Surface Contamination and Cleaning, Vol 1, pp 1-22 Ed KL Mittal 0V SP 2003 Mapping of surface contaminants by tunable infrared-laser imaging DAVID OTTESEN, SHANE SICKAFOOSE," HOWARD JOHNSEN, TOM KULP, KARLA ARMSTRONG, SARAH ALLENDORF and THERESA HOFFARD' Sandia ,Vational Laboratories, P.0 Box 969, .MS 9403, Livernzore, CA 94531-0969 ''Vaval Facilities Engineering Service Center, I100 23rd Avenue, Port Hueneme, CA 93043-4370 Abstract-We report the development of a new, real-time non-contacting monitor for cleanliness verification based on tunable infrared-laser methods. New analytical capabilities are required to maximize the efficiency of cleaning operations at a variety of federal (Department of Defense [DoD] and Department of Energy [DOE]) and industrial facilities. These methods will lead to a re- duction in the generation of waste streams while improving the quality of subsequent processes and the long-term reliability of manufactured, repaired or refurbished parts. We have demonstrated the feasibility of tunable infrared-laser imaging for the detection of con- taminant residues common to DoD and DOE components. The approach relies on the technique of infrared reflection spectroscopy for the detection of residues. An optical interface for the laser-imaging method was constructed. and a series of test surfaces was prepared with known amounts of contaminants. Independent calibration of the laser reflectance images was performed with Fourier transform infrared (FTIR) spectroscopy. The performance of both optical techniques was evaluated as a function of several variables. including the amount of contaminant, surface roughness of the panel, and the presence of possible interfering species (such as water). FTIR spectra demonstrated that a water film up to 7 pm thick would not interfere with the effectiveness of the laser-imaging instrument. The instrumental detection limit for the laser reflec- tance imager was determined to be on the order of a 10-20 nm thick film of a general hydrocarbon contaminant. Keywords: Infrared; tunable-laser: imaging: cleaning; surface contamination. 1. INTRODUCTION Real-time techniques to provide both qualitative and quantitative assessments of surface cleanliness are needed for a wide variety of governmental and industrial applications. The range of potential applications include aircraft, shipboard, vehi- cle, and weapon component surfaces to be coated, plated, or bonded. The avail- 'To whom all correspondence should be addressed. Phone: (925) 294-3526. Fax: (925) 294-3410. E-mail: [email protected] 2 D.O ttesen et al. ability of a convenient analysis technology for on-site, post-cleaning determina- tion of surface contamination will allow more rapid and accurate assessments of the efficiency of chosen cleaning techniques. By developing an on-line technique, processed parts or extracted samples will not have to be sent to a separate labora- tory for analysis, thereby eliminating processing delays. The information provided by the optical method will assist the process operator in distinguishing between specific contaminants and determining subsequent actions to be taken. In this paper we report the development of an infrared laser-based imaging ap- proach that will reduce the use, emission, and handling of waste-stream materials in cleaning operations. This work is supported by the separate development of a hardened, portable Fourier transform infrared (FTIR) reflectance instrument at the Naval Facilities Engineering Service Center (NFESC), Port Hueneme, CA in co- operation with the Surface Optics Corporation. Simultaneous development of an FTIR instrument is complementary in nature to the laser-imaging technique and is described in detail elsewhere [l]. Both instruments will be used primarily for the real-time on-line or nearly on-line detection of contaminant residues on reflective surfaces. In each case, surface contamination is detected by its absorption of a grazing-incidence infrared beam reflected from the surface. The instruments differ in the nature of the information they provide. The laser- based instrument produces images that directly indicate the spatial extent and lo- cation of infrared-absorbing surface hydrocarbon contaminants. In contrast, FTIR instrumentation provides a wide-band spectral measurement of the surface reflec- tance averaged over a small area for nearly all organic materials, and many inor- ganic components. Thus, the laser-imaging system allows the rapid determination of surface cleanliness for organic residues over