“re abekt3) Butea Nee NaS ‘echnical Report No. 9 Review of Commercially Available Particulate Measurement Systems Journal of Parenteral Science and Technology eens tayReview of Commercially Available Particulate Measurement Systems Preface In 1986, the Parenteral Drug Association Research Committee solicited proposals for a grant concerning an engineering related analysis of existing liquid-borne particle measuring equipment, Mr. Julius Knapp (R&D Engineering) and Dr. Patrick DeLuca (University of Kentucky) were co-recipients of the grant. The PDA has elected to publish the completed report of Mr. Knapp and Dr. DeLuca to serve as a possible basis for further discussion among the parenteral technology industry, and to perhaps provide the Research Committee with future constructive research project ideas related to particulate technology. ‘The contents of Mr. Knapp’s and Dr. DeLuca’s report contains technical statements | and conclusions attained by the authors independent of the PDA Association and are not intended to reflect technical positions or interests of the PDA Research Committe, | Board of Directors, and/or the Association at large. Michael S. Korezynski, Ph.D. Chairman, PDA Research Committee August 1987 Vol. 42, Supplement 1988 83Contents Part I—Instrumentation 1. Introduetion ss HI, Common Factors... : Ss III, Optical Systems es . $7 A. Filmy Materials feceeeee S8 B. Refractive Index Effects : $8 C. Conclusions : coo S8 IV. Resistance Modulation Systems (Coulter, etc.) BenBe ap a5H0a50008 so A. Shape Sensitivity ceoset teste s9 B. Filmy and Porous Materials : S10 C. Sample Handling : S10 D. Conclusions Sil V. Present Detection Systems er eos SH ‘A. Comparison poebesnesnsososon6s sul B. Conclusions siz VI. Recent FDA Test Results eects see SIZ VII. Recommended Actions : s13 A. Stop Gap S13 B, Near Term ceo SB VITI, Longer Term Actions : cece si3 IX. Future Perspectives : sects S13 A. All Systems o : cecseeeee si3 B, Coulter System . : see SI C. Optical Detection Systems : : see SB X. Notes cists . : . Sid Appendix I: Commercially Available Particulate Measurement Systems for Parenteral Use Descriptions and Comments si4 Appendix Il: Coincidence Errors in Particulate Counters... . : sis Appendix III: “Ideal” Particulate Measurement S) Si8 Part II—A Selected Annotated Bibliography on Particulate Matter 1. Particle Counting (Size Analysis) . 7 : S19 A. General ppsossupsbeosoueGba cece 819 B. Large Volume Solutions cee : ceveeee $20 €. Small Volume Solutions ceo = S20 D. Injectable Powders seeseeeeee S21 E, Release from Bags sects SD F, Release from Syringes and Needles teeceessseenste SB G. Release from Administration Sets Beco S22 H. Release from Containers... Sospaceosse 2 1, Factors Influencing Particulate Matter ceo 82 1, Methods Evaluation : ere $23 A. General 5 eeerrenen sees SB B. Microscope een : ves $B C. Resistance : cece S24 D. Light Blockage .. - : S25 E. Light Scattering : : : : S25 F, Others... = : $25 IIL Identification cies S26 IV. Clinical Effects : pense seHo0EEo S26 V. Official Limits (Standards)... . -. 826 VI. Visual Inspection . cee S27 VII. Unpublished Articles and Letters .........+ aon see SH sa ‘Journal of Parenteral Science & TechnologyPart |—Instrumentation JULIUS Z, KNAPP* and PATRICK P. DeLUCAT * Research & Development Associates, Inc, Somerset, New Jersey. ' University of Kentucky, College of Pharmacy, Lexington, Kentucky ABSTRACT: An objective review of the parenteral particulate measurement literature and the instrumenta- tion now marketed for the analysis of particulates in parenteral solutions has been completed. This review was undertaken with the goal of resolving the confusion of measurements presently surrounding the determi- nation of particulate contamination in parenteral preparations and to provide an informed basis for anaction plan. The conclusion reached is that the present lack of agreement in these measurements is the result of a comtbination of factors. Most of these factors are traced to instrument limitations and a combination of shortcomings in both sample handling technique and methodology. The review is in two parts: Part 1, “Instrumentation” in which measurement techniques and problems are evaluated and Part II, “A Selected Annotated Bibliography on Particulate Matter.” The Bibliography provides an overview of particulates in parenteral solutions emphasizing instrument evaluations and comparisons. |. Introduction An objective review of the parenteral particulate mea~ surement literature was undertaken to resolve the confu- sion of measurements presently surrounding the determi nation of particulate contamination in parenteral prepa~ rations and to provide an informed basis for an action plan. Relevant information from a thorough search of the scientific literature was combined with current responses from cach insrument manufacturer and some significant preprints from the recent PDA sponsored “International Conference on Liquid Borne Particle Inspection and Me- tology.” This combination of information was utilized to assure that any conclusions reached would be based on current particulate measurement capability The principal problem addressed in this review is the lack of agreement in any of the available measurements of particulate contamination required for the release of par- enteral products. This critical problem is the result of a combination of factors, most of which can be traced to instrument limitations. ‘A major problem shared by all available systems is created by the low sample volume handling capacity which in turn imposes sampling errors on the required tests. In the case of actual particulates with widely varying particle densities, the variability is magnified. Thesé er- rors can be identified as due to: 1, Sample preparation prior to measurement. 2. Losses in the handling of sample particulates within the instrument. In addition to the loss of particulates due to sample ‘manipulation before and during analysis, each of the sys- tems now available suffers in some degree from: 1. Shape dependant signals, 2, Inadequate particle size range. 3. Specification of inappropriate measurement limits, i.e. the number of particles counted are 100 low for “This work was funded bya grant irom the PDA Research Commitee Vol 42, Supplement 1988 measurement accuracy or the concentration of parti- cles is too high for sensor capability A shared problem in all available systems is the inabil- ity to distinguish between particles, microbubbles and insoluble microdroplets. The microdroplets can usually be traced to the use of silicone oil as a packaging component lubricant or to plasticizer droplets that have been extract- ed from flexible parenteral packages. The microbubbles are created in the liquid by either mechanical cavitation or by the thermal release of entrained or absorbed gases. While the small volume handling capacity is a major problem, there are also errors imposed by the sensing means employed. For optical sensing systems, flow prob- lems in the sensing zone will result in random orientation of particles traversing the sensing slit which, for other than spherical particles, results in random selection of the particle dimension measured. The variability imposed by this effect is greatest for flakes and fibers. The results reported by the Coulter instrument start with a volume proportional electrical pulse which is generated as parti- cles pass through a precision orifice in an electrolytic solution. The particle signal is then processed to obtain an equivalent spherical diameter. The volume signal is re- ported to be less affected by particle position than the ‘equivalent optical signal as the particle passes through the sensing zone. The volume measurements of the Coulter system, however, yields equivalent spherical diameters to describe even flakes and fibers. These equivalent diame- ters are small fractions of the required major particle dimension measurement U. Common Factors An essential requirement for measurement comparabil- ity is that the capability of the devices employed be matched in all significant parameters. This match, in particulate measurements, must commence with the re-quirement that the detectors employed have identical ca- pabilities in sizing and counting in the specified size range This includes particulate concentration capability as de- termined by the onset of a specified coincidence error, say 5%. Without this match in measurement capability the evaluation of even ideal suspensions on otherwise ideal instruments will not yield comparable results. The two parameter match requirement can be met by any of the particulate measurement systems in current use according to published specifications. This requirement is as basic as the specification of magnification in the section describing the membrane/ microscope and, similarly, should be listed in the USP XX1 (788) procedure (163). The achievement of data comparability begins with matched detector performance but cannot be accom- plished with this single prerequisite. Comparable data can only be achieved by a specification of rota! instrument system performance requirements; these should be inctud- ced in the (788) procedure. In addition to the publication of these specifications, GMP measurement and certifica- tion of system efficiency through the specified particle size and density range should be required prior to use. Historically troublesome areas have been technique and sampling. Trasen (68), in his discussion of analytical techniques for particulate matter, concludes, “In order to keep sampling errors at a minimum it is important to sample directly from the final packaged container and to sample the entire volume” (emphasis added). Although Trasen was discussing particulate measurement errors in the membrane microscope technique his comment is fully applicable to any of the other measurement techniques discussed. Since the (788) procedure involves sample pooling and ‘manipulation, an essential requirement for analytical ac- curacy is suspension homogeneity. A major factor in the accurate manipulation or transfer of liquid borne particu- lates is their settling time, This is illustrated in Table 1 which lists terminal settling velocities for latex. glass, and stainless steel microspheres in still water at 20 °C. Prior to any discussion of Table I results, specification of the required response range for this destructive test is essential. What boundary should be considered for the test limit stated in an open-ended way as >25 um? If the Pharmacopeial intent is to provide gap-Iree particulate quality assurance from 10 um upwards through the visible inspection range, a workable solution can be defined. TABLE 1. Stokes Law Computation of Terminal Settling Speed for Spheres in 20°C Still Water, mm/sec “Terminal Setting Speed, Particle sm sec Diameter, Stainless am. Latex Glass Steel 10 0.0029 0.38 28 ois 238 50 0.072 9.30 100 0.29 38.00 200 TISisaae 152.00 Densities Catex = 1.05, Borosilica 57. Type 303 Stanien. Steel = 786, 36 Knapp (166), using production tine rejects, selected repre- sentative particulates for holographic measurement. With this methodology, he defined 100 zm as the smallest pro- duction line reject that an experienced inspector, without using magnification, can detect 70% of the time. Borchert (167) used prepared standards with precision fluorescent dyed microspheres. Prior to experimental use, the quanti- ty of microspheres in each container was verified under UV illumination that made each microsphere a point light, thus achieving a detection probability of 1.0. Fol- lowing verification, well designed experiments using stan- ard inspection illumination resulted in data which sup- ported the visibility finding for 