1: \documentclass[prl,aps,twocolumn,superscriptaddress]{revtex4}
2: %\documentclass[prb,aps,preprint]{revtex4}
3:
4: \usepackage{graphicx}
5:
6: \begin{document}
7:
8: \title{Are fluorinated BN nanotubes $n$-type semiconductors?}
9:
10: \author{H. J. Xiang}
11: \affiliation{Hefei National Laboratory for Physical Sciences at
12: Microscale,
13: University of Science and Technology of
14: China, Hefei, Anhui 230026, People's Republic of China}
15: \affiliation{USTC Shanghai Institute for Advanced Studies,
16: University of Science and Technology of China,
17: Shanghai 201315, People's Republic of China}
18:
19: \author{Jinlong Yang}
20: \thanks{Corresponding author. E-mail: jlyang@ustc.edu.cn}
21: \affiliation{Hefei National Laboratory for Physical Sciences at
22: Microscale,
23: University of Science and Technology of
24: China, Hefei, Anhui 230026, People's Republic of China}
25: \affiliation{USTC Shanghai Institute for Advanced Studies,
26: University of Science and Technology of China,
27: Shanghai 201315, People's Republic of China}
28:
29:
30: \author{J. G. Hou}
31: \affiliation{Hefei National Laboratory for Physical Sciences at
32: Microscale,
33: University of Science and Technology of
34: China, Hefei, Anhui 230026, People's Republic of China}
35: \affiliation{USTC Shanghai Institute for Advanced Studies,
36: University of Science and Technology of China,
37: Shanghai 201315, People's Republic of China}
38:
39: \author{Qingshi Zhu}
40: \affiliation{Hefei National Laboratory for Physical Sciences at
41: Microscale,
42: University of Science and Technology of
43: China, Hefei, Anhui 230026, People's Republic of China}
44: \affiliation{USTC Shanghai Institute for Advanced Studies,
45: University of Science and Technology of China,
46: Shanghai 201315, People's Republic of China}
47:
48: \begin{abstract}
49: The structural and electronic properties of fluorine (F)-doped
50: BN nanotubes (BNNTs) are studied using density functional
51: methods. Our results indicate that F atoms prefer to substitute
52: N atoms, resulting in substantial changes of BN layers.
53: However, F substitutional doping results in no shallow impurity
54: states. The adsorption of F atoms on B sites is more stable than
55: that on N sites. BNNTs with adsorbed F atoms are $p$-type
56: semiconductors, suggesting the electronic conduction in F-doped
57: multiwalled BNNTs with large conductivity observed
58: experimentally might be of $p$-type due to the adsorbed F atoms,
59: but not $n$-type as supposed before.
60: \end{abstract}
61:
62: \maketitle
63:
64: Pure boron nitride
65: nanotubes (BNNTs) \cite{BNNT1,BNNT2} are semiconductors regardless of diameter,
66: chirality, or the number of walls of the tubes.\cite{BNNT1} This
67: contrasts markedly with the heterogeneity of electronic properties
68: of carbon nanotubes, and also makes BNNTs particularly useful for potential
69: nanoelectronic applications.
70: From the applications viewpoint, obtaining n-type and/or p-type
71: semiconducting BNNTs is very important for the use in the nanoscale
72: electronic device, such as
73: the $p$-$n$ junction, $n$-type or $p$-type nanoscale field-effect
74: transistors.
75: In this context, the uniformly doped BNNTs
76: obtained through chemical modification would be a
77: prospective object for tailoring the electronic properties.
78: Theoretical studies indicated that H adsorption on B atom or N atom of
79: BNNTs will introduce donor or acceptor states respectively.\cite{BNH1}
80: Previous experimental studies on functionalization of BNNTs have
81: mainly focused on the B-N-C system through the sophisticate control of
82: C content.\cite{BNC}
83: Very recently, Tang {\it et al.} \cite{BNF} have successfully obtained
84: F functionalized BNNTs with highly curled
85: tubular BN sheets, whose resistivity is smaller by
86: three orders of magnitude than that of the undoped BNNTs.
