1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: % INSTITUTE OF PHYSICS PUBLISHING %
3: % %
4: % `Preparing an article for publication in an Institute of Physics %
5: % Publishing journal using LaTeX' %
6: % %
7: % LaTeX source code `ioplau2e.tex' used to generate `author %
8: % guidelines', the documentation explaining and demonstrating use %
9: % of the Institute of Physics Publishing LaTeX preprint files %
10: % `iopart.cls, iopart12.clo and iopart10.clo'. %
11: % %
12: % `ioplau2e.tex' itself uses LaTeX with `iopart.cls' %
13: % %
14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
15: %
16: %
17: % First we have a character check
18: %
19: % ! exclamation mark " double quote
20: % # hash ` opening quote (grave)
21: % & ampersand ' closing quote (acute)
22: % $ dollar % percent
23: % ( open parenthesis ) close paren.
24: % - hyphen = equals sign
25: % | vertical bar ~ tilde
26: % @ at sign _ underscore
27: % { open curly brace } close curly
28: % [ open square ] close square bracket
29: % + plus sign ; semi-colon
30: % * asterisk : colon
31: % < open angle bracket > close angle
32: % , comma . full stop
33: % ? question mark / forward slash
34: % \ backslash ^ circumflex
35: %
36: % ABCDEFGHIJKLMNOPQRSTUVWXYZ
37: % abcdefghijklmnopqrstuvwxyz
38: % 1234567890
39: %
40: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
41: %
42:
43:
44: \documentclass[preprint,12pt]{revtex4}
45: %\newcommand{\gguide}{{\it Preparing graphics for IOP journals}}
46: %Uncomment next line if AMS fonts required
47: %\usepackage{iopams}
48: \usepackage{graphicx}% Include figure files
49:
50: \begin{document}
51:
52: \title[Kondo physics and superconductivity]{Kondo resonance enhanced supercurrent in single wall carbon nanotube Josephson junctions}
53:
54: \author{K Grove-Rasmussen}
55: \email{k_grove@fys.ku.dk}
56: \author{H Ingerslev J\o rgensen and P E Lindelof}
57:
58: \affiliation{Nano-Science Center, Niels Bohr Institute, University
59: of Copenhagen, Denmark}
60:
61: \begin{abstract}
62: We have contacted single wall carbon nanotubes grown by chemical
63: vapor deposition to superconducting Ti/Al/Ti electrodes. The device,
64: we here report on is in the Kondo regime exhibiting a four-fold
65: shell structure, where a clear signature of the superconducting
66: electrodes is observed below the critical temperature. Multiple
67: Andreev reflections are revealed by sub-gap structure and a narrow
68: peak in the differential conductance around zero bias is seen
69: depending on the shell filling. We interpret the peak as a proximity
70: induced supercurrent and examine its interplay with Kondo
71: resonances.
72: \end{abstract}
73:
74: %Uncomment for PACS numbers title message
75: %\pacs{00.00, 20.00, 42.10}
76: % Keywords required only for MST, PB, PMB, PM, JOA, JOB?
77: %\vspace{2pc}
78: %\noindent{\it Keywords}: Article preparation, IOP journals
79: % Uncomment for Submitted to journal title message
80: %\submitto{\JPA}
81: % Comment out if separate title page not required
82: \maketitle
83:
84: \section{Introduction}
85: Single wall carbon nanotubes (SWCNT) have been under intense
86: investigation for more than a decade due to their unique mechanical
87: and electrical properties. They are one-dimensional conductors with
88: two conducting modes and when contacted to electrodes they behave as
89: quantum dots, where phenomena as Fabry-Perot interference
90: \cite{Liang}, Kondo effect \cite{Nygaard} and Coulomb blockade
91: \cite{tans, bockrath} have been observed. The different regimes can
92: be accessed by indirectly controlling the coupling between the SWCNT
93: and the contacts by choice of contact material. In the low
94: transparency regime (closed quantum dot), electron transport is
95: blocked except at charge degeneracy points, where electrons can
96: tunnel through the SWCNT only one by one due to Coulomb blockade.
