In-line optical monitoring of polymer injection molding

In-Line Optical Monitoring of Polymer Injection Molding

ANTHONY J. BUR and FRANCIS W. WANG

Polymers Division National Institute of Standards and Technology

Gaithersburg, Maryland 20899

and

CHARLES L. THOMAS and JOSEPH L. ROSE

Department of Mechanical Engineering and Mechanics Drexel University

Philadelphia, Pennsylvania 191 04

An optical sensor, consisting of optical fibers to transmit light to and from the mold cavity, was constructed for the purpose of measuring the onset of polymer solidification during injection molding. The sensor was used to detect characteris tic fluorescence radiation from a dye which had been doped into the resin at very low concentration. By measuring changes in fluorescence intensity it was possible to detect whether the state of the resin was liquid or solid. We observed that, as the resin cooled in the mold, the onset of solidification was indicated by highly characteristic and distinct changes in the fluorescence intensitypme profile. Application of the method involved the use of a calibration relationship between the fluorescence intensity and temperature of the doped polymer in order to determine the distinct features which characterize the onset of solidification. Injection molding of a glass forming polymer (polystyrene) and a crystallizable polymer (polyethylene) was monitored by this technique.

INTRODUCTION

he use of fluorescent dyes in dopant concentra- T tions to monitor the character of a matrix host material is an accepted scientiflc practice (1, 2). Dyes have been used to probe molecular dynamics by chemically tagging them to the host molecule (3, 41, to measure molecular orientation (5, 6), to measure diffusion constants (71, to monitor polymerization (8), and to monitor the degree of cure of a curing thermoset material (7- 1 1). Application of this science to real-time monitoring of polymer processing has also been achieved (12, 13).

In each case, the fluorescent dye is chosen in ac- cordance with the relationship between its spectral characteristics and the material property being examined. For example, molecular rotor dyes have been used to monitor the cure of a thermoset material (91, and to monitor polymerization (8). These dyes, which were first studied by Oster and Nishijima (14) and subsequently by Lou% (15). display fluorescence spectra whose intensity is dependent on the viscosity and temperature in the dye’s molecular neighborhood. An example of a molecular rotor, (1 -(4-dimethylaminophenyl)-6-phenyl- 1.3.5 hexa-

triene) (DMA-DPH), is shown in Frg. 1. Upon absorb ing excitation energy, a molecular rotor can decay to its ground state via fluorescence radiation or by radiationless decay, i.e. energy transferred to molecular vibrations or rotations. For DMA-DPH, the amount of radiationless decay is regulated by rotations about the chemical bond at the end group as depicted by the arrow in Fig. 1. This intramolecular rotational motion creates potential radiationless decay paths if its relaxation time, T ~ , is shorter than or appro& mately equal to the fluorescence decay time T~ of the dye, usually tens of nanoseconds. For T~ * T~ maxi- mum fluorescence radiation and minimum radiationless decay is observed. Lou@ and coworkers have presented a model describing the influence of intramolecular rotational relaxation on fluorescence radiation from molecular rotor dyes (15). They found that fluorescence quantum efficiency is appro& mately proportional to T~ which in turn depends on temperature through the Debye-Einstein relationship,

Tr= C ( T / T ) (1)

where C is a constant of proportionality, 77 is the

POLYMER ENGfNEERING AND SCIENCE, APRIL 1994, V d . 34, No. 8 671

A. J. Buret al.

Fig. 1 . (d The jluorescent molecule, bis4pyrene) propane (BPp), is shown. The arrows indicate rotational jlexibility about the propane linkage. (b) The jluorescent dye, (1 -(4-dt methy lamino) -6-p heny I - 1,3,5 hexatriene) (DMA-DPH) is shown. The arrows indicate rotational mobility about the endgroup.

molecular viscosity or microviscosity in the neighborhood of the dye, and T is temperature. As temperature increases, 7 decreases resulting in shorter 7,. a larger portion of radiationless decay and less fluorescence radiation.

Another example is environment sensitive dyes which are molecules whose fluorescence spectrum is dependent on factors such as electric polarity and correlation time for molecular rearrangement in its neighborhood (16). In this case, environmental changes are reflected in wavelength shifts of the flue rescence intensity maxima.

