Disappearance of rapid photoresponse in ultraviolet illumination of Ag–Sb–S films

The photoconduction in Ag0.5Sb0.5S films changes anomalously with the excitation energy. Although the usual instantaneous generation and recombination of photocarriers appear in the resistance of the films for the illumination at a wavelength of 633 nm, the photoresponse becomes slow with time scales of minutes when the illumination is performed at a wavelength of 280 nm. The rapid and slow phototransients are mixed for an intermediate excitation wavelength of 375 nm. In the simultaneous photoexcitation at multiple wavelengths, the response is complex instead of a superposition of the rapid and slow behaviors, indicating the mutual interaction in the photocarrier transport. The ultraviolet (UV) illumination can thereby block the rapid response that should be caused by the visible light. Moreover, the resistance can even increase during the illumination. Although the adsorption of molecules at the film surface plays an important role for the resistance, the anomalous properties are unaffected by the surface condition. They are thus suggested to be the bulk properties of the films, plausibly caused by the defects generated in the UV irradiation.


Introduction
Chalcogenide films and nanoobjects are promising for various applications [1,2] such as photovoltaics [3], thermoelectrics [4,5], sensing [6], for memories [7] and as catalysts for water splitting and fuel cells [8][9][10].In particular, the excellent optoelectrical properties in the energy range of visible lights make them attractive as an absorber in solar cells [11,12].The superb properties are available also for the wide bandgap regime [13].Investigations on the optoelectrical properties of the chalcogenide compounds are, therefore, crucial for improving the performance of the devices and expanding the field of the application.
Concerning the optoelectrical properties, the most fundamental processes in the photoexcitation are the creation of electron-hole pairs when the material absorbs the incident photons and the recombination of these photo-generated charges.In principle, the photocarriers can appear immediately as soon as the photons having an energy exceeding the bandgap of the material arrive.The electrical resistance of the material is thus expected to change almost instantaneously.In [14], however, the generation and recombination of photocarriers in the films of transition-metal chalcogenides Ag-Sb-S and Cu-Sb-S were found to be remarkably slow with time scales of minutes and hours.Similar slow photoresponses are known to occur for thin films and nanowires of ZnO [15][16][17][18][19][20][21].The slow transients in the latter are attributed to the modifications in the transport of photocarriers due to environment effects caused by the surface of the films and nanowires and/or the bulk effects associated with crystalline defects [22][23][24][25].That is, the charge transport in separating and uniting the photo-generated electrons and holes is affected by the electric fields induced by the attachment of water and/or oxygen molecules at the surface as well as by the trapping of the holes at deep-level defects [26].
Clarifying the origin of the slow photoresponse for the Ag-Sb-S and Cu-Sb-S films, i.e. whether it is also caused by the mechanism responsible for the behavior of ZnO, is important in employing this class of materials for optoelectronic applications.We focus our attention on the fact that the photoexcitation in [14] was carried out using a fluorescent light.The luminescence spectrum of fluorescent lights spreads over a wide energy range.High intensity lights are emitted from around 480 nm to about 570 nm in wavelength with additional emissions in ultraviolet (UV) ranges 280-315 nm and 315-400 nm and in the infrared regime.Although the photoresponse is expected, in general, not to depend on the excitation energy so long as it is larger than the bandgap of the material, it is worth confirming whether this is indeed the case.In the case of ZnO, the slow response has been shown to be independent of the excitation energy [18,21].
In this paper, the photoresponse of the Ag-Sb-S alloy films is investigated under monochromatic illuminations.Our central finding is that the generation and recombination of photocarriers are almost instantaneous for the illumination in the visible energy range contrary to the previous report.The extremely slow photoresponse emerges when the photon energy is increased to a corresponding wavelength of 280 nm.Such a dependence on the excitation energy indicates that the slow photoresponse of the transition-metal chalcogenides is unusual and different from that of ZnO.We show further that the rapid and slow photoresponses are not independent of each other, giving rise to complicated transient behaviors in the simultaneous illumination of the visible and UV lights.

