Fig. 1.2. The principle of the phase change transition. The phase-change material in the crystalline phase has to be melted to achieve the liquid phase. If the cooling rate is sufficiently high, then the material is quenched in the amorphous phase. To recover the crystalline phase, only a specific thermal profile can provide the energy necessary for the atomic reorganization [9].
the switching phenomenon has not been provided until the early 80’s. Just after the first successful integrations of phase-change materials in silicon based electronics, the future of this technology started to be compromised by the limitation of the technology node:
to guarantee a reliable and power-efficient device, the dimensions required a level of integration not available in that years. This explains the hibernation for about 20 years of PCM technology, just after the realization in 1978 of the first 1 kbit PCM prototype by Burroughs Corporation. Along this period, phase-change materials started their well known success in the optical applications, also thanks to the discovery of Ge2Sb2Te5
(GST) and its phase change properties. With the release of the first rewritable compact disc (CD-RW) in 1990, Panasonic declared the starting of the end of the magnetic tape supports for consumer applications: the first real revolution caused by the phase-change materials in the history of technology.
The principle of the phase transition in a PCM device, is based on the thermal-ly-induced phase change of the material from a crystalline phase, ordered and lowly resistive, to an amorphous phase, disordered and highly resistive, passing from the liq-uid phase, as schematically described in Fig. 1.2 [9]. To achieve the amorphous phase, the phase-change material in the crystalline phase has to be melted (crossing the ma-terial melting temperature Tm), through Joule heating, at high current densities, and then rapidly quenched (RESET operation). Thanks to the switching phenomenon, once reached a specific electric field threshold (ET H), the material in the amorphous phase changes abruptly its conductivity and starts to be highly conductive (what we call “ON-state” of the material). The phase-change material, once in the ON-state, is then crystallized through the Joule heating obtained thanks to the current flow (SET operation). There are many other properties of these materials that are involved in the device behavior, that we will analyze in the next sections introducing the properties of the crystalline phase, of the amorphous phase and of the liquid phase of a phase-change material.
1.2 The PCM cell
Like other resistive memories, the phase-change memory (PCM) cell is a two-terminal device. It bases its functionality on the strong difference in resistivity (up to more than 5 orders of magnitude), between the crystalline phase and the amorphous phase of the
Fig. 1.3. Resistivity as a function of the tem-perature of a phase-change material. The tran-sition from the amorphous to the crystalline phase, is highlighted by the sharp decrease of the resistivity of the material.
Fig. 1.4. General description of a PCM cell.
The active volume involved in the phase tran-sition is highlighted at the plug/phase-change material interface.
phase-change materials. In Fig. 1.3 we can see an example of this transition. The amorphous material is heated, and during the temperature increase, it experiences a decrease of the resistivity at a specific temperature, called crystallization temperature.
The process of crystallization, activated and favorable at this temperature, drastically increases the conductivity of the material, finally crystallizing all the material volume.
Once the material is cooled down to room temperature, the crystalline phase is pre-served. The full process gives rise to the phase-change mechanism, in this case, induced by the external heating of the material.
The PCM cell, as reported in Fig. 1.4, can be basically described as consisting of:
- the bottom electrode;
- the plug conductive element (called also “heater”) with the function to provide the electrical access to the phase-change material, to enable the current limitation, and to contribute to the “heating” of the phase-change material in the different phases of the programming;
- the phase-change material;
- the insulator surrounding the plug;
- the top electrode.
In the standard memory array architecture, the memory cell consists of a transistor access device (1T) and the PCM cell in series (1R). This configuration is called “1T1R”.
In an analytical structure in which only the PCM cell is realized in order to allow the study of the resistive element, the metal levels of the top and bottom electrodes are properly designed in order to provide the electrical access to the cell, and they are called “access lines”.
To program a PCM cell in the RESET state (or high resistance state, HRS), a RESET pulse is applied, consisting of a high current pulse able to raise the temperature above the melting point (Tm) in the active chalcogenide volume, followed by a sharp
1.2 The PCM cell 9
Fig. 1.5. Resistance as a function of the pro-gramming current (R-I), for a phase-change memory device starting from the SET state (open symbols), and starting from the RESET state (filled symbols). The READ, SET and RESET current ranges are highlighted [11].
Fig. 1.6. Programming current as a function of cell voltage characteristic (I-V), typical of a PCM device. The two curves correspond re-spectively to a device starting from a RESET state (filled symbols), and to a device starting from a SET state (open symbols) [10].
trailing edge quenching the same volume into an amorphous state (RESET operation) [10]. The RESET operation can be successful, only if a considerable volume of phase-change material is amorphized over the plug surface, in order to increase considerably the resistance of the device. To program the cell in the SET state (or low resistance state, LRS), two main strategies can be adopted:
- the application of a pulse with the same amplitude of a RESET pulse, but with a trailing edge sufficiently long to guarantee the permanence of the phase-change material in the range of temperatures favorable to the recrystallization;
- a pulse with amplitude lower than the RESET pulse, but higher than the thresh-old voltage, able to provide already during the pulse width, the good crystalliza-tion temperature in the active area of the phase-change material, to achieve at the end of the pulse the final SET state. In this case, the trailing edge can be as fast as in the RESET pulse.
The pulse shape becomes fundamental for the programming of the cell, since the main parameter to play on the material phase-change is the temperature. The temperature in fact, is increased in the device by the current induced Joule heating, and specific pulse shapes, correspond to specific temperature profiles in the cell.
In Fig. 1.5 we report a typical resistance versus programming current characteristic of a PCM device. Each point in the graph corresponds to the final resistance of the memory, achieved after the application of a current pulse of specific intensity. Three main regions are highlighted: READ, SET and RESET. The READ operation enables the sensing of the device, and it has to avoid the disturbing or the change of the resistance state of the cell. Hence, it is performed at low current values. Increasing the programming current the temperature in the cell is risen enabling the recrystallization of the phase-change material, and lowering the final device resistance (SET). When the current intensity allows the partial melting of the phase-change material (melting
current), the rapid quenching allows the amorphisation of part of the material volume, resulting after the pulse application, in an increase of the resistance of the device (RESET). If we consider the characteristics of a device starting from a SET state, and of a device starting from a RESET state, we notice two main differences. The first, is the starting resistance, that is preserved in the READ region. The second is the different resistance achieved in the SET region. In fact, if the device starts already from a SET state, the material does not experience any phase transition till the reaching of the melting current. On the contrary, starting from an amorphous phase, the device, thanks to the crystallization mechanism, decreases its resistance, but this decrease depends as already observed, on the pulse shape (e.g. duration, fall time, etc.). As illustrated in the graph, in this case the pulses applied were not sufficiently long to recover the perfect crystallization of the phase-change material. Once reached the RESET region, the two characteristics overlap, because of the melting of the phase-change material in both cases.
The current as a function of the voltage plots of a phase-change memory (I-V), re-ported in Fig. 1.6, show the overlapping of the SET and of the RESET characteristics (that corresponds respectively to the device starting from the SET and the RESET state) in the SET and in the RESET range of currents. However, coming from the amorphous phase, only a long persistence in the SET current region, allows the recrys-tallization of the phase-change material. Moreover, the RESET I-V curve evidences the main electrical property of a PCM device: the switching phenomenon. The RESET device, once reached a critical threshold voltage VT H, with a typical snap back in the I-V curve, switches to a high conductive state (ON-state), with conduction properties similar to the crystalline phase. Once in the ON-state, the heating induced by the increased current, can allow the recrystallization of the integrated phase-change mate-rial. The switching phenomenon, is the fundamental property that makes possible the phase transition in a PCM device.