Bulk emission and interface probing of thin film CdS by two-photon spectroscopy

Bulk emission and interface probing of thin film CdS by two-photon spectroscopy

Chemical Physics 279 (2002) 249–253 www.elsevier.com/locate/chemphys Bulk emission and interface probing of thin film CdS by two-photon spectroscopy B...

343KB Sizes 0 Downloads 6 Views

Chemical Physics 279 (2002) 249–253 www.elsevier.com/locate/chemphys

Bulk emission and interface probing of thin film CdS by two-photon spectroscopy B. Ullrich a,*, R. Schroeder a,b a

Department of Physics and Astronomy, Centers for Materials and Photochemical Sciences, Bowling Green State University, Bowling Green, OH 43403-0224, USA b Department of Physics, Virginia Tech, Blacksburg, VA 24061-0435, USA Received 11 October 2001

Abstract Two-photon excitation is used to probe the bulk region of differently prepared thin CdS films on transparent substrates. The excitation is performed at room temperature by 200 fs laser pulses at 1.54 eV. By applying the Roosbroeck–Shockley relation, besides self-absorption, it is demonstrated that the emission of films fabricated by spray pyrolysis on Pyrexâ and close-spaced vapor transport on CaF2 takes place according to direct absorption inversion. The sample formed by pulsed-laser deposition on glass shows perfect agreement with the theoretical fit whereas the theory does not describe the emission of the evaporated film on glass. By comparison of top and rear face emission, the interface absorption is probed. Absorption increase at the interface of evaporated and laser deposited samples shifts the rear emission 40 meV towards lower energies with respect to the top emission. No interfacial absorption increase takes place in the other samples. Ó 2002 Elsevier Science B.V. All rights reserved.

The II–VI compound semiconductor cadmium sulfide (CdS) is of considerable interest for light emitting device (LED) fabrication because of two reasons: First, the gap of the material is at 2.45 eV at room temperature, which is close to the highest sensitivity of the human eye and second, thin CdS film preparation is fairly straightforward and has been successfully demonstrated by various techniques, such as spray pyrolysis, close-spaced vapor transport (CSVT), vacuum evaporation, and pulsed-laser deposition (PLD) [1–6]. The application of CdS films in optoelectronics, however, is


Corresponding author.

hampered by the peculiar emission properties of the films. The photoluminescence (PL) of thin film CdS is strongly influenced by the unbalanced stoichiometry, which is typical for CdS and is in general evoked by a Cd surplus. With increasing Cd content, the emission spectra move far below the gap to the red spectral range. Besides stoichiometry problems, independent of the material deposited, the PL properties of thin films depend on surface features and the influence of the interface between the film material and the substrate. It is not possible to probe the bulk PL properties by means of excitations at wavelength at or above the gap since at these energies the penetration depth of the light is typically 0:1 lm. In other words,

0301-0104/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 1 - 0 1 0 4 ( 0 2 ) 0 0 3 8 8 - 9


