Raman Observations Using a Spatial Heterodyne Spectrometer

Raman spectroscopy is a well-known method for analysis and identification of chemical samples. One of the main advantages of Raman measurements is that the samples need no preparation; this makes the technique perfect for in situ measurement applications within the process industry. Raman scattering is a weak process, thus collecting as much light as possible is vital if good quality observations are to be attained.

This issue is compounded when making observations from diffuse targets; as needed for transmission Raman measurements. Furthermore, if a large stand-off distance is present in the observational environment, this can have a major impact on the achievable throughput. This has led to the development of high etendue spectrometers based on static Fourier transform instrument designs (for example, spatial heterodyne spectrometer (SHS), RD1, RD2, RD3). This comprises of the IS-Instruments range of HES spectrometers which have been optimized for Raman applications.

The optical probe design and configuration (the instrument-sample interface) can have a major influence on instrument performance. The probe must accurately collect light from the target sample with a high level of efficiency and exceptional throughput, while eliminating unwanted returns from the sample holder or cuvette. In order to combat these effects, current commercially available systems have used a variety of solutions including: confocal microscope probes (RD4), fiber bundle collection systems used in a spatially offset configuration (RD5), and multi-pass probes for gas and liquid samples.

Furthermore, the probe optics should be well matched to the analyzing spectrometer. The ISI HES range of spectrometers offers both greater depth of field and increased etendue. This demands careful optical probe design and/or selection. For instance, the enhanced etendue provided by the instrument can produce additional challenges in terms of extraneous light suppression; as background light is proportional to the instruments field of view.

As all molecular compounds offer a Raman signature, signal contamination from extraneous material or sample containers can be the main issue. Besides masking the primary signal, these unwanted returns increase the noise in the overall spectrum when adopting an SHS configuration due to the multiplex disadvantage (RD6).

This article explores each of these issues and presents some of the different probe options available. Particular probe designs for stand-off observations at distances up to 1.5 m are also presented. It also analyzes how the data is presented in an SHS and effects of unwanted background light. Additionally, techniques that can be used for enhancing the signal to noise ratio attained with simple design changes are discussed. Key to this is adopting a philosophy where the spectrometer and probe are developed as a single unit instead of separate optics systems.

Background

The overall amount of signal returned by a Raman instrument is given by equation 1.

Equation 1

Where L_P is the laser power, ΔR is the depth of the sample being analyzed, α is the Raman scattering cross section, A is the area of the telescope or collecting lens, ω is the instrument efficiency, and R is the distance to the target. The instrument efficiency is a mixture of the detector quantum efficiency and the optical transmittance. τ is a function of the depth of the sample being analyzed.

This is a revised version of the LIDAR equation and assumes the target sample is a Lambertian scatterer. The equation takes no account of any directionality of the Raman scattering function that can impact the exact return.

The relationship between ΔR and τ is essentially important to the choice of probe being employed for any given application. The amount of light returned is proportional to the number of scattering centers intercepted by the laser beam which is determined by ΔR. Thus, the higher the depth probed by the instrument the stronger the return.

However, the optical depth is also a function of ΔR, thus as the depth increases it is not possible to make no further gains. This effect is demonstrated in Figure 1, where the relative signal strength is calculated for a fixed sample with a range of different optical depth values. Figure 1 demonstrates that the linear relationship is observed with ΔR for an optical depth of just 0.01.

If the probe is designed for penetrating a set distance into the sample with a high optical depth, in a fixed optical arrangement the summation term in equation 1 is negated, and the amount of signal can be decreased! It is also possible to observe this issue when the transmission and collection optics are not perfectly co-aligned.

Figure 1. Relative signal strength as a function of optical depth.

Image Credit: IS-Instruments

The Influence of Samples

The varieties of samples that can be encountered by an instrument differ dramatically, and thus there is no one-size-fits-all probe solution. A particular type of probe setup may be perfect for just a small subset of possible measurements. The probes themselves can differ from simple backscatter arrangements, through to more complicated scanning setups.

