Doped nanocrystals of semiconductors - a new class of luminescent materials

JOURNAL OF

LUMINESCENCE ELSEVIER

Journal of Luminescence 60&61 (1994) 275—280

Invited paper

Doped nanocrystals of semiconductors luminescent materials

—

a new class of

R.N. Bhargavaa*, D. Gallaghera, T. Welkerb ~P/iilips Laboratories, Philips Electronic’s North America Corporation, 345 Scarborough Road. Briarclifi Manor, NY 10510, USA 5P/nlipc GM BH Forschungslahoratorien, A achen Germane

Abstract We report on the unique luminescent properties of nanocrystals of Mn-doped ZnS with varying sizes from 30 to 70 A prepared at room temperature. These nanosize quantized particles yield the best external photoluminescent quantum efficiency of about 18% at room temperature and luminescent decay time at least five orders of magnitude faster than the corresponding Mn2~ radiative transition in the bulk crystals. These luminescent measurements also suggest that the efficiency increases with decreasing size of the nanocrystalline particles. These novel properties may be attributed to electron—hole localization and hybridization of the s—p host states with d-electrons of the Mn impurity.

During the past few years, the preparation and characterization of materials in the nanometer scale has provided not only new physics in reduced dimensions, but also the possibility of fabricating novel materials. Nanosize semiconductor crystallites have changed properties, such as an increased energy band gap, which results from quantum con-

provides an effective energy transfer path and leads to high luminescent efficiencies at room temperature. The key feature in the present work for Mn-doped ZnS nanocrystallites, apart from increasing the energy gap with decreasing size, is that the spatial confinement of the electron—hole at the nanocrystal also profoundly affects the recombina-

finement [1,2]. Here we report on the optical properties of the first realization of ZnS semiconductor nanocrystals doped with Mn isoelectronic impurities [3,4]. The Mn2~ion d-electron states act as efficient luminescent centers while interacting strongly with the s—p electronic states of the host nanocrystal into which external electronic excitation is normally directed. The electronic interaction

tion kinetics of the Mn2” ion d-electron luminescence. More generally, we feel that nanocrystals doped with optically active luminescence centers create new opportunities in luminescent study and application of nanoscale material structures. Nanocrystalline ZnS powder was precipitated with an organometallic chemistry in hydrocarbon solvent. Diethylzinc reacted with hydrogen sulfide in toluene to form ZnS. Bulk powders are usually doped by thermal diffusion of Mn from a salt or carbonate at high temperatures (>1100°C)

*

Corresponding author. Present address: Nanocrystals Technology. P.O. Box 82, Briarcliff Manor, NY 10510, USA.

0022-2313’94/$07.00 ~ 1994— Elsevier Science By. All rights reserved SSDI 0022-2313(93(E0443-2

176

R. \ B/uir~n0

ci

~iI

./oiirmi/ 1

hut since nanocrystallites sinter at extremely low temperatures. they must he doped during precipitation. To dope the ZnS. manganese chloride is reacted with ethylmagnesium chloride to form diethylmanganese in a tetrahydrofuran solvent and added to the reaction, The separation of the particles is maintained by’ coating with the surfactant methacrylic acid. In the coated ZnS : Mn particle system we observe a gradual but significant in2 emiscrease in the luminescent intensity of Mn sion when exposed to exciting 300nm UV light (Ii V-curing) [4]. The photoluminescent efficiency in II V-cured 27 33 A size (diameter) ZnS: Mn nanocrystalline powder is about 1 8% [4]. Our hy— pothesis is that this enhancement in efficiency resuits from the passivation of surface of the ZnS : Mn nanocrystals and is related to the photopolymerizatton of the surfactant. The optical properties of these doped nanocrystals were primarily studied at room temperature with the band-to-band excitation in ZnS being used to excite the Mn2 emission. The subsequent transfer of electron and hole into the electronic level of the Mn ion leads to the characteristic emission of Mn2 in ZnS [5]. This yellow emission observed in the photoluminescence (PL) of bulk ZnS: Mn is associated with the Mn2 + 4T ~—°A 1transition and it peaks at around 2.12 eV (585 nm) with a half width of 0.23 eV at room temperature. The PL and photolurninescenee excitation (PLE) data for three sizes of nanoerystals of ZnS: Mn are compared in Fig. I. These spectra were measured under identical conditions and therefore the PL intensities represent the relative luminescent efficiencies associated with the different sizes of the nanocrystals. The perturbation by the host crystal fields renders the otherwise spin forbidden d d intra-ion transition partially electric dipole allowed (through the mixing of opposite spin states). The large line width of the yellow emission is due to a combination of inhomogeneous broadening and phonon assisted 2 + transitions in[5]. The PL peak of thewhich Mn is transition the nanocrystal is 2.10eV, slightly shifted to a lower energy when compared to the peak in the bulk sample. The PLE spectrum for the hulk ZnS: Mn powder peaks at 3.73 eV (332 nm). This PLE peak should be compared with the room temperature energy band gap of 3.66eV

