Sol-Gel Based Vertical Optical Microcavities with Quantum Dot Defect Layers

FULLPAPER

3772

DOI: 10.1002/adfm.200800784

Sol–Gel Based Vertical Optical Microcavities with Quantum DotDefect Layers**

By Jacek Jasieniak, Cinzia Sada, Alessandro Chiasera, Maurizio Ferrari, Alessandro Martucci, andPaul Mulvaney*

The interaction between CdSe-CdS-ZnS semiconductor quantum dot emitters (QDs) and sol–gel based asymmetric Bragg

microcavities is studied. It is found that the quality factor (Q) of such microcavities is limited by interlayer diffusion. This results

in maximum Q values of less than 100. Despite this, these structures still exhibit optical signatures characteristic of the Purcell

effect, where large modifications to the photoluminescence properties of QDs are observed.

1. Introduction

An optical cavity is an arrangement of mirrors that enables

standing waves of a defined wavelength to be generated.[1]

Conventionally, such cavities have been based on open space

structures in which the standing wave forms in the air gap

between two mirrors and the length of the cavity is much

greater than the wavelength of the standing mode created. This

is particularly advantageous when creating highly coherent and

spectrally narrow cavity modes.[2] More recently, researchers

have begun to explore microcavities.[3–7] Such structures,

which possess cavity length scales of the order of the cavity

mode wavelength, can, unlike their larger counterparts, cause

significant modifications to the optical properties of any

optically active species located within.[8] Semiconductor

quantum dots (QDs) possess high photoluminescence (PL)

stability and can exhibit high PL quantum yields. Con-

sequently, it is of interest to examine how their emission

[*] Prof. P. Mulvaney, J. Jasieniak[+]

School of Chemistry, University of MelbourneParkville, Victoria 3010 (Australia)E-mail: [email protected]

Prof. A. MartucciDipartimento di Ingegneria Meccanica Settore MaterialiUniversita’ di Padova Via Marzolo, 9, 35131 Padova (Italy)

Dr. C. SadaDipartimento di Fisica, Universita’ di PadovaVia Marzolo, 8, 35131 Padova (Italy)

Dr. A. Chiasera, Dr. M. FerrariCNR-IFN Istituto di Fotonica e Nanotecnologie, CSMFO groupVia alla Cascata 56/C, 38050 Povo, Trento (Italy)

[+] Present address: CSIRO Division of Molecular and Health Technol-ogies, Ian Wark Laboratory, Bayview Avenue, Clayton 3168 (Australia)

[**] J. J. acknowledges funding through the Australian postgraduate awardscheme and the PFPC for an overseas travel allowance. P. M. thanksthe ARC for its support through Grant FF 0561486. A.M. thanks MIURfor its support through project PRIN2006031511. Supporting Infor-mation is available online from Wiley InterScience or from the author.

� 2008 WILEY-VCH Verlag GmbH &

may be modified within a microcavity. While the nature of the

microcavity structure can vary, in this paper we focus

specifically on a form of 1D photonic crystal, namely an

asymmetric Bragg microcavity. Such structures, which are

based on periodic variations of the refractive index, possess

both large cavity mode spectral selectivity and small cavity

volumes, enabling large modifications to the optical properties

of dopants located within the active layer.[9,10] Lodahl et al.

were among the first to show that the spontaneous emission of

QDs within photonic crystals could be varied; they employed

inverse-opal, titania-based photonic crystals.[11] More recently

Ganesh et al. incorporated colloidal QDs into a direct 2D TiO2

photonic crystal. By selectively matching the ‘‘leaky’’ modes of

the structure to the emission and absorption of the QDs, a PL

enhancement up to 108 times that of a flat QD film with no

photonic crystal could be achieved.[12] Both of these studies

highlight the viability of utilizingmicrocavities based on colloidal

crystals to modify the optical properties of quantum dots.

In this work we employ wet-chemical methods to grow

the quantum dots, and sol–gel chemistry to fabricate the

grating structures. Consequently, the preparation of the entire

structures is scalable. The geometry that we have utilized is an

asymmetric Bragg cavity. It comprises a Fabry–Perot cavity

composed of a single Bragg mirror, a deposited QD/ZrO2

defect layer, and a confiningAgmirror grown on top. Aswill be

shown, despite its simplicity, such a structure can significantly

alter the optical properties of the QD emitters due to their

coupling to the cavity modes.

2. Results and Discussion

The bottom-up approach to the fabrication of Bragg

microcavities requires careful monitoring of the physical and

optical properties of the resulting structures at each step. In the

following sections we describe each of these characterization

steps sequentially.

