[ E NOTES

An Improved Method of Using the Diamond Anvil Cell For Infrared Microprobe Analysis

D. LIN-VIEN,* B. J. BLAND, and V. J . S P E N C E Molecular Spectroscopy Department, Analytical Chemistry R&D, Shell Development Company, P.O. Box 1380, Houston, Texas 77251

Index Headings: Analysis for paint; Infrared; IR microspectroscopy; Diamond anvil.

INTRODUCTION

Diamond anvil cells are widely used in industrial IR applications. 1,2,~ They provide a convenient way for flat- tening and thinning irregularly shaped samples such as fibers, polymer gels, and inorganic particles. When used in conjunction with an IR-microprobe, they also serve as sample supports on the microscope stage. The typical procedure of using a diamond anvil accessory can be divided into three steps: (1) Press the samples between two diamond surfaces so that they are suitable for spec- troscopic study. In this step the parallelism of the dia- mond surfaces should be checked to avoid damaging the diamond cell2 (2) Obtain the IR spectrum of the flat- tened sample as a "sandwich" between the two diamond plates. (3) Perform a spectral subtraction to remove the background due to the diamond absorptions.

SOME CONSIDERATIONS W H E N USING D I A M O N D ANVIL CELLS

The Interference Fringes. In the existing method of using diamond anvil cells, the IR measurements are done with the flattened sample remaining "sandwiched" be- tween the two diamond surfaces of the diamond anvil cell. The diamond background for the spectral subtrac- tion is measured with the use of a complete cell with no sample in it. Unfortunately, the high degree of parallel- ism between the diamond surfaces often gives rise to

Received 16 February 1990. * Author to whom correspondence should be sent.

interference fringes in IR spectra, especially when the cell is empty (Fig. la). The presence of the interference fringes can cause difficulty in spectral interpretation and increase the uncertainty in semi-quantitative analyses, as the baseline is likely to be ill defined. These problems are particularly pronounced when the analyte bands are weak. Although one can reduce the interference fringes by carefully aligning the two diamond surfaces in such a way that the number of fringes is minimized, the pro- cess is usually very tedious and time consuming.

The Diamond Absorptions. Although diamond is con- sidered quite transparent to IR radiation, the diamond background in the IR spectrum of a diamond anvil cell is unavoidable due to the thickness of the diamond through which the IR beam travels. As illustrated by Fig. la, the major IR bands of a miniature diamond anvil cell (High Pressure Diamond Optics, Inc., IIA diamond) are located in the region of 2300 to 1900 cm -1, where the IR absorptions of triple bonds and cumulative double bonds are expected. In principle, the diamond bands can be removed by subtracting the diamond background out of the sample spectrum since very little chemical interac- tion is expected between the diamond and the sample. In practice, a previous study suggests that a maximum absorbance for accurate difference spectrometry is about 0.7 absorbance units ( a . u . ) . 4'5 An examination of the IR spectrum of a complete cell (Fig. la) reveals that the IR absorptions of the complete cell background in the 2300- 1900 cm -1 region are well above the suggested limit. The spectral subtraction of the diamond background, there- fore, does not yield the most reliable information in this region when the "sandwich" configuration is chosen for sample support. Consequently, it is almost impossible to observe the triple bonds and cumulative double bonds such as nitrile and isocyanates by the IR technique with the use of a complete diamond anvil cell as sample sup- port for IR measurements.

MODIFIED M E T H O D

Our experiences indicate that most samples remain attached to one of the diamond surfaces after the dia- mond anvil cell is disassembled, following step one. In addition, most samples stay thin and flat even though they are no longer held in the "sandwich" configuration. It is thus logical to use only one half of the diamond anvil cell to support the samples for the IR measure- ments in step two. The advantage gained by the modi- fication from the use of a complete cell to a half-cell as sample support is twofold. First, the interference fringes

Volume 44, Number 7, 1990 0003-7028/90/440%122752.00/0 APPLIED SPECTROSCOPY 1227 © 1990 Society for Applied Spectroscopy

D i a m o n d A n v i l C e 1

$ ,

. s ompl ete cel

( b ) h a l f c e l l background o 4 0 0 0 3 0 0 0 2 0 0 0 1 0 0 0

W u m b ( o m -- 1 ) ~ v e ~ e ~ g

FIG. 1. The diamond backgrounds of (a) a complete cell and (b) a half-cell. (Digilab FTS60 FT-IR spectrometer + Digilab UMA-300 IR- microscope, 256 scans, 4 cm 1 resolution, triangular apodization func- tion, 4 × zero-filling.)

caused by the d i amond surfaces are e l iminated, as clearly d e m o n s t r a t e d by compar i son of Fig. l a and Fig. lb . Sec- ond, the I R intensi t ies of the d i amond background in the 2300-1900 cm -1 region are reduced to a level near the r e c o m m e n d e d l imit (0.7 a.u.) for a more accura te spect ra l subtract ion. I t now becomes possible to use the d i amond anvil accessory in some cases to analyze samples containing tr iple bonds and cumula t ive double bonds. Never theless , because of the degraded S /N rat io in the 2300-1900 cm -1 region due to the d i amond absorpt ions , K B r pla tes (or o ther I R t r ansmi t t i ng mater ia l ) are still the p re fe r red choice for suppor t ing samples when the intensi t ies of the ana ly te bands in this region are weak. In addit ion, caut ion should be exercised to ensure the cleanness of the d i amond anvil cell to control cross-con- t amina t i on f rom sample to sample . Preferably , the dia- m o n d background should be measured with the same "hal f -ce l l" which is used to suppor t the sample.

