STM light emission spectroscopy of Si surfaces

STM light emission spectroscopy of Si surfaces

~ ) Solid State Communications, Vol. 84, Nos. 1/2, pp. 173-176, 1992. Printed in Great Britain. 0038-109819255.00+ .00 Pergamon Press Ltd STM LIGHT...

295KB Sizes 3 Downloads 34 Views

~ )

Solid State Communications, Vol. 84, Nos. 1/2, pp. 173-176, 1992. Printed in Great Britain.

0038-109819255.00+ .00 Pergamon Press Ltd

STM LIGHT EMISSION SPECTROSCOPY OF Si SURFACES S. Ushioda Research Institute of Electrical Communication, Tohoku University 2-1-1 Katahim, Aoba-ku, Sendai 980, Japan (Received 10 May 1992 by M. Cardona)

We have measured the visible light emission spectra from the surfaces of single crystal Si by injecting electrons and holes from the tip of a scanning tunneling microscope (STM). The spectra and intensities change depending on the STM tip material, the bias voltage polarity, the carrier type and concentration of the sample, as well as the crystal orientation of the sample surface. Possible emission mechanisms for the observed spectra are discussed.

DEDICATION It is a great pleasure and honor to contribute a paper to the c o m m e m o r a t i v e issue o f the S o l i d S t a t e Communications on the occasion of the 75th birthday of Professor Elias Burstcin. Eli was my Ph.D. thesis adviser at the University of Pennsylvania almost 30 years ago, and ever since he has been my respected teacher and friend for my entire scientific career. I would like to congratulate Eli for the very successful stewardship of Solid State Communications as Editor-in-Chief and for his many important contributions to the solid state physics community in general. To celebrate his 75th birthday and at the same time very successful completion of his service as the first Editorqn-Chief of Solid State Communications, I would like to dedicate this article that reports the first observation of visible light emission from crystalline Si using the STM as the incident electron (hole) source.

and through the prism.6 We found that the spectra are very sensitive to the conditions of the STM tip and the sample surface. In the present work we have measured the STMLES of Si surfaces for the first time. Here, we report the spectral data and discuss possible processes that can produce visible light at Si surfaces. The experimental details will be published in a full length paper.7 EXPERIMENT The experimental setup is illustrated in Fig. 1. A home-made STM was placed on a rotation table in an ultra-high vacuum (UHV) chamber whose base pressure is about l x 1 0 "9 T o r e The image of the STM tip is focused in the entrance slit of a spectrograph through the light collection optics that consists of lenses, a view port, an iris, and a polarization analyzer. The spectra were recorded by a linear array of photon counting detectors, making it possible to record the spectrum between 1.5 eV and 2.7 eV in a single exposure. The spectra we present here are corrected for the energy dependent sensitivity of the spectrograph-detector system. All measurements were made at room temperature. We studied three different Si samples; p-type with a low hole concentration (lx1015 cm'3), p-type with a high hole concentration (3.8x1018 cm'3), and n-type with a high electron concentration (3.5x10 18 era"3). T he sample surface was hydrogen terminated by removing the thermal oxide with a dilute aqueous solution of HF.8 Then it was quickly brought into an UHV chamber, and the chamber was pumped down to 3xllY 9 Torr in 24 hours. Light emitted at 75 ° from the surface normal was collected. Because of the shadow effect of the STM, the actual collection angle was 75"(+11",-5").

INTRODUCTION Light emission from the tip-sample gap of the STM was first discovered in 1988 by Gimzewski et ai. 1 and Coombs et al.-. Glmzewskl et al. measured UV emlsslon from Si and Ta surfaces, using isoehromat spectroscopy. Visible emission spectra were reported for a Ag sample by Coombs et al.2 and for GaAIAs samples by Alvarado et al. 3 and A b r a h a m et al. 4 Both localized surface plasmons (LSP) and surface plasmon polaritons (SSP) that are excited by the incident beam of electrons (or holes) from the STM tip are the source of light from metallic samples. In GaAIAs the iight is emitted mainly by electron-hole pair recombinations across the energy gap. Our group has measured the STM light emission spectra (STMLES) of Au films deposited on a prism coupler.5,6 When the sample is a thin metallic film on a prism surface, SPP can emit light through the prism by conserving the wavevector parallel to the surface. Then the emission angle is sharply defined by the waveveetor conservation condition between the light in the prism and the SPP on the metal film. The angle dependence of e m i s s i o n i n t e n s i t y was the key to i d e n t i f y i n g this mechanism in our previous work. More recently, we have measured the STMLES of Au films both on the tip-side