A probe may well be able to make a given observation but is not the best or most cost-effective solution. For instance, scanning microscope probes are available from most manufacturers and can indeed carry out observations of a number of samples. However, if a sample is predominantly homogenous in nature, or if there are time constraints (as seen in the case of many online observations) this class of system is not perfect, or even necessary. Table 1 offers a guide on selection of a Raman probe to make the best observation of a given type of sample. Samples are considered in relation to the following parameters:

1. heterogeneity
2. volume
3. thickness
4. without or within a container
5. scattering strength
6. opacity

It should be noted that a number of samples fall between these simple classifications and thus final selection should rely on individual circumstances. For instance, a biaxial probe arrangement may be the only solution if the sample is mounted in an unusual setup and the container has highly reflective properties.

The most common probes for Raman measurement are.

• Micro objective
• Confocal
• Biaxial
• Transmission
• Spatially offset
• Ring or multipass arrangements
• Scanning micro objective

Table 1. Type of probes and ability to measure a range of samples.

Table 1 shows the sheer range of possible measurement options that can be adopted. This does not take into account environmental demands such as large stand-off distances, where subtle design changes may be needed.

Image Credit: IS-Instruments

It also shows how common measurement solutions offered by suppliers are not ideal for a wide range of target applications. For instance, a micro-objective probe, with no scanning function is not suited for heterogeneous samples. With this class of target, a transmission or spatially offset arrangement may be superior. Thin film samples are best probed with a micro objective, but complications take place when these samples are weak Raman scatterers. If a sample is greatly opaque, a transmission or spatially offset arrangements may give no response at all.

In the following sections, the design of stand-off Raman probes that may be adopted for online measurement problems is considered for use with a HES spectrometer. Confocal and biaxial Raman probes are considered and their performance is compared. Finally, the specific problem of making observations at > 1 m is examined and a new probe design is then presented. Scanning and ring or multipass techniques are generally more complex and thus are beyond the scope of this article.

Instrumentation

Earlier authors have discussed the strengths of an SHS when making observations of high etendue targets, particularly in applications such as transmission Raman (RD3). However, in a number of applications from laboratory-based testing through to online measurements for the industry, this type of arrangement is not possible.

This article considers measurement situations where the target is at least 50 mm from the target. The probe design is considered with the SHS as a unit, and the susceptibility of the instrument to background light and stray Rayleigh scattering must also be taken into account. Superior performance can be achieved if this can be resolved outside of the spectrometer.

The following three principle probe designs are now considered:

1. A confocal probe incorporating a dichroic mirror, acting to (a) turn the excitation laser and (b) filter the collected Raman signal.
2. A confocal probe integrating an independent laser excitation turning mirror which, due to design limitations, the collected Raman signal becomes partially obscured.
3. A biaxial probe, where the excitation laser optics and the Raman collection optics are independently positioned and aligned.

Collection Probe

The preferred choice of Raman collection probe for a number of applications is a confocal arrangement. This type of probe is easy to use with the excitation and collection optical paths coaligned. This generally needs a dichroic filter to be incorporated into a probe. When using a HES spectrometer, the probe should have a filter inserted ahead of the fiber in order to eliminate any unwanted Raman/fluorescent light. Given the sensitivity of the HES instrument to background light, any unwanted Raman scattering should be removed at source. Thus, the ISI confocal probe uses a focusing mirror to direct and then collect the light from the target rather than a lens which can itself be a significant Raman scattering source.

A schematic of this probe is presented in Figure 2. The benefits of this approach are that no light is lost due to obscuration. This makes the design highly flexible and robust. The design can be adapted to fit a micro-objective allowing for extremely small samples to be probed at reduced focal depth, thereby eliminating unwanted scattering sources (for example, from the sample container).

The probe etendue must be matched to that of the analyzing spectrometer if no light is to be lost. This can become a major limitation in dispersive systems. The ISI confocal probe is fiber coupled, so has the advantage of decoupling the optical systems making the design of both parts more flexible.

Figure 2. IS-Instruments confocal Raman probe.

Image Credit: IS-Instruments

An alternative confocal probe arrangement is presented in Figure 3. Here the dichroic filter is replaced with a turning mirror. This is less efficient than the example provided in Figure 2 as the excitation laser turning mirror partially obscures the return signal and in most cases offers inferior performance. However, this arrangement can provide benefits when attempting to study samples within a container. If the sample has a sufficiently different focus position, then the turning mirror can behave as a spatial filter with respect to the walls of the container.

Figure 3. IS-Instrument mono axial probe arrangement.

