C

Sa

b

a

ARR2AA

KFTZHL

1

htptcaoiamrtepohwfi

((

0d

Thermochimica Acta 530 (2012) 64– 72

Contents lists available at SciVerse ScienceDirect

Thermochimica Acta

jou rna l h omepage: www.elsev ier .com/ locate / tca

alibration and performance of a fast-scanning DSC—Project RHC

am Woutersa,b, Fatma Demira, Linda Beenaertsb, Guy Van Asschea,∗

Vrije Universiteit Brussel (VUB), Physical Chemistry and Polymer Science (FYSC), Pleinlaan 2, B1050 Brussels, BelgiumArtesis Hogeschool Antwerpen, Industriële Wetenschappen & Technologie, Paardenmarkt 92, B2000 Antwerpen, Belgium

r t i c l e i n f o

rticle history:eceived 19 September 2011eceived in revised form9 November 2011ccepted 1 December 2011vailable online 9 December 2011

a b s t r a c t

This paper presents the calibration and evaluation of a Rapid Heat-Cool DSC (RHC), a fast-scanningcalorimeter capable of controlled heating rates of up to 2000 K min−1. As the use of high scan ratesincreases the influence of thermal lag, a thorough analysis of the instrument performance was made.This encompasses a rigorous temperature calibration procedure, a study of the influence of sample massand heating rate on the extrapolated onset temperature, and the analysis of the symmetry of the thermallag in heating and cooling using phase transitions in liquid crystals. The calibration of the heat flow rate is

eywords:ast-scanning calorimetryhermal lagero-heating rate temperature calibrationeat flow rate calibration

evaluated using indium for a broad range of sample masses and heating rates, and verified by heat capac-ity measurements on polystyrene for a wide temperature range. The performance of the RHC proves tobe very good, with estimated errors on temperature and heat flow rate (heat capacity) of 0.1 K and 2%,respectively.

© 2011 Elsevier B.V. All rights reserved.

iquid crystal

. Introduction

In recent years, several forms of fast-scanning calorimeters (FSC)ave been developed [1–10], enabling material scientists to quan-itatively simulate cooling rates employed in industrial polymerrocessing conditions and to study metastable materials, facili-ating process and material optimization. Focussing on polymerrystallization, fast heating can suppress kinetic processes such asnnealing, cold crystallization, or crystal reorganization, enablingne to study the melting of the material as it was crystallized dur-ng cooling. At higher heating or cooling rates, kinetic events canlso be delayed: the glass transition, Tg, and cold crystallizationove to higher temperatures with increasing cooling and heating

ates, respectively, while the melting transition often remains athe same temperature. Shifting kinetic transitions is not only inter-sting for research on polymers but also useful for biochemicalurposes [9,10]. Kinetic events such as evaporation and desorptionf water from proteins and other compounds can now be shifted toigher temperatures, thus enabling the observation of the Tg of theet material [9]. Polymorph conversions, between different crystal

orms of complex pharmaceutical agents may be suppressed, facil-tating separate characterization of each form [11]. And last but

∗ Corresponding author. Tel.: +32 2 6293941; fax: +32 2 6293278.E-mail addresses: sam.wouters@vub.ac.be (S. Wouters), fatma.demir@vub.ac.be

F. Demir), Linda.beenaerts@artesis.be (L. Beenaerts), gvassche@vub.ac.beG. Van Assche).

040-6031/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.tca.2011.12.001

not least, by efficient quenching, crystallization can be avoided formaterials it was not possible for before.

The FSCs developed over the last years can be divided into twocategories: DSCs tuned for higher scan rates and chip calorimeters.In the second type of instruments, existing in both single and dif-ferential calorimeter versions, the calorimeter consists of a smallchip, capable of heating and cooling rates up to 1,000,000 K s−1

[5–8,10]. For these instruments, the sample masses (commonly atthe nanogram level and higher) are usually estimated from the heatgenerated by transitions, from the heat capacity, or from volumeestimations by scanning probe microscopy.

From the side of conventional DSC instruments, instrumentscombining a low mass furnace with low but measurable samplemasses provides for enhanced scan rates and sensitivity. The HighPerformance DSC or Hper DSC developed by Perkin Elmer is typ-ically able to heat and cool at a rate of 500 K min−1 [1–3]. Theinstrument discussed in this work, the Rapid Heat-Cool DSC (RHC)by TA Instruments, allows for controlled heating and cooling rateswell above 1000 K min−1 [9], further extending the range of condi-tions that can be applied.

