1997
Anal. Chem. 1981, 5 3 , 1997-2000
Table V. Slope, Intercept, and Correlation Coefficient Values on p-Bondapak NH, compound phenol p-cresol
interslope cept -1.36 -0.22 -1.44 -0.32
corr’elation coeff 0.994
0.992
5,6,7,8-tetrahydro-2-naphthol-1.54 -0.40 0.994 -1.53 -0.053 0.993 2-naphthol
solvent. With an equation for each phenolic type it should be theoretically possible to optimize the chromatographic system for the best separation of a mixture of the compounds and other classes of compounds that behave chromato graphically similar to the compounds studied. Also, eq 7,8, and 9 should be useful in gradient elution work. These aspect8 and work with other bonded phase columns are presently under investigation.
LITERATURE CITED
where m is the slope from the straight line obtained by plotting log k’vs. log X, (Table I). Linear relationships were also observed for graphs of log of corrected retention volume (V? vs. log (% w/v) and thus the following equation can be used to predict retention volumes: log V’ = intercept - m 1log (% w/v) (9)
Snyder, L. R.; Kirkland, J. J. “Introduction to Modern Liquid Chromatography”, 2nd ed.; Why-Interscience: New York, 1979. Snyder, L. R. Anal. Chem. 1974, 46, 1364. Snyder, L. R. “Principles of Adsorption Chromatography”; Marcel Dekker: New York, 1968. Snyder, L. R.; Poppe, H. J. Chromatogr. 1980, 184, 363. Scott, R. P. W.; Kucera, P. J. Chromatogr. 1975, 712, 425. Scott, R. P. W. J. Chromatogr. 1976, 122, 35. Scott, R. P. W.; Kucera, P. J. Chromatogr. 1978, 749, 93. Scott, R. P. W.; Kucera, P. J. Chromatogr. 1979, 771, 37. Scott, R. P. W.; Traiman, S. J . Chromatogr. 1980, 196, 193. Scott, R. P. W. J. Chfomatogr. Sci. 1980, 78, 297. Soczewinski, E. Anal. Chem. 1969, 41, 179. Golkiewicz, W.; Soczewinski, E. Chromatographla1972, 5 , 594. Soczewinski, E.;Goikiewlcz, W. Chromatographla1973, 6, 269. Soczewinski, E. J . Chromatogr. 1977, 130, 23. Slaats, E. H.; Kraak, *J. C.; Brugman, W. J. T.; Poppe, H. J. Chromatogr. 1978, 149, 2551. Narkiewicz. J.: Jaroniec, M.; Borowko. M.: Patwkleiew. . . A. J. Chromatogr. 1978; 157, 1. Jarnolec, M.; Rozylo, J. K.; Osclk-Mendyk, B. J . Chromatogr. 1979, 179, 237. Jaronlec, M.; Rozylo, J . K.; Jaroniec, J. A.; Oscik-Mendyk, 6. J . Chromatogr. 1980, 188, 27. Jaroniec, M.; Piotrowska, J. HRC CC, J , High Resolut. Chromatogr. Chromatogr. Commun. 1980, 3 , 257. Hara, S.; Hlrasawa, 1% Miyamoto, S.; Ohsawa, A. J . Chromatogr. 1979, 169, 117. Sleight, R. 6. Chromatographia 1973, 6 , 3. Jandera, P.; Churacek, J. J . Chromatogr. 1974, 91, 207. Schabron, J. F.; Hurtubise, R. J.; Sliver, H. F. Anal. Chem. 1978, 50, 1911. Schabron, J. F.; Hurtublse, R. J.; Sliver, H. F. Anal. Chem. 1979, 51, 1428. Callmer, K.; Edhoim, 1.. E.; Smith, 6. E. F. J. Chfomatogr. 1977, 136, 45. Cotter, R. L., Waters Associates Inc., personal communication, 1981. Pearce, P. J.; Simkinf;, R. J. J. Can. J . Chem. 1968, 46, 241. Socrates, G, Trans. Faraday SOC. 1969, 66, 1052. Vogel, W.; Andrussow, K. ”Dissociation Constants of Organic Aclds in Aqueous Solutions“; t3utterworth: London, 1961.