A third example is excimer producing dyes which emit both monomer and excimer fluorescence. For these, the radiated fluorescence energy is distributed between monomer and excimer modes of decay. An example is bis-(pyrene) propane (BPP), shown in Fig. 1 , which contains two fluorescent pyrene molecules joined by a flexible trimethylene link (9, 17-20). When one pyrene molecule absorbs excitation energy at 345 nm, two paths of fluorescence decay are possible, one, by monomer decay at 380 and 400 nm, i.e. the pyrene molecule displays its characteristic fluorescence without interaction with the other pyrene, and two, by excimer decay in the range 450 to 550 nm which occurs when the excited pyrene forms an excimer with its unexcited pyrene neighbor by rotating to a position of close molecular contact. The probabil- ity that the excited pyrene can rotate into the proper position to form an excimer before its own decay occurs is dependent on the r,, the intramolecular rotational relaxation time of the dye, which is propor-

tional to the ratio q/T. We emphasize again that 7 is a microscopic or molecular viscosity. Thus, for excimer producing dyes as well as for molecular rotors, rotational relaxation time plays a prominent role in the production of fluorescence radiation.

The application of fluorescence spectrometry to monitor polymer solidification during injection molding can employ any fluorescent dye which displays distinct spectral changes as the polymer solidifies. We define solidification to mean the transition from liquid to solid either by the formation of a glass or by crystallization. In this work, we used a molecular rotor dye (DMA-DPH) and an excimer producing dye (BPP) which were doped into a glass forming polymer (polystyrene) and a crystallizable polymer (polyethylene). The dye was mixed with the resins at very low concentrations (10 to 100 pprn by weight) before it was injected into the mold. It is assumed that such low levels of dye concentration will not affect the materials properties of the final product.

EXPERIMENTAL PROCEDURE

Materials

The resin materials were polystyrene from Fina Corp., PS 525P1 in pellet form, and polyethylene from Phillips Corp., Marlex TR885 in pellet form (21). The polyethylene contains an antioxidant which causes ultraviolet absorption in this material. (The effect of UV absorption on our fluorescence observations is discussed below.) The polystyrene did not contain any antioxidants. TWO fluorescent dyes were used: BPP and DMA-DPH. Both of these dyes were pur- chased from Molecular Probes and used as received (21). Their molecular structure is shown in Fig. 1. BPP was excited at 345 nm and fluorescence was detected at 380 nm (monomer fluorescence) and over the wavelength regime between 450 to 550 nm (excimer fluorescence). For DMA-DPH, excitation was carried out at 400 nm and fluorescence emission was detected over the range 470 to 530 nm. The characteristic fluorescent spectra of both dyes can be found in a paper by Wang, et aL who used the dyes as microviscosity probes to monitor epoxy curing (9).

Each dye was mixed with the resins so that concentration of dye in the final molded product would be of the order of 10 ppm by weight. Two different methods were used to add the dye to polystyrene and polyethylene. For polystyrene a common solvent, toluene, was used to mix dye with the resin. An initial mixture, 30 g polystyrene with 600 ppm dye, was placed in solution. A thin film of polymer was prepared by spread- ing the solution onto a glass surface and permitting the solvent to evaporate. The film was then cut into small pieces which were dispersed with undoped polystyrene resin pellets in a ratio that would average to 10 ppm by weight of dye with polystyrene.

For polyethylene, which is insoluble at room temperature, pellets were coated with the dye. This was accomplished by pouring a solution of dye ontc polyethylene pellets and subsequently permitting the

672 POLYMER ENGINEERINGAND SCIENCE, APRIL 1994, Vol. 34, No. 8


monochromator. The excitation light energy was transported to the injection mold via one branch of a

solvent to evaporate. The coated pellets were then dispersed with uncoated pellets so that the average dye concentration in the master batch was approximately 10 ppm by weight. Additional mixing of the dye occurred during the screw translation in the machine and during injection into the mold. Uniform mixing of the dye in the resin was not necessary for these measurements because, as described below, we measure changes in fluorescence intensity rather than absolute intensity. Our objective for the mixing procedure was to achieve a dye concentration of a p proximately 10 ppm by weight, a concentration suffi- ciently large for producing detectable fluorescence.

Other ways of mixing dye and polymer are sug- gested. It can be put into a dilute solution and dripped into the resin hopper on the mold machine. I t can be pulverized and then mixed with the resin powder. It can be compounded with the melt resin using stan- dard polymer mixing equipment.