RESULTS AND DISCUSSION We have m e a s u r e d S T M L E S of Si for many different samples and experimental conditions 7 T h e spectra were collected for different carrier types and c o n c e n t r a t i o n s , both polarities of the bias voltage (electron or hole injection), both P- and S-polarizations of the emitted light, different crystal surfaces, (100) and (111), and with different tip materials, W and Pt. All the data presented in this paper were collected, using the STM 173

Vol. 84, Nos 1/2

STM LIGHT EMISSION SPECTROSCOPY

174 Sample

8 E-~

Tip

~p

(100) i/) j

6

hv

4

,,,

\.,

.." View Port

\B! i

(b) Tlp

z(n

(loo) ~

4;J

_z

P o l a r ~

Optical Multlehannel

o..,0,

,,

\ /,

1.5"

2 2.5 ENERGY(e~

hv

3

Fig. 2 Emission spectra of p-Si(100) when electrons are injected into the sample. (a) Low carrier concentration sample: p=lxl015 cm"3. Bias voltage: +25 V (Note that we always measure the bias voltage with respect to the tip. The direction of electron tunneling is indicated by an arrow.) Current: 200 nA. (b) High carrier concentration sample: 3.8x1018 cm "3. Bias voltage: +27 V. Current: 1 ~A. Fig. 1 Experimental setup for STM light emission spectroscopy in UHV.

tip made by elctrochemieally sharpening a piece of W wire. In this Communication, instead of describing the whole range of observed effects, let us concentrate on some of the most interesting aspects of the results. Fig. 2 illustrates the contrast between the low carrier concentration and high carrier concentration samples of pSi(100). Fig. 2a and 2b are the spectra for the low and high carrier concentration samples, respectively. Electrons are injected into the sample in both cases. The bias voltage is approximately the same, while the current was made much higher for the high carrier concentration sample to push the limit. (See figure captions for actual values.) Note that the bias voltage of 25 to 27 volts is very high compared with the usual levels used in the STM measurements. We needed this level of bias voltage to obtain the emission intensity that can be spectrally analyzed. We see that the emission intensity is much lower for the high hole concentration sample than for the low concentration sample. If the light emission process in Si is analogous to the that in GaAs, and light is generated by electron-hole pair recombinations, one would expect stronger emission from the high hole concentration sample than from the low concentration sample, for electrons are being injected. 3 Besides, the energy gap of

Si is indirect and lies in the near infrared, not in the spectral range being measured. Thus the observed emission from Si does not arise from simple electron,hole pair recombinations near the band gap. With the n-type sample, the spectra shown in Fig. 3a and 3b were obtained for electron injection and hole injection, respectively. The emission spectra are quite similar to each other with only slight differences around 2.2 eV. Furthermore, these spectra are similar to that of the low concentration p-type sample shown in Fig. 2. This evidence further suggests that the light emission is not due to electron-hole pair recombinations near the energy gap. When holes were injected, the low carrier concentration p-type sample did not emit significant amount of light (data not shown), but the high concentration n-type sample emits light as seen in Fig. 3b. We found that when a Pt tip is used instead of W, the low carrier concentration p-type sample emits light for both electron and hole injection, and the two spectra are very similar. It is likely that part of the light is emitted by the STM tip. Fig. 4 shows the polarization dependence of the emission spectra for the low carrier p-type sample with the (100) surface. The direction of the electric field for Spolarization is aligned with one of the {110} crystal axes. We see that most of the light is emitted with P-

15

15

(a) Tip

(°)

t

hv

10

(i)

1

'

z

"7'

.

.

.

.

.

.

.

.

.

l"tp,

z w I-

z 10

175

STM LIGHT EMISSION SPECTROSCOPY

Vol. 84, Nos 1/2

(loo) ~ ~

10

i

hv

ds-pot. (~1o) .... [ ]

5

0

1

1.5









I

,







I

2 2.5 ENERGY (eV)

