Achieving ZT=2.2 with Bi-doped n-type SnSe single crystals

Recently SnSe, a layered chalcogenide material, has attracted a great deal of attention for its excellent p-type thermoelectric property showing a remarkable ZT value of 2.6 at 923 K. For thermoelectric device applications, it is necessary to have n-type materials with comparable ZT value. Here, we report that n-type SnSe single crystals were successfully synthesized by substituting Bi at Sn sites. In addition, it was found that the carrier concentration increases with Bi content, which has a great influence on the thermoelectric properties of n-type SnSe single crystals. Indeed, we achieved the maximum ZT value of 2.2 along b axis at 733 K in the most highly doped n-type SnSe with a carrier density of −2.1 × 1019 cm−3 at 773 K.

arXiv:1601.00753v2, and 10.1002/aenm.201500360 as a few examples). The synthesis of Bidoped single crystal SnSe does allow the measurements of properties along crystalline axial directions. The results appear reasonable. However, it is doubtful that the manuscript represents any advance in understanding thermoelectricity or SnSe or influences thinking in the field. As a result, I would recommend the manuscript being published in a more applied journal. Some other comments I have with the manuscript. 1)Error bars are obviously missing in many figures. Even in the cases where the error bars are hard to quantify, an estimation is still better than leaving them out.
2)The authors believe that Bi occupies the Sn sites. It is likely but no supported. 3)Thermal conductivity of different doped samples should be shown. Even though the electronic contribution is small, the effect of doping on phonons at a few atom% can be quite large. 4)The figures in supplemental materials should be improve to similar quality as the main text.
Reviewer #3 (Remarks to the Author) The authors report on unpresented, and extremely high values of ZT reached in n-doped SnSe. This topic could be of large interest for the readership of Nature Communications. However, the paper contains some very unexpected results that raise some important questions that have to be addressed by the authors. Thus, I suggest rejection, yet encourage the authors to resubmit if they address the mandatory points mentioned below: • The authors write that « the single crystal is easily cleaved along a-axis [...] as shown on Figure  1a ». Actually, Figure 1a shows an ingot with a very rough and irregular cleavage, what does not agree with the statement of the authors. Besides, it would be interesting to know the opinion of the authors regarding the orientation of the crystal during the growth. • The carrier concentration in the un-doped sample is found to be 6x10e15 cm-3, what is about 100 times less that what have been measured by Zhao et al. [9]. The authors should comment on this major difference, taking into account that DFT calculations have shown that the vacancy of tin is very probable (very low enthalpy of formation), and does induce a large density of charge carriers (see Bera et al. Phys.Chem.Chem.Phys. 16, 19894 (2014)). • As expected (because of the low enthalpy of formation of the vacancy of tin, which is a "killer" pdoping defect) the doping level reached by the authors is very small (only a few 10e16 cm-3). In other words, the samples are barely n-doped (lower doping level than the "undoped" sample of Zhao et al. [9]), showing a carrier density only about 5x larger than the un-doped samples. This is far of being "highly doped", as claimed by the authors. This major point should be discussed in the revised manuscript, especially taking into account the prior art, both experimental and DFT based. • The authors claim that DFT calculation have forecasted high TE properties for n-doped SnSe. This is true, but only for very high charge carrier concentrations (above 10e19 cm-3). The regime in which the authors operate is very far from the necessary concentrations, and the large ZT values claimed by the authors should not occur in such low concentration of carriers. The author should comment, and explain this very strong difference. • In order to convince the reader about the very high value of ZT claimed, the authors should prove that the ZT values are strongly anisotropic, and measure the TE properties (thermal and electrical conductivity as well as Seebeck) along the three different axis of the single crystal.

