Suppressing Structural Relaxation in Nanoscale Antimony to Enable Ultralow‐Drift Phase‐Change Memory Applications

Abstract Phase‐change random‐access memory (PCRAM) devices suffer from pronounced resistance drift originating from considerable structural relaxation of phase‐change materials (PCMs), which hinders current developments of high‐capacity memory and high‐parallelism computing that both need reliable multibit programming. This work realizes that compositional simplification and geometrical miniaturization of traditional GeSbTe‐like PCMs are feasible routes to suppress relaxation. While to date, the aging mechanisms of the simplest PCM, Sb, at nanoscale, have not yet been unveiled. Here, this work demonstrates that in an optimal thickness of only 4 nm, the thin Sb film can enable a precise multilevel programming with ultralow resistance drift coefficients, in a regime of ≈10−4–10−3. This advancement is mainly owed to the slightly changed Peierls distortion in Sb and the less‐distorted octahedral‐like atomic configurations across the Sb/SiO2 interfaces. This work highlights a new indispensable approach, interfacial regulation of nanoscale PCMs, for pursuing ultimately reliable resistance control in aggressively‐miniaturized PCRAM devices, to boost the storage and computing efficiencies substantially.


(Figure
) Resistance drift behavior of the 5 nm-thick Sb film.S3) Measurements for the activation energy of conduction of the 4 nm-thick Sb film.S4) Energy evolutions of the less-equilibrium (LE) and more-equilibrium (ME) 4 nm-thick amorphous Sb models annealed at 300 K. S5) Coordination number (CN) distribution and Bond angle distribution (BAD) of the LE and ME models of Sb at 87 ps.

(Figure
6. (Figure S6) Amorphous 3-nm thick ME models and corresponding ELF distributions.S7) Peierls distortion for the body and interface parts in the two 4 nm-thick amorphous models.

(Figure
8. (Figure S8) Larger resistance drift of the 3 nm-thick Sb films.9. (Figure S9) Partial electronic density of states (DOS) contributed by heteropolar and homopolar bonds on the interfaces.For each temperature dip in (a), once the measured resistance magnitudes and the changing trend became reliable in the regime from ~10 to ~-30 C, a single straight line was obtained, which was considered as a proper method to deduce the Ec value. [12]The line color indicates the drift time ranging from ~0.

Figure S1 .
Figure S1.A comparison of resistance drift coefficients of diverse PCMs with different geometries.The distribution of resistance drift coefficient (v) for GeSbTe-like, GeTe-like, and SbTe-like PCMs from literature, as well as Sb thin film in this work.The corresponding references are detailed in this figure.

Figure S2 .
Figure S2.Resistance drift behavior of the 5 nm-thick Sb film.Sheet resistance as a function of time for the 5 nm-thick Sb film, measured at room temperature.The whole measured curve of nearly ~20 h duration is divided into multiple segments, each containing ~600 s.The sheet resistance of the 5 nmthick Sb film increased with time steadily within the first ~8 h, whereas afterwards the sheet resistance continuously decreased due to the progressive crystallization.

Figure S3 .
Figure S3.Measurements for the activation energy of conduction (Ec) of the 4 nm-thick Sb film.(a) The cooling and heating cycles (lower panel) and the corresponding sheet resistance of the 4 nm-thick Sb film (upper panel).Repetitive cooling and heating were conducted to the thin Sb sample in the temperature range of -35 to 25 C, for a whole duration about ~6 h.The sheet resistance of the sample was real-time monitored.(b) Arrhenius plot of the measured sheet resistance data.For each temperature dip in (a), once the measured resistance magnitudes and the changing trend became reliable in the regime from ~10 to ~-30 C, a single straight line was obtained, which was considered as a proper method to deduce the Ec value.[12]The line color indicates the drift time ranging from ~0.3 h up to ~6 h.The sheet resistance magnitude of the straight lines continuously increases upon drifting.Here we show some representative lines that were measured at different aging periods.The Ec is Figure S3.Measurements for the activation energy of conduction (Ec) of the 4 nm-thick Sb film.(a) The cooling and heating cycles (lower panel) and the corresponding sheet resistance of the 4 nm-thick Sb film (upper panel).Repetitive cooling and heating were conducted to the thin Sb sample in the temperature range of -35 to 25 C, for a whole duration about ~6 h.The sheet resistance of the sample was real-time monitored.(b) Arrhenius plot of the measured sheet resistance data.For each temperature dip in (a), once the measured resistance magnitudes and the changing trend became reliable in the regime from ~10 to ~-30 C, a single straight line was obtained, which was considered as a proper method to deduce the Ec value.[12]The line color indicates the drift time ranging from ~0.3 h up to ~6 h.The sheet resistance magnitude of the straight lines continuously increases upon drifting.Here we show some representative lines that were measured at different aging periods.The Ec is derived via fitting each straight line, utilizing the equation R(T, t) = R * e E c (T,t) k B T , with R * the pre-factor,

Figure S4 .
Figure S4.Energy evolutions of the less-equilibrium (LE) and more-equilibrium (ME) 4 nm-thick amorphous Sb models annealed at 300 K.The system energy of both models decreased noticeably until ~40 ps, and afterwards became relatively stable.We thus termed the ~40 ps and ~87 ps structures as the initial and relaxed states, respectively.

Figure S5 .
Figure S5.Coordination number (CN) distribution of the (a) LE and (b) ME model at 87 ps, respectively.The CN distribution is separated into two parts by the interface and body counts.Bond angle distribution (BAD) of the (c) LE and (d) ME models at 87 ps.

Figure S6 .
Figure S6.(a) Amorphous ME models of 3 nm-thick Sb film sandwiched by glassy SiO2 dielectrics after annealing at 300 K for 87 ps.(b) Average ELF value along the thickness Z direction of the Sb part in the (a) model.

Figure S7 .
Figure S7.Peierls distortion for the body and interface parts in the two 4 nm-thick amorphous models.(a) and (b) The ALTBC pattern for the body part and the interface part in the LE model at 87 ps, respectively.(c) and (d) The ALTBC pattern for the body part and the interface part in ME model at 87 ps, respectively.In both models, the Peierls distortion is stronger on the interface than in the body, because the count distribution at around r2 / r1 ≈ 1.00-1.07 is larger/marginal for the interface/body part, as indicated by the arrows.

Figure S8 .
Figure S8.Larger resistance drift of the 3 nm-thick Sb films.(a) Resistance drift coefficient (sub v) as a function of time for the 3 nm-thick Sb film, with sub v mostly locating in the range of 10 -3 to 10 -2 after aging for ~4-20 h.For comparison, the sub v for the 4 nm-thick Sb film mainly situates from 10 - 3 to 10 -4 after the same duration of aging.(b) Temporal evolution of 9-level resistant states realized by