a large area, while the spectrally- resolved FTIR method is useful in identifying the specific molecular composition of a surface contaminant at a particular location. The imaging system under development employs a widely tunable infrared- laser illumination source in conjunction with an infrared camera. This approach provides an on-line technique for surveying contamination levels over large sur- face areas in a real-time imaging mode. The laser is broadly-tunable over the 1.3- 4.5 pm wavelength range, thus allowing the detection of many hydrocarbon con- taminants via absorption bands associated with CH-, OH-, and NH-stretching vi- brations. Currently, the detection and identification of surface contaminants on reflective surfaces is conveniently and rapidly done by FTIR reflectance methods. These non-destructive, non-contacting optical techniques identify the chemical constitu- ents of the contaminants, and can yield quantitative measurements with appropri- ate calibration. Infrared optical methods are particularly useful for cleanliness verification since the surface is probed under ambient conditions. More sensitive high-vacuum electron and ion spectroscopic techniques (X-ray photoelectron spectroscopy, Auger electron spectroscopy, and secondary-ion mass spectrome- try) are not suited for on-line application. Tunable IR-laser nzapping of surface contanzinants 3 Commercial instruments that employ infrared reflectance spectroscopy are available for surface analysis and provide both quantitative and qualitative infor- mation on surface coatings. These instruments are limited in their ultimate sensi- tivity to surface contaminants by the nature of their optical design. Infrared radia- tion is focused onto the surface to be analyzed at a near-normal angle of incidence, resulting in a compact hand-held apparatus. The infrared light is col- lected as either specularly or diffusely reflected radiation depending on the roughness and scattering properties of the surface [2, 31. The resulting sensitivity to very thin layers of surface species is limited by poor coupling of the incident electromagnetic field with the vibrating dipoles of the surface molecular species [4-61 in layers less than 0.1 pm thick. In order to maximize the sensitivity of infrared reflectance measurements for absorption bands of thin layers of contaminants on metallic surfaces, theoretical and experimental studies [7-91 have shown that the angle of incidence of infrared radiation on the surface should be increased to at least 60” from the surface nor- mal. This is also true for many thin-film residues on the surface of non-metals, such as dielectrics and semiconductors (although the detectability of contaminant absorption bands under these circumstances depends strongly on the optical con- stants of both surface and substrate, and any absorption features intrinsic to the non-metallic substrate). Additional sensitivity in the reflectance measurement is obtained by measuring only the component of the reflected infrared radiation po- larized parallel to the plane of incidence. This experimental method is variously referred to as, ’*grazing-angle” reflectance spectroscopy or infrared reflection- absorption spectroscopy (IRRAS). We have adapted the technique of “grazing- angle” reflectance spectroscopy to utilize the newly developed tunable-laser source. 2. EXPERIMENTAL The laser-based instrument described in this report offers the capability to rapidly survey large surface areas and to determine the location and extent of residual hy- drocarbon contaminants following cleaning operations. In contrast, a spectro- scopic analysis by an FTIR-based infrared reflectance instrument is able to char- acterize a very broad range of organic constituents and many inorganic species. However, a surface-probing FTIR instrument measures a spectrum at only a sin- gle small area on a sample, thus requiring broad area surveys to be done by se- - quentially probing many points. Even at a rate of 10 seconds per measurement point, this can be a time-consuming process. The rate of measurement by FTIR spectroscopy is constrained by the relatively low spectral brightness (compared to a laser) of the incandescent illumination sources. This makes it necessary to use relatively long integration times to achieve an acceptable signal-to-noise ratio. The tunable-laser-based instrument overcomes these limitations by illuminating a broad surface area with a high-brightness infrared laser. This approach allows a 4 D. Ottesen et al. single-wavelength