Sample preparation
Thin films of Ag-Sb-S alloys were produced by means of the incorporation of transition metals in a hot-wall deposition of chalcogenide films [27].The detail of the preparation method was described previously in [28].Polycrystalline pieces of Sb 2 S 3 and Si substrates covered with a Ag film (thickness 60 nm) were placed in an evacuated quartz tube.A horizontal three-zone furnace was employed to set the temperature of the Sb 2 S 3 source to 400 • C while a descending temperature gradient was staged in the region of the substrates.The films investigated in this work were prepared at substrate temperatures of T s = 340 • C, 358 • C and 383 • C, where T s was adjusted by changing the distance of the substrates from the source.It is pointed out that the annealing effects under strong photoilluminations would be negligible for these films given the high synthesis temperatures.
The alloy formation was conducted for 4 h.The Ag-Sb-S alloys were produced since the Ag atoms from the precursor Dependencies of phototransients in Ag-Sb-S film on illumination wavelength.The film prepared at Ts = 340 • C was exposed to LED lights with λ LED = 280, 633 and 375 nm in the top, middle and bottom rows, respectively.At time t = 0, the illumination was turned on and turned off for the left-and right-hand side columns, respectively.In (a) and (b), the illumination intensity was varied by increasing the operation current I LED of the LED.The rapid transients in (c) and (d) are faster than the measurement interval of 0.1 section In (e) and (f), the duration of the illumination was varied for the red, green and blue curves as indicated.The film was kept in vacuum for the dotted curves, where the duration of the illumination was 10 min.The inset of (d) shows the configuration in measuring the current I.A bias of 10 V was applied between the Au contacts deposited on the film shown as gray areas.
film partially exchanged the Sb atoms in Sb 2 S 3 .Note that Si was chosen as the substrates as it does not participate in the synthesis reaction.Self-regulated reproducibility of the composition and thickness of the Ag-Sb-S films as well as the simplicity are the advantages of the precursor method.Provided that the duration for the alloy formation is long enough for all the Ag atoms to react, an equilibrium composition is realized for the Ag-Sb-S alloy and the growth of the Ag-Sb-S film stops on its own.Here, the Ag atoms can migrate in the alloy over a distance of micrometers [27].
The electrical contacts for the photoconduction measurements were fabricated by depositing a Au layer on the films with a thickness of 100 nm.The contact barrier is expected to be negligible for the Au contacts [14].As illustrated in the inset of figure 1(d), the current flow between the parallel contacts prepared at the two ends of rectangular-shaped sample pieces was monitored to examine the photoconductivity.The width of the pieces was about 2.5 mm.The gap between the electrodes was about 0.7 mm.The bias voltage was set to 10 V.
The photoexcitation was carried out using light emitting diodes (LEDs) at wavelengths of λ LED = 633, 375 and 280 nm.The full width at half maximum of the emission spectra of the LEDs was 15, 9 and 10 nm, respectively.Notice that the wavelengths of 375 and 280 nm correspond to the UV components contained in the spectrum of the fluorescent light.The samples were illuminated with an intensity of ~1 W m −2 at the typical operation current I LED = 50 mA for λ LED = 633 and 375.The illumination was an order of magnitude less for λ LED = 280 nm.For the values of I LED used below, the photo-power changed almost linearly with I LED apart from the sublinear increase for the LED with λ LED = 280 nm for I LED > 50 mA.