Co. KGaA, Weinheim Adv. Funct. Mater. 2008, 18, 3772–3779

FULLPAPER

J. Jasieniak et al. / Sol–Gel Based Vertical Optical Microcavities

Figure 2. Cross-sectional SEM of a six double-layer SiO2–TiO2 Braggreflector deposited on a silicon substrate. The TiO2 layers are composedof titania nanocrystals with radii of between 10 to 20 nm. In contrast, theSiO2 layers possess amorphous structures.

2.1. Characterization of the Bragg Reflector

2.1.1. Spectroscopic Ellipsometry

Spectroscopic ellipsometry was utilized to monitor the film

thicknesses and refractive indices throughout the deposition of

the Bragg cavities. In Figure 1A we show an example of the

ellipsometric characterization of a Bragg reflector with seven

TiO2–SiO2 layers. In this case we optimized the data fit through

a simultaneous error minimization at two incident angles, 65 8and 70 8. This routine allows for a more accurate fitting of the

experimental data, especially for multilayer structures. As

seen, despite the complexity of the system, an acceptable fit to

the data is generated with physically reasonable parameter

values. These parameters, which are shown in Figure 1B, are in

good agreement with the experimentally optimized deposition

thicknesses of 74 and 109 nm for the individual TiO2 and SiO2

layers, respectively.

2.1.2. Field Emission Scanning Electron Microscopy

Cross-sectional field emission scanning electron microscopy

(FESEM) was performed on a Bragg reflector with a total of

six SiO2–TiO2 bilayers on silicon. A typical micrograph is

shown in Figure 2. As expected, the silica layers were found to

be amorphous at these annealing temperatures. In contrast, the

titania layers appeared to be composed of nanocrystals with a

size that varied between �10 and 20 nm in radius. This was in

good agreement with the 14 nm Scherrer radius calculated

from the X-ray diffraction (XRD) measurements (not shown).

The individual layers clearly do not possess well-defined

interfaces. This is due to the surface roughness of each titania

layer. Despite this we estimate that the silica and titania layers

possess an average thickness of (100� 10) and (70� 5) nm,

respectively. These values agree well with those obtained from

modeling of spectroscopic ellipsometry data.

2.1.3. Reflectometry

Reflectometry was utilized to monitor the evolution of

the reflection spectrum following the deposition of each

bilayer. Due to physical limitations, the reflection spectra

could only be obtained at 13 8 to the normal. In Figure 3A

400 500 600 700 8000

10

20

30

40

50

60

70

0

50

100

150

200

250

300

350

400

65

70

/

Wavelength / nm

2 = 117.1

Si

SiO2

TiO2

SiO2

TiO2

6x

A) B)

Figure 1. A) Experimental and fitted values ofC andD for a seven TiO2–SiO2 dreflector grown on silicon. B) The thickness parameters deduced from the a

Adv. Funct. Mater. 2008, 18, 3772–3779 � 2008 WILEY-VCH Verl

the reflection spectra of a Bragg reflector is shown with up

to seven TiO2–SiO2 bilayers grown on a silicon substrate. The

thickness of each layer was in this case optimized such that

maximum reflectance from the grating would occur at

l0¼ 600nm. The high reflectance of the bare silicon substrate

compared to normal glass (R� 4%) is due to the very high

refractive index of the silicon in the visible region (ns� 3.85).

Following the deposition of subsequent bilayers, the reflectance

around l0¼ 600 nm increases drastically due to the reflective

nature of the Bragg grating. A well-defined stop band

66 nm

105 nm

70 nm

109 nm

ouble-layer Braggpplied model fit.

ag GmbH & Co. KGaA

is observed with a spectral width of

�185 nm following the deposition. Due to

sample inhomogeneity and the large collec-

tion angle of the custom reflectometer, some

inherent broadening of the spectral features is

observed. Despite this, the reflectance follow-

ing the deposition of seven double-layers here

exceeds 99%.[13]

In Figure 3B we show a direct comparison

of a slightly different sample with only six

double-layers and the theoretically calculated

reflectance spectra determined through both

T-matrix calculations and through simula-

tions based on the spectroscopic ellipsometry

parameters. The optimized parameters for

, Weinheim www.afm-journal.de 3773

FULLPAPER

J. Jasieniak et al. / Sol–Gel Based Vertical Optical Microcavities

400 500 600 700 800 900 10000

20

40

60

80

100

Reflectance

/ %

Wavelength / nm

Experimental Matrix Method Simulation at 13.5

B)A)

Si, ns= 3.85

SiO2, nL = 1.45TiO2 , nH= 2.146 x 74 nm

107 nm

T-Matrix

Si

SiO2

TiO2

SiO2

TiO2

5 x 73 nm116 nm

67 nm103 nm

Ellipsometry BasedSimulation

C)

8007006005004000

20

40

60

80

100

Reflectance

/ %

Wavelength / nm

Silicon 1 Layer 2 Layer 4 Layer 6 Layer 7 Layer

Figure 3. A) The evolution of the reflectance spectra with increasing number of TiO2–SiO2 double-layers on a silicon substrate. B) A comparison betweenthe experimental reflectance spectra of a six TiO2–SiO2 double-layer stack on silicon, and simulations based on T-Matrix theory and also on the parametersdetermined from spectroscopic ellipsometry. C) A schematic of the optimum parameters utilized in both simulations.

