JX01-243@连续卧式热成型-充填-封口包装机
收藏
资源目录
压缩包内文档预览：(预览前20页/共43页)
编号：492551
类型：共享资源
大小：1.80MB
格式：ZIP
上传时间：2015-11-07
上传人：QQ28****1120
认证信息
个人认证
孙**（实名认证）
辽宁
IP属地：辽宁
30
积分
- 关 键 词：
-
机械毕业设计全套
- 资源描述：
-
JX01-243@连续卧式热成型-充填-封口包装机,机械毕业设计全套
- 内容简介：
-
European Journal of Pharmaceutical Sciences 21 (2004) 493500Determination of the thickness of plastic sheets usedin blister packaging by near infrared spectroscopy:development and validation of the methodMagali Laasonena,b, Tuulikki Harmia-Pulkkinenb, Christine Simardc,Markku Rsnend, Heikki Vuorelaa,aDepartment of Pharmacy, Division of Pharmacognosy, University of Helsinki, P.O. Box 56 (Viikinkaari 5E), FIN-00014, Helsinki, FinlandbPharmia Oy, P.O. Box 387, FIN-00101, Helsinki, FinlandcABB Bomem Inc., 585 Charest Boulevard, East Suite 300, Que., Canada G1K 9H4dDepartment of Chemistry, Laboratory of Physical Chemistry, University of Helsinki, P.O. Box 55 (A.I. Virtasen Aukio 1),FIN-00014, Helsinki, FinlandReceived 21 May 2003; received in revised form 24 October 2003; accepted 17 November 2003AbstractA near infrared (NIR) quantitative analysis method was developed for determining the thickness of PVC-based plastic sheets used aspharmaceutical packs. Samples that can be analyzed are transparent films made of polyvinyl chloride (PVC), PVC coated with polyvinylidenedichloride (PVDC) or PVC coated with Thermoelast(TE) and PVDC. The method, based on a partial least squares (PLS) algorithm, is usedtogether with a previously developed NIR identification method to acquire simultaneously qualitative and quantitative information about thesamples. Validation of the quantitative method was conducted according to the very recent European Agency for the Evaluation of MedicinalProducts (EMEA) guidance on the use of NIR spectroscopy. Suggestions were made for a better statistical evaluation of the calibrationmodel prior to validation. Validation consisted of the study of specificity, accuracy (mean recovery from the reference values was 99.56%),precision (repeatability and intermediate precision were 0.6%), linearity, quantification limit (41H9262m), and robustness of the method. Thisdemonstration of the applicability of NIR spectroscopy as a validated quality control tool for pharmaceutical packaging films will hopefullyfacilitate the acceptance of NIR spectroscopy in pharmaceutical laboratories. 2004 Elsevier B.V. All rights reserved.Keywords: Near infrared spectroscopy; Validation; Pharmaceutical packaging films1. IntroductionThe packs used for pharmaceutical products are an inte-gral part of the dosage form design. They usually consistof a primary packaging, in direct contact with the pharma-ceutical form, and of a secondary packaging that is often apaper-based material. The primary pack is the most impor-tant in terms of protection of the active principle. It must pro-vide protection against climatic (e.g. moisture, temperature,pressure, light), biological (e.g. microbiological, adulter-ation), physical (e.g. shocks), and chemical hazards (Dean,2000), and prevent the loss of active substance through theCorresponding author. Tel.: +358-19159167; fax: +358-19159138.E-mail address: heikki.vuorelahelsinki.fi (H. Vuorela).packaging material. In the case of herbal medicinal productsin solid form, for example, the role of the primary packagingis of prime importance because natural products are espe-cially sensitive to moisture. Blisters are often the packagingforms for tablets and capsules. Ensuring protection presup-pose that the correct thermoplastic resin is chosen and itsquality controlled before thermoforming the blisters.The quality control of plastic sheets should consist, at theminimum, in controlling the appearance, dimensions, den-sity, and identification of the film. However, physical prop-erties such as thickness are also very important to ensurethat the barrier performance of the film will be optimal. Tra-ditional methods for the identification of plastic containersare often laborious procedures. For example, infrared spec-troscopy is the recommended technique in the EuropeanPharmacopoeia (2002) for the identification of polyvinyl0928-0987/$ see front matter 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.ejps.2003.11.011nts494 M. Laasonen et al. / European Journal of Pharmaceutical Sciences 21 (2004) 493500chloride (PVC) for dry dosage form containers. Perform-ing the procedure is rather time-consuming (about 2 h persample), and involves the use of a harmful organic solvent(tetrahydrofuran). In addition, this monograph can not beused in the case of coated PVC films such as PVC coatedwith Polyvinylidene chloride (PVDC). Moreover, it