87: And based on the sole fact that F
88: has excessive valence electrons compared to B and N,
89: the F-doped BNNTs were supposed to be n-type semiconductors.\cite{BNF}
90: However, the behavior of dopants in semiconductors is often far from
91: trivial.\cite{doping}
92: Then does F-doping really induce $n$-type conduction in
93: BNNTs?
94:
95:
96:
97:
98: In this Letter, we conduct a first-principles study on F-doped BNNTs
99: to address this issue.
100: Both adsorption and substitution of F on BNNTs
101: are examined. Surprisingly, neither adsorption or substitution of F on
102: BNNTs results in $n$-type semiconductors, contrasting sharply to
103: previous supposition.
104: Our calculations are performed using the
105: local density approximation\cite{LDA} in the density
106: functional theory as implemented in VASP \cite{vasp}.
107: The electron-ion interaction is
108: described by ultrasoft pseudopotentials
109: \cite{ultrasoft} and the cut-off energy is set to $347.9$ eV.
110:
111:
112: We mainly focus on a typical zig-zag BNNT, i.e., BN(10,0) nanotube.
113: Test calculations indicate that the doping behavior for BN(5,5)
114: NT and BN two dimensional sheet (a model for nanotube with
115: radius $R=\infty$) is qualitatively
116: the same as that for BN(10,0) nanotube.
117: Our supercell for BN(10,0) nanotube contains
118: 80 atoms.
119: First from the optimized structure of the pure BN(10,0) nanotube, we
120: substitute one B or N with one F atom (abbreviated as $S_F^B$ and $S_F^N$
121: respectively). Both relaxed structures are
122: shown in Fig.~\ref{fig1}(a) and (b) respectively. In both cases, F
123: substitution induces significant deformation of the BN layer.
124: Since F is usually in $-1$ valence state, F is
125: bonded with only one adjacent B or N atom along the axis direction.
126: A remarkable difference between the two structures is that F atom
127: doesn't lie in the BN layer and the other two B atoms with dangling bonds
128: come close to each other to form a homoelemental bond in $S_F^N$.
129: Another difference is that the cross section of the tube
130: in $S_F^N$ is no longer a perfect circle
131: as the B atom bonded with F comes outer from the BN plane.
132: To see which substitution is more favorable, we compare the defect
133: formation energy in these two cases.
134: The formation energy is defined as:
135: $E_{form}=E_{tot}-n_B \mu _B - n_N \mu _N -n_F \mu _F,$
136: where $E_{tot}$ is the calculated total energy of the system, $n_B$,
137: $n_N$, and $n_F$ are the number of B, N, and F atoms respectively,
138: and $\mu$ is the chemical potential. $\mu _B$ and
139: $\mu _N$ depend on the growth condition: In the N-rich environment,
140: $\mu _N$ ($-8.629$ eV) is obtained from nitrogen in the gas phase,
141: while a metallic $\alpha$-B phase is used as the reservoir for the
142: B-rich environment ($-7.491$ eV for $\mu _B$).
143: In both cases, $\mu _N$
144: and $\mu_B$ are linked by the thermodynamic constraint:
145: $\mu _B+\mu _N=\mu _{BN}^{tube},$
146: where $\mu_{BN} ^{tube}$ is the chemical potential per BN
147: pair in the pristine BN nanotube.
148: Since both systems has one F atom, the value of $\mu
149: _F$ makes no difference to the difference of the formation energy.
150: So, we don't specify $\mu _F$ explicitly. The calculated formation
151: energies are shown in Table~\ref{table1}.
152: The formation energy difference
153: between $S_F^B$ and $S_F^N$ is $9.778$ ($3.707$) eV in the B (N) rich
154: environment, suggesting F strongly
155: prefers to substitute N in BNNTs. This is not surprising since the large
156: electronegativity of F will result in stronger F-B bond. The
157: homoelemental bond in $S_F^N$ partially saturates the dangling bonds of
158: B atoms, also contributing the stability of $S_F^N$.
159:
160:
161:
162:
163: The band structure of $S_F^N$ is shown in
164: Fig.~\ref{fig2}. For better comparison, the band structure of
165: the pristine BN(10,0) nanotube with the same supercell is also plotted
166: in Fig.~\ref{fig2}.