3774

both simulations are given in Figure 3C. Ideal film thicknesses

were in this case 74 and 109 nm for the TiO2 and SiO2 layers,

respectively. A comparison of the extracted parameters from

both models shows that the calculated average thicknesses are

in excellent agreement with these ideal values. This highlights

the reproducible nature of the sol–gel method for the

development of complex, multilayer optical structures. It is

important to note that both calculations provide a good

description of the stopband region. This is consistent with

relatively even film thicknesses throughout the structure and

only small variations in refractive index of all materials.

Outside of the stopband however, particularly at smaller

wavelengths, the T-matrix calculations fail to adequately

describe the spectral behavior of the experimentally deter-

mined reflectance. This is in contrast to the predictions made

using ellipsometric based parameters. In this case, the more

realistic description of the physical dimensions of the layers in

the system and the inclusion of a frequency dependent

refractive index allows for a significantly better matching

between experiment and theory.

Figure 4. SIMS elemental depth profile of a Bragg microcavity composedof a seven double-layer SiO2–TiO2 reflector, a CdSe–CdS–ZnS/ZrO2 defectlayer, and a Ag mirror fabricated on a silicon substrate.

2.2. Characterization of the Asymmetric Bragg

Microcavities

Asymmetric Bragg microcavity geometries make use of a

metal mirror and Bragg reflector to confine light in a defect

layer. For sol–gel based Bragg cavity applications, the use

of a Ag mirror instead of a secondary Bragg reflector is

essential since high processing temperatures (>250 8C) lead to

quenching of the QD PL.[14,15] In order to use QDs in a sol–gel

Bragg cavity, it is therefore necessary to use a secondary mirror

that is deposited at low temperatures. A polymer-based Bragg

mirror could be used.[16] However, we prefer to deposit a

semitransparent sputteredAgmirror, similar to that utilized by

Finlayson et al. and Rabaste et al.[14,17]

www.afm-journal.de � 2008 WILEY-VCH Verlag GmbH

2.2.1. Secondary Ion Mass Spectrometry

We performed secondary ion mass spectrometry (SIMS) on

a Bragg cavity structure to determine the depth composition

profile. In Figure 4 we show the spatial elemental composition

of Ag, S, Ti, and Si for a typical sample as measured by SIMS.

In this case the S composition was representative of the depth

profile of the CdSe–CdS–ZnS quantum dots that were utilized

in this study. Very similar results were obtained for Zr, Cd, Zn,

and Se, however these were omitted for clarity.

& Co. KGaA, Weinheim Adv. Funct. Mater. 2008, 18, 3772–3779

FULLPAPER

J. Jasieniak et al. / Sol–Gel Based Vertical Optical Microcavities

The repeating structure of the seven bilayer SiO2–TiO2

Bragg reflector is immediately apparent from this measure-

ment. Notably, the SiO2 layer adjacent to the Si substrate

is undetectable here. The film thickness is determined by

analyzing the element signal dynamics. Through a peak fitting

procedure of the individual layers, the full width at half

maximum (FWHM) was taken as the film thickness. In

this manner the average thicknesses of the Ag, ZrO2/

CdSe–CdS–ZnS, TiO2, and SiO2 layers were calculated to

be (34� 7), (269� 24), (95� 16), and (82� 12) nm, respec-

tively. Relatively large errors are inherently associated with

these measurements due to beam-induced broadening effects

and surface roughness.[18] Such effects are particularly evident

for multilayer films such as those measured here. Despite these

factors and considering the difficulty of this technique, these

results are in reasonable agreement with those obtained from

both spectroscopic ellipsometry and FESEM.