Equation 9 is somewhat easier to use than eq 8 because fewer calculations are involved. Equation 7 can also be usled to predict corrected retention volumes; however, a linear relationship was not obtained with ethyl acetate as the strong
RECEIVED for review Nay 13, 1981. Accepted July 3, 1981. Financial support was provided by the Department of Energy, Washington, DC, Contract DE-AC22-79ET14874.
solvent compared to 2-propanol (Tables I and IV), KM should be smaller and thus log [ K m X M / K M ]should be larger. Also K a should be larger for a given phenol with the ethyl acetate compared with the %propanol and thus increase the intercept value. Graphs of the reciprocal of corrected retention volume vs. % w/v of ethyl acetate did not give straight line relationships for the compounds in Table IV,The lines were curved upward suggesting mixed interactions with the stationary phase. The data presented indicate that the models of Snyder, Soczewinski, and Scott approximate the HPLC bonded ]phase systems studied, and additional work is needed to produce a more accurate model. The work of Slaats et al. (15) should be considered in which they determined activity coefficients of solutes and moderators in the mobile phase. However, as discussed below, the linear relationships obtained have considerable utility in evaluating HPLC results. Analytical Considerations. Capacity factors for phenolic type compounds can be predicted readily by using the equation log k’ = intercept - m log X, (8)
Determination of Phosphates by Liquid Chromatography with Inductively Coupled Argon Plasma Atomic Emission Spectrometric Detection Masatoshi Morita” and Takashi Uehiro National Institute for Environmental Studies, 16-2 Onoga wa, Yatabe, Tsukuba, Ibaragi, 305 Japan
Speclation and quantitative analysis of orthophosphate, dlphosphate, and triphosphate are performed by using highperformance llquld chromatography (HPLC) with inductively coupled argon plasma emlsslon spectrometrlc detection (ICP). HPLC is used to separate mixtures of phosphates on an anion exchange column using tartrate magnesium buffer. The ICP is used as a selective detector by observing P I1 emlsslons at 214.9 nm. A detection limlt is 0.5, 1, and :3 pg, respectively, for ortho-, di-, and triphosphate, respectively. Nucleotides such as ATP, ADP, and ATP are also analyzed.
Phosphorus appears in life as various types of compounds: calcium phosphate for skelton, phospholipid and phosphoprotein for material transfer regulation, glycerophosphate for energy transfer, DNA and RNA for storage and transfer of genetic information. I?hosphates are also important as industrial materials such as insecticides,plasticizers, and flame retardants. In the study of biochemistry and environmental chemistry, it is of significant importance to characterize trace amounts of phosphorus compounds. One of the approaches can be a chromatographic charac-
0003-2700/61/0353-1997$01.25/00 1981 Americah Chemical Sooiety
1998
ANALYTICAL CHEMISTRY, VOL. 53, NO. 13. NOVEMBER I981 I t -
Table I. Sensitivity ( S i / N o i s e Ratio)at Different Observation Height and Rf Power' If power,
kW
S/N at ohservn height 13mm 15mm 17 mm 19mm
1.1b
7
1.3*
7 5 6
1.5=
6 7
5 7
3 4
5 5 Signal to noise ratio was calculated for 0.2 x 10.' M orthophosphate solution. Sample gas and coolant gas flow rate was 0.5 and 18 Llmin, respectively. Plasma gas flow rate was 0 Llmin. e Sample gas flow rate was 0.5 L/min. 1.P
Flgm 1. Hiah sensitMty area for P detection. Signal to raise ratio is shown In the plasma when 0.2 X to-' M phosphate was pumped imo at Uw rete of 1.0 mllrdn. Omer parameters are 1.1 kW ti power and coolant, plasma, and sample gas Rows being t8, 0. 0.5 Llmin.
respectively. terization. Orthophosphate and its oligomers and cyclic phosphates were separated by ion exchange or gel permeation chromatography (1-5). Many papers have described the separations of nucleotide by anion exchange high-pwformance liquid Chromatography (617). Fewer articles have referred to the separation of phosphate insecticides or phospholipids (18-24).