167: We can see both pristine and F-doped BN(10,0) nanotubes are
168: semiconducting with a band gap of $3.9$ and $3.2$ eV respectively.
169: Density of states (DOS) analysis shows that the bands related to F 2p states
170: mainly lie in $-4$ to $-8$ eV with respect to the valence top.
171: F substitutional doping induces an unoccupied localized state in the
172: gap, which is mainly contributed by the two B atoms connected with the
173: homoelemental bond as shown in the inset of Fig.~\ref{fig2}(b).
174: However, this state is neither a shallow acceptor state
175: nor a shallow donor state. So we don't expect such F substitutional
176: doping will result in large increase of electrical conductivity.
177:
178:
179: To elucidate the origin of the drastic increase in the conductivity
180: of F-doped BNNTs, the
181: adsorption of F on BN(10,0) nanotube is examined.
182: We find that one F atom prefers to adsorb on the B site (labeled as $A_F^B$,
183: as shown in Fig.~\ref{fig1}(c)),
184: since the adsorption of F on N site is unstable and the F atom will
185: eventually adsorb on the B site.
186: The electronic band structure for $A_F^B$ is
187: shown in Fig.~\ref{fig2}(c). Clearly, the system displays a degenerated
188: $p$-type semiconducting behavior. There are two bands with similar
189: character crossing the Fermi level. Both states are rather delocalized
190: as can be seen from the local DOS of one of the bands (shown in the inset
191: of Fig.~\ref{fig2}(c)).
192: So only 1.25\% F adsorption on BNNTs
193: will lead to drastic increase in conductivity.
194:
195:
196: Besides the doping behavior of one F atom, we also study the BN(10,0)
197: nanotube doped with two F atoms. First, adsorption of two F atoms is
198: examined. We find that the configuration (as shown in Fig.~\ref{fig3}(a))
199: where both F atoms adsorb on two B sites is favourable over the one
200: where the second impurity atom adsorbs on the adjacent N site. This
201: differs fundamentally the adsorption of H atoms on BNNTs.\cite{BNH2}
202: Electronic structure calculations on this configuration
203: indicate that it is a degenerated $p$-type semiconductor with larger
204: DOS in the Fermi level than $A_F^B$, as shown in Fig.~\ref{fig3}(d).
205: Then BN(10,0) nanotube with two substitutional F atoms as shown in
206: Fig.~\ref{fig3}(b) is studied. The DOS of this configuration is shown in
207: Fig.~\ref{fig3}(e), indicating that BN(10,0) nanotube with two
208: substitutional F atoms has similar semiconducting properties as $S_F^N$.
209: Finally, BN(10,0) nanotubes with one substitutional F atom and one
210: adsorbed F atom are investigated.
211: The most stable configuation is the
212: sturucture shown in Fig.~\ref{fig3}(c), where the
213: excess F atom is bonded with one
214: of the two B atoms which form the homoelemental B-B bond in $S_F^N$.
215: The DOS of this configuration is shown in Fig.~\ref{fig3}(f),
216: indicating that there is a half occupied localized state
217: due to the unsaturated B dangling bond at approximately 1.4 eV above
218: the valence top.
219: Thus we don't expect this configration will have high electric
220: conductivity since there is no shallow impurity related states.
221: Besides the most stable configuation, we also consider other
222: metastable configuations where one more F atom
223: adsorbs on other B sites in $S_F^N$. All these metastable
224: structures are found to be p-type
225: semiconductors as $A_F^B$.
226:
227: Based on our calculations, we suggest that the experimentally observed
228: three orders decrease in resistivity of F-doped BNNTs might be due to
229: F adsorption instead of F substitution. One may concern the stability
230: of the adsorption of F atoms on BNNTs since experimental studies
231: indicated that the doped fluorine escapes easily when the tube is
232: exposed to air or under standard beam irradiation in the electron
233: microscope.\cite{BNF}
234: Our results indicate that the adsorption of F on BNNTs is as easy as
235: the substitution of F for N in BNNTs in the B rich environment, and
236: the formation of $A_F^B$ is much easier than that of $S_F^N$ in the N rich
237: environment, as can be seen from their formation energies.