Alloy composition and film morphology
The ω-2θ x-ray diffraction scans obtained from the films are plotted in figure 2. Cu Kα radiation was used as the xray source with a wavelength of 0.154 06 nm.The powder diffraction patterns simulated for the source material Sb 2 S 3 and an alloy Ag 0.5 Sb 0.5 S are shown in the bottom panel, where Pearson's crystal data was used for the simulations.The composition of the Ag-Sb-S alloy was identified to be Ag 0.5 Sb 0.5 S, which is a mineral cuboargyrite.It was also found that an Sb 2 S 3 film overgrew for T s = 340 • C due to a catalytic enhancement of the attachment of Sb 2 S 3 to the Ag-Sb-S surface, see [14] for detail.In addition, the small peak at 2θ = 32.9• suggests a mixture of a small amount of the Ag-Sb-S alloy having a different composition for T s = 358 • C. As a matter of fact, the Ag-Sb-S alloy changes the equilibrium composition with T s [14].The alloy composition for T s = 340 • C cannot be identified as the curve in figure 2 is almost completely dominated by Sb 2 S 3 .Ag 0.5 Sb 0.5 S is nonetheless expected to be the dominant composition according to [14].
In figure 3, scanning electron micrographs of the films are shown as top and cross-sectional views.The Ag 0.5 Sb 0.5 S films on the Si substrates are seen in figures 3(a) and (b).The Ag 0.5 Sb 0.5 S films were granular, where the substrates were not covered completely by the Ag-Sb-S films.(It may be worthy to emphasize that the illumination of the Si substrates does not influence the photoconduction properties of the films since the current flowing through the substrates is negligible compared to that through the films.)The typical thickness of the films is found in the cross-sectional views.The overgrown Sb 2 S 3 layer for T s = 340 • C is recognized in figure 3(c), where the underlying Ag 0.5 Sb 0.5 S film is visible in the gap between the Sb 2 S 3 islands.The overgrowth was unstable because of the competition between the reevaporation of Sb 2 S 3 at the high temperature and its catalytically-enabled surface accumulation.The Sb 2 S 3 layer was consequently highly nonuniform in thickness, spreading between zero and several µm.This means that a large portion of the Ag-Sb-S film was not covered by the Sb 2 S 3 layer.This uncovered part of the Ag-Sb-S film was illuminated in the photoexcitation regardless of the shadowing of the light by the Sb 2 S 3 layer.

