Per-Anders Hansen*,
Joachim Svendsen,
Hanne Nesteng and
Ola Nilsen
Department of Chemistry, Centre for Materials Science and Nanotechnology, University of Oslo, Sem Sælandsvei 26, 0371 Oslo, Norway. E-mail: p.a.hansen@kjemi.uio.no
First published on 20th June 2022
Atomic layer deposition offers a unique set of design possibilities due to the vast range of metal and organic precursors that can be used and combined. In this work, we have combined lanthanides with aromatic aids as strongly absorbing sensitizers to form highly luminescent thin films. Terephthalic acid is used as a base sensitizer, absorbing shorter wavelengths than 300 nm. The absorption range is extended towards the near-UV and blue range by increasing the aromatic system and adding functional groups that have strong red-shifting effects. While terbium and europium provide green and red emission, yttrium allows emission from the sensitizer itself spanning the whole color range from purple, blue and green to red. Many organic dye molecules show very high luminescence quantum yields and several of the molecules and materials investigated in this work show bright luminescence.
The inclusion of organic species into hybrid or nanocomposite thin film is an exciting field as the organic component adds very strong optical absorption and an exceptional flexibility in designing the molecules towards specific absorption and emission ranges. Unfortunately, the choice of organic components in MLD is quite limited. The need to get these molecules into the vapour phase while also having sufficient reactivity reduces the practical range to fairly small molecules. This puts strong limits on the range of optical and electronic properties to choose from.
Reviews on ALD and MLD chemistries7,10 reveals that the organic precursors utilized so far are quite limited beyond a single aromatic ring or a short conjugated chain. There are examples of larger conjugated systems like naphthalene and biphenyl6 showing near-UV activity and even visible absorption.11 These optically absorbing dye precursors require sublimation temperatures of around 250 °C or more, approaching the temperature where such organic molecules start to decompose or polymerize. This illustrates well the difficulty in balancing Vis and NIR active optical properties with ALD/MLD compatibility. An exception to this is our recent work on quinizarin molecules, which sublimes efficiently at 130 °C and produces deep pink films with TMA as cation precursor.12
In this work, we explore a range of mono- and bi-aromatic di-carboxylic acids deposited with lanthanides as cations, as shown in Fig. 1. All three lanthanides have very similar deposition chemistries.13 Terbium and europium was chosen due to their well-known and easily identified sharp emission peaks which are easily distinguishable from organic molecule emissions. In addition, these two lanthanides have large energy gaps from the emitting energy level(s) to the next lower energy level. This allows them to be highly luminescent even in high-phonon matrices such has hybrid materials. Yttrium was chosen as an optically inactive ion that allows characterization of the organic molecules emission properties in the same lanthanide-organic hybrid material environment. In previous works, we have investigated the growth and material chemistry of optically active hybrid materials.5,6,12,14 Here, we focus on the hybrid materials optical properties, absorption and luminescence colour.
Compound name | Short name | Tsub (°C) | |
---|---|---|---|
Ln(2,2,6,6-tetramethyl-3,5-heptanedione)3 | Y(thd)3 | 130 | |
Eu(thd)3 | 150 | ||
Tb(thd)3 | 145 | ||
1,4-Benzenedicarboxylic acid | H2bdc | 200 | |
(1) | 2,3,5,6-Tetrafluoro-1,4-benzenedicarboxylic acid | H2bdc-F | 145 |
2,3,5,6-Tetrabromo-1,4-benzenedicarboxylic acid | H2bdc-Br | 230 | |
(2) | 2-Amino-1,4-benzenedicarboxylic acid | H2bdc-NH2 | 210 |
2,5-Dihydroxy-1,4-benzenedicarboxylic acid | H2bdc-2OH | 225 | |
(3) | Biphenyl-4,4′-dicarboxylic acid | H2bpdc | 250 |
2,6-Naphthalenedicarboxylic acid | H2ndc | 225 | |
(4) | Benzenediol | H2bdo | 130 |
Benzenedithiol | H2bdt | 70 |
In situ quartz crystal microbalance (QCM) analyses were conducted using a 6 MHz AT-cut quartz crystal. The crystal was mounted in a home-made holder and was used to monitor the mass increase, proportional to the change in frequency during the deposition to determine saturation conditions for pulse and purge parameters. The signal was recorded using a Colnatec Eon-LT and processed by averaging over 16 consecutive ALD cycles. The temperature was stabilized for 60 to 90 minutes before any experiments were conducted to ensure a stable temperature and response from the QCM-crystals. Note that the QCM investigations were conducted over an extended timeframe where modifications to the setup have occurred. Thus, the scale of the frequency changes is not necessarily comparable between different materials in this work.