/.iu,iinco

c)i~ (i(htOI

/0)4

.Y’5 2,0)

~

I ig. I. Flie PLF and Pt. spectra 01 5 arlous si/c natlocr\ stiLIIiIiC particles of ZnS: Mn. Note that the position of the dominant Pl.E peak reflects the absorption edge oi the host crsstal -

for the cubic ZnS single crystal. In contrast, the PLE spectrum for about 30 A nanocrystals peaks at 4.68eV (265 nm). This large shift of 1.02 eV to a higher energy is interpreted [6] as being due to an increase in the value of the s p electron band gap in the ZnS nanoerystals from quantum confinement. The exciton Bohr radius (Us) for ZnS is estimated to be 25 A [7]. For ZnS nanoerystallites smaller than soA in diameter (D), the size dependent shift of the energy hand gap (Eg) to higher energies is estimated within the frame work of the effective mass theory [I] and for 30A diameter particle the estimated energy shift is 0.55 eV [7]. This should he compared with experimental value of 1.02 eV. There are several possible causes for this diserepancy: (i) the theoretical estimate of the band gap begins to deviate from the real value as the particle size gets significant smaller than the excitonic Bohr radius~(ii) the size of the particle as estimated from transmission electron microscopy only provides an approximate value of particle size within a given range of size distribution~(iii) the PLE yields an approximate of the s p energy band gap. we The abovevalue results become clearer when study the dependence of peak position and intensity of PLE on size of the nanocrystal. Firstly. the highest energy peak in the PLE spectra is indicative of the band edge absorption of the smallest particle. This is consistent with the observation that in a given sample, the peak of the

RN. Bhargara et al.” Journal of Luminescence 60&6 I (1994) 275—280

PLE spectrum shifts to higher energy under UV exposure [4]. Secondly, the highest energy peak of the PLE spectra in the UV-cured sample has the

277

ing sphere attached to a spectrometer. Such efficiency is to our knowledge the highest observed in nanocrystals and is a direct consequence of

highest intensity, suggesing that

the smallest 2”’ particles are contributing the most to the Mn emission. This is again confirmed by measuring efficiency as a function of particle size and is discussed later. The correlation between the energy band gap of the ZnS host, as measured through PLE spectra

fast capture of the excited electron—hole pairs into the Mn ion impurity,t and a subsequent efficient radiative decay of the d-electron excitation. This contrasts with the circumstance usually encountered with undoped nanocrystals where the surface provides a very effective nonradiative path for the electron—hole pairs [10]. In an undoped

using yellow emission from Mn ion as a probe,

quantum dot, the external luminescent efficiency at

and different sizes of the nanocrystals confirms

room temperature is too poor to be of any practical

that the energy transfer occurs from the s—p dee-

consequence and rapidly decreases with decreasing

tron—hole pair band states to the d-states of Mn ion. This establishes beyond doubt that the Mn ion is an integral part of the ZnS nanocrystal.

size of the particle [11]. The problem is acute in the

Based on earlier experiments performed on bulk

rystal is quite large. Therefore, even with the partial

powders [5], we suggest that Mn2 + is occupying the Zn2 + cation site in ZnS nanocrystals. Recent EXAFS experiments [8] seem to also confirm that Mn is occupying a Zn site. This has also been observed in nanocrystals of ZnMnS [9]. The determination of the band gap of the host crystal by measuring the PLE peak responsible for the impurity emission gives us an elegant and rapid tech-

passivation provided by the UV-cured polymer coating, it is still remarkable that these high luminescence efficiencies are observed in our experiments. This is discussed below using an electron—hole localization model. For undoped nanocrystals, in the limit of a strong confinement case when D < 2aB [1], the electron and hole in a nanocrystal occupy the mdi-

nique to assess the size distribution and their rela-

vidual lowest eigenstate with relatively little spatial

tive contribution to total luminescent efficiency in a given nanocrystal sample with multiple nanosize particles, We now turn to the energy transfer process and its dynamics in the ZnS: Mn nanocrystals. The best external quantum efficiency measured to date in 304. nanocrystals has been about 18%. To measure efficiency, the light of the Xenon arc lamp was dispersed into monochromatic radiation using dual spectrometer system where the excitation and emission intensities were both calibrated. To obtain the value of quantum efficiency, the light output from nanocrystals and standard phosphors of known quantum efficiency were excited at the wavelength of the maximum of the PLE band of the nanocrystal and compared. Integrating the emission spectra yields the number of photons emitted by the sample. Care was taken to account for the absorption in the nanocrystals and standard phosphor by measuring the reflection spectra. Additionally, these efficiency measurements were also repeated in an integrat-