A critical point in the development of DSCs in general, but forFSCs especially, is their calibration and performance. DSC experi-ments are often conducted at the rates equal or close to the oneused during calibration. For FSCs this is in general not the case,as the ability to use largely different rates is one of the main

features that make the techniques interesting. However, using dif-ferent rates than those calibrated for, and especially high rates andlarger masses, leads to important changes in thermal lag of thesystem. These changes are not compensated for using the normal

S. Wouters et al. / Thermochimica Acta 530 (2012) 64– 72 65

F ensorc

copataRaw

2

Hpigtloaatr(td

(1tams

sttSTS

t

ig. 1. (Left) Drawing of a cross section of the RHC cell showing the downsized srucibles and lids.

alibration procedures. A DIN recommendation on a more rigor-us temperature calibration of FSCs has been issued [12]. In thisaper, the results of a similar temperature calibration procedurere presented for a TA Instruments RHC, and the performance ofhis instrument concerning the measurement of transition temper-tures and heat flow rate is discussed. Although only results for anHC will be shown, the different steps involved in the calibrationnd evaluation of this instrument also apply to conventional DSCshen used at higher heating rates or with larger sample masses.

. Experimental

The measurements were performed on a TA Instruments Rapideat-Cool DSC (RHC) using a liquid nitrogen cooling system andurged with neon (12 ml min−1). A drawing of the RHC sensor

s given in Fig. 1. The RHC cell is heated by four quartz halo-en lamps with an almost instantaneous response, overcominghe response problems of conventional heaters (high mass andarge thermal resistance). Combined with the TzeroTM technol-gy [13], an improved resolution with excellent baseline stabilitynd high sensitivity for small transitions is achieved. The TzeroTM

pproach uses an extension of the conventional heat flux DSC equa-ion including terms accounting for the heat capacities and thermalesistances in the instrument [13]. Results for calibration in both T1conventional heat flux DSC mode) and T4P (using the four correc-ion terms and a pan contact resistance correction) modes will beiscussed in this paper.

Samples were enclosed in dedicated aluminium cruciblesdiameter ca. 1 mm, height 2 mm) and lids (Fig. 1). Masses up to

mg can be used when high sensitivity is desired, e.g., to studyhe glass transition. Studies of melting and crystallization are usu-lly performed with sample masses around 0.250 mg in order toinimize sample volume and temperature gradients within the

ample.Temperature calibration was performed using two certified

tandards: indium produced by LGC (purity: 99.99998%) andin produced by Alfa Aeser (purity: 99.9995%). Furthermore,wo organic compounds were used: Adamantane supplied byigma–Aldrich (purity 99%+) and benzophenone being a Mettler

oledo recommended calibration substance, produced by Fluka &igma–Aldrich.

To check the symmetry of the instrument, three liquid crys-alline materials were used: M24 (4-cyano-4′-octylbiphenyl),

s (Courtesy of TA Instruments). (Right) Comparison of RHC prototype and TzeroTM

BCH52 (4-4′-ethyl-4-(4-propyl-cyclohexyl)-biphenyl) and HP53(4-(4-pentyl-cyclohexyl)-benzoic acid-4-propylphenyl ester).These certified secondary standards were supplied by Merck andwere used without further purification. These substances will bereferred to with their product codes. All are traceable to standardreference materials from NIST and PTB.

For heat capacity verification purposes, NIST standard referencematerial SRM-705a polystyrene (Mw = 170–190 kDa) was used. Forthe verification of the transition enthalpy measurements, NIST SRM1475a linear polyethylene (Mn = 18 kDa, Mw = 53 kDa) was used. Inthis paper, endothermic transitions will displayed in the positiveheat flow rate direction.

3. Results and discussion

3.1. Zero heating rate temperature calibration

The temperature calibration was performed over a temperaturerange from −65 ◦C to 230 ◦C using the zero heating rate approachrecommended by GEFTA, the German Gesellschafft für Thermis-che Analyse [14–18]. The solid to solid transition in adamantane,along with the melting of benzophenone, indium and tin were sug-gested for this purpose by GEFTA. Three samples of comparablemass (close to 0.250 mg) were prepared for each material. Adaman-tane was heated from −150 ◦C to 0 ◦C, benzophenone from 0 ◦C to75 ◦C, and tin from 125 ◦C to 300 ◦C, all at 50 K min−1, 100 K min−1,250 K min−1, 500 K min−1. TGA measurements confirmed the sta-bility of the materials in this range. The observed extrapolated onsettemperatures Teo of the transition endotherms were extrapolatedto zero heating rate using a quadratic trend line. Table 1 shows theaverage values for calibration in T1 and T4P mode, along with thestandard deviation. Both T1 and T4P mode are evaluated becauseT1 gives the true thermal lag of the instrument, whereas in T4Pmode it is partly accounted for.

