Z. Phys, D 25, 201-203 (1993)

Atoms, Molecules zo.=°, and Clusters IiJr Fhysik D

© Springer-Verlag 1993

Lifetime measurements of the 3d94 s(

))4p configuration of Cu I

W.E. van der Veer, R.J.J. van Diest, A. D6nszelmann Van der Waals - Zeeman Laboratorium, Universiteit van Amsterdam, Valckenierstraat 65-67, 1018 XE Amsterdam, The Netherlands Received: 2 October 1992

Abstract. The radiative lifetimes of the levels in the 3d94s(©)4p configuration of Cu I are measured. The levels are excited from the metastable 3 d94 s 22D3/2,5/2 levels. The metastable Cu atoms are generated in a p u l s e d hollow cathode discharge, The levels investigated are populated with a 35-ps laser pulse at wavelengths around 220 nm. The laser induced fluorescence signal is detected. The lifetime of the 3 d94s(3D)4p 401/2 level is also determined by direct excitation from the ground state. A comparison with calculated literature values is given. PACS: 32.70.Fw

Introduction The spectrum of neutral copper is a simple doublet spectrum, consisting of the configurations 3d ~° nl. In addition, an inner shell excitation results in the 3d94s nl configurations. In this paper we discuss lifetime measurements of the levels of the 3d94s(1D)4p configuration state [1], (Fig. 1), with final couplings Zp, 2o and 2F. Earlier investigations of the lifetimes of the states in the 3d94s(3D)4p configuration revealed transitions useable for laser operation [2]. In a subsequent experiment we demonstrated the laser action of these transitions in an adapted Cu-vapour laser [3]. Analogously, the transitions from the levels examined in this work to the 3d94s 2 2D3/2,5/z levels can be considered good candidates for a laser scheme in the far UV regime. The transitions are very strong which favours the competition with other transitions to the lower levels. The construction of such a laser presents additional problems: A cavity is needed with a high Q-factor for wavelengths around 220 nm and a low Q-factor at 304 nm, 306 nm, 510 nm and 578 nm. The probability of populating the upper levels in a gas discharge apparatus is expected to be very small, so a more specific excitation mechanism must be employed.

Until recently the lifetimes of the levels in the configuration were difficult to determine. The expected lifetimes are of the order of 2 ns. These expectations are based on calculations by Carlsson [4]. The generation of the very short pulses needed at wavelengths around 220 nm presented the major experimental problem. The use of a mode-locked N d : Y a g pump laser and a synchronously pumped dye laser enables the generation of the short laser pulses. The light is frequency doubled by a fl-BBO crystal to the wavelengths needed. In order to access the levels discussed, a hollow cathode discharge is used to generate atoms in the 3d94s z 2D3/2,5/2metastable states. F r o m these states all the levels in the 3d94s(ID)4p structure can be populated by one excitation step. The transitions shown in Fig. 1 are the strongest transitions. With the same experimental set-up the lifetime of the 3d94s(3D)4p 4D1/2 level is also determined. This level is populated by excitation from the 3 d~°4s zSa/2 ground state (Fig. 2). In a previous experiment [2] we have tried

3d94s(1D) 4p

2D

3d~4sUD)4p

.

1/2 312

<

CuI

3d94s

2p

221.57 221.46 <----

2

=D

m

3/2 '

'

(

2F 512

512

219.98 219.96 <.---

-7/2

(

222.78 223.01 <

. 312 = 5/2

Wavelengths in nm. Fig. 1. The strongest transitions from the 3d94s22D metastable states to the levels investigated

202 1/2

3d~4S(3D)4p

4D

g

312

512 7/2

222.5

3d~4 ~ /

2D 312

5/2

3&°4s

2S.~

Fig. 2. The excitation path to the

3d94s(3D)4p4D1/2level

to measure the lifetime of that level by populating it from the 3d94s 2 2D3/2level. This attempt was unsuccessful due to the small transition probability of the excitation path. In the current work we use a much stronger transition. The calculation of reliable lifetimes and probabilities of transitions to the levels in the 3d94s(1D)4p and 3d94s(3D)4p configuration states is hard. The system consists of two electrons and a hole, which requires extensive calculations. In addition the interaction of the remaining electrons in the 3 d shell with the valence eIectrons cannot be ignored. The measurements presented here can be used as a good test of such calculations.

The experimental set-up In order to measure the lifetimes, sufficiently short light pulses and a fast photo-detection system are needed. The laser system is pumped by a modified mode-locked Nd:Yag laser (Quantel Yg 501.30). The oscillator of this laser produces a train of 35-ps pulses. One half of the intensity of these pulses is coupled-out by a beam splitter and directed via an optical isolator into a Nd:Yag amplifier (Fig. 3). The isolator prevents damage to the oscilla-