Instrumentation

Figure 2 shows a diagram of the measurement system which was used. The light source was a xenon arc lamp. The excitation wavelength was separated from the full spectrum of the lamp using a grating

_ _ - - _- - - - - - . - - - \ bifurcated optical fiber cable consisting of a bundle of fused silica fibers. The common end (the sensing end) of the optical cable was inserted into a hollow ejector pin which was fitted with a sapphire optical window at its end (see Fig. 3). During the molding operation,

r optical fiber

the sensing tip of the ejector pin is flush with the

computer via an analog to digital conversion board. The sampling rate for data acquisition was approxt mately 100,'s. For BPP, two wavelength regimes were detected: using bandpass filters, monomer fluorescence (at 380 nm) and excimer fluorescence, integrated over 450 to 550 nm, were measured sepa- rately, acquired by the computer and displayed as a ratio of excimer to monomer intensity us. time.

Calibration Curve

The calibration curve is a fluorescence us. temperature characterization of the behavior of a dye in the environment of the host polymer. For this experiment, approximately 10 g of doped resin were placed in an oven with the optical fiber and thermocouple as shown in Fig. 4. The objective of the experiment was to observe the characteristics of the dye fluorescence as the resin solidifies. For both polystyrene and polyethylene, the oven was heated to 200°C and then cooled to approximately 60°C in 25 to 30 min. During

Y / - - _ _ -

j i I

ejector pin channel \

optical window \

detector consists of monochromators and/or filters as are appropriate for the analysis of the fluorescence signal. We used photomultipliers in conjunction with current amplifiers to detect the light intensity. The analog output of the current amplifiers was fed into a

n mold resin

molding machine k i filters or

light source

I computer I Fig. 2. A schematic of the optical measurement system inte grated with the injection molding mhine is shown

Y cavity

mold plates

Fig. 3. Detail of the optical f b e r as it is positioned in the mold.

/ light Source optical fiber

L r detector

oven Fig. 4. The experimental setup for obtaining thejluorescence us. temperature calibration cunte.

POLYMER ENGINEERING AND SCIENCE, APRIL 1994, Vol. 34, No. 8 673

A. J . Buret al.

cooling both fluorescence intensity and temperature were measured. Cooling was not carried out as rapidly as it is during the injection molding process, but it was rapid enough to obtain the distinct characteristics of the fluorescence which marked the onset of solidification.

RESULTS AND DISCUSSION

Calibration Measurements

Four calibration experiments were carried out, one for each dye in polystyrene and polyethylene. The results are shown in Figs. 5 through 8.

Polystyrene: For BPP in polystyrene, which is shown in Fig. 5, lex/lm, the excimer to monomer fluorescence intensity ratio us. temperature displayed a knee in the curve at the liquid-to-glass transition. The glass transition temperature, Tg , for polystyrene ( a p

0 0 1 1 1 1 1 1 1 1 1 1 / 1 - 1 1 1 1 1 ~ I , ~ , , I , ~ ~ ~ ~ ~ ~ , , I ~ , , , / ' , , / / , , ~ , , , , ' , I 60 80 1007 120 140 160 180

Temperature O C

Q. 5. I e x / I , , , us. temperature is plotted for polystyrene doped with BPP. Measurements were obtainedfrom cooling curue observations. The glass transition of polystyrene, indt cated by the arrow at approximately 107°C. was obtained from the intersection of the linear extrapolations from high and low temperature. The excitation wavelength was 345 nm.

1 0

h 5 2 0 8 al 4 d

- 0 6 al

E 0 4 rn e, L

z

s o 2

E=,

60 80 100 120 140 160 180 Temperature O C

Flg. 6. Fluorescence intensity us. temperature is plotted for polystyrene doped with DMA-DPH. Measurements were ob tained from cooling curve observations. No evidence of the onset of the glass transition is seen in these data. The excitation wavelength was 400 nm.

1 0 h

E 5 0 8 - V

.2 0 6 ld rx h 0 4 s:

4

2 2 0.2 d -

0.0 I

60 80 100 120 140 160 180 Temperature O C

0. 7. l e x / l m us. temperature is plotted for polyethylene doped with BPP. Measurements were obtainedfrom cooling curve observations. The crystallization temperature is indt cated by the discontinuity at 117°C. The excitation wave length was 345 nm

1 0 1

Temperature O C

Q. 8. Fluorescence intensity us. temperature is plotted for polyethylene doped with DMA-DPH. Measurements were o b tained from cooling curve observations. The small minimum in the curve at 107°C is caused by the heat of crystallization and indicates the crystallization temperature. The excitation wavelength was 400 nm.