reflectance measurement over an area of several square centi- meters to be made on a timescale of less than a second. In order to acquire meas- urements at multiple wavelengths, the laser is tuned and an image is collected at each of the desired wavelengths. While a detailed spectral map of a surface can be generated over the laser tuning range, the primary use of the system is to provide rapid areal surveys at a few key wavelengths that are indicative of hydrocarbon contaminants. The detection sensitivity for several hydrocarbon species at various illumination wavelengths was evaluated in this work, as well as a method to sup- press image noise due to laser speckle while maintaining high illumination inten- sity. 2.1. Quasi-phasematching tunable infrared laser The broadly-tunable infrared laser illuminator is based on a technology called quasi-phasematching (QPM) [lo]. This approach has been exploited to increase the tuning range and power of the infrared light source while reducing its size. For example, continuous-wave (cw) optical parametric oscillators (OPOs) that employ the QPM material, periodically-poled lithium niobate (PPLN), are capable of tun- ing over the 1.3-4.5 pm spectral region while emitting more than 0.5 W of power. This technique has been used to generate tunable infrared laser light for imaging natural gas emissions, and developing laser-based spectroscopic gas sensors [IO], In this work we are extending it to the analysis of hydrocarbon residues on mate- rial surfaces. The limit of the current tuning range of the PPLN-based laser at long wave- lengths is about 4.5 pm (2222 cm-’) due to the transmission characteristics of lith- ium niobate. This property restricts the sensitivity of the chemical imaging system to functional groups containing hydrogen atoms (C-H, N-H, 0-H). Extension of the laser tuning wavelength range beyond 5 pm (2000 cm-’) is desirable to pro- vide specific identification of hydrocarbon and some inorganic molecular species. The light source assembled for the IR imaging sensor is an OPO pumped by a continuous-wave (cw) Nd:YAG laser, as shown in Figure 1 [lo]. An electric field is induced in the OPO’s PPLN crystal by the electric field of the pump laser; these fields interact to form two new laser beams whose frequencies sum to the fre- quency of the pump laser. The reflectivities of the mirrors in the optical cavity are selected to resonate one of the generated waves, while the other wave is simply generated and released from the cavity. The resonated wave is called the signal; the non-resonated wave is called the idler. The exact frequencies of the signal and the idler are determined by the phasematching properties of the crystal (described below), the reflectivity of the cavity, and by any spectrally-selective optics that may be added to the laser cavity (e.g. an etalon). While either the signal or the idler beam can be used for measurements, only the idler is used in the experi- ments reported here. As shown in Figure 1, the OPO used in the imaging sensor is of the “bowtie- ring” design. A diode-pumped, cw, multimode Nd:YAG laser (Lightwave Elec- Tunable IR-laser mapping of surface contaminants Component Surface Focal-Plane Array ! ! - $ 5 Projection lens - - . Overlapped beam segments \;-: , / \ Faceted lens ’/ \\ Rotating diffuser M1 PPLN Crystal Pump dump Figure 1. Diagram of the PPLN OPO and projection optics. tronics, Mountain View, CA) that is capable of generating at least 6 W of output power at a wavelength of 1064 nm is used as the OPO pump source. Two flat mir- rors (M3 and M4) and two curved mirrors (M1 and M2, 50-mm radius of curva- ture), all coated to be highly reflective at the signal and highly transmissive at the pump and idler wavelengths, form the bow-tie-shaped, single-wavelength reso- nant ring oscillator cavity designed to resonate the signal wave. An anti- reflection-coated lens, positioned between the pump laser and the OPO cavity, serves to image the Gaussian pump beam into the PPLN crystal. In this way, a beam waist (E-field radius) of 70 pm is created in the center of the crystal, which itself is centered between the two curved cavity mirrors. During normal operation, the OPO resonates on a single signal mode for minutes at a time, whereupon it hops to another cavity mode. The idler bandwidth is, however, determined by that of the pump beam, which is 10- 15 GHz. The use of the QPM material, PPLN, makes cw OPO operation more tunable and efficient than it would be for a conventional birefringently phasematched