238: Our nudged elastic band (NEB) \cite{NEB} calculations indeed show that
239: the F dissociation in single-walled BNNTs with adsorption
240: of F atoms is easier than that in $S_F^N$ (with dissociation energy
241: $3.44$ eV and $5.86$ eV for $A_F^B$ and $S_F^N$ respectively).
242: However, Tang {\it et al.} synthesized F-doped BNNTs through
243: the introduction of F atoms at the stage of multiwalled nanotube
244: growth \cite{BNF} instead of doping F after the stage of BNNTs
245: growth. We suppose
246: that F adsorption occur in the stage of BNNTs
247: growth, the F adsorbed on the inner walls of the multiwalled BNNTs
248: should be stable.
249: However, our calculations don't exclude the possibility of F
250: substitution: F substitutional doping might also occur along with F
251: adsorption, especially in the B rich environment. In fact, F
252: substitutional doping destroys the six-numbered BN atomic ring, which
253: is in accord with experimental findings.\cite{BNF}
254:
255: To summarize, first-principles calculations have been performed to
256: study the structural and electronic properties of fluorinated BNNTs.
257: F atoms prefer to substitute N atoms, resulting in substantial
258: structural changes of the BN layers but little change in
259: conductivity. The adsorption of F on B
260: sites is more favourable than
261: that on N sites due to the large electronegativity of F.
262: BNNTs with adsorbed F atoms are $p$-type semiconductors with
263: enhanced conductivity than the pristine nanotubes. Our results clearly
264: demonstrate that F-doped multiwalled BNNTs synthesized experimentally
265: are most likely $p$-type semiconductors. The present study also
266: suggests that care should be taken when predicting from intuition
267: doped semiconductors' electronic properties.
268: Further experimental studies on F-doped BNNTs are awaited to
269: confirm our results.
270:
271:
272: This work is partially supported by the National
273: Project for the Development of Key Fundamental Sciences in China
274: (G1999075305), by the National Natural Science Foundation of China
275: (50121202, 10474087), by the USTC-HP HPC project, by the EDF of
276: USTC-SIAS, and by the SCCAS.
277:
278: \clearpage
279:
280: \begin{thebibliography}{99}
281:
282: \bibitem{BNNT1} A. Rubio, J. L. Corkill, and M. L. Cohen,
283: Phys. Rev. B {\bf 49}, R5081 (1994);
284:
285: \bibitem{BNNT2} N. G. Chopra, R. J. Luyken, K. Cherrey,
286: V. H. Crespi, M. L. Cohen, S. G. Louie, and A. Zettl, Science {\bf 28}, 335
287: (1994);
288: A. Loiseau, F. Willaime, N. Demoncy, G. Hug, and H. Pascard,
289: Phys. Rev. Lett. {\bf 76}, 4737 (1996).
290:
291: \bibitem{BNH1} X. J. Wu, J. L. Yang, J. G. Hou, and Q. S. Zhu,
292: Phys. Rev. B {\bf 69}, 153411 (2004).
293:
294: \bibitem{BNC} K. Suenaga, C. Colliex, N. Demoncy, A. Loiseau,
295: H. Pascard, and F. Willaime, Science {\bf 278}, 653 (1997);
296: W. Mickelson, S. Aloni, W. Q. Han, J. Cumings, and A. Zettl, {\it ibid} {\bf
297: 300}, 467 (2003).
298:
299: \bibitem{BNF}C. Tang, Y. Bando, Y. Huang, S. Yue, C. Gu, F. Xu, and
300: D. Golberg, J. Am. Chem. Soc. {\bf 127}, 6552 (2005).
301:
302: \bibitem{doping} S. Limpijumnong, S. B. Zhang, S. H. Wei, and
303: C. H. Park, Phys. Rev. Lett. {\bf 92}, 155504 (2004).