Image Credit: IS-Instruments

Figure 4 presents a decoupled arrangement which could be considered optically biaxial. This setup is more difficult to arrange and is not suited for handheld applications; however, for online monitoring, it can offer the best solution, as it efficiently removes any unwanted light and can precisely target samples within a given container, where only the Raman signal from the target source is collected.

Figure 4. IS-Instrument Bi-axial Probe arrangement.

Image Credit: IS-Instruments

Raman Probes for Challenging Sample Capture

The following section presents a series of Raman measurements explaining the application of customized probe designs. The section also presents configurations to optimize the capture of Raman scattered light into the spectrometer for a series of challenging sampling conditions and sample types.

In each case, the light collection section of the optical probe was coupled through a 1 mm fiber to an ISI HES spectrometer. A 785 nm PDLD Boxx Raman excitation laser was coupled via a 100 µm fiber to the transmission section of the optical probe. The laser power at probe output at the sample was approximately 250 mW.

Measurement of Solid Samples Within a Glass Container

Shown in Figure 5 (a) and (b) are Raman spectra of two freeze-dried samples: A and B, and in both instances measurements were captured through a crimp top glass sample vial container, thus negating the need to remove the sample from the container and expose the material to the ambient environment.

Earlier work demonstrated the sample container gave a strong Raman signal itself and was likely to contaminate the measured spectra. Figure 5 demonstrates the spectra attained with the three probes shown in Figure 2 - Figure 4, demonstrating the strengths and weaknesses of each of the probes. The confocal probe with the dichroic mirror shows an extremely large peak from 1200 cm^-1 – 1600 cm^-1. This is because of the efficiency of the probe collecting the majority of the scattered light both from the target and the sample container. In this instance, the efficiency of the probe is working against the objective to observe just the sample. The spectra recorded by using the confocal mon-oaxial probe shows good potential to capture the spectra. However, both spectra A and B show a baseline lift toward the LHS, indicative of unwanted laser light from the input laser optics interfering with the collected Raman signature. In the confocal probe with a dichroic mirror, this lift has been removed. However, the performance is still greater than the classical confocal.

Figure 5. The acquired spectra of freeze-dried samples through a crimp glass vial container.

Image Credit: IS-Instruments

The biaxial probe resolves both of these issues in this instance. The collection optics are focused only at the target, and by being non-coaligned surface scattering becomes spatially filtered. In Figure 5 (a), the spectra are clearly displayed. However, the mechanical alignment of the system is more challenging and thus some light is lost.

Measurement of a Liquid Sample - Water

The observations discussed above are of solid samples mounted within an optically obscuring container. Liquid samples can also provide challenges for Raman measurement. In this section, an observation is taken into account, where the container is conveniently shaped, and therefore the confocal Raman probe, shown in Figure 2 was used. Additionally, the large depth of field inherent to HES instruments (due to large etendue), permits ΔR to be increased for translucent liquid targets such as water; allowing more photons to be captured and observed.

The resulting raw data observation is displayed in Figure 6, where no data processing has yet been applied. The probe was originally fitted with a 785 nm long pass (LP) filter as labeled. The integration time was four seconds with approximately 250 mW laser excitation power at the target. The water peak can be clearly observed at 1600 cm^-1. Using equation 1, it is possible to compare the observed signal to a simulation of the expected instrument performance.

Using a confocal F2:1# probe, the estimated depth of focus is 10 mm. From (ref) the Raman cross section of the water spectral peak at 1600 cm^-1 is 3.35 × 10^-31 cm²Sr^-1mol^-1 at 785 nm. The instrument used an ANDOR IVac 324 detector with a QE of 30% at 1600 cm^-1, and the optical transmission taking account of the losses with the fiber coupling, filter, and beam splitter is estimated as being 20%. This suggests that 820,000 photons will be observed within the 1 integration time. The detector sensitivity is 6.7 photons per count, thus the detector count rate is 122,000. The actual observed photons are 116,000, which is within 5% of the computed value. This shows that the HES2000 instrument is accomplishing exceptional performance when merged with the confocal probe.