97: When the transparency is increased (intermediate regime) cotunneling
98: of electrons becomes possible, which can give rise to the Kondo
99: effect. Finally in the high transparency regime electrons on the
100: SWCNT are not well defined and the phenomena observed (broad
101: resonances) are due to interference of electron waves (Fabry-Perot
102: interference). The possibility of contacting carbon nanotubes to
103: superconducting leads \cite{Kasumov,Morpurgo,Krstic2003,haruyama}
104: opens up for the interesting study of effects related to
105: superconductivity such as supercurrent and multiple Andreev
106: reflections \cite{Buitelaar} together with the above mentioned
107: phenomena. In the Fabry-Perot regime recent experiments have
108: confirmed that the supercurrent is modulated by the quantized nature
109: of the energy spectrum of the SWCNT, {\em i.e.}, a Josephson field
110: effect transistor with only two modes
111: \cite{Jarillosupercurrent,hij}. Experimental access to the less
112: transparent regimes \cite{Ralph,MSS2006Proc,vanDam,Cleuziou} is even
113: more interesting due to the possibility of probing the competition
114: between effects related to Coulomb blockade and superconductivity.
115: Coulomb blockade generally suppresses the supercurrent and
116: interesting phenomena such as $\pi$-junction behavior has been
117: observed in the closed quantum dot regime in nanowires
118: \cite{vanDam}. For more transparent devices the supercurrent can be
119: enhanced due to Kondo physics \cite{Glazman1989JETPL}. Both
120: supercurrent and Kondo physics are two extensively studied manybody
121: effects in condensed matter physics and SWCNT Josephson junctions
122: thus give a unique possibility to examine their interplay. The
123: supercurrent is predicted to coexist with Kondo resonances provided
124: that the Kondo related energy scale $k_B T_K$ is bigger than the
125: superconducting energy gap $\Delta$
126: \cite{Glazman1989JETPL,Siano2004PhRvL,Choi2004PhRvB}. The importance
127: of the ratio between these parameters has recently been addressed in
128: different measurements \cite{BuitelaarKondo}. In this article we
129: extend this investigation to SWCNTs contacted to superconducting
130: leads in order to experimentally probe the interplay between
131: supercurrent and Kondo physics. We investigate the gate dependence
132: of a narrow zero bias conductance peak interpreted as a proximity
133: induced supercurrent and show that the observation of a supercurrent
134: in Kondo resonances depends on the ratio between the two energy
135: scales ($k_BT_K/\Delta$) with a crossover close to 1 qualitatively
136: consistent with theory.
137:
138:
139:
140:
141: \section{Experimental Details}
142: SWCNTs are grown from catalyst islands consisting of Fe-oxide and
143: Mo-oxide supported by aluminum nano-particles \cite{KongCVD}. Growth
144: is performed by chemical vapor deposition at $850$\,$^\circ$C with a
145: controlled flow of gasses Ar: $1$\,L/min, H$_2$: $0.1$\,L/min,
146: CH$_4$: 0.5\,L/min. During heating the furnace is kept under an Ar
147: and H$_2$ flow, whereas cooling is done in Ar. We reduce cooling
148: time by air-cooling the furnace. The substrate is a doped silicon
149: wafer (used as back gate) with a $500$\,nm SiO$_2$ layer on top.
150: Pairs of superconducting electrodes of Ti/Al/Ti (5/40/5\,nm) are
151: defined directly on top of the SWCNT by electron beam lithography
152: followed by optical lithography to define the Cr/Au bonding pads.