Photoconduction properties
All samples with the three different T s behaved similarly, i.e. the overgrown Sb 2 S 3 layer for T s = 340 • C caused no difference in the photoconduction properties qualitatively.Given the indifference to the overgrowth, we show below mainly the results obtained from the film with T s = 340 • C as this allows us to rule out surface effects from the origin of the anomalous behavior, as we will explain below.Nevertheless, the photoconduction properties of the layer with the highest T s of 383 • C will be additionally presented for evidencing the negligible role of the overgrowth and the qualitative independence on T s explicitly.
It is emphasized that Sb 2 S 3 is semi-insulating, and so the transport properties were almost completely dominated by the Ag-Sb-S layer even if the Sb 2 S 3 layer coexisted.To be specific, the conduction through the Sb 2 S 3 layer was orders of magnitude smaller than that through the Ag-Sb-S layer.In [14], the sheet conductivity σ s of the Ag-Sb-S films was shown to be the same within the measurement accuracy between the samples with and without the Sb 2 S 3 overlayer.It is worthy to note that Ag is incorporated in Sb 2 S 3 films for the application as the absorbers for thin film solar cells to increase the conductivity by orders of magnitude [29].The role of the Sb 2 S 3 layer in the present work is merely the surface passivation.At the same time, the Sb 2 S 3 layer did not prevent the photoillumination of the underlying Ag-Sb-S layer since its surface coverage was merely partial.In other words, we have taken advantage of the porous overgrowth to test the surface effects in the photoconduction.The presence of the Sb 2 S 3 layer reduced the actual electrical-contact area.This may have resulted in a somewhat erroneous estimate of σ s in its calculation.On the other hand, the fact that the Au contact pads deposited on top of the film established the electrical conduction to the Ag-Sb-S layer evidences that the underlying layer was exposed to the surface.
Figure 1 summarizes the photoresponse manifested in σ s of the film.In figures 1(a) and (b), the UV illumination with λ LED = 280 nm was turned on and turned off, respectively, at time t = 0.In the consecutively carried out measurements, the illumination was intensified stepwise by increasing I LED .The film was illuminated for 5 h in each exposure with a constant I LED and kept in dark for about 19 h between the exposures, where the photoresponses roughly saturated.It is important to point out that the film was placed in vacuum since the attachment of the molecules from air at the surface suppressed the conductance, as will be shown later.
Similar to the case of the photoexcitation using a fluorescent light [14], the phototransients are seen to be slow.The transient time in figures 1(a) and (b) is nearly the same between the turning-on and turning-off of the illumination.They did not change with the photoexcitation intensity.The slow photocarrier generation suggested by the slow increase of σ s distinguishes the effect from the persistent photoconductivity in, for instance, GaAs-(Al,Ga)As heterostrucrues [30].Recapturing the electrons released by an illumination at the DX centers requires thermal surmounting of a potential barrier.Although the decay of the photocurrent thus becomes slow, the photocurrent emerges rapidly with the illumination in this case.The slow photoresponses are, therefore, unrelated to the persistent photoconduction phenomenon.
Moreover, the slow photoresponses of the Ag-Sb-S alloys are different also from those observed for ZnO because of an anomalous dependence on the excitation energy.The film was illuminated by a visible light with λ LED = 633 nm in figures 1(c) and (d).As the excitation energy was lowered, the photoresponses changed to be rapid for both turning on and off the illumination.The transients were faster than the measurement interval of 0.1 s, i.e. the photoexcitation behavior is conventional under the illumination of the visible light.When the film was illuminated using a near-UV light with the intermediate wavelength of λ LED = 375 nm, an instantaneous response was followed by a slow transient, as shown in figures 1(e) and (f).For ZnO films and nanowires, the photoresponses do not change and remain slow when the illumination wavelength is varied in a range of 254-656 nm [18,21], which covers both the bandgap and sub-bandgap excitations for ZnO.The optical bandgap of Ag 0.5 Sb 0.5 S was reported to be 1.8 eV [31], which corresponds to a wavelength of 690 nm.The excitation energies in this work are thus larger than the bandgap.Note that Sb 2 S 3 has also a similar bandgap of 1.7-1.8eV [32].
In the phototransients of ZnO films and nanowires, the surface adsorption of H 2 O and O 2 molecules plays a crucial role [15][16][17][18][19][20][21][23][24][25].In this regard, we show in figure 4 that the electrical conduction in the Ag-Sb-S films was affected significantly by the surface condition.Here, the film was exposed to a fluorescent light with the surrounding environment of ambient air in the starting condition.The film was placed in a dark chamber with the atmosphere of air at t = 0. Following the near saturation of the phototransient, the chamber was then evacuated in the interval of t between 7 and 14 days.The increase of σ s during the evacuation indicates that the molecules adsorbed at the surface caused scattering as they plausibly induced random surface charges.When the increase of σ s resulting from the desorption of the surface molecules became slow, the chamber was filled with N 2 gas for t > 14 days.The reduction of σ s due to the N 2 molecules adsorbed at the surface was comparable to that caused by air.For the Ag-Sb-S alloys, it is not necessarily the O 2 molecules that can influence the electrical conduction.The film examined here was prepared at T s = 358 • C, i.e. without the Sb 2 S 3 overgrowth.The behavior of the film originated solely from the Ag 0.5 Sb 0.5 S alloy.
To examine possible influence of the surface condition on the phototransients, the photoexposure with λ LED = 280 nm was executed in air in figure 5(a).The duration of an illumination using I LED = 50 mA was varied to be 5, 21 and 1 h for the red, green and blue curves, respectively.After the illumination for 21 h, the film became more conductive in dark in comparison to before the illumination.The increase in the background value of σ s is attributed to the removal of the surface molecules by the UV irradiation [33].Due to the gradual reattachment of the molecules in air to the surface and their detachment during the UV irradiation, σ s hardly saturated.
Such a surface effect is negligible in figures 1(a) and (b) owing to the vacuum environment, where σ s saturated consequently when the phototransient ceased to continue.Here, the saturation values of σ s before and after the illumination were almost the same for I LED = 50 mA.They were, however, no longer the same for larger I LED .Opposite to the conductivity enhancement that the removal of the surface molecules induced, the conductivity deteriorated after stronger illuminations.The strong UV irradiation is suggested to have ionized the molecules that remained on the surface due to the high energy [34][35][36].It may be even suggested that defects were generated in the film [37][38][39].The atomic bonds in chalcogen compounds are not rigid as well-known for the phase change material Ge-Sb-Te [40], for which a transition between crystalline and disordered phases can be induced by illumination.If the UV irradiation indeed generated defects, the trapping of the photocarriers by the defects could explain the slow response under the UV illumination [23].
The anomalous phototransients were independent from the surface adsorption in contrast to its significant influence on the transport properties.That is, regardless of the photoexcitation wavelength, no remarkable difference was noticed between the situations of keeping the film in vacuum and air apart from the shift in the value of σ s due to the surface scattering.For instance, the slow phototransients were not altered in figure 5(a) by the surface attachment of the molecules in comparison to those in figures 1(a) and (b), as shown for the photocarrier generation by a magnified plot in the inset.In figure 5(b), the film held in vacuum was exposed to N 2 gas instead of air for t > 0. The UV illumination procedure similar to that in figure 5(a) shows that the slow-phototransient behavior is unchanged, apart from quantitative differences due to the nonidentical surface coverage of the molecules, regardless of the species of the molecules attached to the surface, as shown in the inset.In the opposite situation of the photoresponse, the transients became rapid in figures 1(c) and (d) despite the exposure of the film to air.It is indicated, at least, that the surface effect alone cannot induce the slow phototransients.