tot by powerful amplified reflections. The amplified pulse train is frequency tripled by a KDP and a ADP crystal, it pumps a synchronously pumped dye laser [5]. The laser operates at wavelengths around 440 nm. The dye used is Coumarin 120 dissolved in methanol. The pump laser also produces a single light pulse which is frequency tripled by two KDP crystals. This pulse is used to amplify one pulse from the dye laser in a transversely pumped dye amplifier. The beam from the dye amplifier is frequency doubled in a/~-BBO crystal to wavelengths around 220 nm. This light is directed into a vacuum system where it crosses a beam of metastable atoms and excites them to the states of interest. The metastables are generated in a pulsed hollow cathode discharge [2]. The apparatus consists of a water cooled insulated tube with a tungsten anode mounted at one end. A copper cylinder with a bottom mounted at the other end is used as the cathode. Argon at a pressure of 2 mbar is used as a buffer gas. A hole in the centre of the cathode allows the metastables to drift into the vacuum system. The diameter of the cathode (14 ram) and of the hole (0.4 ram) were chosen to obtain an intense beam of metastables and to maintain a reasonably low pressure in the vacuum system (5 × 10-5 mbar). The discharge emits light into the detection chamber. More than 90% of the radiation is emitted at wavelengths larger than 325 nm. Since no commercial filters are available that are transparent at wavelengths around 220 nm and absorb at larger wavelengths, we use a solarblind photomultiplier (Hamamatsu R 166). The quantum efficiency of this tube at a wavelength of 220 nm is 50 times larger than the quantum efficiency at 325 rim. The quantum efficiency decreases sharply for larger wavelengths. The rise time of the photomultiplier is reduced to 800 ps by using only the first six dynodes [6] and a supply voltage of up to 1500 V. The signal is digitized by a Tektronix R7912 transient digitizer and stored in a personal computer. The digitizer is triggered with a fast photodiode which is illuminated by a part of the light from the dye laser.

Results The fluorescence sign als of the levels in the 3 d 94 s (1D) 4 p configuration state and of the 3d94s(3D)4p 4D1/2level are observed. For each level i0 measurements are per-

ML Nd:YA

d~ [ 1064nm optical isolator

Fig. 3. The experimental set-up

203 Table 1. Lifetime of the 3 d 94S (3D)4p

4D1/2 level in Cu I

3d94s(3D)4p

Excitation wavelength (nm)

Measured lifetime ( n s )

Theoretical lifetime (ns) [4]

4D1/2

222.57

23.2 _+0.8

12.6

2F

>,

712 "E

T =2.62 ns

\

t3~

/ JJ

0

Table 2. The radiative lifetimes of the levels in the

configuration state. The solved

I

I

I

I

2

4

6

8

time

10

Excitation wavelength (nm)

Measured lifetime ( n s )

Theoretical lifetime (ns) [4]

2P1/2 2p3/2

221.57 221.46

3.12_+ 0.22 2.76_+0.13

2.62 1.86

3d94s(1D)4p2F7/z level. The

2D5/2

formed and processed independently. Every measurement consists of an average of the signal over 250 laser shots and is corrected by subtracting the signals resulting from stray light of the excitation pulse and light from the hollow cathode discharge. A corrected signal is shown in Fig. 4. An exponential curve is fitted to the corrected signal, which depends on the lifetime of the investigated level and on the characteristic response time of the detection system. The fitted decay time is compensated for the response time, determined by observing the response of the detection system to the laser pulse. N e t w o r k ringing of the electronics in the photomultiplier causes a small modulation of the signal as is seen in Fig. 4. The obtained values are given in Tables 1 and 2, together with calculated literature values [4]. The accuracy of the values is determined from the statistical spread of the fitted compensated values obtained from the measurements. The accuracy of the lifetime values incorporates an additional contribution caused by the correction for the response function of the detection system. The transitions to the 3d94s(aD)4p2D3/2,5/2 levels are not spectroscopically resolved. These measurements contain fluorescence signals from both levels and the measured lifetime is a multiplet value. In Table 1 the measured lifetime of the 3d94s(3D)4pgD1/2 level and the theoretical value are given. Between these values there is a large discrepancy. In earlier investigations of the lifetimes of the levels in the 3d94s(3D)4p configuration state, calculated and measured values differed by as much as a factor 2.75 [2]. The measured value presented here falls within the previously encountered discrepancy between calculated and measured values. In Table 2 the lifetime values of the levels in the 3d94s(1D) 4p configuration state are quoted. The measured values are all significantly larger than the theoretical values, but no systematic trend appears. As noted above for the theoretical lifetime values of the levels in the 3d94s(3D)4p configuration state, the accuracy of

3d94s(1D)4p

are not spectroscopically re-

3d94s(1D)4p

in ns

Fig. 4. The fluorescence signal of the background signal is subtracted

ZDlevels (.)

2F5/2 2F7/2

}2 99

}23+012

222.78 223.01

2.31 _+0.13 2.62_+0.15

-

1.20

1.56 1.24

the values for the levels in the 3d94s(1D)4p configuration state is limited. Although the theoretical values are outside the error limits, they are in the right order of magnitude. The accuracy of the measurements is limited by the bandwidth of the detection system and noise from various sources. More accurate measurements can be performed with faster electronics (bandwidth 1 Ghz or more) or with a completely different detection mechanism that uses the full temporal resolution of the short laser pulses (35 ps). Instead of observing the fluorescence radiation, the atoms can be excited to a higher level or ionized by a short laser pulse. The second laser pulse is delayed variably with respect to the first pulse. The population of the atoms in the higher excited states, or the ions, then depends upon the remaining population of the level under investigation at the time the second laser pulse irradiates the atoms. In this measuring procedure no requirements of response speed are imposed on the rest of the detection mechanism.

References

1. Longmire, M.S., Brown, C.M., Ginter, M.L.: J. Opt. Soc. Am. 70, 1980 2. Veer, W.E. van der, D6nszelmann, A.: Z. Phys. D 17, 159 (1990) 3. Veer, W.E. van der, Diest, R.J.J. van, Kamp, A.B. van der, D6nszelmann, A.: Z. Phys. D 23, 155 (1992) 4. Carlsson, J.: Phys. Rev. A 38, 1702 (1988) 5. Petite, G., Agostini, P., Wu, C.K.: Rev. Phys. Appl. 20, 527 (1985) 6. Beck, G.: Rev. Sci. Instrum. 47, 537 (1976)