153: The first titanium layer of the metallic trilayer ensures good
154: contact to the SWCNT, whereas the thicker middle aluminum layer is
155: the actual superconductor in the device. Finally, the top layer of
156: Ti is intended to stop oxidation of the aluminum. The gap between
157: the source and drain electrode is typically around $0.5$\,$\mu$m. In
158: the same evaporation process a four-probe device is made next to the
159: S-SWCNT-S devices, which is used to measure the transition
160: temperature $T_c=760$\,mK and the critical field around
161: $B_c=100$\,mT of the superconductor. From Bardeen-Cooper-Schrieffer
162: (BCS) theory we deduce the superconducting energy gap
163: $\Delta=1.75k_B T_c=115$\,$\mu$eV. The devices are cooled in a
164: $^3$He-$^4$He dilution fridge with a base temperature around
165: $30$\,mK and we use standard lock-in techniques.
166:
167:
168:
169: \section{Sub-gap structure and supercurrent}
170: \begin{figure*}
171: \center
172: \includegraphics[width=0.8\textwidth]{Fig1}% Here is how to import EPS art
173: \caption{\label{fig1} (a) Bias spectroscopy plot of
174: a SWCNT device at 30\,mK showing a four-fold Coulomb blockade shell
175: structure. The numbers indicate the additional hole filling, where
176: big diamonds correspond to filled shells. Four Kondo resonances
177: K1-K4 are identified. A magnetic field of $180$\,mT is applied to
178: suppress superconductivity in the leads. (b) Bias spectroscopy plot
179: for the same gate range as (a) but with the leads in the
180: superconducting state. A sub-gap structure emerges, most clearly
181: visible at $\pm\Delta/e$ and $\pm 2\Delta/e$ (horizontal green dashed lines). The green dashed rectangles are the gate voltage
182: regions shown in Fig.\ \ref{fig2star}(c) and \ref{fig2star}(d).}
183: \end{figure*}
184: At room temperature the current through the device is gate
185: dependent, which reveals that the SWCNT is semiconducting.
186: Figure \ref{fig1}(a) shows a bias spectroscopy plot at 30\,mK of the SWCNT device in the Kondo regime. It
187: is measured at negative gate voltages and thus transport takes place
188: through the valence band, i.e., tunneling of holes. The
189: superconductivity in the leads is suppressed by a relative weak
190: magnetic field of $180$\,mT ($>$$B_c$). The plot shows Coulomb
191: blockade diamonds \cite{KouwenhovenQDReview} with a characteristic
192: pattern of three small diamonds followed by a bigger diamond as
193: indicated with white lines for hole filling 13-16. Such Coulomb
194: blockade diamond structure indicates a four-fold degenerate shell
195: structure due to spin and orbital degrees of freedom, where each
196: shell contains two spin-degenerate orbitals \cite{Sapmaz}. The
197: filled shells corresponding to the big Coulomb blockade diamonds are
198: marked in Fig.\ \ref{fig1}(a) by the additional number of holes on
199: the SWCNT quantum dot. Due to intermediate transparency contacts
200: significant cotunneling is allowed which tends to smear the
201: features. The charging energy and the level spacing between the
202: shells are estimated as half the source-drain height of the very
203: faintly visible small diamonds and by the additional source-drain
204: height of the big diamonds \cite{Franceschi} giving $U_c\sim 3$\,meV
205: and $\Delta E\sim 4$\,meV, respectively (see arrows Fig.\
206: \ref{fig1}). A very rough estimate of the level broadening
207: $\Gamma\sim 1$\,meV and the asymmetry in the coupling to source and
208: drain $\alpha=\Gamma_s/\Gamma_d\sim 0.4$ are extracted from the
209: current plateaus at negative and positive bias (100\,nA and -60\,nA)
210: at the white arrows. The current is modified due to the four-fold
211: degeneracy which enhances the in-tunneling rate with 4 compared to
212: the out-tunneling rate giving
213: $(e/\hbar)4\Gamma_s\Gamma_d/(4\Gamma_s+\Gamma_d)$ and
214: $(e/\hbar)(4\Gamma_d \Gamma_s/(\Gamma_s+4\Gamma_d)$ for negative and
215: positive bias polarity, respectively \cite{Babic}. Furthermore, four
216: Kondo resonances K1-K4 are identified (green arrows) and labeled in
217: ascending order based on their Kondo temperatures. The Kondo
218: temperatures are estimated by fitting the normal state conductance
219: versus $V_{sd}$ (solid line in Fig.\ \ref{fig2}a-b) to a Lorentzian
220: line shape. The half width at half maximum of each Lorentzian fit
221: yields the estimated Kondo temperature giving $T_K\sim 2$\,K,
222: 4.5\,K, 5\,K and 6\,K for K1-K4. For hole filling 13-15, conductance
223: ridges at low finite bias are seen instead of zero bias Kondo
224: resonances. The exact origin of these features are not fully
225: understood, but they are attributed to inelastic cotunneling through
226: two slightly split orbitals in the shell, which might include Kondo
227: physics as well \cite{Paaske}. We note that these lines are not
228: related to superconductivity.