From the elemental spatial profile one can gauge the

homogeneity of each layer in a given sample. It is immediately

evident that some interlayer diffusion of Ag into the ZrO2/

CdSe–ZnS layer occurs. This effect most likely arises due to the

relatively high porosity (�25%) that exists in these films.[15] It

is also evident that SiO2 diffuses into the TiO2 layers. Once

again one can explain this strong diffusion by considering the

extent of densification in each layer. Using the Bruggeman

Figure 5. A) Experimental reflectance data from a Bragg grating with seventhicknesses of a sputtered Ag mirror. The simulated reflectance data based ointerface (B) and one with a 20 nm exponential grading layer (C). D) A schemat

Adv. Funct. Mater. 2008, 18, 3772–3779 � 2008 WILEY-VCH Verl

effective medium model we have calculated the volume

fraction of voids in SiO2 and TiO2 to be close to zero and

�10%, respectively.[19] Thus, unlike SiO2, which is almost

completely densified at these temperatures, the highly

polycrystalline nature of the TiO2 layer prevents complete

densification and creates significant voids in each layer.

Consequently, diffusion of the SiO2 sol into the pores occurs,

which prevents a well-defined interface from forming. As we

will shortly see, interlayer diffusion, particularly of Ag, has a

drastic impact on the optical properties of the microcavity.

2.3. Ag Mirror Thickness Dependence on the Optical

Properties

In this section we show how variations in the thickness of a

Ag mirror in an asymmetric Bragg cavity influence its optical

response. An understanding of this effect is imperative in

order to determine the physical conditions necessary to

spectrally select the defect mode position.

In Figure 5A we show the reflectance from an asymmetric

Bragg cavity for a 0, 13, 37, and 57 nm Ag mirror grown on the

surface of a Bragg cavity with a deposited QD/ZrO2 defect

layer of thickness 160 nm. For a 0 nm mirror, the reflectance

appears reminiscent of the host Bragg grating with a defined

stopband present and no modal structure within it. Despite

TiO2–SiO2 double-layers, a 160 nm QD/ZrO2 defect layer, and varyingn spectroscopic ellipsometry parameters with a defined QD/ZrO2 and Agic depicting the configuration utilized in the simulation of the Bragg cavities.

ag GmbH & Co. KGaA, Weinheim www.afm-journal.de 3775

FULLPAPER

J. Jasieniak et al. / Sol–Gel Based Vertical Optical Microcavities

3776

this, the configuration still acts as a cavity with the uppermost

interface, QD/ZrO2–air, now being the secondary mirror. In

this case, however, the mirror reflectivity is very low (R� 7%

at normal incidence) and therefore Q is correspondingly also

very low.[1] Thus, littlemodification of the reflectance spectra is

evident. For gradually thicker Ag mirrors it is immediately

observed that a cavity mode forms and gradually develops from

the red side of the stopband. This behavior is unlike that of a

symmetric Bragg cavity configuration where the defect mode

develops at a specific wavelength regardless of the number of

high/low refractive index bilayers. Following the deposition of

a 37 nm thick Ag layer the modal position largely ceases to

blue-shift any further. At this point the quality of the cavity

begins to improve as is evident from an increase in Q� 29 at

37 nm to Q� 50 at 57 nm, while the reflectance of the whole

structure increases significantly, ultimately approaching that of

pure Ag.

In Figure 5B and C we show the results from simulations

based on two different models that only differ in their

description of the QD/ZrO2–Ag interface. For the data

presented in Figure 5B, we have utilized a strict interface

between the ellipsometrically determined 160 nm thick QD/

ZrO2 defect layer and the Ag mirror. Conversely, in Figure 5C

this interface was relaxed with the inclusion of an intermediate

grading layer that varied the Ag composition exponentially

across a 20 nm spatial region. The exact parameters in each

model are given in Figure 5D. The strict interface model is

found to be in reasonable agreement with experimental data.

However, it clearly predicts much higher modalQ-factors than

are observed. In stark contrast, the inclusion of a grading layer

results in excellent agreement between experiment and

simulation. This result supports the conclusions from the

SIMS data, which showed that Ag diffuses into the porous

defect layer.

A comparison of both models suggests that the Ag diffusion

into the defect layer limits the attainableQ-factor of the cavity.

As it is known thatQ/ (ln(R1R2))�1 (where R1 and R2 are the

reflectivities of the cavity mirrors), any change in reflectivity of

either mirror will directly translate into a concordant change in

Q (See Supporting Information for details).[1]

A simple calculation of the reflectivities for Ag

films on strict and graded interfaces reveals

that grading causes a 5 to 10% decrease in the

overall values. This therefore allows us to

qualitatively understand that Q is naturally

decreased as a result of a reduced mirror

reflectivity, which is inherently caused by Ag

diffusion.

Figure 6. A) Angle dependence of the reflectance from an asymmetric Braggmicrocavity grownon a quartz substrate under TM-polarization. Inset: A schematic of the microcavity structureutilized. B) A comparison between the experimentally determined spectral location of the cavitymode and the simulated cavity mode location based on spectroscopic ellipsometry data.

