Solution to the Hole-Doping Problem and Tunable Quantum Hall Effect in Bi₂Se₃ Thin Films

Abstract

ABSTRACT. Bi2Se3, one of the most widely studied topological insulators (TIs), is naturally electron-doped due to n-type native defects. However, many years of efforts to achieve p-type Bi2Se3 thin films have failed so far. Here, we provide a solution to this long-standing problem, showing that the main culprit has been the high density of interfacial defects. By suppressing these defects through an interfacial engineering scheme, we have successfully implemented p-type Bi2Se3 thin films down to the thinnest topological regime. On this platform, we present the first tunable quantum Hall effect (QHE) study in Bi2Se3 thin films, and reveal not only significantly asymmetric QHE signatures across the Dirac point but also the presence of competing anomalous states near the zeroth Landau level. The availability of doping tunable Bi2Se3 thin films will now make it possible to implement various topological quantum devices, previously inaccessible.KEYWORDS. Topological insulator, Bi2Se3, Doping, Interface, Quantum Hall effect 3 Hole (p) doping has been challenging in Bi2Se3 1-3 , one of the most widely studied topological insulators (TIs) [4][5][6][7][8][9][10] . Unlike conventional semiconductor materials, the problem is complicated due to the presence of both surface and bulk states in topological insulators: we have to consider the doping problem of the surface and the bulk states separately. Both the surface and bulk states of Bi2Se3 have a strong tendency toward n-type due to its native n-type defects such as selenium vacancies [1][2][3][11][12][13][14][15][16][17] . In bulk crystals compensation dopants such as Ca and Mn can be used to convert the dominant carrier type from n-to p-type [18][19][20][21][22][23][24][25] . However, such a compensation doping scheme has not been successful in thin films of Bi2Se3. More specifically, we have tried various potential p-type dopants such as Zn, Mg, Ca, Sr, and Ba as compensation dopants, but none of them have so far led to p-type Bi2Se3 thin films; no p-type Bi2Se3 thin films have been demonstrated in the literature either. Only when they were quite thick (~200 nm), we were able to achieve p-type Bi2Se3 films through a complex process 8 . This difficulty in achieving p-type Bi2Se3 thin films has been puzzling, considering the very existence of p-type Bi2Se3 bulk crystals 19,20 . It may be suspected that this discrepancy in doping efficiency between thin films and bulk crystals could be due to the different growth conditions of the two systems, such as growth temperatures, considering that films are, in general, grown at much lower temperatures than bulk crystals. However, it should be noted that even p-type Bi2Se3 bulk crystals tend to become n-type when the crystals are made into thin flakes through, say, the Scotch tape method 26 . Bi2Se3 flakes can remain p-type only if they are relatively thick (> ~150 nm) 24 . All these observations provide evidence that whether they are thin films or bulk crystals, thickness critically affects the doping efficiency of the Bi2Se3 syst...

Pb-doped p -type Bi2Se3 thin films via interfacial engineering

“…However, at much higher doping levels, the carrier type changes back to n-type as shown in the 10% data, and the (n-type) n2D becomes extremely high at 20%. Such a carrier-type reversal behavior at high doping concentrations was previously observed in the Ca-doped Bi2Se3 film study, 15 and indicates that there is a solubility limit for any compensation dopant. Beyond a solubility limit, the dopants start to introduce extra defects going beyond simple Bi substitution, and it is well known that almost all defects in Bi2Se3 behave as n-type dopants.…”

“…At high concentrations, the film becomes n-type due to disorder beyond the solubility limit of Pb in Bi2Se3, accompanied by a higher (than p-type) mobility shown in (b). 15. The mobility of Pb-doped film is substantially higher than that of Cadoped one, which means that Pb-doping is less disruptive than Ca-doping in Bi2Se3.…”

“…In addition, the use of Ca dopant has allowed Bi2Se3 thin films to turn into p-type not only in a relatively thick regime (~200 nm) via a complex process 14 but also in the thinnest topological regime (6 QLs). 15 However, considering that Ca is much lighter than Bi and that spin-orbit-coupling (SOC) strength, which grows fast (to the 4 th power in a hydrogen-like atom) with increasing atomic numbers, is an essential parameter for topological surface states, [16][17][18] substituting light elements for Bi could have an undesirable side effect of compromising the topological properties. In this sense, a heavy 2+ ion will be a much more desirable compensation dopant than such a light element as Ca.…”

“…Despite well-established p-type Ca-doped Bi2Se3 bulk crystals as discussed above, Ca-doped Bi2Se3 thin films have long failed to reach the p-regime, and only last year, the problem was solved with an interfacial engineering scheme. 15 Then, more recently, another interfacial engineering scheme has enabled Ti-doping to convert otherwise p-type Sb2Te3 thin films into n-type. 21 Ti is a potential n-type dopant for Sb 3+ ion in Sb2Te3.…”

“…The most notable feature is that the µs of p-type samples are clearly lower than those of n-type samples, which is consistent with the previous Ca-doped Bi2Se3 study. 15 The very fact that the 10% n-type sample exhibits a higher mobility than the sub 1% p-type samples strongly suggests that the higher mobility of n-type than p-type samples should be an intrinsic band structure effect rather than an extrinsic dopant-scattering effect. This is also consistent with the known band structure of Bi2Se3: the surface band structure is much sharper above the Dirac point than below, 3,9,[29][30][31] thus higher Fermi velocity (or/and lower effective mass) and higher mobility for n-type than p-type.…”