Film thickness and refractive index n(λ) were determined with a J. A. Woollam alpha-SE spectroscopic ellipsometer in the 380–900 nm range. This ellipsometry data was modelled using a Cauchy model. A VASE spectroscopic ellipsometer with range of 280–1000 nm, also from J. A. Woollam, was used to model n(λ) and the extinction coefficient k(λ). These films were modelled with a general oscillator model consisting of several Gaussian peaks. Silicon samples were used for ellipsometry. In both instruments, the sample is modelled by a single layer on top of a thin native silicon oxide layer. This native oxide was measured before each deposition and was generally in the 2–4 nm range. Interface and surface roughnesses were not included in the model as these did not improve the fit. The relation between absorption coefficient α(λ) and k(λ) is given in eqn (1). For readers more accustomed to α(λ), the practical difference between these is a constant scaling factor (2π) a simple wavelength dependent factor (1/λ). FTIR characterization was done using a Bruker VERTEX 70 FTIR spectrometer in transmission mode on single-side polished silicon substrates. Luminescence (PL) and excitation (PLE) measurements was done using an Edinburgh Instruments FLS920 fluorescence spectrometer with a 450 W Xe lamp as excitation source and a Hamamatsu R928 PMT for detection. PL decay measurements utilized an optical parametric oscillator (OPO) system (Opotek HE 355 II) pumped by the third harmonic of a Nd:YAG laser as excitation source. The OPO system was set to 355 nm and a repetition rate of 20 Hz. The detector for decay measurements was the same equipment used for the PLE.
(1) |
The QCM response during a cycle with long pulse and purge times is shown in Fig. 3. The two ideal half cycle reactions are given below. For comparison, the masses of the precursors and leaving groups are: Tb(thd)3 = 709, H2bdc = 166, Hthd = 184 g mol−1. Both the lanthanide and acid pulses show initial rapid frequency change of around 2 seconds followed by a slower evolution. The mass decrease during the acid pulse is due to H2bdc being slightly lighter than the leaving molecule Hthd. The ratio between the ideal reactions are Δm1/Δm2 = 16, while the measured ΔmTb/Δmbdc = 3.5. This indicate that either less lanthanide or more acid is deposited during each half cycle than expected from the ideal reactions.
(2) |
(3) |
Fig. 3 QCM response during a Tb2bdc3 cycle with long pulse and purge times. Tb(thd)3/purge/1,4-bdc/purge = 7/5/20/15. |
The growth rate as a function of deposition temperature is shown in Fig. 4. The growth decreases linearly with temperature up until 300 °C, and becomes almost stable up to 350 °C. The films above 300 °C had a brown-ish colour and was non-luminescent indicating thermal decomposition of either H2bdc or Hthd, or both. The luminescence intensity of the films deposited in the 225–275 °C range did not vary significantly. Thus it was decided to keep the deposition temperature at 250 °C for all films, unless otherwise stated.
Fig. 4 Growth rate as a function of deposition temperature. Measurements from 3 different locations over a 5 cm distance indicates the evenness at each temperature. |
Some hybrid materials are not long-term stable in air, either through absorption and release of for example moisture or oxidation reactions. Tb2bdc3 films were characterized by ellipsometry immediately after deposition and 4 additional times up to 3 days to monitor any changes to the films thickness. The results are shown in Fig. 5. The films show a swelling of about 4% after 3 days with no other apparent changes to the films structure or properties. In fact, films deposited over 4 years ago show identical optical and luminescent properties as freshly deposited films.