2.4. Angle Dependence of the Microcavity

Mode

In the previous section reflectometry was

utilized at a fixed incidence angle of �13 8. Byvarying the Ag mirror thickness the reflected

www.afm-journal.de � 2008 WILEY-VCH Verlag GmbH

phase was found to change accordingly, in turn resulting in a

thickness-dependent cavity mode position. In this section we

discuss the angle dependence of an asymmetric Bragg cavity

grown on a quartz substrate with a thick Ag mirror (100 nm).

This type of geometry allows optical access to the cavity only

through the glass substrate. The thicker Agmirror utilized here

serves to increase the reflectivity within the cavity and yields

higher Q-factors. In Figure 6A the angle dependence of the

reflectance under transverse magnetic (TM)-polarization of

such an asymmetric Bragg microcavity is shown. Varying

the angle of incidence from 20 8 to 70 8caused the mode to

experience a large blue-shift from 638 nm to 557 nm. The angle

dependence of the cavity mode was simulated using the

DeltaPsi2 software package for TM-polarization. This

software inherently accounted for the spectrally dependent

phase change from the metal and Bragg grating. In Figure 6B

the simulated results are shown. An excellent agreement

for the angular dependence of the cavity mode position

clearly exists. Notably, analogous results were attained with

the matrix method. These results highlight the flexibility of

microcavities in fine tuning the spectral position of the cavity

modes.

2.5. PL Active Bragg Microcavities

To study the PL emission from QDs incorporated in

the defect layer of the asymmetric Bragg microcavities, the

instrumental configuration depicted within Figure 7A was

utilized. Here the asymmetric Bragg cavity was grown on

quartz with a 100 nmAgmirror. In this geometry the cavity was

excited from the substrate direction. In Figure 7A the PL

detected at 40 8 from the cavity and from a reference sample

(no Bragg grating between the defect layer and the substrate) is

shown. It is evident that a large modification of the PL profile

has occurred for QD emission with and without the cavity.

Included in this figure is the microcavity reflectance spectrum

at 40 8. The resonance mode is seen to agree well with the PL

emission maximum detected from the cavity. Additionally, the

& Co. KGaA, Weinheim Adv. Funct. Mater. 2008, 18, 3772–3779

FULLPAPER

J. Jasieniak et al. / Sol–Gel Based Vertical Optical Microcavities

Figure 7. A) PL detected from within the cavity and a neat QD film at a detection angle of 40 8with respect to the normal. The excitation source was the 514.5 nm line of a HeNe line focused ata 35 8 incidence angle. A schematic of this experimental setup is shown. The reflectance at 40 8incidence is also included to highlight the coupling between the QD emitters and the cavity.B) The angle dependent PL signal detected from the Braggmicrocavity for a fixed excitation. Inset:The angle dependent Q-factor of the microcavity.

Q-factors determined from both emission and reflection were

similar, with values of 40 and 46, respectively. Both of these

factors support the assertion that the cavity emission has been

modified due to a coupling between the cavity and the QDs. It

is worthwhile noting that uncoupled emission is also observed

in the PL signal. This occurs due to imperfections in the cavity

that allow emission to occur without modification.

To further demonstrate the effects of QD–cavity coupling,

the PL detection angle was varied while the excitation angle

remained fixed. The results, which are included in Figure 7B,

show that the PL maximum strongly blue-shifts from a

Figure 8. A) PL detected at 10 8 from an asymmetric Bragg microcavity and the reference sampleshown in (D). Included is the reflectance spectrum of the microcavity at 10 8. B) Normalized PLspectra of emission from the microcavity, reference film, and a neat QD/ZrO2 film deposited onsilicon. C) A comparison between the experimentally determined and simulated ratio of cavityemission (Icav) and reference emission (Iref). D) Schematic of the microcavity, reference sample,and neat film utilized for this set of experiments.

maximum at 644 nm at 0 8 detection to

585 nm at 50 8. This trend is in accordance

with the angular dependence of the reso-

nant mode. Interestingly, at higher detec-

tion angles, PL is observed at wavelengths

even shorter than the PL Peak from the

reference QD films (Fig. 7A). This is

probably due to the inhomogeneous broad-

ening of the sample, possibly arising from

core–shell nanocrystals that are much

smaller than the mean population or core

nanocrystals that were not adequately

shelled. Notably, with a blue-shift of the

PL a concordant decrease in theQ-factor is

observed (Fig. 7B, inset). Such a trend has

been previously found by Rabaste et al.

who studied similar asymmetric Bragg

microcavities.[14] They showed that this

trend arose from the increased absorption

of the QDs at higher energies, which is

known to have an adverse effect on the Q-

factor (see Supporting Information for

details). In Figure 7B the normalized

absorption spectrum for the QDs is shown.

