Chemical tuning of electrical transport inTi1−xPtxSe2−y

Abstract

The structural and transport properties of polycrystalline Ti1-xPtxSe2-y (x ≤ 0.13, y ≤ 0.2) are studied, revealing highly tunable electrical properties, spanning nearly 10 orders in magnitude in scaled resistivity. Using x-ray and neutron diffraction, Pt is found to dope on the Ti site. In the absence of Pt doping (for x = 0), Se deficiency (y > 0) increases the metallic character of TiSe2, while a remarkable increase of the low temperature resistivity is favored by no Se deficiency (y = 0) and increasing amounts of doped Pt (x > 0). The chemical tuning of the resistivity in Ti1-xPtxSe2-y with Se deficiency and Pt doping results in a metal-to-insulator transition. The simultaneous Pt doping and Se deficiency (x,y > 0) confirms the competition between the two opposing trends in electrical transport, with the main outcome being the suppression of the charge density wave (CDW) transition below 2 K for y = 2x = 0.18. Band structure calculations on a subset of Ti1-xPtxSe2-y compositions are in line with the experimental observations.

Walking freely in the energy and temperature space by the modified replica exchange molecular dynamics method

“…Recently, we proposed a modified REMD method. [24] As shown in Figure 1c, each replica of the molecule stays in one special state (active state or inactive state) all the time except the exchange step. For the replicas in the active state (illustrated by the solid boxes in the figure), their canonical variables are updated by the normal MD method as usual.…”

“…Since the number of the temperatures (replicas) is larger than the number of the CPUs in the computer, the root node has to collect or scatter the canonical variables of the replicas periodically. For example, we perform a REMD simulation for a 12-residue molecule (trpzip2) in a 8-node computer cluster by our previous method [24] in the Generalized Born/ Surface Area (GB/SA) implicit solvent [31,32] (the simulation software is Tinker [33] and the force field is AMBER PARM96 [34] ). The total number of the replicas is 64, among them 32 replicas are active.…”

“…To circumvent this issue, recently we made a modification to the standard REMD method. [24] In the simulation, all the replicas are randomly designated as the active or the inactive replicas at the exchange step. The inactive replicas have the same number of the canonical variables as the active replicas.…”

“…But their canonical variables are frozen during the simulation. [24] Compared to the standard REMD, using the inactive replicas in the simulation can save the calculation quantities, expand the temperature range and reduce the temperature intervals. As we mentioned above, this modification is helpful to the performance of the REMD simulation.…”

“…Without the communications of the canonical variables in the network, the method is simpler and faster than our previous method. [24] In this work, we apply the modified REMD method to some simple models and analyze the simulation data in depth. It shows that the calculations of various thermodynamic quantities in the simulation are more convergent than those in the standard REMD simulation.…”