Fig. 6 shows FTIR spectra from a Tb2bdc3 film taken during the day of deposition and 16 days later. The spectra show that there are no significant differences in number, position or intensities of peaks, indicating no change in bonding during this time. The films show no sign of protonated carboxylic acid groups, which should be clearly visible around 3000 cm−1. FTIR is useful to investigate the type of bonding between the metal and acid group based on the splitting (Δacid) between the asymmetric and symmetric carboxylate stretching peaks.15 A Δacid in the 50–150 cm−1 range indicate bidentate binding, 130–200 cm−1 indicate bridging and >200 cm−1 indicate unidentate binding. The asymmetric carboxylic stretching peak is also split into a double peak, giving two values for Δacid, 151 and 117 cm−1. This indicates that the Tb2bdc3 films contains a mixture of bidentate and bridging binding. An enlarged version zoomed in on these peaks is shown in ESI Fig. 1,† where also the exact wavenumber position of these three peaks are numbered.
Fig. 6 FTIR spectra of a Tb2bdc3 film taken on the day of deposition and after 16 days. The splitting between the asymmetric and symmetric peaks are shown. |
The optical absorption and luminescence properties of Tb2bdc3 is shown in Fig. 7, alongside a photograph of a sample under a 280 nm diode. Transmittance data is obtained with a 28 nm film on silica substrate, where the multiplet absorption from the benzene ring of terephthalic acid is clearly seen at around 250 and 300 nm. The PLE spectrum is obtained by monitoring the 5D4 → 7F5 emission. The two first aromatic absorption peaks are clearly seen in the PLE spectrum. The spectrum is not corrected for the decrease in the Xe lamps light intensity towards deep UV, which causes the apparent red-shift of the high energy peak.
Fig. 7 Transmittance, PL and PLE spectra of Tb2bdc3. The inset shows a glass plate and silicon substrates coated with Tb2bdc3 under 280 nm illumination from a diode. |
Fig. 7 forms the basis for exploring the other organic sensitizers and lanthanides with the aim to obtain absorption towards the near-UV and blue range and different emission colours. Each of the 4 groups of molecules illustrated in Fig. 1 is compared to Tb2bdc3 individually below, while the deposition parameters and obtained growth rates are summarized in Table 2.
Organic | Tdep (°C) | g (nm/cycle) | Parameters (s) | |
---|---|---|---|---|
a All films showed strong gradients. | ||||
bdc | 250 | 0.23 | 1.5/1.5/2/1 | |
(1) | bdo | 175 | 0.12 | 3/1/3/1 |
bdt | 250 | 0 | 1.5/1.5/2/1 | |
(2) | bdc-F | 250 | 0.24 | 1.5/1.5/2/1 |
bdc-Br | 250 | 0.27 | 1.5/1.5/2/1 | |
(3) | bdc-NH2 | 250 | 0.27 | 1.5/1.5/2/1 |
bdc-2OH | 250 | ∼1a | 1.5/1.5/2/2 | |
(4) | bpdc | 275 | 0.26 | 1.5/1/2/1 |
ndc | 250 | 0.27 | 1.5/1.5/2/2 |
Films deposited with Tb(thd)3 and H2bdo showed a growth rate of 0.13 nm per cycle, which is about half of the growth rate of Tb2bdc3. The films were visually transparent like Tb2bdc3, however these films did not luminesce under a 254 nm UV lamp. As enhancing the optical and luminescent properties of lanthanide-hybrid materials is the aim of this work, the H2bdo precursor was not further explored apart from noting that H2bdo and Tb(thd)3 do react and form ALD films. Thus, hydroxy groups can be suitable reaction groups to bind to Ln(thd)3. In fact, H2bdo sublimes at 70 °C lower than H2bdc, which can be a desired property.
Tb(thd)3 and H2bdt did not result in film deposition. Film growth was attempted at 200, 250 and 300 °C. This was somewhat surprising as H2S can be used with Ln(thd)3 to produce lanthanide sulfide films.17
Films produced with these three precursors and Tb(thd)3 all produced luminescent with the same green colour and with similar intensity under a 254 nm UV lamp. To further investigate differences in optical properties, the films optical absorption and luminescence decays were characterized. k(λ) and n(λ) of these films is shown in Fig. 8. Compared to Tb2bdc3, fluorination causes a red-shift in absorption and a lower n(λ) while bromination causes a blue-shift with unchanged n(λ). Fig. 9 show decays of the 5D4 → 7F5 emissions of the three materials. All three can be fitted with a single exponential model. The brominated sample show a lower lifetime than Tb2bdc3 and Tb2bdc-F3. This could indicate higher quenching rates in Tb2bdc-Br3. In summary, bromination offers neither improved growth (high sublimation temperature), more near-UV absorption (weaker and blue-shifted) nor improved luminescence yield (lower Tb3+ lifetime). On the other hand, the main advantage of fluorination is the substantially lower sublimation temperature and a small red-shift of the absorption while the growth and material properties are otherwise very similar to the un-halogenated acid.
Replacing Tb3+ with Eu3+ and Y3+ results in Ln2bdc3 films with very different emissions, shown in Fig. 10. The PLE spectra are very similar to each other, although the lighter Y3+ shows a small red-shift compared to the heavier Tb3+ and Eu3+. Both Tb2bdc3 and Eu2bdc3 produce their regular Tb3+ and Eu3+ emission spectra, showing that bdc is an efficient sensitizer for these two lanthanides. Y3+ doesn't have a partially filled d or f shell, allowing the sensitizer molecule to emit its own photon. Y2bdc3 emits in the near-UV, partly extending into the blue range. This gives the material a violet emission colour, relatively weak compared to Tb2bdc3 and Eu2bdc3.
Fig. 10 Normalized PLE and PL spectra of Ln2bdc3, Ln = Tb, Eu and Y. The emission lines of Eu3+ are named. |
These films show large red-shifts compared to Tb2bdc3, covering the near-UV and even blue range for Tb2(bdc-OH)3 (Fig. 11). However, all depositions with Tb, Eu and Y with these two acids produced non-luminescent samples. The fact that the Y depositions were quenched indicate that the sensitizer acids are quenched due to how they are packed in the structure. Both acids produce strongly luminescent solutions when dissolved. Crystalline MOF structures with H2bdc-NH2 and Tb and Eu has been shown to produce sensitized Ln3+ emission.18 A study by Vogel et al. on MOF structures of Y and H2bdc-OH show that the exact bonding scheme between Ln3+ and the sensitizer acid will have major impacts on the materials optical and luminescent properties.19 Thus, although the films produced in our work are non-luminescent, it is likely that control of the resulting structure and molecule packing can produce luminescent films. We have previously shown that control of crystallinity can be achieved both through in situ modulation by pulsing other species to modify growth mechanics and ex-situ treatments.20,21 However, exploration of how crystallinity and control of growth mechanics can enable luminescent variants is beyond the scope of the current work.
The QCM response per cycle Eu2bpdc3 and Tb2ndc3 is shown in Fig. 12. Tb2ndc3 show clear saturation after 1 and 1.5 s Tb(thd)3 and H2ndc pulses, respectively. Eu2bpdc3 show strong indication of the same saturation values. However, Eu2bpdc3 also show a steady increase with increasing Eu(thd)3 pulse times longer than 1 s, and some growth at 0 s acid pulses. This is likely due to the higher precursor temperature used for H2bpdc. We based the sublimation temperatures on prior sublimations temperatures in ALD-like vacuum conditions. Likely, 250 °C is slightly too warm for this precursor, leading to some precursor vapours seeping into the reaction chamber. This CVD component is quite small for the 2 and 3 s Ln(thd)3 and H2bpdc pulses used for samples, which was also seen in negligible growth gradients.
Fig. 12 QCM characterization of growth saturation as a function pulse durations for Eu2bpdc3 and Tb2ndc3. |
The QCM response during a single cycle of Eu2bpdc3 and Tb2ndc3 using long pulse and purge times is shown in Fig. 13. As for Tb2bdc3 in Fig. 3, both Eu2bpdc3 and Tb2ndc3 show an initial rapid growth for both Ln(thd)3 and acid pulse, indicating surface saturation. Contrary to Tb2bdc3 which experience a mass decrease during the acid pulse, these two depositions experience a mass gain. This is due to the large H2bpdc and H2ndc being heavier than Hthd while H2bdc is lighter, 242 and 216 g mol−1 respectively.
Fig. 13 QCM response during a Eu2bpdc3 and Tb2ndc3 cycle with long pulse and purge times. Ln(thd)3/purge/acid/purge = 30/15/30/15. |
The perhaps most interesting aspects of Ln2bpdc3 and Ln2ndc3 films are their optical properties. These films have absorptions covering most of near UV while being almost an order of magnitude stronger than Ln2bdc3, Ln2(bdc-NH2)3 and Ln2(bdc-OH)3. Unlike Ln2(bdc-NH2)3 and Ln2(bdc-OH)3, these films also result in highly luminescent films. Fig. 14 summarizes k(λ), n(λ), PLE and optical transmittance of selected films, in addition to the PL spectra originating from the aromatic acids.
Y2bpdc3 and Y2ndc3 show strong violet and blue emissions. Their emission spectra are shown in Fig. 14(b). Both consist of broad emission peaks which are normal for aromatic emission. The emission from Y2ndc3 consist of two distinguishable but overlapping peaks at 430 and 480 nm, while Y2bpdc3 consist of one major peak at 395 nm and a smaller shoulder at 475 nm. Replacing Y with Tb or Eu fully removed the aromatic emissions, indicating complete energy transfer to these lanthanides in Ln2bpdc3 and Ln2ndc3. However, while both Tb and Eu were luminescent in Ln2bpdc3, only Eu3+ was luminescent in Ln2ndc3. Tb2ndc3 showed neither aromatic nor Tb emission. The lack of emission from bpdc2− and ndc2− means that there is a complete transfer from these two to something else. ESI Fig. 3† show the emission spectra of Y2bpdc3 and Y2ndc3 along with the ground state f–f absorption transitions of Tb3+ and Eu3+. Both lanthanides show overlapping absorption lines with the emission of both sensitizers. Note that for wavelengths shorter than 400 nm, both lanthanides have dense absorption lines that are excluded in the figure for clarity. The lack of Tb3+ emission in Tb2ndc3 indicated that either Tb3+ is fully quenched or that Tb3+ never received the energy. In both cases, a possible explanation is a low energy intervalence charge transfer (IVCT) state that can accept the energy from either the sensitizer, lanthanide or both. This IVCT and its quenching on Tb3+ is well known in d0 metal oxides.22 Both Tb3+ and Eu3+ are redox active (to Tb4+ and Eu2+, respectively), and so is polyaromatic molecules.23 However, a detailed investigation into the quenching mechanisms is beyond the scope of this work.
The Tb and Eu emission spectra were identical to those from Ln2bdc3 in Fig. 10. In addition, the PLE spectra of Tb and Eu were identical to the PLE spectra of the aromatic emission in Y2bpdc3 and Y2ndc3, thus only the PL and PLE spectra of the Y variants are shown in Fig. 14. The modelled k(λ) and n(λ) were also identical between lanthanides. Comparing the modelled k(λ) in (a) and PLE in (b) show that the absorption and excitation peaks matches well for Ln2ndc3. However, even though both Ln2bpdc3 and Ln2ndc3 show similar absorption at 350 nm, this absorption only produces an excitation band in Ln2ndc3.
In (c) the transmittance spectra on all three Ln2ndc3 materials are shown alongside Tb2bdc3 for comparison. All samples are on silica substrates. The absorption bands 350 and 290 nm match well with the modelled k(λ) and PLE spectre.
Fig. 16 Photo and CIE coordinates of samples from this work under a 254 nm UV lamp. The violet, blue, green and red films as Y2bpdc3, Y2ndc3, Tb2bdc3 and Eu2bdc3. The orange film is a hybrid-fluoride nanocomposite from another recent work,8 showing orange Eu3+ emission sensitized by terephthalic acid layers. |
Footnote |
† Electronic supplementary information (ESI) available. See https://doi.org/10.1039/d2ra03360g |
This journal is © The Royal Society of Chemistry 2022 |