Condensed Matter

Z. Phys. B - Condensed Matter 47, 233-237 (1982)

Zeitsohrift fLir Physik B

9 Springer-Verlag 1982

Inert Gas Condensation of Metal Microclusters J. Mbhlbach, P. Pfau, K. Sattler, and E. Recknagel Fakult~it far Physik der Universit~it, Konstanz, Federal Republic of Germany Received March 19, 1982 Metal microclusters o f lead, indium, bismuth and antimony in the size range between 2 to 500 atoms per cluster are generated by inert gas condensation, extracted into vacuum and detected by mass spectrometry. Resolved clusterpeaks are observed up to Pbllo, Inso, BiT0 and Sb240. Condensation conditions, beam intensities and limitations of the cluster source are discussed.

1. Introduction

2. Experiment

A fundamental question in solid state physics is how many atoms have to be contained in a particle that solid state properties appear; e.g., how many lead atoms are required to form a superconductor, or a metal; or how big a particle must be that its atoms arrange in the geometric structure of the bulk. In the case of metals, a great number of experiments were carried out on small particles. However, the investigations are limited to a lower size of about 20A particle diameter, corresponding to about 500 atoms per particle. In the domain of microclusters, which lies between atom and small particle, information is rare because until recently there was no suitable preparation technique [1]. We adapted the inert gas condensation, which is a standard method for the preparation of small metal particles [2, 3], to the size range of microclusters [1], and combined this generation method with a time of flight (TOF) mass spectrometer [4] for the detection of the clusters. Meanwhile the resolution of the mass spectrometer is improved, the spectra can be recorded at low electron energies, therefore being only little affected by fragmentation or coulomb explosion of small doubly charged clusters [5], and new materials are condensed. Intention of this paper is to present the improved mass spectra of Pb,, Bi n and Sb n clusters. The generation of indium clusters is reported for the first time. Condensation conditions are compared and limitations of the cluster source concerning the class of substances, which can be condensed to clusters, are discussed.

A schematic drawing of the high vacuum assembly is shown in Fig. 1. The source chamber contains the cluster source (O, C, D), the ion source (EO,IO) and the film thickness monitor (TM). Atomic metal vapor effuses from the oven (O) into the condensation cell (C), which is filled with a cold noble gas. Collisions between metal atoms and gas atoms affect cooling and supersaturation of the metal vapor, and growth of metal clusters. Cluster growth is definitively stopped when the clusters pass the differential pumping stage (D), where the transition between laminar and molecular gas flow occurs. The clusters then enter the high vacuum chamber, they pass the ionization region of the mass spectrometer and condense on the quartz crystal of

drift space

~C&~:~

TM ~o~-~

I

I

~ .....

I

Ill

' '-7/grra-t5 II uII I ' ~ " '

~1~

I

i-

.....

I

.I %,,C.

I

drift chamber

,1

rp3

o

TP2

source chamber

Fig. 1. Schematic drawing of the high vacuum apparatus. Explanations are given in the text

234

J. Miihlbach et al. : Inert Gas Condensation of Metal Microclusters

By the sliding valve (SV) the drift chamber can be disconnected from the source chamber, avoiding exposure of the channeltron to air when the oven is changed. Threefold differential pumping (TP1, TP2, TP3) is applied for achieving vacuum conditions required for the proper function of the mass spectrometer: At a He-pressure of 40 mbar at the gas inlet (G), the pressures obtained are 3-10 -2 mbar in the differential pumping stage, 2-10 .5 mbar in the source chamber, and 5.10 .6 mbar in the drift chamber. The residual pressure is 2.10 .7 mbar in the whole apparatus.