.IL,

3

Fig. 3 Emission spectra ofn-Si(100). Carrier concentration: n=3.5x1018 cm-3. (a) When electrons are injected into the sample. Bias voltage: +26 V. Current: 300 nA. (b) When holes are injected into the sample. Bias voltage: -26 V. Current: 300 hA. polarization. To compare with the data of Fig. 4 we show the P- and S-polarized spectra for the same sample but for a different crystal surface (111) in Fig. 5. Again the direction of the electric field for S-polarization is aligned with one of the {110} crystal axes. On the (111) surface both polarizations contribute approximately the same amount of light Thus we see that the polarization characteristics of the emission depends on the crystalline orientation of the sample surface. Since the bias voltage is quite high compared with the energy gap of Si, the electrons are injected into very high energy levels above the bottom of the conduction band, and the holes are injected deep into the valence band. Thus, one expects that the electronic transitions responsible for the light emission are between levels deep inside the conduction or valence band. Hence whatever happens near the band edges depending on the carrier types does not affect the emission spectra. Then why is t h e r e d i f f e r e n c e b e t w e e n the low and high h o l e concentration p-type samples? This may be due to the different degree of band bending at the surface. Another possibility we must consider is that the emission may be due to transitions that involve surface electronic levels. In considering the present observations, we must not forget the possibility that the sample and tip surfaces may be modified by the appreciable current that flows across the v a c u u m gap. T h e r e are r e p o r t s that s u r f a c e modification takes place even at lower current levels.9 However, the spectra are reproducible, and it is clear that the observed spectra contain meaningful information about the sample surface. To check all the possibilities

1.5

2 2.5 ENERGY(e~

Fig. 4 Polarization of the spectra for the (100) surface of the low carrier concentration p-type sample. (a) Ppolarized component; (b) S-polarized component. For both spectra electrons were injected into the sample from a W tip. The emission angle is 75" from the surface normal, and the electric field direction for S-polarization is parallel to one of the {110} crystal axes. Bias voltage: +27 volts. Current: 500 nA.

and identify the light emission mechanism, we must try more experimenting with varied and well-controlled sample surface conditions. The observed emission may be related to the light emission from the surface of porous Si that has received much attention recently. 10 T h e emission mechanism for porous Si is not yet understood either. We have discovered an interesting phenomenon, but we do not yet have answers to many questions raised. CONCLUSION W e have discovered that visible light is emitted from the region between the STM tip and well-defined surfaces of single crystal Si, and measured the visible emission spectra for varied surface and bias conditions in UHV. The emission intensity and spectra depend on the carrier type, carrier concentration, current direction, and crystal orientation of the surface, as well as the tip material. The electronic transitions that produce the observed light have not been identified yet. ACKNOWLEDGMENTS The spectra presented in this paper were measured by Mr. M. Kuwahara. Prof. Y. Uehara has made valuable contributions throughout this work. This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, and Culture.

176

STM LIGHT EMISSION SPECTROSCOPY

REFERENCES 1. J. K. G imzewski, B. Reihl, J. H. Coombs, and R. R. Schlittler, Z. Phys. B72, 497 (1988). 2. J. H. Coombs, J. K. Gimzewski, B. Reihl, and J. K. Sass, J. Microsc. 152, 325 (1988). 3. S. F. Alvarado, Ph. Renaud, D. L. Abraham, Ch. Sch6nenberger, D. J. Arent, and H. P. Meier, J. Vac. Sci. Technol. Bg, 409 (1991). 4. D. L. Abraham, A. Veider, Ch. Scl'~nenberger, H. P. Meier, D. J. Arent, and S. F. Alvarado, Appi. Phys. Lett. 56, 1564 (1990). 5. K. Takeuchi, Y. Uehara, S. Ushioda, and S. Morita, J. Vac. Sci. Technol. Bg, 557 (1991). 6. S. Ushioda, Y. Uehara, and M. Kuwahara, Appl. Surf. Sci. (In press). 7. M. Kuwahara, Y. Uehara, and S. Ushioda (To be published). 8. G. S. Higashi, Y. J. Chabal, G. W. Trucks, and Kfishnan Raghavachari, Appl. Phys. Lett. 56, 656 (1990). 9. M. Ringer, H. R. Hidber, R. Schl6gl, P. Oelhafen, and H.-J. G/intherodt, Appl. Phys. Lett 46, 832 (1985). 10. L. T. Canham, Appl. Phys. Lett. 57, 1046 (1990).

ta)

20

A

!-

z

10

ai

t,-

<~ o

~)

z

.i I_z

Tip .

hv lO

20

1ot

10

0 1.5

2 2.5 ENERGY (eV)

Vol. 84, Nos 1/2

3

Fig. 5 Polarization of the spectra for the (111) surface of the low carrier concentration p-type sample. (a) Ppolarized component; (b) S-polarized component. For both spectra electrons were injected into the sample from a W tip. The emission angle is 75" from the surface normal, and the electric field for S-polarization is parallel to one of the { 110} crystal axis. Bias voltage: +27 volts. Current: 600 nA.