We show another example of the negligible influence of the surface for the case of λ LED = 375 nm by the dotted curves in figures 1(e) and (f).The removal of the surface molecules in the vacuum environment altered neither of the mixed phototransient components.The molecules adsorbed at the surface are thus not responsible for the vast spread in the transient time manifested as the rapid and slow photoresponses.The anomalous responses are presumably the bulk properties of the Ag-Sb-S alloys.Here, we point out again that the enhancement of σ s in vacuum reflects the fact that the overgrowth of the Sb 2 S 3 layer for T s = 340 • C was partial.The weak photoresponse cannot be attributed to the shadowing of the illumination by the overlayer.Concerning the overgrown part of the Ag-Sb-S film, the presence of the Sb 2 S 3 islands made no difference in the fact that the rapid photoresponses could be completely absent, as in figures 1(a) and (b).That is, passivating the film did not change the outcome.We hence rule out extrinsic surface effects from the origin of the anomalous behavior.
It will be worth clarifying the follow issue to avoid confusion.The illumination duration was increased as 10, 30 and 60 min for the consecutively measured photoresponses in air in figures 1(e) and (f).The larger overall values of σ s after each illumination, i.e. the upward shift between the three curves, are attributed to the transient states of the slow photoresponse rather than the alteration of the surface condition by the near-UV irradiation.
Various secondary effects may influence the phototransients.For instance, electrons can leave the surface of the sample under the UV irradiation, charging it to a positive potential.The migration of Ag ions will be thereby affected.Furthermore, the sample temperature increased during the illumination as the LED acted as a heater due to the Joule heating when I LED was large.The sample heating gives rise to the thermal excitation of conduction carriers.Nevertheless, such secondary effects would not disrupt the rapid phototransients.To be specific, the former is a surface effect, and so it does not play a role for the anomalous phototransiensts.For the latter, the transient in the rapid photoresponse is complete before the sample temperature rises due to the heating.
The incident light penetrates less in the material for shorter wavelengths.According to the photo-absorption measurements in [31], the penetration depth for Ag 0.5 Sb 0.5 S is estimated to be 1400, 30 and 10 nm for the wavelengths of 633, 375 and 280 nm, respectively.The illumination with different λ LED hence excited different parts of the alloy film in depth.Given the short penetration for the UV lights, it will be worth emphasizing again that the Sb 2 S 3 layer did not prevent the photoexcitation of the underlying Ag-Sb-S film since the film was covered merely in part.
For unambiguously demonstrating the irrelevance of the partial overgrowth of the Sb 2 S 3 layer in the anomalous dependence on the excitation energy, the film prepared at T s = 383 • C, i.e. the Ag 0.5 Sb 0.5 S layer without the Sb 2 S 3 overlayer, was employed in the photoconduction measurements in figure 6.The photoexcitation with λ LED = 633 nm induced only the rapid photoresponse, as shown for the turning-on of the illumination in figure 6(a) and for the turning-off of the illumination in figure 6(b).As the excitation energy increased, the slow phototransient emerged and the rapid photoresponse became smaller in the relative magnitude, as one finds in figures 6(c)-(f).
Correlation effects occurred when the film was illuminated at a number of wavelengths simultaneously since the charge transports in the rapid and slow responses influenced each other.One clear example is the slow photoresponses under the illumination of a fluorescent light reported in [14].In the inset of figure 4, the rapid photocarrier decay in such a circumstance is shown to be small.Both the generation and recombination of the photocarriers are slow although the spectrum of the fluorescent light is designed to be mainly in the visible energy range.The rapid response that the component of the visible lights should produce is eliminated by the mechanism of the slow transients triggered by the UV components.The blocking effect might imply that the material is made to be transparent The illumination was turned on and turned off for the left-and right-hand side columns, respectively.The illumination at λ LED = 280 nm was performed in vacuum, whereas the film was illuminated in air for λ LED = 633 and 375 nm.The duration of the illumination was 1 h between (a) and (b), 1 day for the red curves and 2 days for the green curves between (c) and (d) and 1 h for the red curves and 2 h for the green curves between (e) and (f).
for visible lights by shining a UV light on it, unless, perhaps, the incident photons vanish by generating phonons.
Even more dramatic and complex photoresponses are revealed in figure 7. Here, the light source was the two LEDs with λ LED = 633 and 280 nm connected to a power supply in series.The operation current was varied as I LED = 30, 50 and 70 mA.(The illumination with λ LED = 633 nm in figure 3 was performed with I LED = 30 mA.)In the simultaneous operation of the LEDs, an instantaneous photocarrier generation took place instead of its blocking by the UV illumination.We need to point out here that it is unlikely that the LEDs operated exactly simultaneously.The threshold voltage as well as the operation voltage of the UV LED are larger than those of the visible LED.The visible LED started to emit photons earlier than the UV LED by a fraction of a second in the rise of the output voltage of the power supply.Although a rapid response is seen to have taken place also when the illumination was turned off, its magnitude was considerably smaller than that when the illumination was turned on.When the driving voltage increased and decreased, the aforementioned relative shift between the two LEDs in starting and ending the light emission was plausibly not the same.The change in the ratio of the rapid and slow components between turning-on and turning-off the LEDs is ascribed to the nonidentical delay in the operation.Surprisingly, the film became less conductive as a slow transient during the illumination.It appears that a quasipermanent gradual reduction of σ s occurred, as indicated by the dotted curves.This suppression of the electrical conduction could be regarded as evidence that the UV irradiation generated defects.Notice, in addition, the presence of small features, which are marked by the horizontal bars, in the illuminations with I LED = 50 and 70 mA.The abrupt increase of σ s at the turn-on of the illumination was followed by a relatively rapid decrease σ s and a subsequent gradual recovery of σ s to the quasi-saturation value.The conduction carriers as well as the plasmons generated by the visible light are apparently transported in the film in a highly complex fashion under the simultaneous UV irradiation.The complication arises as the illuminations by the visible and UV lights play different roles.That is, the visible light only photoexcites electrons and holes, whereas the UV illumination generates defects that trap the photocarriers in addition to creating the electron-hole pairs.The trapping centers were introduced in the system undergoing a diffusion of the photocarriers with a timing determined by the unintentional delay in the operation of the two LEDs.We call our attention to the two-level stepwise changes of σ s at t = 4.7-4.9h.They are similar to the random telegraph noise known to occur for mesoscopic conductors [41,42].Here, the switching is associated with changes in the charge state of an impurity or defect.The induction of such a random fluctuation by the illumination supports our argument that the UV irradiation generated defects.