the film thickness monitor (TM), where the total mass flux of the neutral cluster beam is measured. For the detection in the mass spectrometer, a small fraction of the clusters is ionized by pulsed electrons (pulse width lOOns, repetition time 1 ms, duty cycle 10 -4, energy 20-300 eV), which are focused onto the cluster beam by the electron optics (EO). The ions are accelerated by an electrostatical lens (IO) to a kinetic energy of 2 keV, they enter the drift chamber at the orifice (DO), pass the apertures (B) and (N) and are post-accelerated to a kinetic energy of 7 keV before they hit the copper plate (P) at the end of the drift space. When impinging on (P), the ions induce the emission of secondary electrons which are collected and multiplied (x 109) by a channeltron (CH), enabeling the detection of single ions. The time of flight t of an ion is determined by the time difference between the electron pulse, which provides the ionization, and the registration of the channeltron pulse. As all the ions possess the same energy, this time is proportional to the square root of the ions mass, t~m ~/2. A time to digital converter (TDC) measures the flight times, converts them into digital units and transfers these numbers to the memory of a multichannel analyser, generating complete TOF-mass spectra. A time of flight spectrometer was chosen because of the large mass range covered, Pb4o0~82,900amu being the biggest cluster detected [7]. An upper size limit arises from the decreasing detector efficiency for very big particles, whose velocities are too low to induce secondary electron emission when impinging on (P). Applying a synchronized deflection pulse to the condenser plates (SP) enables the separation of cluster ions of unique mass out of the whole spectrum

3. Generation of MetalMicroclusters Before clusters are generated, the metal is evaporated without He applied to the condensation chamber. As a consequence of the oven geometry chosen and the metal vapor pressures needed, neither Knudsen nor Langmuir conditions are fulfilled. This procedure henceforth is called direct evaporation in this paper.

3.1. Lead The direct evaporation of lead essentially yields atoms, the dimer concentration being less than 1~o. When He is let into the condensation cell, at a pressure of 40mbar resolved cluster peaks up to Pbll o are observed in the mass spectrum (Fig. 2). In the low size range (n<20), the spectrum exhibits a pronounced structure. Relating the peak heights to the stability of the clusters, the positions of the maxima can be explained by a simple packing model E6].

[4]. "C 10

flJ

Pb n

( Inert gas condensation ) ,4,-

9

r-m t-

13 15 17

6

]I~116[ 18

100

30

~II,]|''90 '

20

230

50

60

,

,

360

Fig. 2. Mass spectrum of lead clusters generated by inert gas condensation

7,0

1120

,,

T'ime of fligl~f [IJSeC]

J. Miihlbach et al.: Inert Gas Condensation of Metal Microclusters

235

The condensation parameters for the spectrum in Fig. 2 are given in Table t. When the condensation cell is cooled by liquid nitrogen, the size distribution shifts to bigger clusters, Pb,0 o being the biggest cluster detected under those conditions [7].

The fraction of the bigger clusters is remarkably high: Pbio appears with the highest intensity, and the intensity of Pbso, e.g., amounts to 16 ~ of the Pblo intensity, thus remaining in the same order of magnitude.

u

In m

4o

( Inert gas condensation) o

20

30

10 C

( x85}

' &'O 60 8'0 Fig. 3. Mass spectrum of Indium clusters

I()0

~

189

ILO

1;0 Time

18 19 1

light [ psec

Bi n ( Inerf gos condensotion)

17 >',

of f

~c,

3

CO C'-o

2

-I-i--..,i

4-

(•

)

8 ~"

10 i

JL._JL_

J

s'o

9 1112~3~'4~s~ 1oo

i;o

2;0

2'5o

3;0

Time of flight [ psec ]

Fig. 4. Mass spectrum of Bismuth clusters

~

Sb n (Inert gos condensntionl

~

/

"~

2oo '

30

(x25)

s

16 Izo

/+?0

' I,~4o

'

460

2~0

'

6o 200

Fig. 5. Mass spectrum of Antimony clusters

80 ioo

~,,00

d,~IL, , I

1~o 300

460

' Time of fright [Hsec]

236

J. Miihlbach et al.: Inert Gas Condensation of Metal Microclusters

3.4. Antimony

Table 1. Condensation parameters

Oven temperature (K) Metal vapor pressure (mbar) Helium temperature (K) Helium pressure (mbar)

Pb

In

Bi

Sb

1,320 3 290 40

1,470 0.8 290 35

1,150 3 80 20

920 0.3 80 16

3.2. Indium Indium evaporates essentially monatomic. Figure 3 shows the spectrum which is obtained by inert gas condensation. Resolved peaks are In 1 and In 2 as well as In 9 to Inso. Between In 2 and In 9, the mass peaks cannot be associated with indium clusters, In 3 to In 8 are hidden in the background, which mainly consists of organic molecules. Because indium has a lower vapor pressure than lead, indium has to be evaporated at elevated temperatures to observe cluster formation (Table 1). Since the condensation cell is watercooled in both cases, a higher supersaturation is achieved with indium resulting in the production of bigger clusters: In the case of lead, the intensity decreases from Pbx0 to the bigger clusters (Fig. 2), while in the case of indium an increase up to Ins0 is seen (Fig. 3). The spectrum extends up to Ins0o, but only the resolved part is shown in Fig. 3.