crystal. Simply stated, phasematching is a condition in which all of the interacting waves (i.e., signal, pump, and idler) maintain a specified relative phase relation- ship as they propagate through a nonlinear medium, and is a necessary condition for efficient nonlinear generation. In birefringent materials, phasematching is 6 D. Ottesen et al. achieved by careful selection and/or control of the crystal birefringence, tempera- ture, and beam propagation angles. In a QPM medium, phasematching is designed into the medium during the crystal growth process. Phasematching is achieved by causing the crystal to have a periodically inverting optical axis. The engineering process used to create these crystals increases conversion efficiency by allowing the use of much stronger nonlinear coefficients of the crystal, and frees the system from reliance on bire- fringence thereby increasing tunability. As the light beams cross the crystal-axis- inverting boundaries, any relative dephasing of the waves is corrected. For a crys- tal of a given periodicity, the rephasing is effective for a particular set of pump, signal, and idler frequencies. Some degree of tuning of these waves can be achieved within the crystal phasematching bandwidth (typically 10-20 cm-I). Broader tuning is achieved by accessing a portion of the same crystal having a different periodicity, or by changing the temperature of the crystal. In the present system, two 50-mm-long PPLN crystals (Crystal Technology, Palo Alto, CA) with an aperture of 11.5 mm x 0.5 mm are used as the active me- dium. Each crystal contains eight poled regions with different periodicities. One crystal's periodicities range from 28.5 to 29.9 pm, and of the other crystal from 30.0 to 3 1.2 pm. When operating at a crystal temperature of 148"C, these periods collectively allow tuning of the idler from 2720 to 3702 cm-'. The crystals are mounted in a stacked fashion within a temperature-stabilized copper oven that is attached to a vertical translation stage. Each crystal is tuned by selecting a period using the vertical motion of the stage; horizontal motion of the oven is used to se- lect between the two crystals. The raw output of the OPO contains the idler beam as well as portions of the signal and pump beams and some higher-order (red, green) beams created spuri- ously in the PPLN crystal. Spectral filtering is used to dump all but the idler beam. Prior to illumination of the sample, the idler is passed through a set of pro- jection optics, also shown in Figure 1. The first of these is a ZnSe diffuser (mean - roughness of 3-4 pm) that is mounted on a motor-driven spindle. The diffuser serves to reduce the phase coherence of the idler in order to minimize laser speckle noise in the transmitted beam and viewed by the IR camera in the light re- flected from the sample surface. The cone of radiation leaving the diffuser is col- lected by a ZnSe faceted lens (Laser Power Optics, Murrieta, CA). The faceted lens is formed to contain the equivalent of 16 6.4 mm facets and 16 partial facets around the edge of the lens on a 3.8 cm diameter with an effective f-number of 1.7. It operates as a prism array - the expanded beam is segmented into 32 differ- ent square beamlets that are subsequently overlapped at a distance of 5 cm from the surface of the lens. A ZnSe wire grid polarizer (not shown in Figure 1) is lo- cated at the overlap point, and serves to produce a p-polarized beam for the infra- red reflectance measurement. The square-shaped overlap region is then imaged onto the target using an f4.7, 8.4 cm focal-length ZnSe lens. As a unit, the system converts the Gaussian profile of the idler beam into a uniform square illumination on the sample surface. Tunable IR-laser mapping of surface contaminants 7 The infrared laser light is incident on the sample surface at an angle of 60" from the surface normal, and the specularly reflected component is detected by an InSb focal-plane array (FPA) camera with an infrared macro-lens assembly and an array size of 256 x 256 pixels. The FPA camera is located approximately 0.3 m from the sample surface, and the resulting field of view is 20 x 35 mm. FTIR instruments at both Sandia and NFESC were used to characterize the mid-infrared spectra of contaminated surfaces via optical interfaces for grazing- angle reflectance spectroscopy. The system at NFESC uses a commercially avail- able sampling accessory that permits a variable angle of incidence from 30 to 80", which is convenient for evaluating detection limits for contaminants