229:
230: Figure \ref{fig1}(b) shows a bias spectroscopy plot at low bias
231: voltages for the same gate range as Fig.\ \ref{fig1}(a) with the
232: leads in the superconducting state. A sub-gap structure clearly
233: appears \cite{Buitelaar}. The peaks in differential conductance at
234: $V_{sd}=\pm 2\Delta/e\sim \pm 230$\,$\mu$V are attributed to the
235: onset of quasi-particle tunneling. At lower bias the transport is
236: governed by Andreev reflections, which are possible due to the
237: intermediate transparency between the SWCNT and the superconducting
238: leads \cite{Andreev}. Features at biases $V_{n} =\pm\ 2\Delta/(e
239: n),n=2,3,...$ are expected due to the opening of higher order
240: multiple Andreev reflection processes as the bias is lowered
241: \cite{klapwijk,bratus}. Peaks at $V_{sd}=\pm \Delta/e\sim \pm
242: 115$\,$\mu$V are clearly seen consistent with the energy gap found
243: from BCS theory. Furthermore, for some ranges in gate voltage a
244: narrow zero bias conductance peak is seen. We note, that this zero
245: bias peak is visible in most of the Kondo resonances and is the
246: subject to analysis below.
247:
248: \begin{figure}
249: \center
250: \includegraphics[width=0.75\textwidth]{Fig2}% Here is how to import EPS art
251: \caption{\label{fig2star} (a-b) Bias spectroscopy plot at 30\,mK with the leads in the normal state
252: ($B=180$\,mT) for Kondo resonance K4 and K1, respectively. (c-d) High resolution data with the leads superconducting for the
253: same gate and bias voltages as in the normal state above. The
254: corresponding gate regions are shown by the dashed green rectangles
255: in Fig.\ \ref{fig1}(b). A sub-gap structure appears at bias voltage
256: $\pm2\Delta/e$ and $\pm\Delta/e$ indicated by the green dashed
257: lines. This is particular clear in the Coulomb blockade region (d)
258: and also seen in the green traces, which are $dI/dV$ versus bias
259: curves extracted from the plots. A zero bias conductance peak (red
260: arrow) is also visible throughout Kondo resonance K4 as shown by the
261: red trace in (d) and for some gate voltages in (c).}
262: \end{figure}
263: These above mentioned effects due to superconductivity are more
264: clearly revealed in high resolution data shown in Fig.\
265: \ref{fig2star}(c) and \ref{fig2star}(d) corresponding to the green
266: dashed rectangles in Fig.\ \ref{fig1}(b). Figure \ref{fig2star}(a)
267: and \ref{fig2star}(b) show the same gate and bias regions as (c) and
268: (d) with the leads in the normal state for comparison. Clearly the
269: sub-gap structure is due to the proximity of the superconducting
270: leads. The peaks at $\pm2\Delta/e$ and $\pm \Delta/e$ are mostly
271: pronounced in the Coulomb blockade region as expected due to lower
272: effective transparency \cite{klapwijk} (d) while being more smeared
273: in the high conducting region of K4 (c). Green curves show bias
274: cuts, where higher order multiple Andreev reflection features are
275: faintly visible in (c). A strong gate dependence of the sub-gap
276: structure is observed when the gate voltage is tuned closer to the
277: Coulomb resonances on each side of K1 from the Coulomb diamond with
278: even occupation. This behavior has previously been observed
279: \cite{Buitelaar} and explained by the interplay between multiple
280: Andreev reflection and a resonant level
281: \cite{Yeyati1997PhRvB,Johansson1999PhRvB}. The zero bias conductance
282: peak indicated by the red arrow is clearly visible in Kondo
283: resonance K4 (red trace in (c)) and also present for some gate
284: voltages in (d). Similar behavior of a zero bias peak in Kondo
285: resonances have been observed in other devices \cite{MSS2006Proc}.