Generally, it was found that increased QD

absorption leads to a decrease in the Q

factor.

Adv. Funct. Mater. 2008, 18, 3772–3779 � 2008 WILEY-VCH Verlag GmbH & Co. KGaA,

2.6. PL Cavity Enhancement in Bragg

Microcavities

In the previous section it was shown that

the QD–cavity coupling causes a modifica-

tion of the PL profile. This modification,

which arises due to the Purcell effect, can be

quantified by comparing the relative emis-

sion intensity from a microcavity and an

appropriate reference sample. By utilizing an

asymmetric microcavity grown on Si, the

extent of this enhancement can be deter-

mined. A similar experimental setup was

utilized to that described in the previous

section, a minor difference being that the

laser was incident on the Ag mirror face. To

avoid any anomalies caused by differences in

reflection of the incident light from the

microcavity and a reference sample, the reference was

constructed using the identical geometry as the microcavity,

but without the Bragg reflector.

In Figure 8A, the PL signals emitted from an asymmetric

Bragg microcavity and the reference sample are shown at

a detection angle of 10 8. It is evident that a significant

enhancement of the PL due to the microcavity occurs. In

principle, the reference sample here also acts as a microcavity;

however, due to the low reflectivity from the silicon substrate

the expected enhancement is minimal. In contrast, the high

reflectivity of the Ag mirror and the Bragg grating in the

Weinheim www.afm-journal.de 3777

FULLPAPER

J. Jasieniak et al. / Sol–Gel Based Vertical Optical Microcavities

3778

asymmetric microcavity cause the observed PL enhancement.

The calculated enhancement here at the spectral location of

the cavity emission PL maximum (636 nm) is �16 as compared

to the reference sample. However, the integrated enhancement

of the PL bands is only 6.2. To understand why a smaller

integrated enhancement was observed, the PL spectra of both

samples were normalized and compared to a neat QD film with

no Ag mirror or Bragg grating. Shown in Figure 8B, it is

apparent from this normalization that a modification of the

cavity emission was experienced only in the PL profile of the

asymmetric microcavity. The calculated Q-factor in this case

was 49 (FWHM¼ 12.9 nm), while that of both the neat and

reference films were 24 (FWHM¼ 26.2 nm). Thus one can

deduce that the reduced integrated enhancement has occurred

due to a spectral narrowing of the PL with respect to the

reference sample. This effect is not accounted for when

determining a single peak enhancement value.

To calculate the theoretical enhancement for such a

structure we utilized the formalism of Bendickson et al.[20]

Following an initial determination of the local density of

optical states in each sample and consequently accounting for

the differences in reflectance between the two mirrors, the

ratio of the relative PL signal between the cavity (Icav) and the

reference (Iref) emission was determined (see Supporting

Information for details). In Figure 8C the calculated cavity to

reference emission ratio and the experimentally determined

values are shown. Good agreement between experimental and

theoretical results of the simple theory is found. This confirms

that the observed cavity enhancement results from a

modification of the density of optical states.

Conceptually, the intensity enhancement occurs because the

emission rate into any resonant cavity mode is enhanced due to

the Purcell effect.[21] As this rate is greatly decreased for non-

resonant wavelengths, a large modification in the PL signal is

observed for cavity emission. While for a cavity in a confocal

geometry (i.e., with concave mirrors) a nearly single mode can

be made selectively resonant with the emission of a PL active

species, in coplanar geometries such as the Bragg gratings that

we are treating, the emission from the resonant mode is

strongly angle dependent.[22]

3. Conclusions

Sol–gel chemistry has been applied to the growth of

multilayer Bragg gratings. By utilizing a QD/ZrO2 defect

layer and a sputtered Ag mirror, these Bragg gratings could be

further adapted into planar microcavities. The characterization

of such asymmetric Bragg microcavity structures showed that

Q-factors as high as 80 could be achieved when the resonance

mode was tuned between 600 and 650 nm at near normal

incidence. The simultaneous confinement of light and the

incorporation of PL active species within these microcavity

structures allowed cavity–QD coupling to be realized. Clear

modifications to the PL emission from QDs within the defect

layer were observed. These angle-dependent modifications

www.afm-journal.de � 2008 WILEY-VCH Verlag GmbH

enabled the resonant cavity mode and the cavity-emitted PL

maximum to be tuned.

The results presented here show that the use ofmicrocavities

to modify the spontaneous emission properties of QD emitters

is highly advantageous. Such structures are already utilized in

current light-emitting diodes (LEDs) and vertical light-

emitting laser technology, while their use in controlling the

optical properties of QDs has been limited to date. Improved

cavity design, smaller cavity volumes and higher mirror

reflectivities may allow low-threshold QD-based lasers and

single photon emitters to be realized using sol–gel fabrication

techniques.