Conclusions
A drastic change takes place in the photoconduction in Ag-Sb-S films between the illuminations using a visible light and an UV light.Anomalously, the conventional rapid responses that the photoexcitation with λ LED = 633 nm induces are replaced by extremely slow responses when the excitation energy is raised as λ LED = 280 nm.The presence and absence of the rapid response in the respective wavelengths are unchanged regardless of the intensity of the photoexcitations.The transition between the rapid and slow photoresponses seems to take place gradually as λ LED decreases.The phototransients are unaffected by the surface conditions of the films.The anomalous slow responses are hence speculated to be the bulk properties of the Ag-Sb-S alloys.It is plausible that the UV irradiation generates defects in the Ag-Sb-S films, resulting in trapping the photocarriers to suppress the rapid response.We have furthermore demonstrated correlation effects in the photoconduction, where a rich variety of phototransients emerge in the quasi-simultaneous illumination of visible and UV lights.

Figure 1 .
Figure1.Dependencies of phototransients in Ag-Sb-S film on illumination wavelength.The film prepared at Ts = 340 • C was exposed to LED lights with λ LED = 280, 633 and 375 nm in the top, middle and bottom rows, respectively.At time t = 0, the illumination was turned on and turned off for the left-and right-hand side columns, respectively.In (a) and (b), the illumination intensity was varied by increasing the operation current I LED of the LED.The rapid transients in (c) and (d) are faster than the measurement interval of 0.1 section In (e) and (f), the duration of the illumination was varied for the red, green and blue curves as indicated.The film was kept in vacuum for the dotted curves, where the duration of the illumination was 10 min.The inset of (d) shows the configuration in measuring the current I.A bias of 10 V was applied between the Au contacts deposited on the film shown as gray areas.

