Professional Documents
Culture Documents
Initial Initial
heating cooling Observed T~ reduction
Laser Light absorbed by sample rate rate heating at Full Ha& f-
pulse width Power Power/area Energy Full power full power power power
(@sec) (W) (W/cm ) (10 J) (' K/p, sec) (' K/@sec) (' K) (' K) (' K)
40 0. 3 3 12 0 027
~ 0. 020 ~0 45
0. 75 17 4. 5 0. 15b 0. 11" ~] ~2 5
T~ —
(light off) T~ (light on). Temperatures measured heating effect.
at beginning of light pulse. This quantity includes the "Calculated.
B. Measurements below 7.2 K
275 A FILM T ~ Tc Below 7. 2 'K (T, with no light) an IIt drop across
the sample is obsexved with the 40- psec-pulse
lasex' whose pulse shape 18 a nearly exact x'eplica
of the laser light output recorded by the photode-
- LASER
tector. This behavior requires a response time
of & 5 psec which is a factor of ~ 10 faster than
the thermal response time observed at 7. 8 K.
For the 40-psec-pulse laser (where ET„„«„
-0. 45 K) the IIt drop is observed at temperatures
7, 84K
as far as 2'K below T, (no light). This is well
~F beyond the range in temperature where the heating
effects should be observed. The IB drop at 4. 4'K-
and the laser pulse shape are shown in Fig. 3.
20'. sec Evidence that this is not a new form of the
thermal effect observed at 7. 8 'K lies in the fact
that for rbetween T,' and r, -wr„„„„both the
FlG. 2. 275-A-thick sample resistance at 7. 84 'K and fast response and the slow thermal response can
laser output vs time. Horizontal scale 20 psec/div. Ver- be observed. This is shown in the photograph
tical scale for sample resistance is dB/B = 0. 85&& 10 3/div
or, equivalently, 6Th, «&~ = 0. 14 ' K/div. The laser out- reproductions in Fig. 4. Note that at T= 6. 8'K
put was recorded by a fast photodetector.
= T, —4T„„«„,a slight asymmetry appears in the
pulse with the trailing end slightly larger in am-
plitude than the leading end. This is the shape
expected for the contribution due to heating (see
that shown in Fig. 2. This assumption was con- Fig. 2). Furthermore, the temperature at which
firmed by the results of the thick film experiments this occurs agrees well with that expected for the
(see below). heating effect. As the temperatuxe is raised the
About four samples with thickness 250-300 A trailing edge (heating peak) increases rapidly but
were measured. All showed temperature increases the leading edge peak can be discerned separately*
of 0. 3-0. 5 'K with full power illumination from up to T='7 K. ~
the 40- &sec-pulse laser. This is the magnitude The last pulse in this sequence, at T=7 K,
and uncertainty of the heating effects obtained in allows a further comparison with the exyected
these experiments. results for heating. The laser pulse should in-
Two samples of thickness 1500-2000 A were also crease the sample temperature at the trailing
measured. Heating values in the range 0. 3-0. 5'K
were also obtained. This result shows that the
thermal response of the film to the laser light is
not dependent on the film thicknesses used in
these experiments. (The heat capacity of such P75A FILM
thin films is of minor importance in determining
the thermal response of the system. ) This fact
will be used in distinguishing optical from heating
effects in the superconducting behavior.
For the 6- psec-pulse laser the measured tem- LASER
perature rise in a 270-A film at 7. 7'K was -1'K.
This is consistent with the x'esults of the 40- psec-
pulse laser when account is taken of the difference 5.35 K
in power/area at the sample in the two cases (see
Table I). Because of a much higher electrical
noise level from the 6- &sec-pulse laser the exact 20', sec
shape of the sample heating curve could not be
observed. The values of the initial heating and
cooling rates, given in TaMe I, have been calcula-
FIG. 3. 275-A-thick sample resistance Oovrer curve)
ted by scaling the results of the 40- psec-pulse and laser output (upper curve) vs time. Horizontal scale
laser. These calculations are supported by the 20 IMsec/div. For T=4. 4 'K the sample resistance during
consistency in the measured total temperature laser pulse is 0. 06Bg. For T=5. 35'K the sample resis-
increases for the two lasexs. tance during laser pulse is 0. 158@.