on a variety of surfaces. The optical interface used by the Sandia National Laboratories FTIR instrument was constructed with a fixed 60" angle of incidence with optics exter- nal to the spectrometer. It also differs from the NFESC system in the large solid- angle used both to illuminate the surface and collect reflected light. This feature is particularly useful in the examination of rougher surfaces that cause significant scattering of the infrared beam, with a consequent degradation in both sig- nalinoise ratio and detection limits. Both systems use infrared polarizers to en- hance the sensitivity of the measurements by restricting the surface illumination to p-polarization [4]. Unless otherwise noted, all reflectance spectra presented in this paper are for p-polarized measurements. 2.2. Test sample preparation for calibration In order to evaluate the usefulness of the laser-imaging technique as a cleaning verification method, we prepared a number of test surfaces with well- characterized levels of contamination. These were used to determine detection limits as a function of contaminant species, level of contamination, degree of sur- face roughness, effect of spectral interference, and instrumental parameters such as angle-of-incidence. Seven candidate materials were chosen as contaminant species for evaluation as shown in Table 1. These materials have proven to be particularly difficult to remoke during cleaning operations, and are representative of many other organic contaminants encountered in government and industrial cleaning processes. Detailed measurements on the first four materials have been made in the course of this work and preliminary measurements have been made on the remaining three. A number of metals were chosen as substrates for the target contaminants, based on usage information obtained from military and contractor facilities. These were Aluminum-7075-T6, Titanium 6A1-4V, Steel Alloy 4340, Stainless Steel 304, and Magnesium AZ31B. The metals were fabricated into 3.8 x 12.7 cm flat coupons for laboratory testing and method demonstration. Six surface roughness finishes of the Aluminum 7075-T6 test coupons were obtained, ranging from 80 to 600 grit (600 grit being the smoothest). A profilome- ter instrument was used to examine the surface roughness profiles and provide average R, values. A R, value is an arithmetic average of the absolute deviations 8 D.O ttesen et al. Table 1. Contaminant materials used for preparation of test coupon for calibration Material Description Usage Drawing Agent White soft solid - ester grease Metal drawing, cutting, and lubricating agent Lubricant Brown liquid - paraffin hydrocarbons Rust preventative. cleaner, lubricant. protectant for metals Silicone Silicone Lubricant Mold Release 1 Green liquid - ethanol homopolymer Mold release agent Mold Release 2 Clear liquid - proprietary polymeric Mold release agent resins Solder Flux Yellow liquid - abietic acid or Soldering flux for electrical anhydride and electronic applications Hydraulic Oil Blue liquid - castor oil base Hydraulic systems, shock and strut lubricant MIL-H-5606A AM2 from the mean surface level, in millionths of an inch; therefore, a R, value of 1.5 = 0.00000015 inches (3.8 pm). Due to the nature of metal-shop finishing proc- esses, surface roughness values vary considerably across a given surface area. Finishing operations also result in a directional “grain” parallel to the sample coupons’ longitudinal direction. Surface roughness measurements, therefore, ex- hibit large variations between measurements taken along an orientation longitudi- nal or transverse to the polishing axis. Two surface roughness levels, 600 and 220 grit, were obtained for the other metal alloys. Prior to contaminant application, the aluminum alloy coupons were cleaned with acetone and underwent sonication with a clean-rinsing aqueous cleaner. They were then thoroughly rinsed in distilled water and dried in an oven at 50°C. Once cooled, they were weighed on a microbalance with a precision of 0.01 mg. Two or three weighings were averaged. Both drawing agent and lubricant contaminated AI-7075 coupons were pro- duced by two primary deposition methods - airbrushing and manual brushing. Several other techniques were attempted, including “wire-cator” drawing, coupon spinning, and “manual drop and spread.” These techniques were not used to pro- duce test samples for calibration for these particular contaminants due to the supe- rior results obtained from airbrushing and manual brushing. Three