304:
305: \bibitem {LDA}D. M. Ceperley and B. J. Alder,Phys. Rev. Lett.{\bf 45},
306: 566(1980);
307: J. P. Perdew and A. Zunger, Phys. Rev. B {\bf 23}, 5048
308: (1981).
309:
310: \bibitem {vasp} G. Kresse and J. Hafner, Phys. Rev. B {\bf 47}, 558 (1993).
311:
312: \bibitem {ultrasoft}D. Vanderbilt, Phys. Rev. B {\bf 41}, 7892 (1990).
313:
314: \bibitem{BNH2} X. J. Wu, J. L. Yang, J. G. Hou, and Q. S. Zhu,
315: J. Chem. Phys. {\bf 121}, 8481 (2004).
316:
317: \bibitem{NEB}G. Mills, H. Jonsson, and G. K. Schenter, Surface Science,
318: {\bf 324}, 305 (1995).
319: \end{thebibliography}
320:
321: \clearpage
322:
323: \begin{table}[!hbp]
324: \caption{Defect formation energy (E$_f$) for various doped BNNTs (We regard
325: the BN sheet as a nanotube with radius $R=\infty$).
326: For each of the three nanotubes, the first and second
327: lines are the formation energies in the B rich or N rich
328: environment respectively. Please refer to the text for the
329: definition of $S_F^B$, $S_F^N$, and $A_F^B$.
330: Energy is in eV.}
331: \begin{tabular}{ccccc}
332: \hline
333: \hline
334: & & E$_f$($S_F^B$)+$\mu_F$ &E$_f$($S_F^N$)+$\mu_F$&
335: E$_f$($A_F^B$)+$\mu_F$ \\
336: \hline
337: BN(10,0)&B rich& 5.887 &$-$3.891 &$-$3.846 \\
338: &N rich& 2.851 &$-$0.856 &$-$3.846 \\
339: \hline
340: BN(5,5) &B rich& 5.851 &$-$3.907 &$-$3.995 \\
341: &N rich& 2.868 &$-$0.923 &$-$3.995 \\
342: \hline
343: BN sheet&B rich& 7.196 &$-$2.453 &$-$3.188 \\
344: &N rich& 3.993 & 0.750 &$-$3.188 \\
345: \hline
346: \hline
347: \end{tabular}
348: \label{table1}
349: \end{table}
350:
351:
352: \clearpage
353: \begin{figure}[!hbp]
354: \includegraphics[width=7.5cm]{fig1.eps}
355: \caption{(Color online) Three different configurations of F-doped
356: BN(10,0) nanotubes:
357: (a) one B
358: substituted by one F atom ($S_F^B$), (b) one N substituted by one F
359: atom ($S_F^N$), and (c) one F adsorbed on a B atom ($A_F^B$).
360: Both top view and side view are shown.}
361: \label{fig1}
362: \end{figure}
363:
364:
365: \begin{figure}[!hbp]
366: \includegraphics[width=7.5cm]{fig2.eps}
367: \caption{(Color online) Band structures for (a) pristine
368: BN(10,0) nanotube, (b) BN(10,0) nanotube with one N
369: substituted by one F atom, and (c) BN(10,0) nanotube with
370: one F atom adsorded on one B atom. The local DOS for the states
371: induced by dopants are shown in the insets.
372: The valence top or the Fermi level is taken as zero-enegy point in
373: (a) and (b) or (c) respectively.}
374: \label{fig2}
375: \end{figure}
376:
377:
378: \begin{figure}[!hbp]
379: \includegraphics[width=7.5cm]{fig3.eps}
380: \caption{(Color online) Three different configurations BN(10,0) nanotubes doped
381: with two F atoms:
382: (a) two F atoms adsorb on two B sites,
383: (b) two substitutional F atoms, and
384: (c) one F atom adsorbed on one surrounding B atoms around the
385: substituted N site.
386: Both top view and side view are shown.
387: Total DOS for these three structures are shown in (d), (e), and
388: (f) respectively.
389: The Fermi level (or valence top) is taken as zero-enegy point in
390: (d) and (f) (or (e)).}
391: \label{fig3}
392: \end{figure}
393:
394: \end{document}
395: