary magnets, time-reversal symmetry is broken by aligning large numbers of spins in one direction, which produces a strong external field ^[12]. In altermagnets, the symmetry breaking survives while the spins largely cancel, leaving near-zero net magnetization ^[3]; hexagonal manganese telluride was recently identified as one of these ^[4]. That is attractive for spintronics, where information rides on electron spin ^[13], because a device element with no stray field is a device element that does not talk to its neighbors.

The practical obstacle has been reading the thing at all. The material splits into multiple magnetic domains pointing in different directions, and their overlapping signals obscure the underlying structure, which makes the effect hard to isolate and harder to control ^[5]. Dai's team used the anomalous Hall effect as a proxy: a sideways voltage that appears when current flows through a magnetic material, generated by internal magnetic structure rather than an applied field ^[6]. Manganese telluride produces it spontaneously, which confirms that time-reversal symmetry is broken despite the near-zero magnetization ^[7].

Uniaxial stretching then did two jobs. It forced the competing domains to merge into a single one, which let the group resolve the true magnetic structure for the first time ^[8]. And tuning the amount of strain flipped the sign of the Hall signal outright, switching polarity without disturbing the magnetic order underneath ^[9]. The group calculates that a 1% change in strain is comparable in effect to a temperature change of roughly 150 degrees, which they present as the more practical of the two levers for real devices ^[10]. Taken at face value and assuming the relationship holds linearly, 0.1% strain would stand in for about 15 degrees of thermal tuning ^[14] - the sort of ratio that makes a piezo actuator look reasonable next to a heater.

What is not in the published account is the engineering. It does not give the absolute strain values at which the sign reverses, the temperature at which the magnetic order sets in, or any fabricated device, switching speed, or cycling endurance ^[15]. The authors' own framing is directional: strain-tunable sensors and spin-based components that could operate more practically at everyday temperatures ^[11]. "Could" is doing work in that sentence.

Watch for three things. Whether the flip is repeatable over many strain cycles without the domain structure fragmenting again. Whether the strain window sits inside what thin-film integration on a piezoelectric substrate can actually deliver, rather than requiring bulk crystals in a strain cell. And whether the reported temperature-equivalence figure holds across the operating range, or is a local slope near one working point.