However, the noise in the signal is relatively large, this is because of the multiplex effect when using a Fourier-based device (RD5). To further illustrate this, an 880 nm bandpass filter has been included with 70 nm bandwidth. The filter has a 70% transmission. Although the resulting spectra have a lower signal, the observed signal to noise ratio is greater by a factor of 1.8. Incorporating the light observed from 50 cm^-1 – 700 cm^-1 proves that approximately 3.4 times more photons are observed without the bandpass filter, signifying that this would contribute a factor of 1.84 in additional noise to the capture of water Raman spectra. This is in excellent agreement with the observation. This result demonstrates how the performance of the system can be refined in order to offer optimum performance for a given observational condition.

Figure 6. Raman spectra of water captured using a 785 nm long pass filter and a 70 nm bandwidth 880 nm bandpass filter.

Image Credit: IS-Instruments

Opaque Targets - Graphite

There are situations where a small depth of focus offers the maximum amount of returned light, because of massive attenuation of the sample as specified by Figure 1. One such case is when developing Raman observations of graphite. Graphite is an opaque material as presented in Figure 7

Figure 7. Example of a Graphite block.

Image Credit: IS-Instruments

The IS-Instruments range of HES spectrometers is usually designed to collect light from a 1 mm diameter spot. Matching the collection probe to this spot size typically offers a depth of field of several mm. This is more than sufficient for most solid and liquid samples to provide a good Raman signal, assuming a laser power in excess of 100 mW at 785 nm. Graphite is recognized for having a large Raman cross section (RD7).

As with the water observation, the confocal Raman probe was used. However, when making observations of the graphite poor quality Raman spectra were obtained as presented in Figure 8. The general background slope observed from 400 cm^-1 – 2000 cm^-1 is indicative of light being reflected back from the surface and exciting the fiber. The expected peaks are at ~ 1300 cm^-1 and 1600 cm^-1 although visible are extremely weak in comparison. It was noted that the surface did have a major reflective sheen.

The observation was repeated on a roughened surface of the graphite, with dust from the graphite being present. The captured Raman spectra are displayed in Figure 8 (b) where both of the predicted peaks are clearly observed.

Figure 8. Graphite Raman observations on (a) smooth surface and (b)roughened surface.

Image Credit: IS-Instruments

This indicates that the surface itself has had an impact on the measurement. Analyzing the results in more detail, the actual Raman peaks are similar in relative strength and shape in both cases; however, in the roughened case the background lift is far less. Furthermore, the length of time to attain the data was in excess of 10 seconds despite the large scattering cross section of graphite. This can be explained by the important optical depth of sample with surface penetration being estimated as being no more than 50 µm. Additionally, when observing the smooth surface the total amount of light collected is considerably greater than when measuring the roughened face.

The smooth surface has a reflective sheen, and this creates a lot more Rayleigh scattered photons. This reflective sheen is indicative of the face comprising of specular reflective properties. Thus, a major increase is observed in the amount of unwanted laser light being collected, and this, in turn, excites the glass with the probe generating the unwanted background response. In this case, a micro-objective with a much smaller etendue may offer greater results as these unwanted photons are spatially filtered and thus do not contaminate the observation.

The Stand-Off Challenge

In a number of process applications, interfacing a Raman system to a manufacturing facility can be a challenge. It is often not possible to have the system in close proximity to the target sample. In this case, if there is a line of sight, a stand-off probe can be used. However, this offers significant challenges to the spectrometer. As shown in equation 1, the strength of the signal is α 1/ R² and furthermore, the instrument etendue can become a restrictive factor especially if the distance to the target is not fixed. The ISI HES range of spectrometers has a factor of up to 500 gain in the etendue (RD3) over classical dispersive systems, providing it a clear advantage in making observations of this nature.

The extended distance is likely in most cases to need an extended integration time to attain the spectrum from the target. This, in turn, will increase the prominence on any undesirable background signals. One issue frequently not considered is the Raman response from the lenses used within the probe. For integration times beyond 10 seconds, this signal can be significant, thus any stand-off Raman instrument must consider this effect in the design.

In a stand-off probe, it is desirable to increase the size of the receiving optics in order to maximize the collected signal. However, the filter technology can limit the maximum diameter that can be used to < 100 mm diameter. For stand-off applications, a monoaxial design as shown in Figure 9 is usually adopted. It should be noted that in this design the turning lens is mounted behind the main collection lens. For bigger apertures, this lens has a thickness in excess of 10 mm. This can result in additional Raman or fluorescent response being emitted as demonstrated in Figure 9.