can onlybe used to control the chemical properties of the sample andadditional procedures have to be used to control the physicalproperties.Compared to traditional methods, NIR infrared spec-troscopy is a very versatile tool. Its advantages over tradi-tional analysis methods can be summarized as follows: (1)high speed of analysis (less than 2 min for a duplicate sam-ple); (2) in most cases it does not require sample prepara-tion; (3) molecular level chemical and physical informationare available simultaneously; (4) training an operator toperform the analysis takes less than 20 min. Moreover, NIRspectroscopy has already proved its usefulness in control-ling the identification of plastic materials and in measuringcertain physical properties of plastics.NIR spectroscopy can be used for the on-line iden-tification of food packaging waste e.g. PVC, Polyethy-lene (PE), Polyethylene terephtalate (PET), Polypropylene(PP) or Polystyrene (PS) (Feldhoff et al., 1995, 1997;Schilling and Ritzmann, 1995) or for the off-line identifica-tion of PVC-based sheets for pharmaceutical applications(Laasonen et al., 2001). It can also be used for determina-tion of the thickness food packaging films (Davies et al.,1985; Miller et al., 1993). However, so far NIR has notbeen used to determine the thickness of plastic sheets forpharmaceutical packs.The physical phenomenon responsible for the correlationbetween the thickness of the sheets and the NIR spectra canbe explained on the basis of the BeerLambert law. Thislaw states that the fraction of radiation (A) absorbed by asample at a given wavelength is proportional to the concen-tration (c) of molecules in the sample, the absorptivity (a),and the thickness (b) through which the radiation passes,i.e., A = abc. However, caution must be taken when us-ing this formula. In fact, the BeerLambert law does notalways adhere to true absorption measurements. One of thereasons is that it does not take into account the effect ofscattering on the absorption, and assumes complete absenceof reflection. So, in the case of diffuse-reflectance or trans-flectance measurements, the KubelkaMunk theory may bet-ter fit the actual absorption as it also takes into accountthe scattering coefficient of the material (Osborne et al.,1993).The present study aims at demonstrating that NIR spec-troscopy can be used as a combined analysis tool forPVC-based films for pharmaceutical blistering applica-tions: using the same spectra, qualitative (identification ofthe film) and quantitative (determination of the thickness)properties of the films can be analyzed simultaneously. Val-idation of the method was performed according to the veryrecent EMEA guidance (February 2003). However, somesuggestions were made for a better statistical evaluation ofthe calibration model prior to validation.2. Materials and methods2.1. MaterialsThe plastic films were collected over a period of twoyears from a large number of batches, several suppliers, andseveral countries, so that the batches included in the cali-bration set can be considered sufficiently representative tocover the normal variation of these films. The five suppli-ers were: Perlen Converting AG (Switzerland), Aerni-LeuchAG (Switzerland), which is the production unit of KlcknerPentaplast GmbH & CO. KG, Paskel International SA (Mex-ico), Solvin SA (Belgium), and Riflex Film AB (Sweden).The samples collected for the calibration and validationset were of several different types. The so-called “true” sam-ples (A, B, C- types) were samples of the required quality(based on polymer type and thickness criteria) for blisteringpurposes. The other types of samples (At, Bt, and Ct- types)were collected in order to increase the thickness range ofthe collected true samples. The sample description was thefollowing:A-type: clear transparent rigid 250H9262m PVC film; At-type:identical to A-type except for the thickness of the PVClayer (other than 250H9262m); B-type: clear transparent 250H9262mPVC film coated with 40 g m2of PVDC, with a nominalthickness of 273H9262m; Bt-type: identical to the B-type exceptfor the thickness of the PVC layer and/or the amount ofPVDC coating (nominal thickness was other than 273H9262m);C-type: the last type of true sample was clear transparent250H9262m PVC films coated with 5 g m2of Thermoelastand90gm2of PVDC, with a nominal thickness of 308H9262m;Ct-type: identical to the C-type except for the thickness ofthe PVC layer and/or the amount of TE and PVDC coating(nominal thickness was other than 308H9262m).We also obtained a range of samples (D-type samples ofpolymer composition other than the A, B or C samples, e.g.polypropylene (PP) or PVC/PE/PVDC films. These filmswere only used to control the specificity of the method. Atotal of 193 batches of plastic film were used in the study.Most of the samples were the same as those used previ-ously for the development of a NIR identification methodfor PVC-based films (Laasonen et al., 2001).2.2. Infrared analysis and thickness measurementsThe identity of all the samples was confirmed by in-frared spectroscopy in the range of 4000400 cm1(Spec-trum One, Perkin-Elmer).The film thickness of all the samples (average of sixmeasurements per batch) was measured using a cali-brated digital micrometer (Digimatic micrometer, Mitutoyo,Japan). All sample thicknesses were in accordance with thentsM. Laasonen et al. / European Journal of Pharmaceutical Sciences 21 (2004) 493500 495specifications. Precision of the thickness reference methodwas assessed by calculating the standard error of laboratory(SEL) (EMEA, 2003).SEL =radicalBiggsummationtextmi=1(x1x2)2mwhere m is the number of batches, and x1x2is the differ-ence between the values measured by different operators.SEL was evaluated on three different days and the averagevalue was found to be 5.0H9262m.2.3. Near infrared spectroscopic measurementsAs the spectroscopic measurements were performed inthe same way as for the identification method of the plasticsamples (Laasonen et al., 2001), the spectra are usable forboth the identification of the film and the determination of itsthickness. The configuration used a Fourier transform nearinfrared (FTNIR) spectrometer MB 160 DX (ABB Bomem,Inc., Quebec, Canada) and the Powder Samplir reflectanceaccessory. The software package was from ABB Bomem,Inc.: Grams 32 version 4.04, PLSPlus/IQ version 3.03, andAIRS version 1.54. For spectral acquisition, each film wasplaced on the beam of the Powder Samplir Accessory with aSpectralon 99% reflective standard (Labsphere, Inc., NorthSutton, New Hampshire, USA) located on top of the film.The diffuse reflectance mode was used. The combined pro-cedure (identification and thickness measurement) can beperformed in less than 2 min for a duplicate sample.2.4. Model developmentSystem suitability tests, including frequency, spectralquality, spectrophotometric noise, photometric linearity,and precision testing, were performed regularly on the NIRspectrometer before any analyses were performed.The samples were divided into calibration set and valida-tion set. The division was performed such that the calibrationset (CS) included transparent films covering the largest pos-sible variation in thickness (approximately 190340H9262m),and derived from a range suppliers. The D-type films wereonly included in the validation set (VS) for the specificitystudy. As a result, CS contained 18 batches, and VS 175batches. This calibration set was constructed by setting themeasured thickness to each spectrum and then implement-ing the data into the PLSPlus/IQ software.3. Results and discussion3.1. Pre-processing and data analysisThe quality of the calibration was optimized by choosingpretreatment options according to the same criteria as de-scribed in Laasonen et al. (2001) but for a different objective.In the present method, the pretreatment options were cho-sen such that the parameter to be calibrated was the samplepathlength (thickness), i.e. a physical property of the sample.The aim was therefore to control the light scattering with-out affecting the effect of the thickness on the spectra. As aresult, the pretreatment options chosen were only the meancentering and 9-point second derivative of SavitskyGolay.The method was primarily optimized for the spectral re-gion selection. Three tools were used to perform the selec-tion:(1) Raw spectra from samples of different thickness wereplotted and the regions showing the best correlation wereinvestigated (Fig. 1).(2) Second derivative spectra of different thicknesses wereplotted and the regions showing the best correlationwere investigated. (Fig. 2).