100 um. In the face of their studies, data from the destructive tests, covering the range from 10- to 100-sm particles, will supplement the nonde~ structive visual or optical particulate tests, with secure data from 100 um upwards, to provide gap-free particu- late quality assurance testing for parenteral products. To assure realization of this range of data from either optical ot resistance modulation instruments, such as the Coulter, requires information on the effect of the various densities normally encountered over the full size range being con- sidered, Accepting the 100-um size as an upper limit for the (788) destructive tests provides a basis for review of the Table I data. The 8.71 mm/sec terminal settling speed for 100-um glass spheres and the 38.00 mm/sec terminal seitling speed for 100-m stainless steel spheres define the agitation speeds required for the maintenance of uniform suspensions. They also establish the length of time in which an unstirred solution can still be considered reason- ably uniform. Close consideration of the agitation limits implied could suggest that total container sampling is a more practical response Addressing the case of actual particulates in test solu- tions where particulate densities and sizes cannot be speci fied prior to test, handling must be controlled to maintai adequate suspension of the worst case particulate that can bbe encountered. In the usual case, it is at least the han- dling (both in agitation and operation speed) that will keep glass particulates in suspension. At present, the (788) pharmacopeial processing is such that errors ean be introduced in the act of combining container volumes, extracting the required aliquot, aspirating the sampled volume into the liquid handling system, and finally effect- ing movement of those particles through the sensing zone of the instrument that have not been trapped on tubing walls or in eddy currents. Vigorous agitation, just under that for which cavitation results, will reduce the nonuni- form sampling error; but the residual volume left unana~ lyzed in the instrument sample container remains a proba- ble error source (175). Inadequate flow velocity, the length, material, and cur- vature of the tubing in the sample handling system be- (ween pickup and measurement point contribute to the loss of the larger and heavier particles. The recent FDA, side-by-side test (179) of particulate measurement sys- ems showed that the Climet system (180) with liquid processing rate of 100/mL per min (compared to ranges from 4.6 0 20 mL/min for the other instruments tested) Journal of Parenteral Science & Technologyhad significantly better carrying capacity for the larger particles than most other units evaluated. Settling and trapping errors, even for 10-um latex spheres used in instrument calibration, have been reported by Grant (174). In the recent past, the contamination resulting from Kraft paper use in sterile areas has been eliminated by the introduction of plastic packaging materials. However, the use of plastic materials has brought with it a correspond ing increase in “floaters”: particulates with densities low enough that they can be trapped in the meniscus of the solution. Since current systems and standards do not ad- dress this problem, it could also contribute to data vari- ability. ‘A common problem in all fluid stream particulate mea- surement systems now available is their departure from the Pharmacopeial required measurement of maximum linear pasticle dimension. Both optical and resistance modulation measurement can accurately determine spherical and near-spherical equivalent diameters. The Physiologically important measurements of flakes and fi- bers can, however, be in gross error. It has been argued that given knowledge of the shape of the particulate that should be evaluated, calibration adjustments can be made to obtain accurate results. The core problem with the evaluation of production particulates in parenteral solu- tions is that there can be no a priori description of the particulates that may be encountered. Measurement dis- crepancies resulting from shape differences, therefore, remains errors, Problems can persist even with the avail- ability of ideal analyzers: those inherent in the selection of inappropriate limits for instrument operation, The statis- tics of counting are well established. The establishment in the (788) procedure of constant volume for analysis leads to an error of 19.50% for a 95% confidence interval in the determination of acceptance for the count of >25-m particulates and 6.17% error in the >10-zm particulates for the same interval. 95% confidence limit errors for the ‘SVI range from 'f 10 100 mL are shown in Table UI. A review of Table II results clearly shows that a fixed test volume specification is inappropriate for all container sizes in this test. The error resulting from this source can be controlled by the specification of a practical tora! error limit. A tolerance of 5% for this measurement at the maximum allowable particulate concentration for parti- cles >25 um would define the test volumes required. Since TABLE Il__Count Error for Fixed Sample Volume Determinations this is a mandated limit test, accuracy at the compendial test limits is essential for any review of the quality of a tested bateh Another measurement error (122, 175) that can now be easily controlled is that due to the coincidence of multiple particles in the instrument sensing zone. The magnitude of the coincidence error is determined by the ratio of particles to sensing detector volumes and is usually speci- fied per mL of sample liquid. A review of the initial analysis, therefore, permits the selection of any desired error tolerance by dilution adjustment of the particulate concentration. Delly's report (87) that 81 to 96% of all particulates in some samples occur in the size range below 10 am where they can contribute, through coincidence, to the data variability now seen, takes on added significance. Incorrect batch rejections can result from these coinci- dence errors which have been shown to be controllable witha minimal change in measurement technique. A com- plete derivation and discussion of coincidence errors is included in Appendix II. It should be noted that the vol- ume of the sensing zone in optical counters is a clearly defined geometrical volume. In resistance modulation counters, due to divergence of field lines in the three dimensional liquid medium, the sensing zone requires em- perical determination and is a multiple of the orifice vol- ume. This multiple varies with the specific design chosen and can range from 2.5 upwards. Ml, Optical Systems At the 1969 Liquid Borne Particle Metrology Confer- fence sponsored by the New York Academy of Sciences, two of the contributors were the High Accuracy Products Corp. (109) and Royco Instruments, Inc. (118). These Pioneers are now combined into the HIAC/ROYCO In- struments Division of Pacific Scientific. Their basic parti- cle detection concepts are still in use. Both instruments were designed within the constraints of the bulky low complexity electronics of the era and functioned with minimal optical signals. HIAC selected light blockage; ROYCO employed a somewhat more complex forward scatter lighting principle. The surviving design is that pioneered by HIAC. Most present instru- ments follow the light blockage design concept pioneered by HIAC. Particle passage through a narrow sensing zone at right angles to the flow stream reduces the light energy as Required in (788) Particle Counting Particles X 10° 95% CA. Pooled Pooled Tn 10 mb % Error Volume 310 335 310 335 310 335 20 400 40 200 20 0.22 1.38 20 200 20 100 10 0.62 1.95 20 100 10 50 5 087 276 50 100 10 20 2 138 4.36 10 10 100 100 10 10 1 195 617 20 10 200 100 10 5 05 276 an 50 10 500 100 10 2 02 436 13.79 100 10 1000 00 10 1 a1 6.17 195 Vol. 42, Supplement 1988delivered to the detector. The signal generated is 2 de- crease in sensor output proportional to the area of the particle shadow projected on the photosensor. Far forward seatter systems, such as the Royco, direct illumination from the light source is blocked: only the light scattered by the particle is, ideally, collected by a large area photosen- sor. The amount of light scattered is proportional to the area of the particle viewed by the sensor: the signal pro- duced is seen as an increase at the photosensor output. For both types of systems, only spheres can generate signals independent of the orientation of the particle as it passes through the photodetection zone. For spheres, the ‘mathematical extraction of an equivalent diameter can be replicated with good accuracy. For irregular shapes, platelets and fibers. the signal extracted varies randomly with particle orientation as it passes through the detector. The report of an equivalent diameter for these particles cannot be related to the maximum dimension desired without prior knowledge of the shape of the particle mea~ sured (88). When production particulates are considered, it is clear that no effective particulate measurement is possible except for spherical particles. In both systems pulse height analysis was used to measure particle size The basic design is well suited for low flow applications Due to the relatively low sophistication level of the elec- tronics, amplifier response time contributed to the mea- sured coincidence errors, In both systems random particle motion across the sens ing zone, due to turbulent flow, limited measurement accuracy 10 spherical particles. Flakes and fibers were either measured with gross error or were disregarded if the detected signal was below the threshold of the instru- ment The eighteen-year period since the 1969 meeting has seen modular replacement of system elements. Today, integrated circuits and microcomputers are in customary use in cach of the commercially available particle measur ing systems. With the availability of reliable, low-powered as and solid state lasers, forward seatter augmented de tection systems are once again in the marketplace The two basic optical designs have been perpetuated with little or no change in all present day systems. The processed signal is limited by its primitive optical charac ter and the turbulent flow conditions in the sensing zone, The turbulent sensing volume flow randomizes the parti- cle dimension scanned, thus limiting the utility of the data A. Filmy Materials A major factor in the decision to use optical counters in the US, was the existence of traces of the 5-HMF poly- mer dextrose breakdown product in parenteral solutions, Quoting from the 1980 July-August issue of Pharmaco- peial Forum in which the following interim notice was published, “For dextrose containing solutions do not enu- merate morphologically indistinct material showing little or nosurface relief and presenting a gelatinous or film-like appearance. Since this material consists of sub-units of 1 zum or less in linear dimension and is liable to be counted only after aggregation on the membrane Following an industry collaborative study, the HIAC- Royco was qualified as an alternative to the membrane/ microscope evaluation of parenteral particulates. AS a result of this problem, Supplement No, 3 to USP XX issued on February 15, 1982 excluded the 5-HMF poly- mer particles from consideration as a rejectable particu- late or allowed the use of an electronic particle counter. The extension of optical counter usage from special pur- pose tool to facilitate continued production of dextrose containing solutions into its present usage as a device considered usable for total particulate quality evaluation in parenteral products had unfortunately not been ade~ quately studied prior to the