2192 L. R. TE 8 TARDI
T= 7. 02 'K the film is driven fully normal for most
of the duration of the laser pulse. At 7. 08 K the
275A FILM
laser pulse raises the sample temperature to
7. 43 'K and the sample must remain in the normal
state for a time necessary to achieve a cooling
of 0. 23 'K or 0. 23 'K/0. 35 'K of the total tempera-
ture rise. Prom the thermal behavior shown in
I ASER
Fig. 2 this time is = 50 p. sec, which is in reasonable
agreement with the results in Fig. 5.
Note also from Fig. 5 thatbelow V. 2'K the
response is slow and first appears at the trailing
edge of the laser pulse as expected for heating
7.02 oK effects.
An important conclusion of these results is that
the heating effects for T& T, are also found for
T& T, . This excludes as possible explanations (i)
changes in the Kapitza thermal boundary resistance
6.97 oK at the film-substrate interface between the normal
and superconducting states and (ii) other anomalous
temperature dependences of the thermal properties
6.87 K
which would cause the heating below T, to be grossly
6.80 K different than the heating above T, .
5.10 'K C. Nonlinear Effects
20'. sec Although the transition in zero light is only a
few millidegrees wide, the photoinduced resistive
state shows a very broad superconducting to full
FIG. 4. 275-A-thick sample resistance vs time normal-state transition. This is shown in Fig. 6
(20psec/div). Normal-state resistance is 6 vertical which gives the ratio of normal-state resistance,
units. R/R„, observed during the 40-psec-pulse laser
as a function of the initial temperature. Note that
for the full laser output the resistive state with
end of the pulse to V. 47'K. With an initial cooling fast time response appears just above 4 K and is
rate of 0. 02'K/itsec (see Fig. 1 or Table I) the observed up to = 6. 8'K where the heating peak
sample should remain in the normal state for = 13
psec after the completion of the laser pulse. This
is confirmed in the results.
The most important evidence elucidating this
effect comes from the thick-film experiments. For 2000A FILM
film thicknesses d- 1500-2000 A, one has d = 2$
and d= 85, where g and 5 are the superconducting
coherence length and optical penetration depths,
respectively. For these films superconduction
will occur along the "back" half of the film thick- LASER
ness presumably uninfluenced by the direct inter-
action with light or with the effects of light carried
over the coherence length. However, as found in 7.08 K
LASER
DaTA FOR 40 p. Sec HEATING y
PULSE LASER
ll
FULL HALF
POWER
0—
1.
0.8—
: ':::
HEATING PEAK
OBSERVED
1700 A 275 A
FIG. 6. Sample 8/RN versus temper-
ature during illumination by the 40-psec-
FILM FILM
K pulse laser beam. Temperatures are
0.6— those recorded in zero light.
I
/
0.4
0.2
appears. The transition to the full normal state Also shown in Fig. 6 are the results for a thick
occurs abruptly after this. film different from that discussed above. Again
When the intensity of the 40- &sec-pulse laser only the heating effect is observed below T, .
is reduced by 50% (by a calibrated transmission D. 6-@sec-Laser Results
filter) the data represented by the dashed line are
obtained. Both the measured temperature rise The sample voltage signals obtained with the
above T, and the range of the heating peak below 6- &sec-pulse laser showed "ringing" at the
T, are reduced by a factor of 2. However, the leading edge and a peak at the completion of the
range of the anomalous photoinduced resistive laser pulse. These are electrical pickup trans-
state is reduced by more than a factor of 2. In ients from the laser power supply. There was
the range of the anomalous behavior the slope of also a slight delay (& 1 psec) of the sample signal
the data line for the full laser output is one-fourth relative to the laser pulse which was introduced by
that obtained for the 50% laser output. The anom- the signal amplifier.