(3) During the calibration using the PLS algorithm, thecorrelation spectrum was plotted together with the cal-ibration set spectra. The correlation spectrum showsthe correlation of the absorbance at every wavelengthto the thickness of samples. Regions for which the cor-relation coefficient was almost equal to 1 indicated agood correlation between the spectral absorbance andsample thickness.The selected region was 58715647 cm1and 43973996 cm1and contained 83 data points at a resolution of16 cm1. This region mainly contains absorption bands ofthe CH stretching first overtone, CH deformation 2ndovertone, and a number of useful combination bands: CHstretching and CH deformation, CH2symmetrical stretch-ing and =CH2deformation and, finally, CH stretching andCC stretching (Osborne et al., 1993).The method was constructed using the PLS model andcross-validation. The number of significant PLS factorsfor each method was chosen as described by Haaland andThomas (1988). Using this criterion, the resulting numberof factors was found to be eight, which is relatively highand leads to a potential risk of overfitting the model, e.g.modeling the system noise. Nevertheless, the model per-formance was found to be suitable for this application (seesection 4 and validation results), and therefore the factornumber was not modified.The PLS Loadings of the first three factors were inves-tigated. They showed characteristic NIR features of vinylicstructures (Fig. 3). Factor 1 exhibited main absorptions atabout 5778 and 5825, assigned to CH stretching, first over-tone (CH2groups), and 4335 cm1, assigned to the com-bination band of CH stretching +CH deformation. Fac-tor 2 exhibited main absorptions at about 5778, 4328, and4373 cm1. These bands can respectively be assigned toCH stretching, first overtone (CH2groups), and combi-nation bands of CH stretching +CH deformation (CH2and CH2groups). Factor 3 exhibited main absorptions atabout 5778, 4050, 4250, and 4297 cm1. These bands canrespectively be assigned to CH stretching, first overtonents496 M. Laasonen et al. / European Journal of Pharmaceutical Sciences 21 (2004) 493500Absorbance,Log (1/R)At-type sample, 300 nmA-type sample, 250 nmAt-type sample, 200 nmAt-type sample, 300 nmA-type sample, 250 nmAt-type sample, 200 nmAbsorbance,Log(1/R)4400 43004200 4100 400.8160005900 58005700 560.3(a)(b)Fig. 1. Absorbance spectra from transparent PVC samples of differentthicknesses in the 60005500 cm1(a) and 45004000 cm1(b) regions.A-type samples are clear transparent rigid 250H9262m PVC film. At-typesamples are clear transparent rigid PVC with thicknesses other than250H9262m.(CH2groups), combination bands of CH stretching +CCstretching, CH deformation, second overtone, and combi-nation bands of CH stretching +CH deformation. Otherfactors are not described as their contribution to the modelis more limited.These first three factors characterize, using differentloading values, the vibrations present in PVC and PVDCstructures. The spectral differences in loading values can beexplained by the simultaneous presence in the calibration setof samples containing only PVC, samples containing PVCand PVDC, and some containing PVC, PVDC, and TE. Inaddition, the physical properties of the samples vary as someSecond derivative ofLog (1/R)Second derivative of Log (1/R)Bt-type samples, 220 nmB-type samples, 273 nmBt-type samples, 330 nmBt-type samples, 220 nmB-type samples, 273 nmBt-type samples, 330 nm4350 4300 4250 4200 4150 4100Wavenumber / cm-1Wavenumber / cm-1-.03-.02-.010.1-.01-.0050.00559005800 57005600(a)(b)Fig. 2. Second derivative spectra from transparent PVC/PVDC samples ofdifferent thicknesses in the 60005600 cm1(a) and 44004000 cm1(b)regions. B-type samples are clear transparent 250H9262m PVC film coatedwith 40 g m2of PVDC, with a nominal thickness of 273H9262m. Bt-typesamples are identical to the B-type except for the thickness of the PVClayer and/or the amount of PVDC coating.of the samples are monolayer, whereas others are bi- ortri-layers. The samples also came from a number of supplierswho used different manufacturing processes. This explainswhy several factors are needed to extract all the chemicaland physical information present in the calibration set.3.2. Calibration model performanceThe quality of the developed model was checked by evalu-ating the calibration equation, and calculating, among others,the standard error of cross calibration (SECV), and the num-ber of outliers. The results of the calculations were the fol-lowing: prediction bias =0.03H9262m and SECV = 2.36H9262mand the F-test ( = 0.01) was applied to determine the sta-tistical significance of outliers. No outliers were found inour calibration set.ntsM. Laasonen et al. / European Journal of Pharmaceutical Sciences 21 (2004) 493500 497Factor 3Factor 2Factor 16000 5500 50
- 温馨提示:
1: 本站所有资源如无特殊说明，都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
2: 本站的文档不包含任何第三方提供的附件图纸等，如果需要附件，请联系上传者。文件的所有权益归上传用户所有。
3.本站RAR压缩包中若带图纸，网页内容里面会有图纸预览，若没有图纸预览就没有图纸。
4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
5. 人人文库网仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对用户上传分享的文档内容本身不做任何修改或编辑，并不能对任何下载内容负责。
6. 下载文件中如有侵权或不适当内容，请与我们联系，我们立即纠正。
7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
人人文库网所有资源均是用户自行上传分享，仅供网友学习交流，未经上传用户书面授权，请勿作他用。
川公网安备: 51019002004831号