4. Experimental

Chemicals: 5-amino-1-pentanol (AP, 95%), zirconium (IV) iso-propoxide (70 wt% in 1-PrOH), titanium (IV) isopropoxide (97%),tetraethyl orthosilicate (TEOS, 99%), glacial acetic acid (AAc,99.8%), and acetylacetone (ACAC, 99%) were purchased fromSigma–Aldrich. Hexane, chloroform, propanol, ethanol, and methanolwere all of analytical grade and purchased from Univar. All chemicalsand solvents were used without further purification.

Synthesis of CdSe–CdS–ZnS Quantum Dots: CdSe nanocrystal(NC) preparations were carried out by established methods reportedby van Embden et al. [23]. Shelling of the NCs with CdS–ZnS wasfurther performed using the recently developed successive ionic layeradsorption and reaction (SILAR) method [24,25]. Metal rich surfaceswere created to ensure high PL and high photostability [26].

Preparation of Substrates: Silicon or fused silica substrates wereprepared according to the method of Siqueira Petri et al. Briefly,the substrates were initially ultrasonicated in dichloromethane for15min [27]. They were then immersed into a solution of NH3 (25% byvolume), H2O2 (30% by volume), and distilled water at a volume of1:1:5 at a temperature of 70 8C for 20min. As a final step the substrateswere washed with distilled water and dried under a stream of N2.Cleaned substrates could be stored in distilled water for several weeks,however freshly cleaned substrates always resulted in more repro-ducible results.

Preparation of Sol–Gel Bragg Cavities: The sol–gel Bragg cavitywasmade in two stages. In the first stage a Bragg reflector was grown ona desired substrate with alternating SiO2 and TiO2 layers. Subse-quently, in the second stage a defect layer and a semi-transparentmirror were introduced.

SiO2 Sol-Solution Preparation: Silica thin films were prepared via asimilar method to that described by Fardad et al. [28]. Briefly,tetraethylorthosilicate (TEOS, 5mL, 0.110mol) and EtOH (5mL)were mixed in a round bottom flask for 10min. A solution of H2O(1.934mL) and HCl (81mL, 1M) was gradually added drop-wise over20min. The final solution was heated to 70 8C over 15min in air, whereit remained for a further 2 h. Upon cooling to room temperature thesolution was aged for 24 h before use. Prior to deposition, a givenquantity of the sol was diluted with EtOH where it was stirred for10min.

TiO2 Sol-Solution Preparation: The titania thin films were preparedvia a method described by Bahtat et al. [29]. This initially involvedrapidly adding glacial acetic acid (2.65mL, 46mol) to titanium tetra-i-propoxide (Ti(i-PrOH)4, 97%, 2.35mL, 7.7 mmol) under vigorousstirring. After 30min i-PrOH (0.8mL) was added, at which point thesolution was stirred for a further 10min. This sol solution was usedwithout further dilution.

Thin Film Growth: Thin film deposition of both silica and titaniawas performed by spin-coating. The intended sol solution was filtered

& Co. KGaA, Weinheim Adv. Funct. Mater. 2008, 18, 3772–3779

FULLPAPER

J. Jasieniak et al. / Sol–Gel Based Vertical Optical Microcavities

through a 0.2mm syringe filter directly onto an appropriately treatedsilicon or fused silica substrate. Films were spun for a total of 30 s.The exact deposition speed for the silica and titania films was pre-determined for a given sol-solution through a calibration. For thethicknesses required, generally for silica this was �4000 rpm, while fortitania it varied between �4000 and 5000 rpm. Deposition of all filmswas performed under a dry nitrogen environment (humidity< 30%).Following the deposition of each film, annealing took place in twostages. Firstly, the films were baked on a hot-plate at 200 8C for 2min toremove any volatile solvent molecules and/or organics. Secondly, thefilms were placed in a furnace at 1000 8C for 1min to densify thematrix.These films cooled by convection. The defect layer in the Bragg cavitywas composed of a ZrO2/CdSe–CdS–ZnS thin film that was preparedanalogously to that previously described [15]. Heat treatment of the filmwas performed at 250 8C to ensure a balance between densification andPL retention. The appropriate film thickness was chosen throughsimulation of the Bragg cavity with known parameters of the Braggreflector. The volume fraction of QDs remaining in the films was�0.1%. A Ag mirror (ProSciTech) of a defined thickness was sputtercoated directly on the sample. This was performed in an Emitechsputter coater model K575A. Ellipsometry measurements were madeon a HORIBA Jobin Yvon Uvisel spectroscopic ellipsometer withmodeling performed on the integrated software package DeltaPsi2.