286: We interpret the zero bias conductance peak as being due to a
287: proximity induced supercurrent running through the SWCNT. The
288: magnitude of the supercurrent is estimated by the area of the peak
289: giving a typical value in the order of $I_m\sim 0.2$\,nA. A similar
290: analysis of a zero bias conductance peak as supercurrent has
291: successfully been carried out for an open quantum dot \cite{hij},
292: {\em i.e.}, without Coulomb blockade effects. We note that an
293: alternative interpretation of the origin of the zero bias peak based
294: on quasiparticle current (multiple Andreev reflections) with a
295: cut-off for higher order multiple Andreev reflection processes has
296: also been suggested \cite{vecino}, but in this article the
297: supercurrent interpretation will be pursued. The value of the
298: measured supercurrent is highly suppressed compared to its
299: theoretically expected value for one spin-degenerate level with the
300: deduced asymmetric coupling ($\alpha=0.4$) at resonance
301: $I_0^{res}=2e\Delta/\hbar~(\Gamma_{min}/\Gamma) \sim 8$\,nA in the
302: broad resonance (short) regime $\Gamma>\Delta$, where
303: $\Gamma_{min}=min(\Gamma_s,\Gamma_d)\sim 0.3$\,meV
304: \cite{Beenakker2004condmat}. Despite having two spin-degenerate
305: levels in the SWCNT, only one level is available due to Coulomb
306: blockade.
307:
308: The suppression is partly explained within an extended resistively
309: and capacitively shunted junction model due to interaction of the
310: SWCNT Josephson junction with its electrical environment. The
311: electrical environment for this device is the low (serial)
312: resistance between the Josephson junction and the bonding pads as
313: well as the capacitance between the source and drain bonding pads
314: via the backgate. The low serial resistance turns out to play a
315: crucial role making the device an {\it underdamped} Josephson
316: junction \cite{hij,Steinbach,Vion,Jarillosupercurrent}. This means
317: that the full value of the critical current is never measured due to
318: thermally activated phase-slips. A more intuitive understanding is
319: obtained by the mechanical analog to the Josephson junction of a
320: fictious particle in a tilted washboard potential, where the tilt is
321: given by the current and the phase difference between the
322: superconductors is "running" as the fictious particle moves in the
323: potential. When the particle stays in a minimum of the washboard
324: potential (constant phase), a supercurrent is seen. However, as the
325: current increases the washboard potential is tilted and at the
326: critical current the particle can slide into the next minimum. If
327: thermal excitations are significant this process happens at lower
328: current than the critical current. Since the junction is
329: underdamped, the friction of the particle is low and it thus easily
330: acquires a "run away" phase, that suppresses the supercurrent, which
331: results in a much lower measured supercurrent. In contrast to the
332: reported underdamped carbon nanotube Josephson junction in the
333: Fabry-Perot regime \cite{hij}, the effect of Coulomb blockade and
334: single hole tunneling also contribute to the suppression of the
335: supercurrent, because Cooper pair transport is a two particle
336: tunneling process. A quantitative analysis of the suppression due to
337: Coulomb blockade is outside the scope of this article. This issue is
338: being addressed in a separate work, where the on-chip electrical
339: circuit of the SWCNT Josephson junction has been modified and the
340: effects of Coulomb blockade thus can be compared to theoretical
341: calculations \cite{pi}. Finally, we note that noise effects giving
342: rise to thermally activated phase slips can be analyzed in the
343: overdamped case \cite{Ambegaokar}.