proximately 107°C) was defined as the intersection of two linear extrapolations of the curve from high and low temperatures. This is an operational definition of Tg for this measurement and is somewhat higher than Tg obtained from a fit of the Williams-Landel- Ferry (WF) equation to our data (see Appendix). F'lgure 5 shows that, as temperature increased above Tg, the slope of the curve increased markedly. This is because the excimer intensity increased and the monomer intensity decreased upon raising the temperature above Tg. As the microviscosity of the resin decreased for T > Tg , the increased rotational freedom of the dye permitted the generation of more excimer fluorescence at the expense of monomer fluorescence. Below Tg. the near zero slope in the curve indicates that T~ is nearly insensitive to changes in temperature. In fact, after making background corrections to the data of m. 5, we determined that the

674 POLYMER ENGINEERING AND SCIENCE, APRIL 1994, Vol. 34, No. 8


excimer intensity in the range 60 to 85°C was zero within the sensitivity of our measurement. Below Tg, excimer formation is severely restricted because the free volume cell size in the glass is not large enough to accommodate motion of the large pyrene ring. This concept is supported by an extensive study by Victor and Torkelson who examined the role of free volume in limiting photochromic transitions and BPP excimer fluorescence (22). Also, in a series of papers (23-261, Bokobza, MoMerie, and co-workers using probes similar to BPP demonstrated that the temperature and pressure dependence of intramolecular excimer production is dependent on free volume and that the temperature dependence of correlation times obeys the characteristic WLF expression. They con- cluded that intramolecular conformational change of the fluorescent probe is controlled by polymer seg- mental motions involved in the rubber to glass transition. The distinguishing feature of the data of Fig. 5, which was employed during real-time observations of injection molding, is the near zero slope of the curve for T < Tg which indicated that polystyrene was in the glassy state.

The calibration curve for DMA-DPH, shown in Fig. 6, did not display the same sensitivity to the onset of the glass transition as did BPP even though both dyes respond to their molecular environment with similar functional dependence, i.e. dependence on T~ which in turn is proportional to q/T. The difference in their behavior below Tg is attributed to the microscopic viscosity q which is peculiar to each molecule. DMA-DPH, whose end group requires much less free volume for rotational movement, is not so constricted by the glassy state as is BPP. Whereas BPP fluorescence is insensitive to changes in temperature below Tg, DMA-DPH changed significantly, and was employed in the injection molding experiments as a probe to monitor the changing temperature environment for T < Tg.

Polyethylene: In Fig. 7, data obtained from the cooling curve experiment for BPP doped into polyethylene are shown in a plot of the ratio IJI,,, us. temperature where I,, is the excimer fluorescence integrated over the wavelength range 450 to 550 nm and I,,, is the monomer fluorescence at 380 nm. The crystallization temperature T, is clearly indicated by the sharp discontinuity in the curve at 117°C. The positive and negative slopes below and above T,, respectively, originate from two competing phenomena: intramolecular excimer formation and dissociation, and differences in optical transmission for the 380 nm monomer fluorescence and the 450 to 550 nm excimer fluorescence. In the melt, T > 117°C. excitation light energy is transmitted deep into PE relative to transmission in crystalline PE which was limited due to light scattered by the microcrystals. For melt PE, the 380 nm fluorescence was disproportionately absorbed over the optical path within the bulk compared to absorption by the longer wavelength excimer fluorescence. We suggest that this was due to the antioxidant or to the presence of carbonyl and hy-

droxyl groups in PE. By measuring the relative trans- missions of 380 and 490 nm lght in melt PE, we confinned that 380 nm light was indeed absorbed to a greater degree than 490 nm light. For T > 1 17°C. we observed that 380 nm absorption increased more rapidly with decreasing temperature than the 490 nm absorption due to the increase in 380 nm ab- sorbers as the density of melt PE increased. At T, (1 17"C), the sharp drop in the ratio le,/lm occurred as light scattering from the crystallites causes the excitation and detection of fluorescence to be concen- trated at the surface. For T < 1 17°C optical absorp tion for 380 nm and for the 450 to 550 nm range is comparable because the optical path is short and confined to the near surface.