B. Ullrich, R. Schroeder / Chemical Physics 279 (2002) 249–253

common PL spectra reflect the surface emission or, if excited through a transparent substrate, the emission of the interface. A way to avoid absorption and to measure the bulk properties, i.e., the material between the surface region and the interface, is to excite the thin film with photon energy considerably below the gap causing twophoton (anti-Stokes) transitions, which result in gap emission of the bulk region. Indeed, in this article, we introduce two-photon spectroscopy as a powerful tool to probe bulk emission, interface features and to a certain extent the stoichiometry of thin film materials by studying various thin CdS films on different transparent substrates. Concerning two-photon excitation, one has to distinguish between coherent and non-coherent processes, i.e., one-step or two-step excitations. The latter was reported for CdS crystals due to excitation with a low power continuous-wave (cw) He–Ne laser [7]. PL induced by coherent twophoton processes was published for CdS crystals excited by various pulsed laser sources starting with the ‘‘classical’’ work almost 40 years ago [8]. The subject still attracts research activities employing ultrafast laser sources [9]. We stress that the appearance of commercial femtosecond (fs) lasers with pulse widths of about 100 fs during the last decade paved the way for experiments based on ‘‘pure’’ coherent anti-Stokes transitions since typical thermal relaxation times of semiconductors are in the range of several hundred femtoseconds ( 400 fs) and the irradiation of the material with ultrafast pulses causes carrier effects without gap alterations because of thermal effects [10]. The samples were excited with a Coherent laser system consisting of a 20 W argon pump laser, femtosecond oscillator (Mira) and amplifier (Rega). The system delivers a pulse width of 200 fs at 805 nm (1.54 eV) with a repetition rate of 249 kHz resulting in a pulse energy of 3 lJ. By focusing the beam to a diameter of about 100 lm, an intensity of 190 GW cm2 is achieved. The samples investigated were prepared by spray pyrolysis on Pyrexâ , CSVT on calcium fluoride (CaF2 ), vacuum evaporation on glass, and PLD on glass. The reactive spray deposition, CSVT and vacuum evaporation of CdS films is described elsewhere [2].

The PLD film was fabricated by infrared PLD as described in [11]. The thickness of the films was found by standard transmission spectroscopy and is 10 lm for spray pyrolysis, 10 lm for CSVT, 1 lm for evaporation, and 2 lm for PLD. The PL spectra were recorded with an Ocean Optics fiber optics spectrometer with a resolution of 1 nm. The experiments were performed in reflection (RE) and transmission (TR) geometry by collecting the PL from the front and the rear of the films, respectively. All experiments were carried out at room temperature. For the interpretation of the results, it is necessary to introduce the penetration depth dp of effective two-photon excitation. By defining the intensity decay with I ¼ I0 =e, as in the case of single-photon excitation, we find, dp ¼

1 ðe  1Þ; bI0


where b (¼6.4 cm GW1 ) [12] is the two-photon absorption coefficient of CdS and I0 is the impinging laser intensity (¼190 GW cm2 ). With these parameters, we find dp ¼ 14 lm. The comparison of dp with the thickness of the films shows that two-photon excitation probes indeed the bulk emission of the evaporated and the PLD film since the films are well penetrated by the excitation. For the spray and CSVT sample, however, the influence of absorption processes is expected. Curves (a) and (b) in Figs. 1–3 show the twophoton excited PL (TPL) of the film fabricated by spray deposition, CSVT and PLD in TR and RE geometry, respectively. The solid line shows the theoretical fit of the TPL intensity according to the Roosbroeck–Shockley relation [13] 2

IðhmÞ /

ðhmÞ aðhmÞ expðhm=kTc Þ  1


which expresses the direct inversion of absorption into emission. In Eq. (2), hm represents the emitted photon energy, a the absorption coefficient, k the Boltzmann constant, Tc the carrier temperature, which exceeds the lattice temperature T, as known from PL investigations of GaAs [14]. The absorption coefficient is calculated by the density of states (DOS) and Urbach’s tail. By introducing the continuity conditions, the cross-over energy be-

B. Ullrich, R. Schroeder / Chemical Physics 279 (2002) 249–253

Fig. 1. TPL of the spray sample measured in (a) transmission and (b) reflection geometry. The solid line represents the theoretical fit.

Fig. 2. TPL of the CSVT sample measured in (a) transmission and (b) reflection geometry. The solid line represents the theoretical fit.

tween both absorption regions is found to be at Ecr ¼ kT =ð2rÞ þ Eg [15]. Hence, for hm P Ecr , the absorption takes place according to the DOS and is expressed by the well-known formula aðhmÞ ¼ Aðhm  Eg Þ1=2


and for hm 6 Ecr according to the modified Urbach tail [15]


Fig. 3. TPL of the PLD sample in (a) transmission and (b) reflection geometry. The solid line represents the theoretical fit.