commencement of its extend- ced usage. Delly’s comments in this regard are especially apropos (87). He evaluated the membrane/microscope method of particulate analysis against the optical counter and reported that thin filmy materials could be found in a variety of apparent shapes and sizes in both methods as a result of the handling that the film experienced. In the electronic counter, thin filmy materials in the counter’s turbulent flowstream could appear crumpled as balls, scrolled up in needle or tubular shapes, and in flat or plate- like appearance. He also commented that these films can be substantially thinner than the wavelength of light used. Under these conditions, little edge effect would be no- ticed. Due to their gel-like nature, these particles tend to take on the refractive index of the medium in which they are immersed. To quote from Delly: “Presented edge on or fore-shortened to the detector, these particles still have some absorption color and will be seen, but will be counted in the smaller category.” Delly’s conclusion concerning optical counters is as valid now as when originally pub- lished. Iti in part, “. .. particles in solution have at least some chance of being considered in all aspects, and al- though still not perfect, at least this method is more accu- rate where unusual geometries are involved. B. Refractive Index Effects A problem shared by all counters employing optical detection is that particulates whose refractive index matches that of the suspending fluid cannot be detected. This problem has been discussed by Lloyd and Freshwater (168) in the use of a HIAC instrument and summarized by Russell (179). As Russell summarized Lloyd and Akers data, a difference of 41.25% in refractive index between particulate and solution is required to avoid this, kind of refractive index limitation, Within this band the resulting errors can overwhelm those due to any other source, ©. Conclusions There is unquestioned utility and economy available from the use of the particle detection systems listed in Appendix I for the monitoring of parenteral product qual- ity. The present usefulness of these systems is limited by inadequate specification of the instrumentation employed and by the sample handling method described in the (788) method of analysis. Blanchard and his co-workers, (53) and more recently Barber (173) have observed and Journal ot Parenteral Solonce & Technologyreported the serious error contribution that suspension nonuniformity generates in the measurement of particu- lates. Kushner (175) also discusses the serious impact of sampling errors traceable to nonuniform suspensions, Minimization of handling errors and losses, selection of test volumes that will result in data within acceptable error limits and avoidance or reduction of the error result- ing from coincidence will reduce the present range of data variation seen, Pulse height analysis, now uniformly em- ployed in optical detection systems, limits the maximum size of particles that can be measured. In addition, particle shape and the refractive index difference between particle and suspending liquid will affect measurement accuracy. Prior tose, the refractive index difference between parti- cle and suspending liquid must be shown to be more thatn £1.25%. In addition, Schroeder and Deluca observed that calibration with similar shapes must be utilized to obtain best results (89), The available optical particulate analyses systems are therefore suitable for relative indica~ tions of product quality. IV. Resistance Modulation Systems (Coulter, etc.) At the 1969 Conference, Kinsman (93) noted that, “particle counting is not easy but it can be done.” The instrument Kinsman discussed followed Coulters original patent by 16 years. While improvements have been made in the Coulter counter over the years, the basic concept unaltered. A cylindrical orifice in an electrolyte wi change resistance as individual particles enter the orifice displacing current conducting electrolytes. The output signal pulse is nominally proportional to the volume of the particle as it sweeps through the orifice. Since its intro- duction, the Coulter has become the analytical instrument of choice in diverse areas from blood analysis to the evalu ation of industrial powder sizes; specialized adaptations have also been used to evaluate fiber lengths. The present availability of integrated circuits and microcomputers has reduced the instrument size and increased its capability ‘The change from a constant voltage system to a constant ‘current design and the inclusion of a flow transition zone at the orifice throat appears to have extended the particle size range for linear response that can be achieved with a single orifice. Linearity of sphere measurements up to 77 10 80% of orifice diameter were shown to be possible (106- 108). Coulter's conservative choice of new limits are from 2 t0 60% of orifice diameter; the older limits were 2 to 40%. The conservative 2% lower particle size threshold is established by the electronic instrument noise: operation to.1'4% is often possible A. Shape Sensitivity Resistance modulation instruments have evolved from empirically defined devices to those whose response can now be analytically calculated (100). The special chal- lenge that must be evaluated in the analysis of parenteral particulates is that the capability of this type of instru- ‘ment must respond equally to the full range of randomly occurring contaminant types and sizes. The only particu- Jate measurement specified in the XX] Pharmacopeia isin Vol. 42, Supplement 1988 the membrane/ microscope procedure for LVI. USP XXI uses the effective linear dimension. This selection is be- lieved to be based on the potential blood vessel blocking capability of particulates. The capability of these instru- iments to generate the required measurement is discussed below. Karuhn and co-workers (95), reviewing pre-1969 expe- riences with the Coulter, reported that they and others experienced significant broadening of the distribution of, latex calibration spheres over that determined by Dow Chemical using a microscopic technique. Karun also quoted experienced Coulter operators as saying that “mass balances performed on various types of industrial powders (ic., mass calculated to have passed through the orifice to the mass determined by a weight difference) generally was in the order of 1.31:1 for nonextreme shapes while in the case of flakes it was often greater than this.” Davies and co-workers (97) have shown that excepting spheres “particles of identical volumes but different shapes generated different pulse heights depending on the magnitude of their approach diameter to the orifice.” Experimental data indicated that the pulse height is also dependant on the flow streamline followed by the particle through the sensing zone (95, 97). Since flow in the sens- ing zone is turbulent, data scatter results It must be appreciated that Davies based his conclu- sions on experimental data from a scaled-up model of the Coulter orifice. This analog model was used due to the ‘complexities which are encountered in analysis of electri- cal conduction in a three-dimensional volume. Davies also investigated the effect on the pulse height of the volume signal for orifice entry positions of varied shapes with equal volumes. His data shows a marked proportionality between the approach diameter of the particle and the recorded signal. Of special interest is the difference in signal reported for a rectangular shape with orifice entry both in the flat and end-on positions. The signal ratio for these two conditions is 3.37/1.49 = 2.26. The diameters recorded for these two measurements will vary by 31.2%. This is the kind of variability that is possible in lake and platelet measurements due to this effect alone. When a cube and sphere of equal volume are measured in this system, the diameter recorded for the cube will be 1.8% ‘ereater than that for the sphere. Lloyd (101, 102) also used a large scale analog of the Coulter-type system to demonstrate volume linearity for differing particle shapes traversing the orifice on a central streamline. The volume measurements were linear within the same particle shape but the slopes of the straight lines relating volume to signal magnitude differed as particle shapes differed. For the limited number of cylindrical shapes investigated, the linear slopes varied from 0.70 to 2.49 of the spherical calibration curve slope (101). When Lloyd studied @ range of shapes closer to spheres he found 4 range of linear slopes from 0.85 to 1.37 of his spherical calibration curve, Some of Lloyd's cylindrical shapes had proportions similar to those encountered in fibers (i.e. length/diame- ter ratios of 5:1, 6:1, and 7:1). In his studies, the cylinder traversed the measuring orifice on a central streamline. 80TABLE III. Review of Fiber Measurement Effects on Resis Example: Sum Diameter Analog Mode, K,= 1.38 Physical Dimen- Mag. Factor =(K) sion, Magnified L(USP XX1) ylinder Lid Ke Ky ik) ODD 1096 0.70 089 335 298 2 160 116 108 422 443 100 3196 12 112 483 SAL 15.0 4 234 169 119 531 632200 527 197 12s $72 11s 280 6 308 220 130 608 790 © 300 7 344 249 135 640 Bos 350 s 359 260 138 669 923400 o 395 280 142 696 988 450 Wor 421 305 145 721 1043500 20" 640 463 1.67 9.09152 100.0, To illstrate this effect, measurements on a Sam diameter fiber of varied lengths are compared with the USP XXI requted dimension showing large differences. This magnification partially compensates for ‘he minimization of fiber length i his typeof measurement. The analog ‘model datas from Lloyd (101, 102):theastershed data ws extrapolated ‘The extrapolation i used on the bass of the linearity of the Tog/log plot of slope against the ratio of fiber length to diameter ‘Notations: per diameter Kk phere diameter magnified sphere diameter Ly Kj =slope flinear fiber volume curve vs, signal ouput = normalized fiber volume slope Tength of fiber linear dimension /) = linear dimension magnifier Table {11, calculated from the physical dimensions and Lloyd's data, shows that a fiber with a length/diameter ratio of 5:1 will be evaluated as 125% of the fiber length determined on an equivalent volume basis. Considering the linearity of Lloyd’s data, results have been extrapolat- ced to include length/diameter ratios up to 20:1. At 10:1 length/diameter ratio the extrapolated data shows a fiber length magnification of 145%, Similarly, at 20:1 length diameter ratio Table III shows fiber length magnification of 169%. This magnification compensates in part for an otherwise drastic undersizing of fibers in resistance modu- lation systems. Lloyd's data is limited t0 the examination of fibers traversing the orifice on a central streamtine. Considering the turbulent nature of low through the orifice and the consequent randomness of both fiber position and path during the traverse, Lloyd's work cannot be considered definitive analysis of fiber measurements in Coulter-type systems. An additional limitation that should be noted is that the geometry of Lloyd's analog system was not mod- eled after a particular commercial device, therefore, the resulting magnification ratios can be expected to show variation as orifices with varying diameters and lengths are employed B. Filmy and Porous Materials Coulter-type measurements of filmy or porous materi= als should also be considered when parenteral contami nants are evaluated, In measurements of porous materials 10 (nylon and fly ash are examples) the dimensions recorded are several times that due to the skeletal volume of the particle (91). Filmy gels such as those encountered in the 5-hydroxy furfural glucose breakdown product are detect- ed in Coulter-type systems on