Temperature dependence of sputtered cluster yields

Experimental

Part of the experiments were performed with an experimental setup (setup 1) which has been described in detail elsewhere [4]. In short, a double focusing mass spectrometer consisting of a magnetic sector and an electrostatic prism is used to detect positive ionic clusters which are sputtered from a clean silver sample by 8.5 keV Ar+ ions. The ion current density was about 3 mA/cm² with an incident angle of 45°. The base pressure in the chamber was about 10^-10 mbar.
The second part of the experiments was performed with a reflectron time-of-flight mass spectrometer (setup 2, base pressure 10^-9 mbar) which has also been previously described [5]. Here, the bombarding energy was 12.6 keV with an ion current density of about 30 |iA/cm . During data acquisition, the primary ion-gun was used in a pulsed mode with a pulse length of 10 |is at a repetition rate of 10 Hz. In order to reduce the effect of surface contamination due to residual gas adsorbates during the acquisition of a mass spectrum, a cw sputter cleaning cycle of 10 sec was performed after every 500 shots. The technique employed for the detection of positive secondary ions and neutrals with this instrument is described elsewhere in this volume [6]. The excimer laser used for post-ionization was operated at X = 193 nm and focused to dimensions of about 1.2 x 2.2 mm in directions perpendicular and parallel to the surface, respectively. Note that in this particular experiment the photoionization efficiency is not saturated, and the measured signals of post-ionized neutrals therefore depend on the laser power density. In order to eliminate the influence of this dependence, the pulse energy of the laser was kept constant within a few percent.

3. 
   Results and Discussion
   

temperature (°C)
Fig. 1 a) and b) show the secondary ion signals measured with both experimental setups as a function of the target temperature for selected cluster sizes. In order to demonstrate the temperature dependent variation, the data for a specific cluster size was normalized to the respective signal obtained at the lowest temperature. First, it is seen that the signals tend to increase with increasing temperature, the effect being more pronounced for larger cluster sizes. Although details of the curves show slight differences, this qualitative trend as well as the overall magnitude of the effect (a factor of two for Ag₇) is comparable in both figures. In view of the large difference between the two experimental setups we consider this a reassuring agreement. The only exception is given by the monomer ions, which appear to increase in Fig. 1a but slightly decrease in Fig. 1b. The reason for this difference is unclear at the present time. Fig. 2 shows the relative yields (i. e. the yield normalized to that of the monomer ions) of cluster ions as a function of the cluster size for two values of the target temperature. In order to compare the data taken with the two setups, it is important to note that setup 1 detects the flux, whereas setup 2 detects the number density of secondary ions within a volume located approximately 1 mm above the surface [6]. In order to convert to flux, the data obtained with setup 2 must therefore be corrected for the average inverse velocity (vof the sputtered particles. From our previous measurements of the velocity distribution of sputtered silver clusters, we assume that varies with cluster size as и⁰'⁸ and corrected the signals measured with setup 2 accordingly. It is seen that the resulting yield distribution measured at a temperature of about 650 °C shows a good agreement between both setups


temperature (°C)

Fig. 1 Yield variation of secondary cluster ions vs. target temperature measured with a) setup 1 and b) setup 2 (see text)

cluster size n
Fig. 2 Relative yields of secondary cluster ions vs. cluster size measured with (1) setup 1 and (2) setup 2.
.

The important observation made in Fig. 2 concerns the difference between the cluster yield distributions measured for high and low temperature, respectively. From the presented data, it is apparent that the relative contribution of larger cluster ions becomes more abundant at higher temperature. Note that this trend is reproduced in both setups and should

therefore be regarded as independent of the particular experimental details. In principle, two possible causes may lead to an increased abundance of larger cluster ions from the heated surface, namely either i) an enhanced formation probability or ii) an increased ionization probability of larger sputtered species. In order to distinguish between these possibilities, it is necessary to compare the mass spectra measured for secondary cluster ions with those of the respective sputtered neutral species. As explained above, we have done this by post-ionizing the neutral clusters using an intense pulsed UV laser. The resulting temperature dependence of the corresponding signals of post-ionized sputtered neutral clusters are shown in Fig. 3. It is immediately seen that the yields of sputtered neutrals do not increase with increasing temperature, but instead exhibit a small decrease, which is most pronounced for Ag atoms and Ag2 dimers. Practically no temperature dependent variation is observed for larger clusters. From our previous work, we know that the ionization probability of silver clusters sputtered from a clean silver surface under conditions comparable to those applied here is small (below 10 % for all clusters). This means that the neutral yields reflect the partial sputtering yields of the clusters (regardless of their charge state).
As a consequence, we conclude that the collisional processes leading to the formation of a sputtered cluster do not depend on the temperature of the bombarded surface. The temperature effect observed in Fig. 1 and Fig. 2 must therefore be attributed to an enhanced ionization of larger clusters at elevated temperatures. In order to illustrate this, we plot in Fig. 4 the ratio between the signals of secondary ions and post-ionized neutrals as a function of the target temperature. Since the post­ionization efficiency can be regarded as constant and, in particular, independent of the surface temperature, the resulting values are proportional to the ionization probability a+ of the sputtered clusters. The data in Fig. 4 show that the values of a⁺ increase with increasing temperature, the effect being the more pronounced the larger the cluster.
The physical origin of the increased ionization probability with increasing target temperature cannot unambiguously be identified at the present time. One possible explanation involves the residual gas contamination of the surface. It is known that already small amounts of electronegative species present at a metallic surface may lead to a drastic enhancement of the formation probability of positive ions [7]. If the residual gas pressure in the vacuum chamber rises with increasing temperature, one might therefore expect an increase of the measured ionization probabilities. Indeed, we find a relatively strong increase of mass spectrometric signals detected, for instance, for oxygen containing mixed clusters like Ag3O2 both in the secondary ion and the neutral spectrum above a temperature of 300 °C. However, we would intuitively expect the effect of an oxygen contamination to be largest for the monomers and dimers, since these species exhibit the lowest ionization probabilities (< 10"⁴ [5,8]) when emitted from the sputter cleaned surface. Interestingly, practically no temperature dependence is observed for the ionization probability of the monomers. A second cause of surface contamination is given by the segregation of impurities within the silver target itself. Inspection of the mass spectra shows a signal of indium which strongly increases with increasing temperature. Due to the apparent lack of theoretical models describing the ionization mechanism of a sputtered cluster, it is not clear at the present time how such an indium coverage would influence the ionization probability of silver atoms and clusters of different sizes.
References

1. 
   H. E. Roosendaal in Sputtering by Particle Bombardment I, ed. R. Behrisch (Springer 1981) 219.
   
2. 
   W. Szymczak and K. Wittmaack, Nucl. Instr. Meth. B 82 (1993) 220.
   
3. 
   S. W. Rosencrance, N. Winograd, B. J. Garrison, and Z. Postawa, Phys. Rev. B 53 (1996) 2378.
   
4. 
   N. Kh. Dzhemilev, U. K. Rasulev, and S. V. Verkhoturov, Nucl. Instr. Meth. B 29 (1987) 531.
   
5. 
   M. Wahl and A. Wucher, Nucl. Instr. Meth. B 94 (1994) 36.
   
6. 
   A. Wucher, R. Heinrich, and C. Staudt, these proceedings.
   
7. 
   A. Benninghoven, F. G. Rudenauer, and H. W. Werner (1987)
   
8. 
   R. Heinrich, C. Staudt, M. Wahl, and A. Wucher, these proceedings.
   

Fig. 3 Yield variation of sputtered neutral clusters vs. target temperature

temperature (°C)

50

CO

_1