Figure 2 .
Figure 2. X-ray diffractometry on Ag-Sb-S films.The ω-2θ scans are shown for the synthesis temperature Ts of 340, 358 and 383 • C. The peak associated with the 004 reflection of the Si substrates is indicated.The curves are vertically offset for clarity.Simulated curves for Ag 0.5 Sb 0.5 S and Sb 2 S 3 are shown in the bottom panel.

Figure 3 .
Figure 3. Scanning electron micrographs of Ag-Sb-S films.Surface and cross-sectional views of the films are shown for the synthesis temperatures Ts = (a) 383 • C, (b) 358 • C and (c) 340 • C. The films were prepared on Si substrates.The scale bars show the length of 1 µm.

Figure 4 .
Figure 4. Variations of sheet conductivity of film caused by molecular adsorption associated with surrounding environment.The film synthesized at Ts = 358 • C was kept in the atmosphere of air.The film under the illumination of a fluorescent light was placed in dark at time t = 0.The inset shows the initial part of the photocarrier decay.As the phototransient became negligible, the air was removed by a vacuum pump.The film was then exposed to N 2 gas when the desorption of the surface molecules became slow.The film was kept in the darkness throughout this procedure in t > 0.

Figure 5 .
Figure 5. Effects of UV irradiation under exposure of film to gases.The film examined in figures 1(a) and (b) was kept in the surrounding environment of (a) ambient air and (b) N 2 gas instead of the vacuum environment in figure 1.The duration of the UV exposure with λ LED = 280 nm and I LED = 50 mA was varied as indicated by the gray rectangles.The initial response when the illumination was turned on at t = 0 is shown in the insets.The three measurements shown as the red, green and blue curves in (a) were not carried out consecutively.

Figure 6 .
Figure 6.Phototransients in Ag 0.5 Sb 0.5 S film with no Sb 2 S 3 overgrowth.The film prepared at Ts = 383 • C was exposed to LED lights with λ LED = 633, 375 and 288 nm in the top, middle and bottom rows with I LED = 50, 50 and 30 mA, respectively.The illumination was turned on and turned off for the left-and right-hand side columns, respectively.The illumination at λ LED = 280 nm was performed in vacuum, whereas the film was illuminated in air for λ LED = 633 and 375 nm.The duration of the illumination was 1 h between (a) and (b), 1 day for the red curves and 2 days for the green curves between (c) and (d) and 1 h for the red curves and 2 h for the green curves between (e) and (f).

Figure 7 .
Figure 7. Photoresponses under simultaneous illumination of visible and UV lights.The current I LED supplied to the visible (λ LED = 633 nm) and UV (λ LED = 280 nm) LEDs connected in series was varied as 30, 50 and 70 mA in the photoexposures.Each illumination was performed for 1 h, as indicated by the gray rectangles.The film prepared at Ts = 340 • C was held in vacuum during the measurements.