Figure 2 compares the I R spec t ra of an unknown pa in t chip ob ta ined with the " sandwich" configurat ion (Fig. 2a and 2b) and the IR spec t rum of the same pa in t chip acqui red with the modif ied half-cell me thod (Fig. 2c). T o obta in Fig. 2c, we sub t rac ted the spec t rum of the f la t tened pa in t chip suppor t ed on a half-cell by a half- cell d i amond background. No interference fringe is ob- served in Fig. 2c. T h e improved spect ra l sub t rac t ion ac- curacy due to the use of the half-cell me thod allows the observa t ion of the IR bands a t 2192 and 2156 cm -~, in- dicat ive of the presence of d icyandiamid. Figure 2a and 2b were genera ted by subt rac t ing a spec t rum of a "sand- wiched" sample with a comple te cell background and a half-cell background, respectively. Li t t le useful spectra l in fo rmat ion is ob ta ined in Fig. 2a and 2b in the 2300- 1900 cm -1 region because of the intensi t ies of the dia- m o n d absorp t ions in the " sandwich" configuration. In- terest ingly, the in ter ference fringes which are quite ob- vious in the region above 2000 cm -~ in Fig. 2a are less

0 - 4 0 0 0 : 3 0 0 0 2 0 0 0 1 o o o

T i ' a v e n u m b e r | ( o r e - - l )

FIG. 2. The IR spectra of an unknown paint chip: (a) "sandwich" configuration, complete cell diamond background; (b) "sandwich" con- figuration, half-cell diamond background; (c) half-cell sample support, half-cell diamond background. (Digilab FTS60 FT-IR spectrometer + Digilab UMA-300 IR-microscope, 256 scans, 4 cm -~ resolution, trian- gular apodization function, 4 x zero-filling.)

a p p a r e n t in Fig. 2b. Th i s observa t ion suggests t h a t the in ter ference fringes in Fig. 2a are in t roduced via the spect ra l sub t rac t ion process f rom the d i a m o n d back- ground of an e m p t y comple te cell, which is more ame- nable to the fo rmat ion of in ter ference fr inges t h a n a comple te cell wi th a " sandwiched" sample due to the large difference be tween the refract ive indices of dia- m o n d and air. Th is resul t indicates tha t , even if the na ture of the sample prohibi ts the use of a half-cell as the sample suppor t and the " sandwich" configurat ion is called for, the in ter ference fringes or iginated f rom a com- plete cell d i amond background can still be avoided by the use of a half-cell d i amond background for spect ra l subtract ion.

In conclusion, the appl ica t ions of d i amond anvil cells for IR -mic rop robe analysis can be improved by using only one half of the d i amond cell as the sample suppor t in the s tep of I R measuremen t s . T h e modif ied m e t h o d e l iminates the in ter ference fringes caused by the full cell configurat ion and improves the spect ra l sub t rac t ion ac- curacy in the d i amond absorp t ion region.

1. J. E. Katon, P. L. Lang, D. W. Schiering, and J. F. O'Keefe, "In- strumental and Sampling Factors in Infrared Microspectroscopy," in The Design, Sample Handling, and Applications of Infrared Microscopes, P. B. Roush, Ed., ASTM STP 949 (American Society for Testing and Materials, Philadelphia, 1987), p. 49.

2. N. R. Smyrl, R. L. Howell, D. M. Hembree, Jr., and J. C. Oswald, "Industrial Problem Solving Using Microvibrational Spectroscopy," in Infrared Microspectroscopy, R. G. Messerschmidt and M. A. Harthcock, Eds. (Marcel Dekker, New York/Basel, 1988), p. 211.

3. D. W. Schiering, "Infrared Microspectroscopic Solutions to Man- ufacturing and Quality Control Problems," in Infrared Microspec- troscopy, R. G. Messerschmidt and M. A. Harthcock, Eds. (Marcel Dekker, New York/Basel, 1988), p. 229.

4. P. R. Griffiths and J. A. de Haseth, Fourier Transform Infrared Spectrometry (John Wiley & Sons, New York, 1986), p. 343.

5. R. J. Anderson and P. R. Griffiths, Anal. Chem. 47, 2339 (1975).

1228 Volume 44, Number 7, 1990