The behavior of DMA-DPH in polyethylene is shown in the fluorescence us. temperature cooling curve of Fig. 8. The onset of crystallization of polyethylene was observed at 109°C where the fluorescence intensity first decreased and then increased abruptly at 107°C. The decrease in fluorescence was due to the heat of crystallization which was released during the liquid- to-crystal transition. As polyethylene crystallized, the dye molecules, which were too large to be incorpo- rated in the crystal, were displaced to the amorphous regions. As the heat of crystallization was conducted away from the crystallites, the dye experienced an increase in temperature with the result that T~ of the DMA-DPH endgroup decreased causing an increase in radiationless decay and a decrease in fluorescence intensity. The abrupt increase in fluorescence at 106 to 107°C upon crystallization was due to a combina- tion of effects: the increase in density of the material permitting more dye to come within the view of the optical sensor and the change in index of refraction which causes an increase in the background reflected light. Below T,, the slope of the curve reflects the change in T~ with temperature. The magnitude of the crystallization effect observed in this calibration curve is proportional to the cooling rate: the faster the cooling rate, the larger will be the production rate of heat energy from crystallization and the greater will be the momentary fluorescence response.

Real-Time Injection Molding Observations

Polystyrene

In-situ, real-time measurements of fluorescence during injection molding for BPP and DMA-DPH in polystyrene are shown in Figs. 9 and 10 respectively. Here, fluorescence intensity and pressure levels are plotted us. time during the injection mold cycle. The time scales of the two sets of data are coincident within 0.5 s and we assume that the cooling rates were the same. The pressure levels indicate the clamping pressure used to lock the mold halves to- gether and pressures exerted by the hydraulic ram on the polymer as it was injected into the mold and held in the cavity during the cooling period. The changes in the pressure data serve as event markers in the mold cycle. Considering Fig. 9, the mold was

POLYMER ENGINEERING AND SCIENCE, APRIL 1994, Vol. 34, No. 8 675

A. J. Buret al.

1140

-120 + -

0 10 20 30 40 50 6 0 70 Time (seconds)

Fg. 9. Pressure (dashed curve) and l e x / l , (solid curve) for bis-(pyrene) propane (BPP) doped into polystyrene are plotted against time during injection molding. Here, I,, is the excimer intensity and I , is monomer intensityfrom BPP. The excitct tion wavelength was 345 nm. Polymer solidi$cation is indt cated by the zero slope of theJuorescence us. time curve at 20 s. Separation of the specimen from the optical sensor corresponded to the small increase at 27 s. The dashed lines give the pressure of the hydraulic drive and indicate the cycle state. The data are for one mold cycle for which the hold pressure was applied for 20 s.

1 0

4 3

Z O t 7 a, -4 r: "0 6 a,

E 0 4 v) a, L

2

s o 2

t;: 0 0

Fig. 10. Pressure (dashed curve) andJuorescence intensity (solid curve) of DMA-DPH doped into polystyrene are plotted against time during injection molding. The excitation wave length was 400 nm. The time scale for these data is coincident with that of Fig. 9. There is no indication of polymer solidi@cation at 20 s. Separation of the specimenfrom the optical sensor occurred at the abrupt increase at 27 s. The dashed lines give the pressure of the hydraulic drive and indicate the cycle state. The data are for one mold cycle for which the hold pressure was applied for 30 s.

closed with the application of clamping pressure (147 MPa) at t = 5 s, polymer fill pressure (3 1 MPa) was applied at 7 s and fill was complete at approximately 9 s, as the polymer cooled, hold pressure was applied to the polystyrene from 9 to 39 s. Hold pressure was released at 39 s and the polymer cooled further under zero applied pressure.

The fluorescence signal plotted in Fig. 9 is the excimer to monomer intensity ratio I e x / I m of BPP us. time. Initially the mold was empty; the signal o b

tained was background from the excitation light which reflected off the wall of the mold and back into the collection fibers. As the mold filled (7 to 9 s), this background decreased and the characteristic signal from the fluorescent dye was observed. At t = 20 s, the slope of the curve became approximately zero as the polystyrene cooled below its glass transition. We interpret the slight increase at 27 s an optical effect which occurred when the specimen pulled away from the sensor as it cooled and the reflection of light from the sapphire window/polystyrene interface changed. We note that the slope of the curve remained zero after release of the packing pressure indicating that the polystyrene remained in the glassy state. In this case, the mold remained closed while we obtained data for long times. In normal operation, releasing the packing pressure and opening the mold would occur sooner.