correlation. These electrons and holes trapped separately at different locations recombine through a combination of radiative and nonradiative tunneling and show long decay time [1]. The efficiency of such electron—hole recombination at room ternperature is rather low and surface recombination probably remains the dominant nonradiative path. Additionally, the surface recombination may also dominate due to a phonon bottle-neck phenomenon among the discrete quantum states of the dot [11]. However, in doped nanocrystals we expect efficient energy transfer of the electron—hole pairs to the impurity and it should depend on the size of the nanocrystals. Indeed, in case of doped nanocrystals

small particle limit (such as the one considered

here), since the surface to volume ratio of a nanoc-

The transfer rate ofelectrons from ZnS host to Mn2~excited states is not experimentallyknown in the bulk. We observe the rise time of Mn2~’luminescence to be less than SOOps in . ZnS : Mn nanocrystals which when coupled with observed high luminescent efficiency suggests that the electron—hole transfer to Mn2* energy level manifold is fast and efficient.

RN. Matien, a a!. Jounwi vi I.wulnesamr (sOdA! ‘1Q94. 275 28(1

27K

3C~

•\uI,uu

g

I

‘°

0I___l_. 10

I

I

I

I•I

20 30 40 50 50 70 50 90 100

Diomew~langitroms)

Fig. 2. Ia) Schematic representation of the capture process and the times measured in 35 A nanocrystals. 1W Variation of external luminescent efficiency as a function of the size of the nanocrystals. Note that there is apt to be a significant error margin in the P1. efficiency measurements, but the values ate consistant with relative emission intensities measured separately.

of size less than the exciton Bohr radius, we observe a large increase in efficiency with decreasing size of the nanocrystaL We believe this effect is related to amodulation nanocrystal. the hole recombination process of For electron capture rates in shown in Fig. 2(a), the internal efficiencyq is simply given by q = tj

1/(Tj’ + tp~~),

Tg and rj~jare the radiative and nonradiative rates related to Mn and are primarily surface states. The band-to-band recombination rate is slow [1] and is neglected. The capture rate Tjt is proportional to the number density of Mn2 • at the Zn site within the nanocrystat2 hence, it is inversely proportional to the volume of the nanocrystal (i.e. tjt x D ~)for the case of a single Mn ion within a nanocrystal. The capture rate

where

TNJ should depend on the number of surface atoms per unit volume which is inversely proportional to the size of the particle (i.e. rjj x W l). Thus, the The capture rate at the impurity I t1 = Nc~where si is the cross-section, &~is the thermal velocity and N is the number of impurities per unit volume. We make an assumption that in these nanocrystals there is one impurity per nanocmtal, based on our doping concentration. More than one impurity would decrease the luminescent efficiency ~ ~

tration quenching.

internal efficiency can be expressed as I 2’ I + P1) where ft is related to the ratio Ta/T,a. In the case of all surface related nonradiative recombination. where 1)—’ x. q-.0 and in the limit D—’O. the internal efficiency goes to unity. We have plotted the functional dependence of efficiency on the size of the doped nanocrystal in Fig. 2(b) as well as the measured values of the efficiency for 75.55 and 30 A nanocrystals. The presence dan impurity within the nanocrystal and the localization of electron and hole wave function due to the confinement in a nanocrystal results in an enhancement in the efficiency. In bulk crystalline material, the partially spinforbidden Mn2 fl 1. bA transition has a lifetime of about 1.8 ms [5]. In our nanocrystals. we find dramatic shortening of this decay time by several orders of magnitude. Fig. 3 shows the transient yellow luminescence for 30A size particles, measured by exciting the host ZnS nanocrystal above the band gap by a picosecond pulsed laser [3]. The laser pulse is also shown as a reference. In the inset we have plotted the luminescence dccay on ‘1=

a semilog scale. Two distinct decay time constants of 3.7 and 20.5 ns are identified, suggestingthat two

RN. B/IargaL’a Ct 0/. / Journal at Luminescence 60&61 (1994) 275--280 14

~—~-~-—

I

F

I

F

I

I

279

I

°~ lime (nanoseconds)

Fig. 3. The time decay of the light emission from 30 A nanocrystalline ZnS: Mn. The last pulse is shown by dots as a reference. Inset: the time decay is separated into its exponential decay components.