3.3. Bismuth In contrast to lead and indium, the direct evaporation of bismuth already yields clusters up to the tetramer [8], which act as parent particles for the condensation. At an electron energy of 70eV, an intensity ratio of

I(BiO:I(Bi2):I(Bi3):I(Bi4) = 1 : 0,6 : 0,015:0,018 is observed. Condensation in helium results in the production of clusters up to Bi28o, resolved peaks up to BiT0 (Fig. 4) are recorded. In the low size region, a trimer sequence (6, 9, 12, 15) of antimagic numbers 1-9] attracts notice, whose interpretation is open to question so far. The oven temperature is essentially lower than in the case of lead; therefore, to obtain a supersaturatio~a high enough for a size distribution similar to that of lead in Fig. 2, the condensation cell has to be cooled by liquid nitrogen while with watercooling only an enrichment of the trimers and tetramers is achieved [10].

Like bismuth, antimony is a semimetal, whose covalent character is even stronger, which manifests in the predominant emission of Sb 4 under direct evaporation conditions. This leads to a tetramer sequence dominating the mass spectrum of antimony clusters (Fig. 5). The peaks which do not belong to the tetramer sequence are assigned to fragment ions formed during the electron impact ionization of the neutral Sb4,-clusters [-11]. The structure in the tetramer sequence can be explained by a simple tetrahedra packing model [,12]. As in the case of bismuth, the walls of the condensation cell are LN2-cooled. The evaporation, however, occurs at a lower oven temperature, thus an almost monotonously decrease of the intensity, starting with the monomer unit, Sb4, to the bigger clusters is obtained. The spectrum shows resolved peaks up to Sb240 and extends up to Sb500. 4. Generation Conditions The condensation parameters employed to obtain the cluster spectra (Figs. 2-5) are listed in Table 1. The oven temperature is measured by a thermocouple at the bottom of the oven, the metal vapor pressure is taken from a table of equilibrium vapor pressures [,,13]; this should be a rather good approximation, as the orifice area is small compared to the free metal surface in the oven (orifice diameter 2 mm, oven diameter 20 mm). As the true helium temperature is unknown, the temperature at the walls of the condensation cell is given. The helium pressure is measured at the gas inlet at room temperature. It is seen from Table 1 that for the generation of metal clusters, vapor pressures between 0.3 mbar and 3 mbar are required; the difference between oven temperature and helium temperature has to be about 1,000 K. By these conditions, the cluster source until now is restricted to substances which reach a vapor pressure of 1 mbar between 900 K and 1,500 K. The upper temperature limit is rather a consequence of the source design than of principle nature: With manganese, at an oven temperature of about 1,800 K, an extremely high mass flux was displayed at the film thickness monitor, but no clusters and no atoms were observed in the mass spectrometer. Presumably clusters were generated with masses extending the range of the spectrometer. It should be possible to get smaller clusters by keeping the walls of the condensation cell at a higher temperature. A source design which enables higher gas temperatures is under construction.