A possible interpretation of the constant level of fluorescence in the data of Fig. 9 for t > 20 s is that the temperature of the polystyrene was not changing and therefore the fluorescence was constant with time also. The fact that the temperature was still decreasing for t > 20 s is seen from the real-time fluorescence for DMA-DPH in polystyrene shown in Fig. 10. As with the calibration experiment we did not observe a distinct change in the DMA-DPH fluorescence signal to indicate the onset of the glass transition. According to the BPP data of Fig. 9, the liquid- to-glass transition occurred at approximately 20 s and, as expected, there is no indication of this transition in the DMA-DPH data of Fig. 10. Instead, we observed that the fluorescence continued to change monotonically at t = 20 s indicating that the temperature of polystyrene continued to decrease. At t = 27 s , separation of specimen from the sapphire window produced a discontinuous change in reflection at the sapphire/air/specimen interface. The fluorescence intensity began to plateau at approximately 55 s and the slow change thereafter was a reflection of the asymptotic approach of the polystyrene temperature to the ambient temperature of the mold.

Polyethylene

In Figs. 1 1 and 12 the real-time fluorescence and pressure data for BPP and DMA-DPH in polyethylene are shown. The time scales for both fgures are coincident within 0.5 s and we assumed that the cooling rates were approximately the same. After the mold fill was complete at t = 6 s we observed that fluorescence displayed uneventful change until crystallization occurred at t = 23 s. The release of the heat of crystallization transpired over the approximate time period 23 to 26 s after which the curves showed monotonic changes due to cooling. For t > 26 s some of the fluorescence intensity changes were caused by the increase in density accompanying crystallization and scattering of light in the crystallized polyethylene. We interpret the abrupt increase, which occurred at 34 s in the DMA-DPH curve (Fig. 12). as a change in optical reflection at the sapphire/specimen interface



1 0 7 h

E Y

\ O 8 I v

0 2 0 6 a p: h O 4 5 z $ 0 2 C -

0 0

Time (seconds) 151s. 1 1 . Pressure (dashed curve) and lex/l,,, (solid curve) for bis4pyrene) propane (BPP) doped into polyethylene is plotted against time during injection molding. Here, I,, is the excimer intensity and 1, is monomer intensrtyfrom BPP. The excitu tion wavelength was 345 nrn Polymer crystallization is indicated by the small plateau at 26 s. Separation of the spect men from the optical sensor o c c ~ ~ ~ e d at 35 s. The dashed lines g iw the pressure of the hydraulic driue and indicate the cycle state. The data are for one mold cycle for which the hold pressure was applied for 20 s.

1 I

151s. 12. Pressure (dashed curve) andfluorescence intensity (solid curue) of DMA-DPH doped into polyethylene is plotted against time during injection molding. The excitation wave length was 400 nm. The time scale for these data is coincC dent with that of Rg. 1 1 . Polymer crystallization is indicated by the small plateau at 26 s. Separation of the specimenfrom the optical sensor occurred at 35 s. The dashed lines give the pressure of the hydraulic drive and indicate the cycle state. The data are for one mold cycle for which the hold pressure was applied for 30 s.

which accompanied the separation of the specimen from the sensor upon shrinkage. For both curves, the fluorescence intensity approached a plateau for long times as the temperature of polyethylene approached the ambient mold temperature.

Comparison of Fig. I I with the calibration curve for BPP/polyethylene, Fig. 7, indicates that there was no time region during the injection mold cycle for which ultraviolet absorption effects were observed. We interpret this to mean that a crystalline skin forms almost immediately at the surface of the mold

and that the observations are from the near surface. Since the temperature of the mold was held near 23°C by circulating water coolant, rapid crystallization occurred. From 10 s to 100 s, led&, decreased monotonically except for changes at 26 s (crystallization) and 35 s (specimen/sensor separation). The major differences in the experiments generating the two curves (Figs. 7 and 1 I ) were the time scale and the magnitude of the temperature gradients.

SUMMARY AND CONCLUSIONS

We have demonstrated that fluorescence spectrometry can be used to monitor the onset of phase transitions in a polymer resin which occur during cooling of an injection molded polymer product. Two fluorescent molecules, an excimer producing dye (BPP), and a molecular rotor dye (DMA-DPH), were used at dopant levels of concentration in a crystallizable polymer, polyethylene, and a glass forming polymer, polystyrene. The fluorescence response of the doped polymers were examined by generating cooling curves in the temperature regime of the respective phase transition for each resin. In polyethylene, the fluorescence intensity us. temperature profile from both dyes showed distinct changes upon crystallization. The o h served effect was attributed to a local temperature increase in the molecular environment of the dye caused by the heat of crystallization. Absorption of 380 nm fluorescence in the BPP/polyethylene speci- mens complicated the interpretation of the lex/l,,, us. temperature cooling curve data; the onset of crystallization was clearly observed, however, and application of the measurement concept to injection molding of polyethylene was successfully carried out. For polystyrene, the behavior of the two dyes during the liquid-to-glass transition was different for each. BPP, the larger of the two molecules, showed a marked decrease in excimer fluorescence below Tg due to the limited free volume available for intramolecular rotation of the large pyrene ring (22-26). On the other hand, DMA-DPH, whose fluorescence intensity is also dependent upon intramolecular rotation, showed no distinguishing features on a temperature scan at the glass transition. The difference in the behavior of the two dyes is attributed to their molecular size in relation to the size of free volume cells which are available for intramolecular rotation.