rffiffiffiffiffiffi n r o kT ðhm  Ecr Þ ; exp aðhmÞ ¼ A 2r kT


where Eg is the gap energy, the constant A represents a for hm  Eg and r is a dimensionless phenomenological parameter, which is a measure for the intrinsic character of the absorption process. The fit parameters are summarized in Table 1. The most striking difference of the spectra in Figs. 1–3 is that the TPL maximum of spray and CSVT sample does not depend on the geometry. The separation of the TR and RE spectrum of the PLD sample is caused by the increased absorption at the CdS/glass interface as pointed out in previous papers [11,16]. The reason of the interfacial absorption increase is still under investigation but it seems that it is caused by the formation of clusters of 50 nm on the glass surface at the beginning of the PLD process before an effective deposition takes place. These clusters cause an unbalanced stoichiometry and/or grain boundaries at the interface, which increase the local absorption. Samples fabricated by spray deposition do not show an interfacial absorption increase since the reactive liquid delivered onto the substrate allows an easier nucleation and film deposition than the bombardment of the substrate with ablated atoms and ions. Also the film fabricated by CSVT on CaF2 does not show an interfacial absorption increase since apparently the physical transport


B. Ullrich, R. Schroeder / Chemical Physics 279 (2002) 249–253

Table 1 Fit parameters used in Eqs. (2)–(4) and peak position of the TPL for the various films Preparation method

A ðcm1 ðeV1=2 ÞÞ


kT ðmeVÞ

kTc ðmeVÞ

Eg ðeVÞ

Peak position of TPL (eV)

Spray deposition CSVT PLD

2 105 2 105 2 105

1.20 1.20 0.95

25 25 25

45 45 40


2.335 2.375 2.365 (TR) 2.405 (RE) –

2.36 2.40 2.38 2.43 2.38 2.41

supports smooth nucleation and the growth procedure takes place in more homogeneous ways than in case of PLD. Furthermore, CdS grows epitaxially on CaF2 since the lattice constants are reasonably matched [2]. Self-absorption of the green-blue emission leads to the deviation between TPL and theory at the high-energy side in Figs. 1 and 2 [11]. With increasing film thickness, the self-absorption becomes more and more visibly dominant by narrowing the full width at half maximum (FWHM) of the emission spectrum. The FWHM of a CdS platelet with a thickness of 50 lm is 50 meV [11], which is clearly below the FWHM of 69 meV of the spray and CSVT sample. The different decay of curves (a) and (b) at the high-energy side of the TPL of the spray and CSVT sample is most likely created by light scattering and nonradiative recombination at the interface, respectively. The agreement between measurements and theory at the low energy side is excellent in Figs. 1–3 indicating that the absorption edge is very well modeled by Urbach’s tail. The TPL of the PLD film in Fig. 3 shows a very good agreement with the theory. Remarkably, shape and FWHM of the emissions are not influenced by the geometry. Therefore, at the interface, only an absorption increase takes place but no effective radiative recombination occurs. A completely different result was seen measuring the evaporated film. Fig. 4 shows that the TPL is much broader than the spectra of the other samples. This is an indication of a strongly unbalanced stoichiometry causing a considerably different recombination behavior than that expressed by Eqs. (2)–(4). The peaks of the TR and RE spectra are separated pointing to an increased absorption at the interface. The TR

(TR) (RE) (TR) (RE)

Fig. 4. TPL of the sample fabricated by vacuum evaporation measured in (a) transmission and (b) reflection geometry.

spectrum is shifted by 40 meV to lower energies, as in case of PLD. The spectra, however, are not only shifted but exhibit also a clearly different FWHM, 269 and 334 meV for the TR and RE spectra, respectively. This indicates considerable influence of the surface and interface on the emission properties. Finally, it is worthwhile to discuss the energy difference of the TPL maxima, which are summarized in Table 1. The values scatter from 2.36 to 2.43 eV. All these values are below the gap of CdS at room temperature [17]. The shift to lower energies is caused by self-absorption of the TPL [10]. Besides self-absorption, additional effects such as optical doping influence the peak position. The spray sample exhibits a rather low dark electron concentration of n  1016 cm3 . At the applied experimental conditions, the excited electron con-