Received: June 9, 2008Revised: September 2, 2008

Published online: November 13, 2008

[1] A. Yariv, Quantum Electronics, Wiley, New York 1989.

[2] F. L. Pedrotti, L. S. Pedrotti, Introduction to Optics, 2nd ed., Prentice-

Hall, Englewood Cliffs, NJ 1993.

[3] Y. Yamamoto, R. E. Slusher, Phys. Today 1993, 46, 66.

[4] N. Le Thomas, U. Woggon, O. Schops, M. V. Artemyev, M. Kazes, U.

Banin, Nano Lett. 2006, 6, 557.

[5] A. M. Vredenberg, N. E. J. Hunt, E. F. Schubert, D. C. Jacobson, J. M.

Poate, G. J. Zydzik, Phys. Rev. Lett. 1993, 71, 517.

[6] C. B. Poitras, M. Lipson, H. Du,M.A.Hahn, T. D. Krauss,Appl. Phys.

Lett. 2003, 82, 4032.

Adv. Funct. Mater. 2008, 18, 3772–3779 � 2008 WILEY-VCH Verl

[7] H.-J. Eisler, V. C. Sundar, M. G. Bawendi, M. Walsh, H. I. Smith, V.

Klimov, Appl. Phys. Lett. 2002, 80, 4614.

[8] K. J. Vahala, Nature 2003, 424, 839.

[9] E. Yablonovitch, Phys. Rev. Lett. 1987, 58, 2059.

[10] E. F. Schubert, Light-Emitting Diodes, 2nd ed., Cambridge University

Press, Cambridge 2006.

[11] P. Lodahl, A. F. van Driel, I. S. Nikolaev, A. Irman, K. Overgaag, D.

Vanmaekelbergh, W. L. Vos, Nature 2004, 430, 654.

[12] N. Ganesh, W. Zhang, P. C. Methias, E. Chow, J. A. N. T. Soares, V.

Malyarchuk, A. D. Smith, B. T. Cunningham,Nat. Nanotechnol. 2007,

2, 515.

[13] This value is with respect to bulk aluminium, which was utilized as a

reference sample. Aluminium within the spectral region of the stop-

band has an absolute reflectance of 97%.

[14] S. Rabaste, J. Bellessa, C. Bonnand, J. C. Plenet, L. Spanhel, Eur.

Phys. J. B 2004, 42, 47.

[15] J. Jasieniak, J. Pacifico, R. Signorini, A. Chiasera, M. Ferrari,

A. Martucci, P. Mulvaney, Adv. Funct. Mater. 2007, 17, 1654.

[16] J. Yoon, W. Lee, J. M. Caruge, M. Bawendi, E. L. Thomas, S. Kooi, P.

N. Prasad, Appl. Phys. Lett. 2006, 88, 091102.

[17] C. E. Finlayson, D. S. Ginger, N. C. Greenham,Appl. Phys. Lett. 2000,

77, 2500.

[18] K. J. Kim, D. W. Moon, Appl. Phys. Lett. 1992, 60, 1178.

[19] D. A. G. Bruggeman, Ann. Phys. 1935, 24, 636.

[20] J.M. Bendickson, J. P.Dowling,M. Scalora,Phys. Rev. E 1996, 53, 4107.

[21] E. M. Purcell, Phys. Rev. 1946, 69, 681.

[22] I. Abram, J. L. Oudar, Phys. Rev. A 1995, 51, 4116.

[23] J. van Embden, P. Mulvaney, Langmuir 2005, 21, 10226.

[24] J. J. Li, Y. A. Wang, W. Guo, J. C. Keay, T. D. Mishima, M. B.

Johnson, X. Peng, J. Am. Chem. Soc. 2003, 125, 12567.

[25] J. van Embden, J. Jasieniak, D. E. Gomez, P. Mulvaney, M. Giersig,

Aust. J. Chem. 2007, 60, 457.

[26] J. Jasieniak, P. Mulvaney, J. Am. Chem. Soc. 2007, 129, 2841.

[27] D. F. Siqueira Petri, G. Wenz, P. Schunk, T. Schimmel, Langmuir

1999, 15, 4520.

[28] M. A. Fardad, E. M. Yeatman, E. J. C. Dawnay, M. Green, F.

Horowitz, J. Non-Cryst. Solids 1995, 183, 260.

[29] A. Bahtat, M. Bouazaoui, M. Bahtat, C. Garapon, B. Jacquier, J.

Mugnier, J. Non-Cryst. Solids 1996, 202, 16.

ag GmbH & Co. KGaA, Weinheim www.afm-journal.de 3779