Real-time, in-line fluorescence monitoring of injection molding was carried out. A n injection mold machine was instrumented with optical equipment em- ploying an optical fiber to monitor fluorescence from the mold cavity. We demonstrated that, for BPP doped into both polystyrene and polyethylene, changes in the fluorescence intensity were easily recognized and clearly marked the onset of polymer crystallization in polyethylene and the Tg in polystyrene. Also for DMA- DPH doped into polyethylene, the onset of crystallization was clearly seen in the real-time injection mold data. But for DMA-DPH in polystyrene, we found that fluorescence intensity was not an indicator for the onset of the glass transition; this dye, however, can

POLYMER ENGlNEERlNG AND SCIENCE, APRIL 1994, Vol. 34, No. 8 677

A. J. Buret al.

be used to monitor temperature changes of polystyrene while cooling in the mold. We emphasize that the application of this measurement method does not require measurements of absolute fluorescence in- tensities or background corrections to the data because the detection of solidification is camed out by observing distinctive changes in the fluorescence intensity/time profile.

In order to take full advantage of the information obtained from this new measurement technique, a feedback loop to control the mold cycle is being developed. Also, other measurements based on fluorescence and optical methods to monitor rheological properties and temperature are being developed and will be the subject of future publications.

APPENDIX

In high viscosity environments, the ratio I e x / I m for BPP is a function of the rate constants of excimer formation or l / ~ ~ , of monomer and excimer decay, and excimer lifetime ( 18). Previously published observations of the temperature dependence of l e x / I m for BPP in toluene solution were shown to obey an Arrhe- nius activation law in the low temperature/high viscosity regime, T < 20°C (18, 27). The Arrhenius behavior indicates that all rate constants associated with &,/Im have the same activation energy and are controlled by the same process. In a polymer whose viscosity is many orders of magnitude larger than that of toluene, we assume that the low temperature/high viscosity condition also holds. For T > Tg, we expect that polymeric rate processes will be described by the WLF equation (28, 29). After background corrections were made to the data of Fig. 5, our measurements of for BPP in polystyrene were examined on a plot of 1n( l ex / Im) us. reciprocal temperature. As shown in Fig. 13, a fit of the WLF equation to the data was observed in the temperature range 102 to 152°C. Deviation from the WLF function below 102°C implies that Tg is approximately 102°C

WLF Dependence of Intensity Ratio

WLF 22.56(T-T~)/(41+T-T~) Ts=378 "K 10

2.2 2.4 2.6 2.8 3 .O \

lOOO/T (Kelvin)-'

Rg. 13. A plot of ln/R(T) / R(T,)] us. 1 / T is shown for BPP in pclystyrene. R(T) = lex/l, , , a t temperature T. The solid line is thef i t of the WLF equation to the data using WLF constants C, = 22.56 and C, = 41.O"Kfrom R e j 29.

for this specimen, somewhat lower than the operational value ( 107°C) obtained from the linear extrape lations of the Fig. 5 data. Above 152°C. the low temperature/high viscosity condition no longer applies and deviation from the WLF dependence results from an increase in the rate of dissociation, i.e. the rate at which excimer converts back to monomer ( 18). Below 152°C. the dissociation rate is small compared to the excimer lifetime rate and the process dynamics are dominated by the rate of excimer formation. The WLF dependence of our data for T > Tg supports the view that intramolecular excimer formation of BPP reflects the local dynamics of the polymer molecule. We interpret these observations to mean that the BPP molecule must completely cooperate with the polymer molecular dynamics in order to carry out its own motion. This view is in agreement with the work of Bokobza and co-workers who measured the temperature dependence of the excimer formation rate constant of meso-bis[ 1 -(2-pyrenyl)ethyl]ether (B2PEE), an excimer forming dye similar to BPP (23). They observed that, when B2PEE was doped into several amorphous polymers, their data followed the WLF function for T > Tg (23).