the basis of electrolyte displacement, For instance, a mass equivalent to a 25-zm sphere with solids content of 5, 10 and 20% will generate volume signals respectively of spheres with diameters of 9.2, 11.6 and 14.6 um. It can be deduced that the essential difference between these materials is the relative displace- ment of electrolyte. The low solids content of the filmy gel can generate signals below the noise level of the Coulter instruments for smaller particulate volumes. C. Sample Handling The Coulter uses a cylindrical orifice, whose length is about 75% of its diameter, to generate the particulate signal. The orifice size specified for pharmaceutical prod- uct testing has high fluid resistance due to its small size. At the recent FDA tests (178) the flow rate recorded for the Coulter instrument was 4.6 mL per min which was the lowest flow among the instruments compared; the highest ‘was 100 mL per min, This low flow rate can be compensat- ed for by vigorous agitation within the instrument to mini- mize selective elimination of larger particulates prior to their evaluation. Using an earlier model of the present Coulter, Bungay and Krebs (92) reported that stirring rate variation caused a 20% increase in count when higher stirring speeds were used. Concerning the need for vigor- ‘ous agitation to avoid this error, Lines (181) comments as follows: “the standard two bladed stirrer operating at 1020 rpm ina flat bottomed sample container will suspend, 40-um glass spheres in a saline solution.” Lines cautions that irregular shaped pieces of glass may require different agitation velocity to achieve the same results. The stan- dard two bladed stirrer operating at 900 rpm in a special round bottomed accessory beaker can suspend glass spheres of 200 am diameter. Lines concludes that for 300- ‘zm glass spheres the Coulter four-bladed stirrer operating «at 1320 rpm in the accessory round bottomed beaker with vertical baffle is required to maintain suspension unifor- mity, If consideration is extended to include suspension of the occasional stainless steel fragment that may occur, Stokes law estimates indicate 92-um stainless steel capa- bility when 200-1m glass spheres are suspended and 138- mn stainless steel capability when 300-zm glass spheres are adequately suspended. Without adequate suspension these particulate measurements cannot be made. Consid- ering the irregular nature of particles in parenteral prod- ucts, agitation to suspend at least the 200-um glass spheres would appear to be a conservative choice of opera~ tion to ensure measurement of the occasional heavier par- ticles that can appear. The agitation required to maintain uniform suspensions of the wide range of particulates ‘encountered has not been adequately addressed and is a problem for all types of particle analyzers in present use. The agitation must, however, be below the limit at which cavitation occurs since bubbles are counted as false partic- ulates. Lines (103) reports that the Coulter counter re- Journal of Parenteral Science & Technologysponse can theoretically be affected by particle shape, resistivity, ete.: *...it is recommended that the instru- ‘ment be calibrated with the material under test.” In the usual circumstances of product testing, neither the size range nor the material of the contaminants are known, Test data, as with the optical systems discussed, are suit- able for relative indications of product quality. The present design cannot be considered as approaching the functional requirements for a referee method, D. Conclusions Based on the particle measurement used is USP XI (163), the signal generation and computations used by the present Coulter particle analyzer generate excellent to usable measurements for spherical and near spherical par- ticulates respectively. Measurement of flakes, platelets, and fibers have been found to generate variable data. In this type of measurement, an equivalent spherical diame- ter is computed from a pulse height measurement of the particles volume, The measurement of nonspherical parti- cles have also been shown to be affected by the profile of the particle as it enters the sensing zone and the stream- line followed through the orifice. Inadequate detection of filmy gels, such as those of the S-hydroxyfurfural glucose breakdown product, must also be considered in any evalu- ation of the present Coulter system (or its commercially available equivalent) for quality assurance measurements of parenteral particulates. The low flow rates in Coulter type instruments require small volume sample processing which increases measurement error potential. The avail- able resistance modulation particle detection systems for parenteral particulate analysis are, therefore, suitable for relative indications of product quality. V. Present Detection Systems A. Comparison Tables IV and V have been prepared to provide an objective comparison framework for optical and resis- TABLELY. Fiber Measurement Result Comparison for Opti- cal and Resistance Modulation Systems Showing Major Differences from the USP XXI Measure- ment Specification USP XXI Fiber Physical Resistance Optical -—_Linear 11/4; Dimensions Modulation. Measurement Dimension Ratio “de Lr Dm Dy DG DED sl oS$ 2 98 123 5 126 25 10 $0 196 245 0 252 50 2 100 391 490 20 505 100 1 $30 123 178 5 178 50 10 100 247 388 10 357 100 2 200 493 715 20 714 200 21 5 100 155 259 5 258 100 10 200 311 519 10 505 200 20 400 62.1 1037 20 174 400 Notations 4; = fiber diameter B_ » linear dimension Dj = minimum optical measurement of equivalent spherical diameter ‘D3 = maximum optical measurement of equivalent spherical diameter ‘Day = resistance modulation equivalent spherical diameter based on volume caleulation Dig magnified fiber measurement showing effect of Lyd ratio tance modulation particle measuring systems with which parenteral particulates can be evaluated. In these tables are listed typical nonspherical particulate shapes and sizes that can be encountered in production material. Three fiber groups with length /diameter ratios of 5:1, 10:1, and 20:1 are listed in Table IV. Relatively thin arrowhead and square shapes have also been selected to represent chips and flakes and are shown in Table V. ‘The performance of these particulates against the USP XXI specified maximum linear dimension will now be ‘considered. For optical systems operating under the best TABLE V. Measurement Result Comparison for Optical and Resistance Modulation Methods with Particle Shapes Representative of Chips and Flakes Showing Major Deviations from the USP XXI Lineae Dimension Resistance Physical Dimensions Modulation USP XXI article Shape t Lyla b Di in./max.) B ‘DE___Linear Dimension Arrowhead 5 30 25 17.9(143/32.4) 178 218 50 15 100 50 35.9(33.5/435) 33.7 55 100 20 200 loo 71.8(57.4/1290) 7141110 200 Square 5 25 = 18.1 (14.5/32.8) 126 282 354 10 50 36.3 (29.0/65.7) 25.2 564 70.3 20 100 = 726 (8.11314) 50S 1128 14.4 Notations: Like ‘= cqual sides of triangular chp shape o sides of square base of rangular chip shape b base of triangular chip shepe o minimum opis! measurement of equivalent spherical diameter ob ‘aximum optical measurement of equivalent spherical diameter Dom = resistance modulation equivalent spherical measuremeat based on volume calculation Don (min, fax. Vol 42, Supplement 1988 ‘estimate of eect of particle orientation in traverse of sensing zone sitconditions, that is with laminar flow conditions through the sensing zone and with oriented particles, the dimen- sions recorded are listed in the “Optical Measurement” columns of Tables IV and V. The three groups of fibers (5:1, 10:1, and 20:1 length/diameter ratios respectively) are evaluated as 50, 35, and 25% of the required dimen- sion. When the flow through the sensing zone is adjusted to be turbulent, all positions of the fiber in the sensing zone become equally probable and the spread of measured values for the same sequence of fiber groups becomes: 20 10 50, 10 t0 35, and 5 t0 25%. When resistance modulation measurements are con- cerned, consideration of their variability begins with com- putations of the diameter which will define an equivalent spherical volume. These computed diameters are listed in the column titled “Resistance Modulation.” When Lloyds multiplier is used to estimate the dimension recorded for 2 fiber with a length /diameter of 5:1, the result is 49% of the desired linear dimension. The extrapolated values for 10:1 and 20:1 length/diameter ratios are 36 and 26%, respec- tively, of the USP XXI required dimension. Since orifice flow is turbulent, additional measurement variability will result from the randomness of the path followed through the sensing zone (97). Since Lloyds data is not related to any commercial device, the computations cited must be considered crude estimates. For the two shapes representative of chips and flakes shown in Table V, the optimum optical results for the arrowhead shape is 56% of the required dimension; simi- larly, for the square shape 80% of the USP XXI dimension is achieved. When turbulent flow through the sensing zone is selected, measured dimensions for the arrowhead be- tween 36 and 36% of the maximum linear dimension become equally probable. Similarly, for the square shape, fa range of values between 36 and 80% become equally probable in turbulent flow. Examining resistance modula- tion data for the chip and flake category of particulates, for the dimensions of the arrowhead shape shown, the spherical equivalent dimension evaluluated is approxi- mately 36% of the USP XX1 dimension; for the square shape listed the equivalent spherical dimension is 1% of the size required. When the effect of turbulent flow on particle position as it traverses the orifice is considered, Davies’ (97) data on the ratio of rectangle flat and end-on ‘measurement variability can be used to estimate the effect fon the measurement results. The results will, with equal probability, be distributed as follows: for the S-m arrow head shape 33.5 t0 43.5% of the desired dimension and for the square shape 47.4 to 61.7%. An additional variability of 3 to 5% is contributed to this data variation by the random selection of the particles pathway through the orifice. B. Conelusions Neither the resistance modulation nor the optical detec tion systems surveyed evaluate the maximum linear part cle dimension as described in USP XXI (163). In the optical instruments available, measurements are based on the maximum projected cross-sectional area of the parti- sr cle as it traverses the sensing zone. For a laminar flow system delivering oriented particles through the sensing zone, the measurements are simply related to the parti- cles’ shape and are listed in the D*apica column of Tables IV and V. No presently available system makes this type ‘of measurement. Instruments with fully developed lami- nar flow in the sensing zone will deliver this type of mea~ surement for fibers and will show some degree of improve- ment when chips and flakes are measured. When turbu- Tent flow is encountered, the resulting randomness of particle position in the sensing zone results in the full range of measurements shown in Table IV from D-opict to Depa with equal probability. In resistance modulation systems, the simplistic as- sumption that linear volume measurements forall particle shapes is obtained with spherical calibration has been shown to be incorrect. The divergence of field lines in the electrolytic detection system used in these instruments significantly affects nonspherical measurements. To the extent that the data used in Tables 4 and 5 yield valid estimates, the shape sensitivity of resistance modulation systems increases the equivalent diameter of nonspherical particles. The