B. Ullrich, R. Schroeder / Chemical Physics 279 (2002) 249–253

centration is 1019 cm3 per pulse and gap narrowing due to optical doping takes place [18]. The peak without carrier excitation lies at 2.40 eV, which corresponds to the peak position of the CSVT sample. The film fabricated by infrared PLD exhibits a high electron concentration of n  1018 cm3 in the dark [19]. Therefore, the excitation does not change the electron concentration and optical doping is unlikely. Hence, for the PLD sample only self-absorption lowers the energy of the emission peak explaining the high position with respect to the other films. Thus, the peak position of the TPL does not necessarily represent a quality measure for the film but mixture of thickness, dark concentration of carriers and possible many body effects due to optical doping. Concluding, two-photon spectroscopy on various thin CdS films was performed. Theoretical analysis of the spectra shows that thin film CdS is capable of emitting green light according to the Roosbroeck–Shockley relation if sufficient electron–hole pairs are injected into the bulk area. This is a highly notable result since the direct inversion of the fundamental absorption edge in emission is not straightforwardly expected for thin film material. Stoichiometric films with a thickness clearly below dp are required in order to fulfill the Roosbroeck–Shockley equation above the gap owing to self-absorption. We further demonstrated how to probe interface properties by purely optical means. The existence of increased interfacial absorption is revealed by the shift of the TR and RE spectra of the TPL and stresses the difference of the interface formation of the various preparation methods. Samples formed by evaporation and PLD exhibit an interface with clearly


increased absorption. On the other hand, films fabricated by spray pyrolysis and CSVT do not show a geometry dependence of the TPL peak indicating a smooth interface without an interfacial absorption increase. References [1] B.J. Feldman, J.A. Duisman, Appl. Phys. Lett. 37 (1981) 1092. [2] C. Bouchenaki, B. Ullrich, J.P. Zielinger, H.N. Cong, P. Chartier, J. Opt. Soc. Am. B 8 (1991) 691. [3] H.S. Kwok, J.P. Zheng, S. Witanachchi, P. Mattocks, L. Shi, Q.Y. Ying, X.W. Wang, D.T. Shaw, Appl. Phys. Lett. 52 (1988) 1095. [4] H. Ezumi, S. Keitoku, Jpn. J. Appl. Phys. 32 (1993) 1783. [5] B. Ullrich, H. Ezumi, S. Keitoku, T. Kobayashi, Appl. Phys. Lett. 68 (1996) 2985. [6] H. Sakai, T. Tamaru, T. Sumomogi, H. Ezumi, B. Ullrich, Jpn. J. Appl. Phys. 37 (1998) 4149. [7] B. Ullrich, N.M. Dushkina, T. Kobayashi, Jpn. J. Appl. Phys. 36 (1997) L682. [8] R. Braunstein, Ockman, Phys. Rev. 134 (1963) A499. [9] J.-F. Lami, C. Hirlimann, Phys. Rev. 60 (1999) 4763. [10] B. Ullrich, R. Schroeder, W. Graupner, H. Sakai, Opt. Express 9 (2001) 116. [11] B. Ullrich, R. Schroeder, IEEE J. Quantum Electron. 37 (2001) 1363. [12] T.D. Krauss, F.W. Wise, Appl. Phys. Lett. 65 (1994) 1739. [13] W. van Roosbroeck, W. Shockley, Phys. Rev. 94 (1954) 1558. [14] P.Y. Yu, B. Welber, Solid State Commun. 25 (1978) 209. [15] B. Ullrich, C. Bouchenaki, Jpn. J. Appl. Phys. 30 (1991) L1285. [16] B. Ullrich, H. Ezumi, S. Keitoku, T. Kobayashi, Appl. Phys. Lett. 68 (1996) 2985. [17] V.V. Sobolev, V.I. Donetskikh, E.F. Zagainov, Sov. Phys. Semicond. 12 (1978) 646. [18] B. Ullrich, R. Schroeder, Semicond. Sci. Technol. 16 (2001) L37. [19] B. Ullrich, H. Sakai, Y. Segawa, Thin Solid Films 385 (2001) 220.