Because high-performanceliquid h m a t o g r a p h y (HPLC) has high resolution for various types of Phosphate compounds, it should be a good method for trace characterization of phosphorus if it is connected to a phosphorus-specific detector. Spectroscopic methods including emission and absorption spectrometry can detect phosphorus selectively. Julin et al. used flame emission detector for a selective detection of phosphorus and sulfur in HPLC (4). The aensitivity was seriously affected by the chemical form of phosphorus. Yoza et al. indirectly detected diphosphate and triphosphate in the chromatographic elutant by monitoring complexed magnesium with atomic absorption spectrometry (5). Recently developed inductively coupled argon plasma (ICP) and direct current argon plasma (DCP) emission spectrometry have advantage over the above two methods. It is highly sensitive and independent of the chemical form of phosphorus. The present paper describes the availability of ICP for HPLC detector. EXPERIMENTAL S m I O N Reagent. Potassium dihydmgem phosphate (Wako SR grade), tetrasodium diphosphate decahydrate (Wako SR grade) and pentasodium triphosphate (Wako P grade) were obtained from Wako Chemical Co. Nucleotides were obtained from S i The reagents were used without further pdication Other ehemieals were commercially available highest grade materials. HPLC-ICP. A Waters Associate liquid chromatograph was connected to anion exchange columns. HPLC column packmgs examined here were p-Bondapak-NH,, Nagel SA. The outlet of liquid chromatograph was connected to cross flow nebulizer of the ICP spectrometer (JarreU-Ash-Atom-Comp750) with Teflon in. X 5 in.). Elutant flow rate was set at 1mL/min. tubing Phosphorus was monitored at 214.9 nm (second order). The analog signal was taken out though the profde mode of a computer program. In order to filter noise, we installed a 50-kQresistor and 20 pF condenser before the recorder giving an approximate time constant of 1 s. RESULT AND DISCUSSION ICP Optimization. Optical and plasma parameters may be essential for the high sensitive detection of phosphorus. Figure 1illustrates the obaervation paitions relationship with
7 7
6 6
Table II. ICP Responae Factor for Various Phosphorus Compounds response compounds factor KH,PO, Na,P20,.10H,0 Na,P,O,, (NaP03, C,,H,,N,O,PNa.6HZO C>O~32N,O,OP, C,.H,,N,O,,P,Na,.3H,O
1.00 1.01 1.00 1.02 0.98 1.03 1.02
punty 1.00 1.02 1.00 0.93 1.14 0.98 0.97
sensitivity. High sensitivity was obtained in the middle area of the plasma. Table I shows the signal/noise ratio at various radiofrequency power levels and at various observation heights in the plasma. At the eane time, however, background noise level also increased. Thus there is an optimal set of operational parameters for the highest sensitivity. The highest S/N ratio for the 1.0 mL/min flow of 0.2 X 1W' M phosphate was 7. The highest S/Nratio is obtained at several sets of parameters. We used the following operational parameters in the experiments thereafter: rf power 1.3 kW, coolant gas 18 L/min, sample gas 0.5 L/min, plasma gas 0, observation height 15 mm. Response Factors for Different Phosphorus Compounds. Efficient use of the ICP in the quantitative analysis of different phosphorus compounds is facilitated when the sensitivity of the ICP is independent of the molecular form of a phosphorus compound. This was examined for several phosphorus compounds. Aqueous solutions of several phosphorus compounds were analyzed by the ICP, by observing P line. The phosphorus content was verified by ICP after converting to orthophosphate by wet digestion. Digestion was made by heating with concentrated nitric acid for 30 min. Response factors were calculated as the detector response per gram of phosphorus present, relative to potassium dihydmgen phosphate (Table 11). The results in Table I1 indicate that the detector response is substantially independent on the chemical form of the phosphorus compounds. On the other hand, it was noticed that reagent purity could give serious error for some phosphates. Metaphosphate gave smaller P content and AMP employed here showed higher P content than expected from the chemical formula For the other five phosphates, reagent impurity seemed negligible. Separation of Orthophosphate, Diphosphate, and Triphosphate. Several solvent systems were checked for the separation of these three phosphates. Although p-Bondapak-NH, is a weak ion exchanger, high ionic strength was necessary to elute triphosphate from the column. Furthermore gradient elution is necessary to see three phosphate peaks in a single chromatogram of reasonable analysis time. The high salt concentration of the eluant and the gradient elution are not favorable for ICP spectrometry.
ANALYTICAL CHEMISTRY, VOL. 53, NO. 13, NOVEMBER 1981
Chromatogram of orthophosphate, diphosphate, and triphosphate. Column is p-Bondapak-NH, and the eluant was oxalate-Mg buffer at the flow rate of 1.0 mL/min, 50 pL of mixed solution which contains M each phosphate.
1999
Flgure 2.
Table 111. HPLC Variables column fi-Bondapak-NH, Nucleosil N(CH,), eluant
Y----"(PO3)"
-2
Figure 3. Chromatogram of phosphate, diphosphate, triphosphate,and metaphosphate. Eluant was Tris-sulfate buffer (pH 8.2). I/,,
in. X 1 ft in. X 1 ft
1
1 0 - 3 ~50pl ~
I
tartaric acid (0.1 M)t MgSO, (0.01 M)t NaOH (0.1 M )
buffers oxalic acid (0.1 M)t MgSO, (0.01 M) t NaOH (0.1M)
ammoniumformate (0.2 M) + Mg80,. (0.01 M) + . ~
HCl(O.1 M) acetic acid (0.2 M)t NaOH (0.1 M ) tris(hydroxymethy1)amine (0.3 M)+ H,SO4 ( 0 . 0 7 5 M)
High concentration salt solutions tend to clog both torch and nebulizer. The present system cannot torelate solutions containing more than 3% solute for 8 h of operation. 'The gradient elution causes a base line drift due to physical and spectral interferences. To overcome this difficulty, we added the appropriate amount of magnesium ion to the eluant. 'When the pll is around 2, the major ionic form of phosphate, diphosphate, and triphosphate are HzPO,, H3P207-+ H2P20T2-, (and H3P301:- + H2P301$-, respectively. Mg2+added at the llevel of 0.01 M forms complexes with diphosphate and triphosphate and neutralizes their negative charges, resulting in a simple elution from the column with rather low concentration buffer. Figure 2 shows the chromatogram of phosphate, diphosphate, and triphosphate with oxalate buffer elutant containing magnesium ion. Without addition of 0.01 M Mg, elution was not observed. Table I11 shows the columns and eluants which are examined for good separation. Among the eluants examined, oxalate Mg gave the best resolution, tartarate + Mg gave the second best, but formate Mg and acetate + Mg gave poor resolution. Oxalate Mg buffer should be freshly prepared as it is subject to decomposition. The column should be rinsed after usage. Calibration curves prepared by injecting an aliquot of different concentrations showed a linear relationship with a dynamic range of 104-10-2 M. The practical detection limit was estimated from the chromatogram taking the peak height to noise level ratio at 2. It was 0.5 pg of P, 1 pg of P, and 3 pg of P for mono-, di-, and triphosphate, or 3 X g P/S. With the use of basic eluant, such as ammonium or Tris buffer, it was noticed that triphosphate was rapidly decomposed to orthophosphate and diphosphate during chromatography. Figure 3 shows the chromatogram of orthophosphate, diphosphate, triphosphate, and metaphosphate using Tris buffer as an eluant. One may reasonably expect triphosphate peak to appear later than diphosphate. However, no peak appeared in such a later retention time range and, instead, two peaks which had the same retention time as orthophosphate and diphosphate appeared. The area of the first and the second peak had the ratio of 1:2 indicating the mole ratio of 1:l. This suggests that the third P-0 bond of triphosphate is readily cleaved at the column head giving 1 mol of phosphate and 1mol of diphosphate. Diphosphate is
+
+
+
Column head conlcentration technique for low concentration phosphate analysis. Flgure 4.
Table IV. Elution Recovery of Orthophosphate from HPLC Column at Different Sample Concentrations emission intens recovery, concn, M integrated % 10-3
114
10-4
125
100 109
10-5 10-6
108 100
87
95
~
stable enough to give a single peak during chromatography. When triphosphate aqueous solution was put at room temperature for 1week, significant deterioration was observed. Basic condition and the presence of p-Bondapak-NH2might accelerate decomposition rate. Analysis of a Low Concentration Sample. When the ionic strength of the water sample is not high, phosphate can be trapped at the column top and be eluted with eluant buffer. By use of this concentration technique, dilute solutions can be analyzed. Orthophosphate aqueous solution (lo4, and lo4 M) was pumped into the column at the rate of 1.5, 1.5, and 3.0 mL/min, respectively. Since too much time is required to load the water sample, a faster loading rate is chosen for the most dilute sample (IO4 M). The chromatogram is shown in Figure 4. Recovery from the column is listed in Table IV. A significant base line depression is noted before the phosphate peak for and lo* M samples. It is corresponding to the loading t h e of dilute phosphate sample. Base shift is brought to by the emission of C and Mg in eluant buffer. A fairly high recovery was obtained for lo4 M phosphate solution but the peak width of the chromatogram broadened, suggesting that lo4 M is the limit of the present method. For further analysis to the lower concentrations, another preconcentration method is necessary. Separation of Adenosine Phosphates. The given experimental conditions using p-Bondapak-NH2and tartarate + Mg2+buffer is also applicable on the analysis of nucleotide. Figure 5 shows a separation of adenosine phosphate, diphosphate, and triphosphate. These three adenosine phosphates were well separated with these columns. However,
2000
Anal. Chem. 1981, 53,2000-2003 AMP
i
ATP
Figure 5. Separation of adenosine monophosphSte, dlphosphate, and triphosphate, with tartarate-Mg buffer.
adenosine cyclic 2,3-monophosphate and adenosine cyclic 3,5-monophosphate showed almost same retention time as adenosine phosphate. For the separation of AMP, cyclic AMPs, and orthophosphate, an eluant with weaker ionic strength is necessary.
LITERATURE CITED (I) Ohashi, S.;Tsuji, N.; Ueno, Y.; Takeshita, M.; Muto, M. J. Chromatogr. 1970, 50, 349. (2) Ueno, Y.; Yoza, N.; Ohashi, S. J. Chromatogr. 1970, 52, 481. (3) Kura, Genichiro; Ohashi, Shigeru J . Chromatogr. 1971, 56,111. (4) Julin, B. G.; Vandenborn, H. W.; Kirkland. J. J. J. Chromatow. - 1975,
(6) Virkola, P. J. Chromatogr. 1970. 51, 195. (7) Brook, A. J. W. J. Chromatogr. 1970, 47, 100. (8) Murakami, F.; Rokushika, S.; Hatano, H. J . Chromatogr. 1970. 53, 584. (9) Kirkland, J. J. J . Chromatogr. Scl. 1970, 8, 75. (10) Brown, P. J. Chromatogr. 1970, 52,257. (11) Pennington, S. N. Anal. Chem. 1971, 43, 1701. (12) Drobishev, V. I.; Mansurova, S. E.; Kulaev, I. S. J. Chromatogr. 1972, 69,317. (13) Shmukier, H. W. J. Chromatogr. Scl. 1972, IO, 139. (14) Gabrlei, R. F.; Michalewsky, J. J . Chromatogr. 1972, 67, 309. (15) Henry, R. A.; Schmk, J. A.; Williams, R. C. J. Chromatogr. Sci. 1973, 7 1 , 360. (16) Baker, D. R.; Williams, R. C.; Steichen, J. C. J. Chromatogr. Sci. 1974, 12,501. (17) Kreis, W.; Greenspan, A.; Woodcock, T.; Oordon, C. J. Chromatogr. Scl. 1976, 14,331. (18) Anderson, F. S;Murphy, R. C. J. Chrowtogr. 1976, 721,251. (19) Henry, R. A.; Schmit, J. A.; Dieckman, J. F. Anal. Chem. 1971. 43, 1053. (20) Frei, R. W.; Lawrence, J. F. J . Chromatogr. 1973, 83, 321. (21) Seiber, J. N. J. Chmmatogr. 1974, 94,151. (22) Lawrence, F.; Renauk, C.; Frei, R. W. J . Chromatogr. 1976, 127, 345. (23) Erdahl, W. L.; Stolyhwo, A.; Privett, 0. S. J . Am. 011 Chem. SOC. 1973, 50,513. (24) Jungalwala, F. B.; Evans, J. E.; McCluer, R. H. Blochem. J . 1976,
755, 55.
112,443.
(5) Yoza. N.; Kouchiyama, K.; Miyajima, T.; Ohashi, S. Anal. Lett. 1975, 8, 641.
RECEIVED for review March 9,1981. Accepted July 23,1981.
Ion-Exchange Chromatographic Separation of A/-Nitrosodiethanolamine in Cosmetics Yoji Fukuda,* Yoshlhlro Morlkawa, and Isao Matsumoto Shiseido Laboratories, 1050 Nippa-Cho, Kohoku-Ku, Yokohama, Japan 223
A cleanup method is presented for the analysis of N-nitrosodiethanolamine (NDELA) as a contaminant in cosmetic products and their ingredlents. NDELA was found to be speclficaily adsorbed on a strongly bask anion exchange resin (OH-Form) in 80 % ethanol and quantitatlveiy recovered by elution wlth 10% acetic acid/ethanol. Samples were examined in the presence of various ionic and nonlonlc polar compounds used in common cosmetic formulations, and recoverles were found to be satisfactory (80-100%). The detectlon llmit of NDELA obtained wlth a Thermal Energy Analyzer (TEA, registered trademark of Thermo Electron Corp., Waltham, MA) attached to a high-performance liquid chromatograph (HPLC) was 1 X IO-' g (1 ng). Detection limits at the 10 parts-per-blillon (ppb) level were obtained for NDELA in cosmetic products.
With regard to the detection and determination of nitrosoamines in our environment, a selective and highly sensitive method has been developed by Fine et al. (1, 2) using a chemiluminescence detector called the Thermal Energy Analyzer (registered trademark of Thermo Electron Corp.). However, to improve the accuracy of nitrosoamine determination in complicated matrices, this detector must be used in conjunction with a procedure to isolate nitrosoamines from samples. The determination of nitrosamine in cosmetic products is difficult, and for this reason cosmetic products were analyzed.
A cleanup method used for the first time for the analysis of NDELA in cosmetic products was an adsorption chromatography using silica gel (3). However, this method is known to be affected by the type of cosmetic products and their components, resulting in poor recovery and reproducibility. Mitchell and Rahn (4), Rosenberg and co-workers (5),and Fellion and co-workers (6)have reported HPLC methods using a UV detector, instead of the TEA. Ion-exchange chromatography has been used for the separation of various ionics, or mixtures of ionic and nonionic compounds such as surface active agents (7), and also applied to cosmetic analysis (8). At the beginning of this study, the authors thought that NDELA might be adsorbed by a cation exchanger because of its structure, but NDELA was found not to be adsorbed. Later, it was found that NDELA was specifically adsorbed on a strongly basic anion exchange resin (OH form) and the NDELA was quantitatively recovered by elution with an acetic acid/ethanol solution. This means that most of ionic and nonionic components in cosmetic products can be removed by this method. This paper describes the conditions of ion-exchange chromatography required for the separation of NDELA from cosmetic products and the ingredients to be analyzed. Recoveries from various compounds commonly used in cosmetic formulations are also discussed. EXPERIMENTAL SECTION Materials. NDELA. Reagent NDELA (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan) was purified as follows; Approximately 0.5 g of NDELA was dissolved in 10 mL of 1-propanol
0003-2700/81/0353-2000$01.25/00 1981 American Chemical Society