amount of this increase is such that for nonspherical particles no clear superiority between optical and resistance modulation instruments can be established at the present time. No present manual or automated system achieves ac- curate measurements when filmy particulate materials are evaluated. ‘The minimization of sampling errors can be addressed by evaluating the total contents of containers. To achieve this goal in practical terms requires improved sample volume handling capabitity. When the presently available systems are evaluated from this point of view, an impor- tant benefit is seen in the design of the sensing zone of optical detection systems. Fora given particulate concen- tration handling capacity (Le., for a specified coincidence error), the cross sectional area for flow of the sensing zone of an optical instrument can be much larger than in the resistive modulation system. This increase in cross sec- tional area also minimizes the possibility of blockage dur- ing use and generates less fluid resistance than the cylin- drical Coulter orifice. For liquids with low cavitation threshold or those with higher viscosity levels, the flow resistance consideration may well guide the final choice of instrument selected. VI. Recent FDA Test Results It is impossible to distinguish significant operational benefits among any of the particle detection systems now available in any search of the literature or as a result of the side-by-side tests just completed by the FDA (178). The only resistance modulation instrument tested was the Coulter. The optical systems evaluated included: CLI- MET, HIAC, KRATEL, and MET-ONE. The HIAC was used with @ Russell Laboratories sensor. The particle measurement systems unit was not included due to the manufacturers’ scheduling difficulties. Appendix 1 in- cludes details and comments for the listed particle mea~ Journel of Parenteral Science & Technologysurement systems. Each system listed uses modest modifi- cations of the primitive optical design pioneered by HIAC and ROYCO and discussed in 1969. Resolution and re- producibility of size determinations are required in the (788) procedure for 10- and 30-um particulates. Valida- tion of the capability of each system over the full range of particle sizes that can be measured with it as in a counting efficiency determination, has been inadequately consid- ered. This inadequacy in the validation could prove to bea source of the troublesome differences between measure- ‘ments made by presumably comparable systems for >25- um particles. Specifications for agitation during sample handling and processing on the selected particle analyzer are considered to bea critical prerequisite for data accuracy. VII. Recommended Actions A. Stop-Gap Proposal The economic and regulatory value of the measure- ments available from automated particle counting sys- tems are generally acknowledged. With the sources of data variability described above, some approximation of a referee method is essential. This initial referee method can be provided by the measurement of full container volumes on systems with validated particle size linearity and reso- lution from 4 to 100 um whenever the use of the present sampling and handling described in (788) results in a batch rejection. B. Near Term Materials and Techniques: Much could be done to im- prove test result comparability with adequate recognition of sampling errors and handling requirements necessary to maintain uniform sample suspensions, The testing of particulate measurement systems for equivalent perfor- mance depends on the availability of identical suspensions with precisely maintained homogeneity. Any differences in either suspension constitution or homogeneity (when solution aliquots are used) will reduce comparability of test results. Data for particulate instrument system evalu- ation cannot be in any closer agreement than the differ- ences in the test suspensions employed for their evalua~ tion. Development of the techniques neccesary for the preparation of standard suspensions and the maintenance of their homogeneity is therefore an early priority. It must be stressed that prior to the test of the instru ‘ments, the handling and sampling techniques required to maintain sample accuracy from the container until entry into the measuring device is an essential prerequisite Containers prepared with selected spheres having the re- quired range of density and size (say 200 um of glass) can be used to develop the handling necessary to assure accu- rate maintenance of suspension uniformity throughout the required handling. Instrumentation: When the necessary handling tech- nique has been adequately described, the validation of competitive detection systems can commence. Consider- ing the importance of these tests to both the pharmaceuti- cal industry and its regulators, the test protocol should be Vol. 42, Supplement 1988 developed in cooperative interaction. The tests should be comprehensive enough to establish the essential instru- tment qualities required over the entire size and particle density range required: 1. Sample handling capability. 2. Size resolution. 3. Size reproducibility. 4. Coincidence onset. Enough testing to accumulate statistically valid results for each of the systems in each category listed is essential. It must be recognized that the time and labor required for method development and testing are substantial. Any at- tempt at a “quick look” isa disservice to the industry and its regulators. VIII. Longer Term Actions For a longer time frame response, it is recommended that specifications for a referee method system be pre- pared. Following review and acceptance of these specifi- cations, a request for proposal should be circulated to selected vendors in a publicized, open competition, Evalu- ation of the responses, tests of the developed system, and the emergence of a secure referee method should then follow in sequence. IX. Future Perspectives A. All Systems ‘As discussed, sampling the entire contents of a contain- er would remove one major element that results in data variability. For routine use, this requirement must be supported by either higher sample handling capability than is commercially available at present or methodology which will reduce volume requirements by concentration ‘of the particulate contaminants prior to test. B. Coulter System A recent telephone discussion with Coulter's chief de- sign engineer in England, R. W. Lines, revealed that Coulter holds patents that could be used to evaluate the maximum dimension in fibers and flakes as required in the ‘mandated particulate test. The design change requires the creation of laminar flow stream through the sensing vol- ume and detection of the passage times of particles through it. A market request for the investment time and labor to achieve this result has not yet appeared. Improve- ment in the volume handling capability of these instru- ‘ments should also be considered. C. Optical Detection Systems There are two major deficiencies in the basic optical signal presently used. The first results from the fact that only a single imaging plane is used in the present genera- tion of optical particle analyzers. With this limited capa- bility, the dimension reported for nonspherical particu- lates is randomly determined by the particles orientation as it is swept past the detection slit and it, therefore, seldom approaches the maximum dimension required. A flake can be recorded as a sphere with its diameter deter-mined by the projected area presented by its thickness and its length; the diameter reported will be. in this case, substantially less than that required. When the flake tran- sits the scanning slit in a perpendicular position, the di- mension calculated can approximate the major dimension specified in USP XXI. In the extreme, a fiber could be reported as a sphere whose diameter is determined from that of a dise equal to the fibers’ diameter. In each ease a second viewing plane at right angles to the one in present use could be used to improve the quality and information content of the optical data, The second deficiency is that due to sole use of pulse area particle measurement without any use of transit time measurement. This type of signal analysis limits the maximum length of particles that can be measured to the sensor slit width, The use of transit time measurement, however, requires control of liquid flow conditions in the sensing zone. Alternatively, direct measurement of particle size using the recent improve- iments in imaging technology and computation in the right angled viewing system described above could satisfy the requirement for a standard referee method that would provide information concerning three dimension particle shape as well as its major dimensions. X. Notes, Appendix I lists commercially available particle mea- suring system descriptions and comments. Appendix I treats the problem of coincident count error and reduces it to an easily avoidable problem in the course of particulate measurements. An earlier version of this paper was presented as part of the H. K. Kushner, L Abramson, T. Barber, and J. Z. Knapp paper titled, “Im- plications of Sampling Theory,” presented at the Interna tional Conference on Liquid Borne Particulate Inspection and Metrology, May 11-13, 1987. Appendix 111 isan attempt at specifying a particulate analysis system which will eliminate the concept limita- tions discussed in this Report. Appendix I: Commercially Available Particulate Measurement Systems for Parenteral Use—Descriptions and Comments Climet Instruments Co.. P.O. Box 1760, Redlan 92373 Model C1-1000 The Cl-1000 instrument with the C1 150-150 light obscu- ration detector and the separate C1 1000 sampler unit ‘were among those evaluated by the FDA. The sampling unit for small volume injectables uses a 10- or 20-ml syringe to draw the sampled liquid directly through the detector. This minimized path is believed to be the reason for the outstanding large particle detection demonstrated for this unit, The Cl 150-150 detector is designed for operation up to particle concentrations of 4370/mL. with 5% coincidence error. A new three-fold higher concentra- tion detector isin final development CA su Coulter Electronics Inc, 13960 N. W. 60 St., Miami Lakes, FL 33014 Model ZM/P The Coulter Model ZM/P with 140-ym orifice was in- cluded in the FDA equipment evaluation. The use of the Coulter was aborted due to the lack of agreement between its data and the other instruments present. Since the sig nal processing method chosen for the Coulter yields data in gross error for measurements of flakes and fibers it is judged undesirable for the evaluation of parenteral partic- ulates. The Coulter was the only nonoptical instrument included in the evaluation: it employs resistance modula- tion to detect particulates. HIAC/ROYCO Inst. Die., Pacific Scientific, 2431 Linden Lane, Silver Springs, MD 20910 Model 4103 ‘The HIAC/ROYCO system recommended for contami- ant monitoring in batch samples is their 4103 system. ‘This includes the Model 4100 counter, the Model 3000 syringe operated sampler and a selected detector for the particle size range of interest. The recommended sensors for (788) determinations are the HR-60HA for 1.5~30-n particles or the HR-120HA for 2-100-y particles. Since the HR-120HA provides gap free data when its data is integrated with the results of a visual or nondestructive optical inspection, it should be of greater interest. A flow rate of £1 mL is specified with this detector to assure specified accuracy. This detector has 5% coincidence error at a concentration of 6000 particles/mL. Kratel Instruments GMBH, D 7250, Leonberg, Stuttgart, West Germany Boblinger Strasse 