344:
345: \section{Kondo physics and supercurrent}
346: \begin{figure}[t!]
347: {\center
348: \includegraphics[width=0.8\textwidth]{Fig3}% Here is how to import EPS art
349: \caption{\label{fig2} $dI/dV$ versus $V_{sd}$ for Kondo resonances K1-K4 shown in Fig.\
350: \ref{fig1}(a)
351: illustrating the effect of superconducting leads. (a-b) Solid lines and circles are with the
352: leads in the normal and superconducting state, respectively. (c) The four Kondo resonances K1-K4
353: for small bias voltages with the leads in the superconducting state. The measured supercurrent given by the zero bias peak area
354: is zero for the lowest value of the ratio $k_BT_K/\Delta$, while it increases as this ratio is increased.
355: }}
356: \end{figure}
357:
358: We now return to the Kondo resonances K1-K4 and analyze the
359: interplay between Kondo physics and supercurrent. The solid curves
360: in Fig.\ \ref{fig2}(a) and \ref{fig2}(b) show bias cuts with the
361: leads in the normal state through the center of Kondo resonance K2
362: and K4, respectively (see Fig.\ \ref{fig1}(a)). The circles
363: correspond to the behavior when the leads are superconducting. In
364: both cases an enhancement of the differential conductance is
365: observed for bias voltages between $\pm 2\Delta/e$ and a zero bias
366: conductance peak is present due to supercurrent. The measured
367: supercurrent is largest for the Kondo resonance with the highest
368: Kondo temperature (K4), {\em i.e.}, broadest Kondo resonance. In
369: Fig.\ \ref{fig2}(c) bias sweeps at the center of all four Kondo
370: resonances are shown with the leads in the superconducting state.
371: The Kondo temperature normalized by the superconducting energy gap
372: ($k_B T_K/\Delta$) is the important parameter and is given for each
373: resonance in the figure. It is seen that the measured supercurrent
374: vanishes for the lowest ratio (blue circles) while it emerges and
375: increases as the ratio is increased. The crossover is close to
376: $k_BT_K/\Delta\sim 1$.
377:
378: \begin{figure*}[t!]
379: \center
380: \includegraphics[width=0.98\textwidth]{Fig4}% withonlyres Here is how to import EPS art
381: \caption{\label{fig:PeakvsVgate} (a) The measured supercurrent is zero for K1 and shown to increase as a function of
382: $k_BT_K/\Delta$ (K1-K4) . An exponential fit to the data points is given
383: by the solid curve as guideline, where the Kondo temperature is based on the width of the resonance (black diamonds).
384: The red squares are the Kondo temperature obtained from the temperature dependence
385: (K1, K3 and K4). (b) Measured supercurrent versus gate voltage in the range with the Kondo resonances K1-K4,
386: where the additional hole number is given for filled shells as in Fig.\ \ref{fig1}(a).
387: A finite supercurrent is seen in the center of the broader Kondo resonance K2, while being zero in the center of narrowest Kondo resonance
388: K1. This behavior does not reflect the normal state conductance
389: $G_N$ shown for the same gate range below in (c).}
390: \end{figure*}
391: To illustrate this point more clearly, the measured supercurrent
392: versus $k_BT_K/\Delta$ is plotted in Fig.\ \ref{fig:PeakvsVgate}(a)
393: for the four Kondo resonances analyzed above (black diamonds). The
394: solid curve shows a guideline to the eye based on an exponential fit
395: to the available data points. The overall trend is qualitative
396: consistent with existing theory, which predicts a suppression of the
397: supercurrent in the socalled weak coupling regime ($k_BT_K \ll
398: \Delta$), while the supercurrent coexists with Kondo resonances in
399: the strong coupling regime ($k_BT_K\gg\Delta$)
400: \cite{Choi2004PhRvB,Siano2004PhRvL}. The red squares show the
401: measured supercurrent where the Kondo temperature is extracted from
402: the temperature dependence for completeness (only data available for
403: K1, K3 and K4) \cite{GoldhaberGordon1998Nature}.