Reviewer #1:
We thank the reviewer for recognizing the significance of our work. We appreciate very much the reviewer's constructive comments and suggestions, which are very helpful for us to sharpen the messages better in the revised manuscript. Some of your comments we would like to detail explain as below: 1) The for 4% Bi, and 2.5×10 16 cm -3 for 6% Bi at 300 K.
2) The referee commented that "There is no reference scale on for the insert picture in Fig. 1a; one should be provided by the authors. The reader cannot determine if these crystals are 0.1 mm, 1 mm or 10 mm in length." Answer: We replaced the sample photo in Fig. 1(a). In the new photo, a ruler is shown to indicate the size of Bi-doped SnSe single crystal. 3) The referee suggested the title change from "Achievement of ZT=2.1 in Bidoped n-type SnSe single crystals" to "Achieving ZT=2.1 with Bi-doped ntype SnSe single crystals". Answer: The title is changed in the revised manuscript as the referee suggested.
Additionally, we measured the ZT values along all three crystal directions in the revised manuscript; in fact, we achieved the ZT value of 2.2 along b axis at 733 K (we obtained ZT = 2.1 along c-axis in the previous manuscript). Thus, we updated this ZT value of 2.2 in the title of revised manuscript. 3

Reviewer #2:
We appreciate very much his/her constructive comments and suggestions.
Below we address his/her specific comments in detail.

1)
The referee commented that "While the result will be interesting to the thermoelectric community, it should not claim "no success result on n-type doping of SnSe has been reported by the community thus far" (see APL 108, 083902, arXiv:1601.00753v2, and10.1002/aenm.201500360

as a few examples)."
Answer: We again appreciate referee's comment on this point. As a matter of fact, there are some reports on n-type SnSe. However, all of them reported about n-type doped polycrystalline SnSe, and they were not successful in achieving ZT values comparable to ZT = 2.6 of p-type SnSe: ZT = 1 in Iodine doped SnSe 1 and ZT = 0.7 in BiCl 3 doped SnSe. 2 In order to address the referee's comment, we revised the sentence on line 61, page 3 as following; FROM: "However, no successful result on n-type doping of SnSe has been reported to the community thus far. Here, we report that n-type SnSe single crystals were successfully synthesized by doping Bi into Sn sites for the first time." TO: "However, no successful experiment results on n-type SnSe with ZT values comparable to ZT = 2.6 of p-type SnSe have been reported to the community thus far. Here, we report that n-type SnSe single crystals were successfully synthesized by doping Bi into Sn sites." Additionally, we removed the phrase "for the first time" in line 18, page 1.
We added a below sentence in line 56, page 3.
"There are a few reports on doped n-type SnSe, which are polycrystalline with ZT values of below 1; ZT = 1 in Iodine-doped and 0.7 in BiCl 3 -doped SnSe. 1,2 " 2) The referee commented that "Error bars are obviously missing in many figures. Even in the cases where the error bars are hard to quantify, estimation is still better than leaving them out." Answer: We determined the error levels of three transport properties (electrical conductivity, Seebeck coefficient, and thermal conductivity) as below.
 For electrical conductivity, = = where ρ is resistivity, R is resistance, l is length of sample, and a and b are width and height of sample.  3)

Average of electrical conductivity
The referee commented that "The authors believe that Bi occupies the Sn sites. It is likely but no supported." Answer: We carried out microscopic study on un-doped (p-type) and Bi-doped (n-type) SnSe using low temperature scanning tunneling microscope (STM) operated at 79 K under ultra-high vacuum environment (< 10 -10 Torr). All samples were cleaved in-situ to obtain clean b-c planes of SnSe. Figure r2  calculations. Single Sn atom is removed from the supercell, and then the Sn vacancy is occupied by single Bi atom. As shown in Fig. 3r(c), the unique shape is well reproduced in the STM simulation. Therefore, we confirmed that Bi atoms are indeed occupying the Sn sites of SnSe.
We added this information in the revised manuscript.

4) The referee commented that "Thermal conductivity of different doped samples should be shown. Even though the electronic contribution is small, the effect of doping on phonons at a few atom% can be quite large."
Answer: Thermal conductivities of different Bi doping content samples were measured from 300 to 800 K, following referee's comment. Results are very similar each other, which are added in Fig. 3(d) of revised manuscript.