23 The Model 100 with the HIB light obscuration detector was evaluated by the FDA. It is a small compact unit using a stepper motor driven syringe sampling unit. The fluidic design eliminates the pressure pulsations resulting from the stepper motor drive. The short path length be- tween sampling syringe and detector resulted in large particle detection performance almost as good as the Cli- mel. The detector design is very similar to that of the Russell unit discussed below. It is rated for a concentra- tion of 22,222 particles at a 5% coincidence error Met-One Inc., 481 Grants Pass, OR 97529 Model 214 ‘The Model 214 with the Model 211 detector uses forward scatter light employing 2 solid state laser diode ina system offering calibration automation. The benefit of the laser use in the 5- to 100-x particle range is its 25,000-hr stable lifetime, The detector is designed for a coincidence error of Sat a concentration of 3390 particles/mL. A detector with 10,000-particle/mL capability is in final develop- ment. The Met-One Model 250 Automatic batch sampler designed for use with this system has 10-120-mL sample capacity and can apply either pressure vacuum as the driving force Particle Measuring Systems Inc. 1855 8. $7th Court, Boulder, CO 80301 IMOLV/SOPS 100 System Due to scheduling difficulties on the part of the manufac- turer, this laser based instrument was not included in the Journal of Parenteral Science & TechnologyFDA evaluation. This instrument operates with a maxi- ‘mum concentration of 3000 particles/mL for a specified size range of 2-150 w. With the 10-mL. syringe mounted on the sampler flow ranges of 20-168 mL/min are avail- able. At these high Now ranges, superior large particle detection performance is expected Russell Laboratories, 3314 Rubio Crest Dr., Altadena, CA 91001 Designs and manufactures light obscuration detectors only. The FDA favors the RLV 1-50H Model for use with their HIAC system, Model PC320. This detector uses a tungsten source operated to provide extended life. The design provides self-alignment when lamps are replaced, The RLY 1-50H Model could be used with @ concentra- tion of 22,222 particles/mL at which a 5% coincidence error will occur. Appendix Il: Coincidence Errors in Particulate Counters Introduction The error in particulate size measurement due to the simultaneous presence of more than one particulate in the detector volume can be potentially troublesome. This er- ror comes from evaluating a concentration of particulates ‘beyond the accuracy limits ofthe selected detector. ‘The presence of multiple particles in the detector vol- ‘ume results in the addition of the individual signals. The coincident signal sum cannot be distinguished from the signals of single, larger particles. These coincident signals can contribute to the rejection of a batch of acceptable material. This error can be easily recognized and avoided The analysis below reviews this error and relates its mag- nitude to the number of single particles inthe test solution With this information available, the user can evaluate the quality of the data being generated and can take correc- tive action, when required, to maintain the degree of accu- racy specified for the assay. The bibliography listed indi- cates the troublesome potential of this problem (1-7) Following a general treatment, coincident count error ima detector with typical dimensions are computed. It can be seen that examination of the initial count results can be sed to determine the adjustment of particulate concen- tration required for operation within any desired error. The effect of a massive overload of particulates below the range of clinical or regulatory interest is also examined. Analysis Expressing particle concentration/mL in terms of mul- tiple detector volumes/particulate C= 10" /nV, where C, = particulates/mL, n = multiple detector vol- umes/particulate, and Vg = detector dimensions in ym. The average number of particulates to detector volumes is designated lambda and is the reciprocal of Vol. 42, Supplement 1968 d=1/n Since the number of detector volumes/mL is large and the ratio of particulates to detector volumes is fixed, the Pois- son approximation can be used to analyze the probability of particulate coincidence within the detector volume (8, 9). Expressing the particulate detector volume as K, the probability of K occurring in such a volume is: K (particulates/detector volume) = (Se™)/K! Table A.I lists the results of using the commonly select- ed ratio of ten cell volumes/particle. It can be seen that over 90% of all cell volumes are empty, approximately 9% have single particles, 0.45% have two particles, and 0.016% have three particles in them It is convenient to have an expression for the sum of all possible coincidence effects for any selected operating condition. Since the sum of all terms in the series, from zero to infinity is 1, deducting from | the probability of an empty detector and that of one particle/detector, results in the total number of coincidence occurrences for a se= lected ratio of particles to detector volumes; it is: K(T) = 1% de® =1-e%U +d) In Table A.t, the computations are based on ten cell volumes/particle (or 0.1 particle/10 cell volumes). The sum of all coincidence combinations is 0.00467884. This sum is only 3.42% larger than the two particle coincidence probability of 0.00452419 as shown in this table. The probability of multiple particles in the detector volume increases with particle count. The coincidence effect is expressed as a ratio of multiple particles counted to those of the single particles in the solution. These ratios are shown in Table A.II ‘A convenient way to utilize these results is to evaluate the coincidence effect errors in a detector of typical di- mensions. Selecting typical dimensions 100 X 100 X 1000 jum and calculating the coincidence of two and three parti- cles in the detector as a percentage of the single particle ‘counts. This has been accomplished in Table A.{II. The final column in Table A.III lists the sum of all coincident ‘counts as a ratio of single counts for the particulate con- centration selected. For the selected detector volume, coincidence errors are seen to increase with increasing particulate concentration as shown by cither the values of decreasing n or increasing lambda. Review of the data in the last three columns of ‘Table A.JIL show that the greatest error contribution is in TABLEA.L Probability of Zero, One, Two, and Three-Parti- ee Coincidence Probability of Keparticlesin ny Remarks 090483742 Empty detector 0.09048374 One particle 0.00852419 Two particles 0.00015081 Three partcies = General case 815TABLE All. Counts to Single Particles General Solution for the Ratio of Coincidence NO) e* Ld Zeroes Ratio NI) Ae NQ) . Net/2 Ld Doubles Ratio. NO)” de? 2 NG) eA/6 ‘Triples Ratio MI) deh 6 Ae Ko D jeneral Ratio See General Rati TOOFM LUE gy Coine de a cidence Ratio TABLE A.lII. Coincidence Errors Expressed as a Pereent of Single Particle Counts for a Detector Whose ‘Volume is 10? um? (NQ/NDNGY/NI) NTNU Lambda ER = Rnb 5 0.20 20,000 10.000 0.667 10.701 10 0.10 10,000 5.000 0.167 5.171 2 008 000-2800 0042 2.542, 50 002 2,000 1.000 0.007 1.007 too 0.01 1,000 0.500 0.002__—0.502 the coincidence of two particles. Three particle coinci- dence is at least an order of magnitude smaller. The last column, showing total coincidence error, is very litle dif- ferent than the combined error due to double and triple coincidence. For a detector with half the volume of the detector used in Table A.IIL, the particle concentration values for the same error are twice those shown, For a detector with twice the volume of this detector, particle concentration values are half those shown in the table. In the example of Table A.IIl, Ro, Rs, and Rr are expected ratios which will be seen as an average of many trials. Due to the small count numbers that these ratios determine for double and triple coincidence counts, the 95% C.L. limits can be large. As an example, when 1% coincidence operation is selected from this table a maxi mum concentration of 2000 particles/mL. is shown to be the measurement limit. The 95% confidence limits for a count of 2000 particles in 1 mL is 4.36%. The 0.007% ‘occurrence of double coincidence for this example shows that only 14 double coincidence counts will occur in a 1- mL sample volume. For this small number of particles counted, a wide 95% confidence limit results; in this case, 452.9%. One requirement for data uniformity is the selection of a specified coincidence error at a particulate concentra~ tion designed for the mandated test. The (788) test speci- fications as presently written require measurement capa~ bility to 20,000 particles/mL for evaluation of Ys-mL containers. This measurement requirement is commer- cially available, The technical soundness of this test speci- S16 fication is beyond the limi The relationship between coincidence errors and parti- cle concentration shown below has been extended to in- clude the effect of massive concentration overloads of particulates well below the range of clinical or pharmoco- peial interest. This type of concentration overload might result as a consequence of the extraction of plasticizer from parenteral bag material In an extreme case, n could well be equal to 1. For clarity in the following discussion, the diameter of the interfering droplet is assumed to be | zm. The concentra- tion of these droplets, for the detector used in the example above, is 100,000/ml.. The coincidences resulting and the effective diameter of the merged particle are shown in Table AV. The effect of coincidence on the evaluation of particle size varies with the type of detector employed, For Coulter counter type instruments, the particle volume is measured and the diameter is calculated from the sensed volume. The total particle volume resulting, ¥, is: V, = w(D,/6[1 + (D,/D)'1 of both reason and this analy- where D; is the particle diameter and Dz is the diameter of the interfering droplet. For a ratio of 10:1 between these diameters, the measurement effect is negligible. For optical counters, particle size is evaluated by mea- surement of the projected area of the particle. The coinci- dence effect can be calculated from the sum of the areas involved, The total area, Ay, A,= (DY /4{1 + (Dy/D,)") where D, and Dz are, as above, the particulate and the droplet diameters, successively. The effective diameter of multiple coincident droplets is examined in Table A.1V where the slow rate of increase of the effective droplet coincidence signal is clearly evident. ‘The effect ofthe coincident signal on the measured diame- ter of 10- and 25-zm particles is examined in Table A.V for both the increase in size and the probability that this increase will occur given the volume of the detector and the concentration of 10,000 particulates/mL chosen for this example. In Table A.V it can beseen that the disturbing effect of the droplets on the measured diameter of even the 10-zm diameter is negligible, When 2 doubling of the initial TABLE A.IV. Coincident Count Effects for Massive Over- load of I-um Droplets. Detector Volume 107 1am? Droplet Concentration 100,000/m Normalized Equivalent Count Ra Count __Diameter, nam y= Na/N, = 0.500 50,000 1at 0.166 16,667 13 0.0866 a6 2.00 0,00833 833 223 R 0.00140 140 244 R 0.00020 20 265 Ry = Ne/Ni = 0.000025 25 2.83 Journal of Paranteral Science & TechnologyTABLE AY, ‘The Effect of an Overload of 1-um Droplets on the Accuracy of Measurement of 10-and 25-um Particles and on the Probability with which this Increase is Observed Coincident Equivalent Probability Lum Diameter, Size Inerease,% of Droplets TO um 254m Increase, % 1 1.000.499 0.080 36.788, 2 141 0.995 0.160 18.394 3 173 1490239 6.131 4 200° «19803191833 3 223 246 03970307 6 244 293 0.475. 