404:
405: Finally, Fig.\ \ref{fig:PeakvsVgate}(b) shows the measured
406: supercurrent as a function of gate voltage in the range including
407: the four Kondo resonances. It is strongly gate dependent
408: illustrating effects of the four Kondo resonances. Figure
409: \ref{fig:PeakvsVgate}(c) shows the zero bias conductance $G_N$ with
410: the leads in the normal state for the same gate range. The
411: supercurrent {\em does not} directly reflect the normal state
412: conductance $G_N$ as in the case of an open quantum dot
413: \cite{Jarillosupercurrent,hij}, where the supercurrent is uniquely
414: determined by the normal state conductance. This is most clearly
415: seen by comparing Kondo resonance K1 and K2. No supercurrent is
416: present in the center of the Kondo resonance K1 despite the high
417: normal state conductance, while a finite supercurrent is present in
418: Kondo resonance K2 with equally high normal state conductance in
419: contrast to the behavior expected in the open quantum dot regime.
420: Similarly, the Kondo resonances K3 and K4 have almost the same
421: normal state conductances, but very different measured supercurrent.
422: These observations support the above analysis that the interplay
423: between Kondo physics and superconductivity has the fraction
424: $k_BT_K/\Delta$ as the important parameter for the observation of
425: supercurrent in Kondo resonances, i.e., the bias width of the Kondo
426: resonances and not only their normal state conductance is important.
427: We also note that the measured supercurrent in diamond 4, 8 and 12
428: is zero due to Coulomb blockade. The supercurrent in the high $T_K$
429: Kondo resonances can thus be view as being enhanced from the
430: suppressed values by Coulomb blockade and the single spin due to the
431: formation of Kondo resonances. Furthermore, the overall behavior of
432: the supercurrent of these four Kondo resonances also indirectly
433: indicates a $\pi$ to 0 transition of the current phase relation of
434: the Josephson junction as a function of the parameter $k_B
435: T_K/\Delta$ [17,18]. Similar gate voltage behavior has been
436: observed in Ref.\ \cite{Cleuziou}, but the authors do not show the
437: magnitude of the supercurrent as a function of Kondo temperature. We
438: end by noting that for large Kondo temperatures compared to the
439: superconducting energy gap (K4), the observed gate voltage
440: dependence of the supercurrent resembles the behavior of a
441: superconducting single electron transistor, i.e., the supercurrent
442: obtains a sharp maximum at the gate voltage corresponding to the
443: center of the odd diamond \cite{Joyez}.
444:
445: \section{Conclusion}
446: In conclusion SWCNTs have been contacted to superconducting Ti/Al/Ti
447: leads creating SWCNT Josephson junctions. We observe sub-gap
448: structure due to multiple Andreev reflections and a narrow zero bias
449: conductance peak. The zero bias peak is interpreted as a proximity
450: induced supercurrent with a suppressed magnitude due to the
451: underdamped nature of the Josephson junction and Coulomb blockade.
452: We examine its interplay with Kondo resonances and the measured
453: supercurrent is shown to coexist with Kondo resonances which have
454: high Kondo temperatures compared to the superconducting energy gap,
455: while being suppressed when the Kondo temperatures becomes
456: comparable with the superconducting energy gap, qualitative
457: consistent with existing theory.
458:
459: \section*{Acknowledgement}
460: We like to thank Karsten Flensberg, Tom{\'a}\v{s} Novotn{\'y} and
461: Jens Paaske for fruitful discussions. Furthermore, we wish to
462: acknowledge the support of the Danish Technical Research Council
463: (The Nanomagnetism framework program), EU-STREP Ultra-1D and CARDEQ
464: programs.
465:
466: \section*{References}
467: \bibliography{bibl}% Produces the bibliography via BibTeX.
468: \bibliographystyle{unsrt}
469:
470: \end{document}
471: