Far Infrared Spectroscopy of CsNiCl3 Crystals

The primary goal was to modernize the far-infrared interferometer, transitioning it from a manual system to one controlled by a computer, enabling real-time data acquisition. This involved integrating a microcomputer to control the interferometer, collect data, and perform real-time spectral computations using the interferogram. This real-time processing offered a significant advantage: low-resolution spectra could be generated quickly using only a few data points from the interferogram, with resolution improving as more data was collected. This enabled experimenters to assess experimental conditions (optimizing parameters such as sample thickness, concentration, and cutoff frequency) and make real-time adjustments without the previously necessary wait for complete interferogram acquisition. The efficiency gains were substantial, particularly given the often-limited supply of liquid helium, making each experimental run critical. This real-time capability drastically reduced experimental time and avoided wasted runs on suboptimal samples.

A crucial part of the project was designing and implementing the computer interface for the interferometer. This included developing software to control the instrument's operation and calculate spectra in real time. The interface boards, designed and built by Physics Department technicians, were driven by the KIM bus of the graphics board. The graphics board itself provided sufficient resolution (312 x 198 pixels) to display standard spectra (250 wavenumbers wide, one-wavenumber resolution). Furthermore, the system incorporated eight-bit digital-to-analog converters (D/A) to facilitate plotting the spectra on an x-y chart recorder. This intricate software and hardware integration was paramount in achieving the real-time spectral analysis capability. The efficiency gains from immediate feedback contrasted sharply with the previously substantial delays (hours or even overnight) experienced when using the interferometer's older, queue-based system.

The far-infrared interferometer functioned as a two-beam interferometer, recording intensity as a function of the path length difference between the two beams. The core of the spectral analysis relied on the Fourier transform of this interferogram (the recorded intensity versus path difference) to generate a plot of intensity versus frequency (the spectrum). While theoretically, the transform should ideally cover an infinite path difference, practically, finite sampling intervals and path differences were employed. The real-time system efficiently calculated this Fourier transform, converting the collected interferogram data into usable spectral information. This process, refined through the project's upgrades, was central to the real-time capabilities of the improved interferometer.

A key phase of the project involved the design and construction of a highly sensitive He3 cooled bolometer integrated with a superconducting magnet system. This system was specifically engineered to conduct Zeeman studies on magnetic crystals, particularly those under active investigation by the solid-state research group. The combination of the He3 cooling and the superconducting magnet provided the necessary sensitivity and high magnetic fields for the experiments. The system's design aimed to facilitate studies of magnons in quasi one-dimensional antiferromagnets, with a structure similar to CsNiCl3. A thorough understanding of the phonons within these crystals was recognized as a prerequisite for a comprehensive magnon study. The development of a theoretical framework for the transmission of off-axis electromagnetic radiation through a non-isotropic dielectric supported by an isotropic substrate was also a necessary component of this phase.

The creation of the He3 cooled bolometer necessitated considerable development in the design of suitable He3 refrigerators and associated components. A charcoal pump was a critical component of the cooling system. Initial designs for the charcoal pump experienced issues with inconsistent cooling performance due to uneven thermal contact between the activated charcoal granules and the container wall. These issues, manifested as large fluctuations in pumping speed and temperature, were resolved by improving thermal contact within the charcoal pump through the addition of a perforated brass sleeve. This confined the charcoal granules to a narrower space against the container wall, simultaneously reducing the charcoal volume from 100ml to 10ml and enhancing thermal efficiency. For longer experimental runs, a heater wire was implemented around the charcoal can, allowing for controlled heating above 40K when needed.

To accommodate the superconducting magnet, a substantial Dewar system was designed and built. The liquid helium Dewar required significant length to facilitate pumping the liquid helium bath down to 1.7 Kelvin. The Dewar's dimensions (inner diameter 15cm, outer diameter 22cm, length 104cm) were constrained by the university's glassblowing capabilities, leading to the construction of the outer liquid nitrogen cryostat in stainless steel within the Physics Department workshop (inner diameter 30cm, outer diameter 38cm, length 95cm). This cryostat incorporated superinsulation within its vacuum wall. A significant aspect of the Dewar design involved incorporating a ballast volume (3.5m of 100mm bore PVC sewer pipe) to mitigate temperature fluctuations in the liquid helium bath arising from pressure fluctuations of the rotary pump (450 liters/minute). Measures to reduce vibrations from the pump included clamping to a concrete shear wall and the use of copper bellows with rubber discs. The system achieved a helium evaporation rate of 0.4 liters/hour at 1.6 Kelvin.

An unforeseen challenge emerged concerning the influence of the magnetic field on the bolometer element. A significant, irreversible signal decrease was observed above a critical magnetic field (0.019 Tesla), dropping by a factor of five at 0.037 Tesla, rendering the detector largely unusable at higher field strengths. The cause was suspected to be linked to the critical magnetic field of superconducting indium (0.028 Tesla), present in the solder used to attach the electrical leads to the bolometer element. The transition from the superconducting to a normal state in the indium solder is likely to affect the amount of radiation absorbed, thus altering the element's temperature and subsequently the signal. This problem was eventually addressed by implementing magnetic shielding, allowing the system to operate effectively at fields up to 5 Tesla.

The research utilized the Bridgman-Stockbarger technique to grow single crystals. This method involved sealing dried and purified powders (CdCl2 type crystals) in quartz ampoules and then slowly lowering them through a sharp temperature gradient. The gradient itself varied from 50 to 70 degrees Celsius per centimeter, with lowering rates ranging from 0.6mm to 4mm per hour. This resulted in crystal growth times ranging from 2 days to 2 weeks. A standard lowering rate of 1mm per hour and a temperature gradient of 60 degrees Celsius per centimeter were typically employed. The specific purification process involved passing dry HCl or HBr gas through the molten material to remove hydroxyl ions. The process to prepare the salts for crystal growth used a stoichiometric mixture of AX and MX2 type salts, followed by the passage of halogen acid gas (e.g., hydrogen bromide for bromide compounds) to eliminate hydroxyl ions before sealing in quartz tubes. The growth of RbCoBr3 proved to be particularly challenging.

Due to the hygroscopic nature of many of the crystals, sample preparation was carried out inside a dry box maintained at less than 15% humidity. Far-infrared spectra were recorded using a Grubb-Parsons Model IS-3 Fourier transform spectrometer equipped with a He3-cooled, doped germanium bolometer. All samples were immersed in liquid helium at 1.7 K. Transmission spectra were obtained using unpolarized radiation at 60° and 0° (axial) incidence relative to the sample normal, and reflection spectra were acquired with 45° incidence. Background spectra were carefully subtracted to account for the contribution of the sellotape and varnish used in sample mounting and to minimize systematic errors in spectral interpretation. The non-collimated beam had an approximate spread of 90 degrees on either side of the beam direction. Initial attempts to obtain reflection spectra using thin films in direct contact with a metal substrate proved problematic, yielding inconsistent results. This involved compressing a layer of crystal between aluminium plates and peeling off excess material; however, this resulted in fractured samples resembling powdered samples, thus producing unreliable data.

To overcome the challenges of sample preparation for reflection measurements, a sellotape stripping technique was developed and employed to produce thin films of the crystals, with the crystal c axis oriented perpendicular to the film plane. This approach proved much more reliable. This technique allowed for the production of thin films suitable for transmission experiments, facilitating the unambiguous measurement of Eu(TO) and A2u(LO) infrared-active mode frequencies for various CdCl2-type crystals (CdCl2, CdBr2, CdI2, MgCl2, MnCl2, FeCl2, CoCl2, NiCl2, and CoBr2). However, due to the crystal structure’s inherent ease of cleavage, creating sufficiently thin films with the c axis in the plane of the film proved impossible, preventing direct measurement of Eu(LO) and A2u(TO) mode frequencies using this method. Analysis of powder samples revealed a preferred orientation of the crystallites (small platelets roughly 20µm across), which is consistent with observations from electron microscopy. This preferred orientation explains why some previous studies using powders had difficulty in resolving the A2u mode.

The research began by deriving expressions for calculating far-infrared transmission through thin films of anisotropic single crystals at non-normal incidence. This theoretical work extended previous models, such as those by Born and Huang (1954) for normal incidence transmission through isotropic dielectrics and Berreman (1963) for non-normal incidence. The new expressions account for the anisotropic nature of the crystals under investigation, providing a more accurate framework for interpreting experimental data. These expressions were crucial for analyzing the experimental transmission and reflection spectra obtained from thin films of CdCl2-type crystals. The work highlighted that for very thin films deposited on a metal surface, the vibrational mode parallel to the surface is not excited because the electric field component parallel to the metal is zero at the surface. This technique has been used by Berreman (1963) to observe the LO mode of alkali halide crystals.

Transmission spectra were collected using unpolarized radiation incident at angles of 60° and 0° (axial) with respect to the sample normal. Reflection spectra were obtained with radiation incident at 45°. In all cases, the beam was not collimated and had a wide angular spread (approximately 90°). The data analysis focused on identifying the infrared-active modes, specifically the Eu(TO) and A2u(LO) modes. The results were compared to previous work utilizing powder samples embedded in media like polyethylene or nujol. The study noted that these prior methods can yield misleading results due to the finite size of the crystallites and the influence of the embedding medium (Frohlich, 1948; Martin, 1970), where the absorption maxima fall between the transverse and longitudinal optical mode frequencies of the single crystal. The analysis demonstrated that the frequencies of the infrared-active modes are strongly influenced by the sample preparation method and the angle of incidence. The sellotape technique proved superior for obtaining clear and reproducible results for the Eu(TO) and A2u(LO) modes.

The research compared its findings on infrared-active modes with existing data obtained using Raman spectroscopy. Raman-active mode frequencies, determined through polarization studies (Lockwood, 1973), were readily identifiable, unlike infrared-active modes which have been previously reported as broad overlapping bands due to the strong absorption characteristics of these modes in bulk samples. This strong absorption has necessitated previous researchers to study powdered samples, resulting in broadened and shifted absorption peaks compared to the single crystal values (Lockwood 1973, Pollini et al 1980, Anderson et al 1981). The research used a theoretical fit to the experimental data for CdBr2 and reported the linewidths of infrared-active modes (less than 15 cm-1 FWHM), which are notably broader than those reported for the Raman-active modes. The study also directly compared their low-temperature infrared results with room-temperature reflection data from Lucovsky and White (1977), demonstrating good agreement considering the expected temperature-dependent frequency shifts (Campbell, 1978).

The study acknowledged the inability to directly measure the Eu(LO) and A2u(TO) mode frequencies due to the difficulty in preparing sufficiently thin films with the crystal c axis in the plane of the film. The paper suggests two indirect methods for future work to determine these unmeasured frequencies. The first involves measuring the anisotropic dielectric constants at high and low frequencies, allowing deduction of the mode frequencies using the Lyddane-Sachs-Teller relation adjusted for anisotropic crystals (Richman et al., 1963). The second involves measuring the reflectivity from a bulk crystal, as demonstrated by Lucovsky and White (1976) for CdI2. The research emphasizes that the analysis of the obtained spectra provides valuable insight into the anisotropic nature of the studied crystals and the strong forces within the basic 1-Cd-I sandwich structure, contrasting with the weaker van der Waals bonding between these sandwiches.

In magnon studies using far-infrared spectroscopy, sample thickness optimization is critical. The sample needs to be thin enough to allow sufficient transmission in the region of interest (including phonon absorption), yet thick enough for any discrete magnon absorption lines to appear clearly above the background noise level. This balance is essential for obtaining reliable and interpretable data on magnon modes. For instance, in the case of RbCoBr3, this optimization was crucial for successful measurements. This consideration highlights the importance of careful sample preparation and experimental design in obtaining accurate results in far-infrared spectroscopy, particularly when dealing with weak absorption features such as magnon lines. The selection of an optimal thickness avoids wasted experimental time and effort on samples that do not meet the required sensitivity.

The research presented results from far-infrared spectroscopic analysis of phonon modes in CsNiBr3. The observed spectra clearly showed absorption due to three modes in sigma polarization and two in pi polarization. The transverse optical (TO) mode frequencies were identified at the points of maximum absorption, which, in regions of zero transmission, were located near the low-frequency edge. The longitudinal optical (LO) mode frequencies of the transverse modes were found in the pi-polarized spectrum at the high-frequency edge of the broad absorptions seen in the sigma polarization spectra. This analysis provided key information about the lattice vibrations and helped in understanding the phonon structure, a vital component for future magnon studies in similar materials. The thin sample (less than 0.1mm thick) facilitated phonon analysis by allowing sufficient transmission of radiation through the sample.

The influence of altering the monovalent ion on phonon frequencies was investigated. The results showed that replacing rubidium with caesium had minimal impact on the overall frequencies of the absorption lines. While a slight effect might be present on the lower two absorption lines, it was concluded that any mass change due to substitution was largely compensated by a corresponding adjustment in the involved force constants. This analysis provides valuable insights into the bonding and interactions within the crystal lattice and how changes in ionic mass affect vibrational behavior. The investigation into the subtle changes in the crystal properties caused by ion substitution was crucial for further exploration into the phonon and magnon behavior of these quasi one-dimensional systems.