0.051 7 265 3.450.560 © 0.0072 8 2833930639 0.0009, droplet diameter is reached and coincidence witha 10-zm particle measurement occurs the effect is seen as a 2% increase in measured size. When the coincidence effect on 25-uin particles is considered the effect is considerably smaller; the increase due to the doubling of the droplet signal to 2 um increases the measured diameter by 0.319%, Even at the eight-fold coincidence shown in this table, a projected occurrence of 1/100,000 in this exam- ple, the measurement effect is seen to be only 3.93% at 10 hm and 0.693% for 25-um particles. As the coincident Groplet signal grows in size, the probability that it will affect computational accuracy is seen to rapidly diminish Iti, however, reasonable to expect that there could be droplet growth during product storage and consequent effect on the controlled particulate size range. This discus- sion has also neglected the possibility that turbulent flow conditions in the sensor could lead to inelastic droplet collisions and, therefore, droplet growth during measure- ment, For those who considered particulate analysis in the 2- to 5-zm region to be an adequate indicator of the particu- lates below the mandated measurement span, this has clearly been shown to be ineffective. A choice between data from laser-based instrumentation or chemical analy- sis for quality assurance control of this condition appears tobe indicated. The presence of particles below 5 um in major dimen- sion, while of interest as process control indicators, are not of pharmacopeal or present clinical concern, The data analyzed in Table A.IV supports the conclusion that the ‘major effect of high concentrations of particle in this size range would be a distorted analysis of particles in the 5. 10-1 range. This distortion is unlikely to affeet the com- pendial test results. Error contributions from particles between S and 10 jm can be counted as double or triple their correct size and thus enter into the controlled count of particles greater than 10 um. Those particles in the size range between 10 and 25 wm have greater potential effect on compendial measurement limits than the I-zm droplets whose effect has just been reviewed. Coincidence errors in this size range should be regarded with concern since the control limit is ten-fold lower than that for the smaller particles ol 42, Supplement 1988 For instance, the coincidence of two particles over 17.73 um in diameter will result in the detection of a 25-um particle signal. Other particle pairs in this size range whose coincidence will result in a 25-um particle signal are: 3 and 24.82, 4 and 24.67, 5 and 24.5, 10 and 22.9, 15 and 20. Conclusions The effect of the error in particle size evaluated duc to concentration has been reviewed. The range of particle sizes that can contribute to this kind of error is from 5 to those less than 25 um in major dimensions. The false transformation of particle sizes merits careful consider- ation and adjustment by dilution when it occurs. Theonset of this kind of error can be determined by a review of the concentration of particles measured in the test solution against data such as in Table A.III. The methodology described facilitates selection of particulate size analysis within any selected coincidence error limit. Particulate ‘measurements that exceed a specified error limit from coincidence effects should be considered invalid. This is especially so since it is impossible to deduce a valid correction factor and avoidance of the error is so simple. The analysis and conclusions presented are valid for ‘those suspensions in which the distribution of particulates can be adequately described as an average per unit volume of liquid, This requirement can only be satisfied by sti ring of the solution through the entire measurement peri ‘od at a rate to achieve suspension of the full range of particle sizes and densities to be found within it. The absence of any review of the rate at which particulates enter the detector can seriously impair the accuracy of the ‘measurements that have been made. A system adjusted to operate with a 5% coincidence error when the solution is well stirred can achieve an unacceptable 25% coincidence error if the particulate burden is concentrated in the last 20% of the solution to be evaluated because of sedimenta- tion. Considering the present lack of attention to this problem it may well multiply the major sampling errors that have been recently described (10). Acknowledgments Dr. Lee Abramson’s clarity and helpfullness were es- sential to the ideas presented in this paper. Bibliography 1, Wales, M, and Wilston, J. Ni, “Theory of coincidence in particle eounters." Ree. Set. Instrum, 32(10), 1132 (1960) Kubitschek, H. E, “Loss of resolution in Coulter counters,” Reo Sei Instrum, 383), $16 (1962). 3, Berg, Rober I, “Sensing zone methods in fine patil size analy sis" Mat Res. Standards, 83). 119 (1963) 4. Prince, L. Hand Kwolek, W. F., “Coincidence corrections for parle size determinations with the Coulter counter,” Rev. Se. Instrum, 36 646 (1965), Pisani, JF. and Thompson, G. H., “Coincidence corrections for parle size determinations with the Coulter counter” Ret. Sc ‘rsa, 36,634 (1963). 6 Lieberman, A. "Flow rate and conentaton effects in automatic patil counters” Proc. Natl. Cov. Fluid Power, Chieago (1975) 17, Raaseh, 1 and Umbaver, H.,“Ertors inthe determination of par- tcl see dsteibutions caused by coincidences in optical particle counters” Pert. Char. 1 85 (1984), 878, Personal communication, Lee Abramwon, Statistical Consultant, Wastington, D.C ipele, R.- and Myers, R., Probability and Statistics for Eugi- inert and Scientists, 2d ei, MeMillan Pub, Co. Ine, New York NY. 1978, 10, Kushner, H. K. Abramson, L. Rand Knapp, 1.2. "Implications cof sampling theory,” Presented atthe May 1987 Meeting on Liquid Borne Particle Inspection and Metrology in Washington. D.C Appendix Ill: “Ideal” Particulate Measurement System ‘The “ideal” Particulate measurement system described below is, necessarily, a compromise. The specifications are a balance between capability and cost increase. They have drawn heavily on the best that has been achieved in the commercially available instrumentation and have in- cluded extrapolations that would add additional desirable capability. To achieve enough instrument volume to justi- fy the expenditure of development funds, the require- ments for SVI and SVI have been merged ‘An essential next step is the solicitation of design/ price responses from responsible instrument manufacturers followed by a joint user/vendor review of the proposals This joint review should be followed by the emergence of instrument system specifications that will be incorporated into the Pharmacopoeia (788) particulate quality mea: surement requirements. An essential future action is the establishment of a framework within which design asser- tions can be objectively evaluated against performance to avoid perpetuation of the present difficulties Sample Processing Capability ‘Sample procesing capability should be a minimum of 100 mL/min for each input channel. The input channel should be designed and tested to provide fully developed laminar flow in the detector. The >100-mL /min volume process- ing capability needed to eliminate container sampling for the bulk of SVI requirements should be available through use of a multiple input system with up to 5 input channels, The multiple input channels will share microcomputer control, analysis, and data output capability 318 ‘System Capability 1 The sample handling sizing and counting efficiency of the entire particulate analysis system should be in ex- ess of 98% from 4 to 100 um for particle densities from 1.05 to 8.03 at the specified flow rate. 2, System resolution should be at least 5% for the smallest particle size in the measurement range. A 128 channel analysis capability is required for this extended range. 3. The sensor should be capable of analyzing 6500 parti- cles/ mL with a maximum of 5% coincidence error. 4, Sample handling containers and agitators which are compatible with the particle sensor should be provided. The performance of these accessories should be certi- fied to provide homogenous particulate input over the entire range of particle sizes (4-100 zm) and densities (1.05-8.03), 5. The system should have the capability to analyze and display the (788) described maximum particle dimen- sion for spheres, irregular shapes (ACFTD), flakes, and fibers. 6. Thesystem should be capable of automated calibration swith monodisperse spheres. 7. The system should be capable of 2.5% flow accuracy and -£1% volume accuracy using a stepper motor sy- ringe pump. Liquid pressure pulsations should be fil- tered to 1% at all flow velocities required for particu- late analysis. 8. The signal-to-noise ratio should be 100:1 for all mea- surements from 10-100 um. At the 4-um lower mea- surement limit the signal-to-noise ratio should be no lower than 20:1 9. Isolation should be provided on power and signal input lines to ensure that particle sizing and counting shall be unaffected by electrical noise spikes. System Maintenance 1. The system assembly should be modular with certified mean time between failures of 2500 hours. 2. Remote diagnostic capability should be available to facilitate user replacement of plug-in or bolt-on mod- ules, 3. The sensor light source should be rated for either 3000- hour life or be capable of user field replacement and alignment without special tools or fixtures within a one hour period Journel of Parenteral Science & TechnologyPart II—A Selected Annotated PATRICK P. DeLUCA*, BICE CONTI", and JULIUS Z. KNAPPt jiography on Particulate Matter * University of Kentucky, College of Pharmacy, Lexington, Kentucky. 1 University of Pavia, Pavia, Haly. Research & Development Associates, Inc, Somerset, New Jersey ABSTRACT: This selected Annotated Bibliography on Particulate Matter represents a summary of the information obtained from a literature search on particulate matter. This information was necessary 10 develop an unbiased professional evaluation of the instrumentation now marketed for the analysis of particulates in parenteral solutions. As such, special emphasis has been placed on instrument evaluations and comparisons. |. Particle Counting (Size Analysis) A. General 1 Physics of particle size analys 3, (1954, ‘A compilation of papers on: a) a comparison of methods for particle size analysis, b) scattering absorption of light, c) pho twestnction measurements on spherical particles ) the signi cance and application of shape factors, e) a particle profile est, Strip for microscopesly assessing the accuracy of sizing iregu larly shaped particles, fa survey of the automatic couating and sizing methods, ) th theory of particle sizing and counting by trace scanning, ) some fundamental aspects of particle count ing and sing by ineseans,i the automatic size analysis of dust ‘deposits by means of an illumination sit, the automatic count- ing of red blood cell, and ) testing a counting machine, 4) A comparison of methods for particle size analysis (p. 21) ‘An investigation af sedimentation methods of incremental iy, to analyze particle sizes over range of 2-83 and 20.76 um (Course dusts). The following table surnmarizes the methods; Brit. J. Appl Phys. Suppl, Particle SizeRange, Concentration, Accuracy Method um % & Pipette nts 1 High Hydrometcic 0-25 001 215 035 oor 225 Diver 0-25 O01 21 Root Diver 10-53 001 15 Manometric ‘Open 1s als Closed 533 a3 Photoextinction® 1-25 = 46 Electrical Resistance? 