levels of draw- ing agent were applied by airbrushing to three A1 test coupons for each of the six surface finishes, creating a suite of 18 panels. Varying concentrations of drawing agent in water were prepared for the airbrush solutions. Similarly, four levels of lubricant were applied to four A1 test coupons for each of six surface finishes, creating a suite of 24 panels. Manual brushing was used for all but the least con- taminated samples, which were airbrushed. Lubricant solutions for both tech- Tunable IR-laser mapping ofsurface contaminants 9 niques were prepared using pentane as the solvent. Similar methods were used in preparing calibration samples with the mold release, solder flux, and hydraulic oil samples. All contaminated coupons were gently heated in an oven at 50°C for several days to remove both semi-volatile and volatile components. This served to stabi- lize the contaminants, allowing for quantification by weighing. Once the weights became stable, the coupons were cooled and weighed to determine the amount of contaminant present on the surface. When not being weighed or examined, the coupons were kept in a desiccator. 3. RESULTS AND DISCUSSION Grazing-angle incidence reflectance spectroscopy acts to enhance the detection sensitivity for thin layers of residue predominantly through improved coupling of the electric field intensity of the incident beam with the vibrating dipoles of the surface contaminant layer perpendicular to the metallic surface. Some additional enhancement of the infrared absorption spectrum will also occur due to a length- ening of the effective path length through the absorbing thin film layer [4-61. If the optical properties of both thin film and substrate are known (or can be de- termined), the reflection-absorption spectrum can be calculated as a fknction of film thickness and angle of incidence. This capability is particularly useful for in- terpreting experimental data and designing optical instrumentation. Computer codes written at Sandia [7] performed these calculations for a variety of materials. 3.1. FTIR measurements FTIR reflectance data for the full drawing-agent sample set were obtained at NFESC and Sandia using angles of incidence of 75 and 60" for average film thickness ranging from 0.1 to 1 ym, and aluminum substrates with surface finish ranging from 600 to 80 grit. Since the surface finishing operation produced a highly directional roughness, measurements were made both longitudinally and transversely with respect to the polishing grooves. R, values were determined at NFESC using profilometer measurements, and resulted in surface roughness val- ues of 0.3 to 1.5 pm for the longitudinal direction, and 0.5 to 6 pm for the trans- verse direction. The FTIR reflectance spectra were normalized using the uncoated back of a panel as a clean reference standard, and the intensity data are presented as either reflectance or -log reflectance in the following discussion. The C-H stretching vi- brations near 2900 cm-' proved to be generally useful in quantifying instrument response since these frequencies are well isolated from atmospheric interference due to water vapor and carbon dioxide. However, the baseline for these reflec- tance data was often non-linear. A simple single-point measurement of intensity was therefore not sufficient to determine the instrument response function. 10 D. Ottesen et al. Optical constants (n and k) were derived for the contaminant C-H stretching vibrations using the Sandia reflectance code and a dispersion model to calculate a fit to the experimental data for one of the test coupons [7].R eflectance-absorption spectra for the 2800-3000 cm-' range were calculated for 1-pm thick films of a specific hydrocarbon contaminant on an aluminum surface at either 60 or 75" an- gle of incidence. This function was then used as a linear variable in conjunction with a second-order polynomial to produce a least-squares fit of the experimental reflectance data for the test coupons. An example is shown in Figure 2 for the longitudinal measurements of three thicknesses of drawing-agent contaminant at 0.40 4 - 0 . 9F~ilm 0.35 0.30 0.25 - 0 - --- 0.4 pn Film C Least-squares Fit Q 0.301 U 0 -0.65 - - --- 0.1 pm Film 0 Least-squares Fit -0.702 -C0I al -0.75: IC 8 n U m -0.80: - 0 I -0.85 r 8 n I o n 8 1 8 m 8 8 I rn 8 I Figure 2. Linear least-squares fit of experimental reflectance data for drawing-agent contaminant on 600 grit polished aluminum surfaces. Average film thickness: (Top) 0.9 pm. (Middle) 0.4 pm. (Bot- tom) 0.1 pm.