different recombination centers may be involved. In PLE spectra, frequently several peaks are also observed which suggests that we have distinct and discrete size distribution. The different sizes of partides, as observed in PLE spectra during the UV-curing process, probably contribute to these two separate decay times [3]. We have carefully examined any remaining longer lived components in the PL, and conclude that the occurrence of the dominant radiative decay on a nanosecond timescale is indeed a characteristic of the small ZnS: Mn nanocrystals. On the other hand, we have not been successful in measuring the capture rate of the electron—hole pairs into Mn2 + ion d-electron states. This is limited by our ability to resolve the rise time of the yellow luminescence, and because the energy transfer occurs on a picosecond time scale. We now argue that the decay time measurements of the d-electron luminescence is consistent with a strong sp—d electron system in the nanocrystals. Given the high external quantum efficiency of the yellow emission, the extraordinary shortening in the d-electron decay time must include a significant degree of hybridization of the s—p host states with

those of the d-electron. It is unlikely that modified crystal fields, given the proximity to a surface in a crystal, can alone be responsible for such admixing. Instead, we believe that the electronic confinement experienced by the s—p states and the corresponding increase in their spatial overlap with the localized d-electron states promotes the process of hybridization. An immediate benefit of the mixing is, of course, an enhanced energy transfer (capture) rate of the electron—hole pairs at the d-electron sites. The enhancement of efficiency in Mn-doped ZnS nanocrystals is not only consequence of the faster energy transfer to Mn2 + ion but also is connected with the change of the decay rate of Mn emission. The faster decay time as a result of strong sp—d mixing due to localization of electron and hole on Mn2 + is necessary for this enhancement. If the lifetime of Mn2 + emission remained as long as in the bulk crystal (1.8 ms), faster transfer rate would not result in high efficiency because light output would saturate at very low levels of excitation. We have also studied the quantum confinement effects in Tb3 + in ZnS nanocrystals. These doped nanocrystals were prepared by precipitation within a polymer film and quantum confinement effects

28(1

R./V. Bhargaca ~'t a/.

,Iournal ol l.tonim',~cence (idN61 , 1994: _'~/~" 2:~#

results suggest that "'doped zero-dimensional quantum confined structures" with proper surface passivation can provide a new class of luminescent materials for the future.

35 P[ L

PL

54.5

25

2O

304

Acknowledgements

S

L

250

300

__,

350

400

i

i

i

i

450

500

550

600

We would like to thank J. Racz for photoluminescent spectroscopy and X. Hong and A. Nurmikko of Brown University for the time decay measurements. 650

Wovelength (nm)

References Fig. 4. The PI. and PLE spectra of nanocrystals of Z n S : T b 3~ embedded in polyeth3,1ene oxide. The shift of the band gap corresponds to about 60/)k particle size.

[I] 1, Brus, J. Phys. ('hen3. 90 11986) 2555 and references

resulting in the band gap shift were observed. In Fig. 4 we have plotted the PLE and PL spectra for this sample. The PLE peak corresponds to a larger band gap energy of 4.05eV in these quantum confined nanocrystals. In summary, we report large external luminescent efficiencies in doped nanocrystals accompanied by shortening of lifetime of the transition associated with the impurity. These results are explained on the basis of the interaction of the s-p electron -hole of the host (ZnS) and the d-electrons of the impurity (Mn). Further, theoretical and experimental work is needed to understand the detailed physics of these doped nanocrystals. These

[2] Y. Wang and N. Herron, J. Phys. ('hem. 95 (1991) 525 and references thereto. [3] R.N. Bhargava, D. Gallagher, X. Hong and A. NtJrmikko, Phys. Rev. Lett. 72 11994)416. [4] D. Gallagher, W. Heady, J. Racz and R.N. Bhargava, J. ('rystal Growth, to be published [5] H.E. Gumlich, J. Lumin. 23 11981) 73: see also W. Bussc, H.E. Gumlich, B. Meissner and D. Theis, J. Lumin 12 13 [1976) 693. [6] L. Brus, J. Quantum Electronics QE 22 119861 1909. [7] R. Rossetti, R. Hull. J.M. Gibson and L.E. Brus. J. Chem. Phys. 82 11985) 552. [8] Y.I_. Soo, Z.H. Ming, S.W. Huang, Y.H. Kao, R.N. Bhargava and D. Gallagher, Phys. Rev., submitted. [9] Y. Wang, N. Herron, K. Moiler and T. Bein, Solid Slate C o m m u n . 77 [1991) 33. [10] J. V i l m s a n d W.E. Spicer, J. Appl. Phys. 36 11965) 2815. [11] H. Benisty, C.M. Sotomayor-Torres and C Weisbuch. Phys. Rcv. B 44 119911 10945.

therein.