REFERENCES

1. J. R. Lakowicz, Principles of Fluorescence Spectroscopy,

2. D. Philips, ed.. Polymer Photophysics, Chapman and Hall,

3. T. Sasaki, M. Yamamoto. and Y. Nishqima, Macro

4. B. Valeur and L. Monnerie, J. Polym Sci., Polym Phys.

5. J. P. Jarry and L. Monnerie. J. Polym Sci. Polym Phys.

6. D. 1. Bower, K. K. P. Korybut-Daszkiewicz, and I. M Ward,

7. F. W. Wang and E. Wu, Polym Comm, 28, 73 (1987). 8. F. W. Wang, R. E. Lowry. and W. H. Grant, Polymer, 25,

690 ( 1984). 9. F. W. Wang, R. E. Lowry, and B. M. Fanconi, Polymer,

27, 1529 (1986). 10. F. W. Wang, R. E. Lawry, W. J. Pummer, B. M. Fanconi,

and E. Wu, in Photophysics OfPolymers, C. E. Hoyle and J. M. Torkelson, eds., ACS Symposium Series 358, Ch. 33, p. 454, American Chemical Society, Washington, D.C. (1987).

Plenum Press, New York (1983).

London (1985).

molecules, 2 1, 6 10 ( 1988).

Ed., 14. 11 (1976).

Ed., 18. 1879 (1980).

J. AppL Polym Sci., 28. 1195 (1983).

11. R. L. Levy, Polym Mater. Sci. Eng., 50, 124 (1984). 12. A. J. Bur, J. Shibata, T. K. Trout, F. W. Wang, and C. L.

13. A. J. Bur and F. M. Gallant, Polym Eng. Sci., 31. 1365

14. G. Oster and Y. Nishijima, Fortschr. Hochpolym-Forch, 3. 313 (1964).

15. R. 0. Lout@ and B. A. Arnold, J. Phys. C h e m , 86, 4205 (1982).

16. J. B. Birks, ed., Organic Molecular Photophysics, Vol. 2, Ch. 4 and 9, J. Wiley and Sons, New York (1975).

17. M. L. Viriot, M. Bouchy, M. Donner, and J. C. Andre, Photobiochern Photobiophys., 5, 293 (1983).

18. K. A. Zachariasse, G. Duveneck, and R Busse, J. Am C h e m Soc., 106, 1045 (1984).

19. K. A. Zachariasse. R. Busse, G. Duveneck, and W. Kuhnle, J. Photochern, 28, 237 (1985).

20. K. A. Zackariasse, W. Kuhnle, U. Leinhos, P. Reynders, and G. Striker, J. Phys. C h e m , 95. 5476 (1991).

Thomas, Polym Eng. Sci., 29, 1759 (1989).

( 199 1).



2 1. Identification of a commercial product is made only to facilitate experimental reproducibility and to describe adequately experimental procedure. In no case does it imply endorsement by NIST or imply that it is necessar- ily the best product for the experiment.

22. J. G. Victor and J. M. Torkelson, Macromolecules, 20. 2241 (1987).

23. D. P. Jing, L. Bokobza. L. Monnerie, P. Collart, and F. C. DeSchryve, Polymer. 31. 110 (1990).

24. D. P. Jing, L. Bokobza, P. Sergot, L. Monnerie, P. Collart and, F. C. DeSchryver, Polymer, 30. 443 (1989).

25. L. Bokobza. C. Pham-Van-Cang. L. Monnerie, J. Vanden-

POLYMR ENos#EERWG AND SCIENCE, APRIL 1994, Vol. 34, No. 8

driessche, and F. C. DeSchryver, Polymer, 30.45 (1989). 26. L. Bokobza, C. Pham-Van-Cang, C. Giordano, L. Mon-

nerie, J. Vandendriessche, F. C. DeSchryver. and E. G. Kontos, Polymer. 28. 1876 (1987).

27. M. J. Snare, P. J. Thistlethwaite, and K. P. Ghiggino. J. A m C h e m Soc., 105.3328 (1983).

28. M. L. Williams, R F. Landel, and J. D. Ferry, J. A m Chem Soc., 77. 3701 (1955).

29. F. R Schwarzl, Encyclopedia of Polymer Science and Engineering. Vol. 17. p. 643, H. F. Mark, N. M. Bikales, C. G. Overberger, and G. Menges, eds., John Wiley and Sons, New York ( 1989).

679

Documents

In-line optical monitoring of polymer injection molding