5)
The referee commented that "The figures in supplemental materials should be improved to similar quality as the main text." Answer: The quality of figures in supplemental information has been improved. <Figure r4. Thermal conductivities of Bi-doped SnSe single crystals> 9 Reviewer #3: We appreciate very much the reviewer's constructive comments and suggestions.
As encouraged by the referee, we address his/her specific comments in detail below.

1)
The referee commented that "The authors write that « the single crystal is easily cleaved along a-axis [...] as shown on Figure 1a ». Actually, Figure  1a shows an ingot with a very rough and irregular cleavage, what does not agree with the statement of the authors. Besides, it would be interesting to know the opinion of the authors regarding the orientation of the crystal during the growth." Answer: We replaced the old sample photo in figure 1(a) as the referee suggested.

Opinion in single crystal orientation:
We observed that SnSe single crystal prefers to grow along the c-axis as shown in Fig. r5. The crystal orientation is added in Fig. 1(a)  plane and each layer along a-axis is weakly coupled by Van der Waals interaction. The lack of chemical bonding might be the reason why SnSe is not likely to grow along the a-axis. Along b-c plane, it is not easy to exactly know why SnSe grows along c-axis. During growth process, crystal orientation depends on many parameters such as the crystallographic orientation of the seeds and atomic bonding. In the case of SnSe, however, the bonding structure is very complicated along b and c-axes. Therefore, we can only guess that SnSe seedcrystals are preferentially oriented along the c-axis due to thermodynamic effects at the solidification point of growth process.

2)
The referee commented that "The carrier concentration in the un-doped sample is found to be 6x10 15 cm -3 , what is about 100 times less that what have been measured by Zhao et al. [9]. The authors should comment on this major difference, taking into account that DFT calculations have shown that the vacancy of tin is very probable (very low enthalpy of formation), and does induce a large density of charge carriers (see Bera et al. Phys.Chem.Chem.Phys. 16, 19894 (2014) We added this data in figure 2  We carried out microscopic study on un-doped (p-type) and Bi-doped (ntype) SnSe using low temperature scanning tunneling microscope (STM) operated at 79 K under ultra-high vacuum environment (< 10 -10 Torr). All samples were cleaved in-situ to obtain clean b-c planes of SnSe. Figure r8( Fig. r8(a)) shows bright intensity at the Sn site, but dark intensity at the Se site with a rectangular unit cell. It is reported that this contrast difference is attributed to the upward buckling of Sn atoms and the dominant contribution of Sn 5p states. 5 On the contrary, as shown in Fig. r8 calculations. Single Sn atom is removed from the supercell, and then the Sn vacancy is occupied by single Bi atom. As shown in Fig. 9r(c), the unique shape is well reproduced in STM simulation. Therefore, we confirmed that Bi atoms are indeed occupying the Sn sites of SnSe.
We added this information in the revised manuscript.

5)
The referee commented that "In order to convince the reader about the very high value of ZT claimed, the authors should prove that the ZT values are strongly anisotropic, and measure the TE properties (thermal and electrical conductivity as well as Seebeck) along the three different axis of the single crystal." Answer: Based on referee's suggestion, we conducted thermoelectric measurements along all three crystal directions for the high ZT sample with carrier concentration of 2.1×10 19 cm -3 at 773 K, as shown in figures r10 and r11 in revised manuscript.
We achieved ZT=2.2 along b-axis, which is higher than ZT=2.1 along c-axis (the original manuscript's value). The title has been changed to "Achieving ZT=2.2 with Bi-doped n-type SnSe single crystals". Note that maximum ZT value along aaxis is 0.8. The anisotropic trend is similar with p-type SnSe. However, the optimum temperature is 773 K in Bi-doped SnSe, whereas 923 K in p-type SnSe.