alous depressions of T, (i. e. , corrected for the With the full power of the 6- &sec-pulse laser
heating effect) are in the approximate ratio 3. 5/1. the near full normal-state resistance was induced
The shape of the curve, R/R„vs T, however, at temperatures down to & 3 'K (see Fig. 7). (The
is influenced somewhat by the transverse non- laser pulse heating at 7. V'K was-1'K. ) Only
uniformity of the laser light intensity over the for T&6. 2 K did the sample remain in the normal
sample length. This was due in part because the state for a short time after the completion of the
laser beam diameter was only slightly larger than laser pulse. This behavior for T&6. 2 K was con-
the length of the measured sample. A further sistent with that expected for heating and calculated
complication arises from (- 500-1000 A) polishing from the initial cooling rates shown in Table I.
grooves on the surface of the sapphire substrate With the intensity of the 8-psec-pulse laser
within which oblique illumination will occur. The reduced by a factor of 2 the onset of the resistive
measured R/R„ is therefore an average value for state occurred at about 4. 5 'K. In this case one
the sample between the potential probes. An could study the photoinduced resistive state where
estimate of the nonuniformity of illumination is R/R„&1. A puzzling result of this study was the
given by the width of the superconducting transition occurrence of delay times of several microseconds
which is induced only by the heating effect of the between the beginning of the laser pulse and the
laser beam. For the thick film, and for the thin onset of the resistive state. This is shown in Fig.
film where the heating "peak" could be discerned, ' 'K).
8 for T = 5. 15 'K (bT„,~„,g- —, ,
is e superconducting ener a
ergy gap. No
measurements ave b een made with g
have
e lasers described above. '
e lain the results
Phenomenologieally y one can explain
nec in the conversion of optical
o en-
ergy, absorbed b y th e electrons to phonons ulti-
o e hhea, t sink.. Th'
This effect couM
6.5 oK arise if th e electron-phonon relax
relaxation time after
the optical-absor tio
su ercon-
e optical penetration depth and thee supercon-
ducting coherencee len h. For thicker films onl y
eng th.
thee th ermal effect is seen. Th' ld 6.85 K
No explanation wn ~o The
for these effect s ls kn own. FIG. 275-A-thick sample resist
magnitude of thee laaser power is not exec ti
"
large (photon flux -10"/ cm ' sec).. The electr ic 18 8 dlvlslon8
ma -state resistance
].5' Kand 4 dlvlslons for T= 6. 85 K
for T=s.. 15
DESTRUCTION QF SUPERCONDUCTIVITY BY LASER LIGHT 2195
1.0-
0.2— l
r
0 I I
and taking n as the carrier concentration and A. as the superconducting gap 2~. Experiments'6 show
the area of the laser beam, then the average in- that these quasiparticles decay (at least in part)
crease in energy per carrier after time v is -Pv/ by emission of phonons of energy 24 even when
nAd. For the experiments described above P/A the excitation energies are several times 24.
=3 to 17 W/cm~, r 10 ~ s-ec (the delay time ob- Our experimental 7™ 10 sec may be the decay
served with the 6-psec-pulse laser), n= 2. 54 time v', J. in the optical-absorption process which
x10 ' d=275 A, and one calculates an in-
cm, describes the conversion of the excited electron
crease in energy for the electron gas of 3 to 18 - energy (after being shared by many carriers) to
K due to the absorption of the laser beam. This, phonons, somewhat similar to quasiparticle de-
presumably, leads to the destruction of the super- cay. ' Shortly after photon absorption, then, the
conducting state. optical energy may be shared by a number of
Physically one might expect the energy of the carriers all of which are excited by 24. These
absorbed 2-eV photon to be shared initially with "pumped" electrons would be quite numerous,
other electrons very rapidly, say, in a time 7'«, amounting to 10-50% of the total number of free
followed by a conversion to phonons over a longer carriers. The destruction of superconductivity
time 7', ~ and finally transmitted as heat to the would occur when the number of excited electrons
surroundings in a thermal response time. (Re- exceeds some critical value.
garding the assumption v', & & t„ -
it is noted that the
highest single phonon energies are Debye temper- ACKNOWLEDGMENTS
ature -
0. 01 eV for Pb compared to 2 eV for the -
photons. ) Normally one expects v„and r, ~«10 ' The author is indebted to J. J. Hauser for
sec and very little heating of the electron gas preparation of the films, A. G. Chynoweth for
above the bath temperature occurs. In the super- helpful suggestions related to the proposed ex-
conducting state longer lifetimes are expected" planation, and William Royer for technical assist-
-
for quasiparticles excited (by millivolts) across ance.