15-40 = - = Measures optical density ofa suspension from which size vejght analysis ean be made ‘Measurement ofthe change in resistance ata suitable sam pling depth, gives a dust concentration curve 1) Scattering absorption of light by particles (p. 64). Survey of theories about scattering and absorption of ight by particles, Discussion of Mie theory and Mie effect for large and very small particles (004 jim) using Raleigh scaitering, The seatcering coefficient fluctuated with decreasing amplitude inthe diffrac- tion region, The instrument was calibrated by measuring the ‘ross sectional area of particles using axial ight extinction, the calibration was simpler for opaque particles than for transparent particles, Dr. Contihelda Post Doctoral Scholar appointment atthe University of Kentucky, "Astersked articles sre surnmarized Vol, 42, Supplement 1988 Inthe Raleigh region the sensitivity of measuring smal pri les in the presence of large ones wat increased by varying the Screening coefficient with wavelength, For Irum particles xrays were used as the illumination 6) Photoextinetion measurement on sphercol parties (p. 71) Using incident radiation a total seatering coefficient iealelat- co sing the folowing equation, he tog “2 optical density (&q.) ‘optical density = K log e"*7 (Eq.2) isthe otal scaterng ooefcient and canbe calculated wsing tithe the theories of Mie or Vandetulst. It depends onthe ato, ofthe radius of particles, rand the wavelength ofthe radiation {5 well asthe reactive index ofthe parle relative to the Surrounding medium, Examples are given using Barium sulfite and Lyeopodium pyriforme spores. 4) The significance and application of shape factors in particle Size anatyts (5.82). By Ineasuring the thickness of individual fubseve particles with the optiea! microscope and the dimen- sions of the projected images, surface and volume shape factors ‘were calculated. An expression was derived for the ratio of the particles projecte diameter tots Stokes’ ameter. ‘A large variation inthe shape facors between particles wes foand for col dust and for particles of sme other materials “The largest partis of ssubsive fraction ofa powder have shape factors different fom those ofthe smaller sizes. ©) A particle profile test srip for assessing the accuracy of Siving regularly shaped particles with microscope (p. 105) ‘A patil profile test strip thal can be viewed microscopically Bives a realistic impresion of black dust and « quantitative estimation of observer bias. Tecan be used Tor assessing the rors of sizing irregularly shaped particles, using a globe and citcleeyepicegraticule f) Surcey ofthe automatic counting and sizing of particles (p 121): Presents the principles on which some automatic counting devices are based. Particles were counted by seaming two spots Problems encountered included randomness ofthe sample coin. cidence and counting very small prices. 8) Theory of particle sing and counting by tack scanting (p. 125), A microscopic method using» photoelectric detecting system with high speod pulses allowed forthe measurement of projected area, Problems encountered included the counting nd Sizing of nonspherical parle, coieidence, and over. 1) Some fundamental axpects of particle counting and sizing 2by line soon (p. 133). A method of using line seans (0 size articles which approach the limit of optical resolution as eseribe, ‘Scanning by spot—Use ofa simple on-off detector Scanning by spots—Use ofa simple on-off detector Scanning by sit—Use of simple on-off detector By relating derivable information to observed daa of frequency interception of patiles by lines scans and combination of ines Scans ood accuracy was reported. S191) The automatic size analysis of dust deposits by means of an Ittwmization slit (p. 143). A encroscopte comparison of the ‘rious methods for automatic counting and sie analysis of Sample produet pulverized to microscope sizes using photogra- ‘py. The principle was to detect and count diserte light im- pulses using a photocell. Erors were attributed to fractional fotcteeption of particles. overlap coincidence, and very small particles. Statistical solutions to these problems were described, More precise size distributions were reported Oy Wsing (wo images at diffecent magnifications. An apparatus based on sut- face area determination was described 5) The automatic counting of red blood cells (P- 147). An Apparatus in which the blood cells are scanned by mecha ‘oscillation ofa microscope stage was described. The novel shape ofthe scanning aperture was found to influence the errors ars ing from various sources. The results in counting blood cells were reported reproducible within 3%. i) Testing @ counting machine (p. 161). Particle sizing and counting machin results are compared withthe results obtained ‘when sizing with 2 microscope. A machine which measures the size frequency distribution of chords oblained from moving siains obtained by spraying a black dye at uniform velocity in Front of alight sensitive detector was compared withthe miero- ‘cope. From the total numberof chords in each of several ranges it tae posible to compute the approximate number of circular stains from which they were derived. The counting machine measured quickly, with an accuracy better than 1% ofthe range ‘which was felt tobe lower than that associated with eye count ing Groves, M.J.""The size distribution of particles contaminating parenteral solutions,” Analyst, 94, 992 (1969). ‘One of the earlier studies in which the log-log distebtion was tested. Forty-five samples of injection BP from and it wos dificult ocorelate microscope and Coulter Count for particles <15 jim. Te was almost imposibie 1 distinguish small particles with refractive index similar tothe membrane jon within a batch was noted. To avoid the effect of ‘vacation within a batch an index of cleanliness was sugeested 3. Apt, B:K.."Pariculate matter in intravenous infusions, East. Pharm. 15(176), 27-31 (1972) 4. Emnert, ., "Studies on the problem of particulate mater in parenteral products,” Sten. Farm. Tidskr, 78, 129-139 (Feb, 1973). 5. Akers, R. J, Lyd, P. J. and Scares, B. “European Sympo- ‘ium on particle size measurement." DECHEMA-Monogr., 79, Pat 8, 191-208 (1973), This was a review of the history of automatic microscope Particle size analysis systems, The semiautomated Zeiss-Endter particle size apparatus was described in which ercles are super- Imposed o macroscopic images. It was reported to be relatively fee of erors. An automatic microscopic technique using a Leitz CClassimat, Zeiss Videomat was deseribe. This involved genera tion ofa Fectangulr rater formed by progressively moving & scanning soto” sit over the image and was reported 10 be quite Stitable fr routine measurement. ‘A digital computer particle size analysis system was the ‘Quantimet 720 scanner system with & high quality vidicon or plumbioon camera system capable of resolving afield of 880 x {S88 picture points with high sabity, This ublized TTL digital logic circuitry and a FORTRAN subroutine to evaluate particle parameters, The disadvantages of automatic image processing systems were reported. These included inability to sole problem of {ouching and overlapping particles of arbieary shape. A ‘pen was used to identi specitic images 6, Davis, PJ. Dasley, R. W. and Patel, N.. "Particulate contami: ration in parenteral solutions.” (Procesdings) J. Pharm. Phar ‘macol. 31, Suppl. 59P (Dec. 1979) 7. Longe, &. 1, "Particulate contamination in selected parenteral frogs,” Can. Anaesth, Soe. J, 21(1), 62 4, (1980). . Andrews, D.. "Particulate contamination in thiopental solu sions." Can. J. Hosp. Pharm. 31031). 3 (1988), ‘Seealso Reference 67 B. Large Volume Solutions 9, Lines. R. W., “Counting of particulate matter in parenteral solutions: 1. Survey of the literature,” Bull. Parenter. Drug’ Assoc. 244). 113-17 (Review) (1967), Davis.N. M. Tureo,S..andSivelly EA study of particu ‘matter in iy. infusion fluids.” mt. J. Hosp. Pharm. 21, 82 826 (Oat. 1970), Darby, TD, aad Ausman, RK. “Particulate matter in polyi- nyt chloride intravenous bags (Cont.),” NEW. Med. 290, 579 (1975). Hayashi, T. “Occurrence and size distribution of maiter in parenteral solutions,” Yakuzaigaku 40, (Galy 15,1980), Japanese “Messerschmidt, W.. "Particulate contamination in infusion so- lutions,” Krankenhauspharmazie 1, 2), 24-26 (Oct. 1980), German. Caramella, C., Montanari, L, Pavanetto, F. and Ponci, R “Research on particle contamination of injectable solutions.” Farmaco, Ed. Prat. 36, 148-16| (Mar. 1981), lalian, 'A comparison of the Coulter Counter with mierascope using 35 commercial solutions, Al solutions met the USP LVP limits, ‘while only 9 complied with British Pharmacopoeia mis, Using microspheres asa ealibrant no difference was found between the ‘microscopic and Coulter Counter methods. With ACFTD as a calibrant, differences of 25% were found atthe 10am size: the iference increased with increasing size of paris. I was. concluded that the Coulter Counter should be limited to 10-am size partes. Broussali-Tsiftsoglou, T.. and Iconomeu-Petrovitch, N. G., "Methods for detection and quantitative determination of par- ‘cule matier in parenteral solutions,” Pharm, Delt. Epistem. Ekdosis,8(1), 73-89 (1982). Budgen, C.J, and Frost, L, "Reuse of glass containers for Irrigation solutions.” N.2. Pharm. 3:1) 45-46 (1983). “Montanari L. Pavanetto,F and Pon, R. “Investigation of foceign particle contamination in high volume injectable so} tions and in powders for injectable solutions. Proposals for Ital- Tan reolatons" Farmaco Ed. Pra37,397-407 (De. 1983), alin ‘Sixteen LP batches from twelve manufacturers end ten dry powders from nine manufactures were studied using the Cou ter Counter. Eleven of the sixteen LVPs passed USP limits, while four ofthe sixteen passed the British Pharmacopeia. Besedina, 1. V."Conductometrie method for determination of the range of particle dimensions in injection solutions,” Farmat- siya (Moscow) 33(1), 30-22 (1988), Russian Maines-Nutt, RF ané Muaton, T-1., "Particle size measure: ment in intravenous Nui.” J, Pharm. Pharmacol, 36(8),534— 6 (1984). Lage varacons were observed ina comparison ofthe HIAC and Coulter Counter for monitoring LVPS. Since variations ‘ould not be explained on bass of shape factors, it was coneluded the correlation ofthe twe methods was not posible Coulter counts were an order of magnitude higher at am and from 016 to 5X at 5 gem. The differences between HIAC and Coulter were not comparable with salt and sugar slutions. For small particles, the HIAC was found to count les than the Coulter due to forward scattering For particles with refractive index close tothe solution, the HLAC again counted les. Meserschmidt, W., "Extent and accepiable limits of partion: late matter in infusion solutions,” Krenkenhauspharmaste, 5, 277-282 (Sep, 1984), German, See lso References, 64, 90.and 98, C. Small Volume Solutions a 2. 2. Lines, R. W., “Counting of particulate matier in parenteral solutions” I. Partcie counting in small containers.” Bull. Par- enter Drag Assoc, 21(8), 118-123 (1967). ‘One of the early studies on particles using the Coulter ‘Counter. Good reproducibility was found in the various solations studied. Handling and sampling were more eritcal with small ‘volume parenteral. Dungan, D. J, “Particulate contamination in pharmaceutical preparations for injection,” Aust. J. Pharm. 40,($84),S59-S64 (Aug. 1968), Somerville, T.G., and Gibson, M., “Particulate contamination ‘inampuls," Pharm. J. 211, 128-130 (Aug. 18 1973) ‘This was a comparative analysis of S0-am particles ia am- poules using a Polarized light viewer (Allen type 208/2) and a ‘Coulter Counter (Medel T, 70:gm orifice). Both techniques showed a spread of eadings but sherewasa reported comparison Journal of Parenteral Science & Technology