J. Mtihlbach et al.: Inert Gas Condensation of Metal Microclusters

5. Intensity Considerations As seen in the cluster spectra, a high portion of larger clusters is obtained, which is one precondition for the systematic investigation of cluster properties. Concerning the total intensity, at the film thickness monitor an enhancement of the total mass flux by a factor of 100 to 1,000 is measured under inert gas condensation conditions related to the mass flux observed under direct evaporation conditions. Most probably a helium flow characteristics establishes in the condensation chamber, which centers the cluster beam, consequently reducing the number of particles which are lost because of being scattered to the walls of the condensation cell. Such gas flows have been observed by light scattering at small metal particles generated by inert gas condensation [14]. Typical condensation rates are 1 A/s related to an area of 0.5 cm 2 of the quartz crystal. This should be high enough to prepare probes by matrix isolation within reasonable time. The advantage of this matrix isolation technique compared to the usual technique [15] arises from the knowledge of the TOF-size distribution, which tan be monitored simultaneously to the preparation of the matrix. Clusters of different sizes would be present in the matrix. Therefore, the question arises how big the cluster is which gives the response in an experiment. It should be possible to answer this question by comparing experiments at probes with different size distributions: In the case of bismuth, e.g., the ratio of dimer intensity to tetramer intensity can be varied by two orders of magnitude [10]. Therefore, the intensity ratio of dimer response to tetramer response is expected to vary by the same amount. Comparing the changes of the cluster intensity ratios in the mass spectra with the intensity ratios of the cluster responses should enable the assignment of measured quantity to cluster size at least for the smaller clusters (n< 10), for larger clusters only minor changes of the relative intensities are obtained. A promising approach to get rid of matrix effects is to measure physical properties at the neutral cluster beam in vacuum in coincidence with mass detection in the spectrometer, which in principle is possible in experiments which end in the ionization of the observed particle (e.g. multi-photon-ionization, UPS, EXAFS).

6. Summary Metal microclusters of lead, indium, bismuth and antimony in the whole size range between atom and

237

small particle are grown by inert gas condensation, extracted into vacuum and detected by TOF-mass spectrometry. Limitations to substances with vapor pressures lying in a region which is roughly outlined by the vapor pressure curves of antimony and indium are given by the special design of the cluster source and should not be of principle nature. The intensity of the neutral cluster beam is high enough to carry out experiments within reasonable time. This work was partly supported by the Deutsche Forschungsgemeinschaft.

References 1. Sattler, K., Miihlbach, J., Recknagel, E.: Phys. Rev. Lett. 45, 821 (1980) 2. Granquist, C.G., Buhrmann, R.A.: J. Appl. Phys. 47, 2200 (1976) 3. Yokozeki, A., Stein, G.D.: J. Appl. Phys. 49, 2224 (1978) 4. Sattler, K., Miihlbach, J., Recknagel, E., Reyes-Flotte, A.: J. Phys. E, 13, 673 (1980) 5. Sattler, K., Miihlbach, J., Echt, O., Pfau, P., Recknagel, E.: Phys. Rev. Lett. 47, 160 (1981) 6. Miihlbach, J., Sattler, K., Pfau, P., Recknagel, E.: Phys. Lett. 87A, 415 (1982) 7. Mfihlbach, J., Recknagel, E., Sattler, K.: Proc. 4th Int. Conf. on Solid Surfaces. Vol. I. p. 692 (1980) 8. Miihlbach, J., Pfau, P., Recknagel, E., Sattler, K.: Surf. Sci. 106, 18 (1981) 9. Magic numbers for the stability of atomic systems have been defined in: Echt, O., Sattler, K., Recknagel, E.: Phys. Rev. Lett. 47, 1121 (1981) 10. Pfau, P., Sattler, K., Mfihlbach, J., Pflaum, R., Recknagel, E.: (submitted for publication) 11. Pfau, P., Sattler, K., Mtihlbach, J., Recknagel, E.: (submitted for publication) 12. Sattler, K., Miihlbach, J., Pfau, P., Recknagel, E.: Phys. Lett. 87A, 418 (1982) 13. Margrave, J.L.: The Characterization of High Temperature Vapors, p.481. Fig. A1. New York, London, Sydney: John Wiley & Sons Inc. 1967 14. Kasukabe, S., Yatsuga, S., Uyeda, R.: Jpn. J. Appl. Phys. 13, 1714 (1974) 15. With the usual matrix isolation technique, atomic metal vapor and an inert gas are simultaneously deposited onto a cold substrate. Clusters are grown in the matrix, the size is varied by changing the metal concentration, or by heating the matrix, thus providing diffusion of the atoms in the matrix. Therefore, there is no direct method to determine the size distribution J. Miihlbach P. Pfau K. Sattler E. Recknagel Fakultiit ftir Physik Universit~it Konstanz Biicklestrasse 13 D-7750 Konstanz Federal Republic of Germany