A. H. Dayem and J. J. Wiegand, Phys. Rev. 155, 419 ~The transmittance was measured for white light at
(1967); P. W. Anderson and A. H. Dayem, Phys. Rev. room temperature. Some changes may occur at low tem-
Letters 13, 195 (1964); I. I. Eru, V. A. Kashchei, and peratures (partly from the increase in electrical conduc-
S. A. Peskovatskii, Zh. Eksperim. i Teor. Fiz. 58, 778 tivity) but changes in the optical parameters due to super-
(1970) [Sov. Phys. JETP 31, 415 (1970)]. conductivity should be negligible at these frequencies.
L. P. Gor'kov and G. M. Eliashberg, Zh. Eksperim. For thinner films the resistance ratio p(300 K)/p(8 K)
i Teor. Fiz. 54, 612 (1968); 55, 2430 (1968)[Sov. Phys. decreases and the magnitude of d lnR/dT for T just above
JETP 27, 328 (1968); 28, 1291 (1969)]; S. A. Peskovat- T~ becomes too small to allow an accurate measurement
skii, ibid. 58, 897 (1970)[31, 482 (1970)]. of the film heating during laser illumination.
3Further details are given in L. R. Testardi, W. A. This is of the order of magnitude of the thermal re-
Reed, P. C. Hohenberg, W. H. Haemmerle, and G. F. sponse time expected for the copper block (see Fig. 1).
Brennert, Phys. Rev. 181, 800 (1969). The amplitude of the leading edge peak is obviously
2196 L. R. TE S TARD I
being increased by the heating contribution as T increases. 6-@sec laser pulse. A greater change would occur if
The shape of the sample signal obtained with the 6-psec- each photon led directly or indirectly (e. g. , by phonon
pulse laser showed some irregularities which could not emission) to much more than one excited electron state.
be accounted for by transient pickup or the electronic (See further discussion in Sec. IV. )
circuitry. A complicating feature of these experiments For a free electron in the alternating electric field of
is that the mean free path of thermal phonons in the sap- the laser the amplitude of the oscillatory velocity would
phire substrate was -1 cm. Ballistic phonon effects be ™1 cm/sec. The dc critical velocity for the destruction
(complicated by a possible dimensional resonance in the of superconductivity is 105 cm/sec.
substrate) may occur. These effects also hinder a calcu- It would also be important to determine whether bal-
lation of the thermal response time of the system. The listic phonons in the sapphire substrate (from the heat
thermal diffusion distance for sapphire in 10 psec is- pulse) are involved in this effect.
sapphire phonon mean free path and this is comparable to ~4J. R. Anderson and A. V. Gold, Phys. Rev. 139,
or greater than the dimensions of the substrate. These A1459 (1965).
phonon effects may be important for the superconducting l~J. R. Schrieffer and D. M. Ginsberg [Phys. Rev.
state of the film and the effects discussed in this paper. Letters 8, 207 (1962)] have calculated the recombination
~A Kapitza thermal boundary resistance which depends time for thermally excited quasiparticles in Pb at l. 44 K
on light intensity, of course, is not precluded. to be -0. 04 @sec. bee also A. Rothwarf and M. Cohen,
One may also question the assumption, so far implicit, Phys. Rev. 130, 1401 (1963). Quasiparticle lifetimes
that the observed effect in the normal state is due to or- -
of psecs have been observed in Al [see K. E. Gray,
dinary heating. A. R. Long, and C. J. Adkins, Phil. Mag. 20, 273
Within the skin depth the photons are absorbed at a (1969)].
rate -10 4-10 ~/cm sec. If each photon were absorbed R. C. Dynes, V. Narayanamurti, and M. Chin, Phys.
by an electron whose excited state lifetime was greater Rev. Letters 26, 181 (1971).
than several microseconds the equilibrium carrier con- Other processes may contribute to the experimental
centration would be altered by less than 1% during the delay time.