Figure 9. Illustration showing effect of a large FOV on collected Raman or fluorescence response.

Image Credit: IS-Instruments

The solution to this issue was to shift the turning lens to a position, ‘after’ the main collecting lens. This has the effect of removing any internally excited photons. Figure 10 and Figure 11 display IS-Instruments Ltd stand-off probe built around this concept.

It is possible to clearly observe the folding mirror after the main light collects elements. The probe uses a 100 mm diameter lens, with an adjustable focus from 50 mm to 1500 mm. The laser etendue is controlled in order to match the collecting spectrometer (i.e. 1 mm spot diameter at the target).

Figure 10. Stand-off Raman probe: Stand-off distance 500 mm – 1500 mm.

Image Credit: IS-Instruments

Figure 11. Schematic showing IS-Instruments 500-1500 mm stand-off probe.

Image Credit: IS-Instruments

The performance of the probe was tested by analyzing a sample of paracetamol at a stand-off distance of 1 m. The resulting spectra are displayed in Figure 12. Observations were carried out with the help of a HES 2000 instrument, with an integration time of 0.5 seconds and 1 second respectively, with an average of 10 observations as presented in Figure 12 (a). The observed signal to noise ratio is exceptional. The high etendue provided by the HES instruments permits the device to maximize the return from the target. The difference between a single shot and the averaged observation is shown in Figure 12 (b).

Figure 12. Paracetamol Raman spectra measured at a distance of 1 m using ISI stand-off Raman probe. (a) average of 10 observations at 0.5 and 1 second integration times. (b) 1 second integration time with single shot and 10 average observation.

Image Credit: IS-Instruments

The spectra presented have had no processing applied, and display no residue fluorescent or Raman signal from the probe demonstrating the exceptional performance that can be attained with the HES 2000 even at a distance of 1 m and an integration time of less than 1 second.

Summary and Conclusions

• Customizing the probe specification to the demands of the sample and sampling environment is demonstrated to be very critically important for optimizing the performance of the Raman measurement.
• A series of probe types are presented and their suitability for capturing Raman spectra from a series of challenging samples and sampling conditions is discussed.
• A comparison of three different optical probes employed for capturing Raman spectra of a freeze-dried sample within a glass vial highlighted the strengths and weaknesses of each, specifying in this instance that a biaxially configured arrangement was optimum.
• Raman spectra of a translucent liquid and an opaque solid are also presented, where probe specifications for each were optimized in order to capture the best measurement.
• A new confocal monoaxial optical probe is also presented, where the undesirable effects brought about by optical component induced Raman capture is addressed, and where the partially obscuring excitation laser turning mirror is placed externally from the Raman collection optics. This probe has been established to reliably capture Raman spectra at a stand-off distance of up to 1.5 m.

The Raman probes and HES2000 spectrometer are now available from IS-Instruments.

References

[RD1] J. Harlander, R. J Reynolds and F. L. Roesler., “Spatial Heterodyne Spectrometer for the exploration of diffuse emission line a far ultraviolet wavelengths” Astro Phys Journal 396 (1992)

[RD2] M. J. Foster, J. Storey and M. Zentile “A Spatial-Heterodyne Spectrometer for Transmission-Raman Observations” Optics Express Vol 25 Issue 2 (2017)

[RD3] M. J. Foster, J. Storey, P. Stockwell and D. Widdup “Stand-off Raman spectrometer for identification of liquids in a pressurized gas pipelines” Optics express 3968 (2015).

[RD4] https://en.wikipedia.org/wiki/Confocal_microscopy

[RD5] P Matousek; IP Clark; ERC Draper; MD Morris; et al. (Apr 2005). "Subsurface probing in diffusely scattering media using spatially offset Raman spectroscopy". Applied Spectroscopy. 59 (4): 393–400

[RD6] J. M. Harlander, “Spatial heterodyne spectroscopy: interferometric performance at any wavelength without scanning,” Thesis (Ph.D.) University of Wisconsin Madison (1991).

[RD7] S. Reich and C. Thomsen “Raman spectroscopy of Graphite” Philosophical Transactions: Mathematical, Physical  and Engineering Sciences  Vol 362 (2004)

This information has been sourced, reviewed and adapted from materials provided by IS-Instruments, Ltd.

For more information on this source, please visit IS-Instruments, Ltd.