Fig. 2 displays the results of the four point temperature calibra-tion for T1 and T4P mode. The slopes of the calibration lines areclose to unity. The absolute deviations from linearity are less than±0.5 K.

To evaluate the dependence of the calibration on sample prepa-

ration and to study the possible influence of crucible position insidethe cell on calibration, the melting of five indium samples of approx-imately 0.250 mg was measured 5 times for each sample, each timeremoving both sample and reference from the instrument. Table 2

66 S. Wouters et al. / Thermochimica Acta 530 (2012) 64– 72

Table 1Results of four point zero heating rate temperature calibrations, showing the literature value of the melting temperature Tm and the extrapolated onset temperatures Teo

after extrapolation to zero heating rate for measurements in T1 mode and T4P mode (average Teo and standard deviation � based on 3 samples).

Material Literature T1 mode T4P mode

Tm (◦C) Teo (◦C) � (K) Teo (◦C) � (K)

Adamantanea −64.53 −64.47 0.04 −64.50 0.11Benzophenoneb 48.10 45.21 0.33 44.91 0.13Indium 156.60 150.55 0.11 150.28 0.09Tin 231.93 222.57 0.17 222.39 0.06

a Although little mass loss was noted during TGA experiments, multiple RHC measurements on a single adamantine sample will lead to measurable sample losses due tosublimation. When calibrating with the highly volatile adamantane, it is recommended to use fresh samples only.

b Benzophenone displays polymorphism: rhombic �-crystals melt around 48 ◦C, while the metastable monoclinic �-crystals, formed at the high cooling rates common inRHC, melt around 25 ◦C. It is recommended to use fresh samples or to create conditions that allow sufficient time for the formation of �-crystals to permit the use of datafrom subsequent heating cycles.

Table 2Reproducibility of measurements of the melting of indium at 250 K min−1 (T1 mode). Average and standard deviation � for the extrapolated onset temperature Teo, themelting enthalpy �hm and the peak temperature Tp , based on 5 repeated measurements for each sample (removed from and replaced in the instrument). The last lines givethe averages 〈xi〉 and standard deviations � for all 25 experiments.

Sample Mass (mg) 〈Teo〉 (◦C) � (K) 〈�h〉 (J g−1) � (J g−1) 〈Tp〉 (◦C) � (K)

1 0.248 157.68 0.01 29.06 0.34 159.00 0.042 0.246 157.72 0.02 28.57 0.11 159.33 0.073 0.261 157.75 0.05 28.32 0.34 159.06 0.054 0.249 157.80 0.06 27.49 0.25 159.18 0.105 0.245 157.70 0.11 28.22 0.08 159.46 0.17

〈xi〉 0.250 157.72 0.05 28.33 0.23 159.21 0.09

� 0.006 0.05

y = 0.97x - 0.35R2 = 1.00

y = 0.97x + 0.01R2 = 1.00

-100

0

100

200

300

3002001000-100

Tm literature / °C

T eo /

°C -

T1 m

ode

-75

25

125

225

325

T eo /

°C -

T4P

mod

e

-0.20

-0.10

0.00

0.10

0.20

3002001000-100

Tm literature / °C

devi

atio

n T e

o / K

T1T4P

Fig. 2. Four point temperature calibration: (top) experimental zero heating rateextrapolated onset temperatures Teo for T1 and T4P mode against melting temper-atures from literature; (bottom) plot of residuals.

0.57 0.18

shows the results for all samples as well as the pooled result basedon the 25 measurements. The standard deviation � on the extrap-olated onset temperature is 0.05 K, indicating that the results arehighly reproducible. The average standard deviation on the melt-ing enthalpy for one sample is about ±1% and the overall average(28.3 J g−1) agrees with the literature value (28.71 J g−1 [19], usedfor the calibration) within the experimental error (0.6 J g−1 or 2%).

3.2. Sample mass & heating rate effects

In order to fully profit from the capabilities of RHC, it is highlydesirable to be able to work at heating and cooling rates that deviatefrom those used for calibration. This however, leads to the need forcompensation of the thermal lag, as the use of different massesand scan rates also lead to changes in thermal lag (the temperaturedifference between sample and temperature sensor underneath thesample pan). In order to perform quantitative measurement undersuch conditions, the thermal lag should be accounted for in thecalibration or by post-measurement corrections.

For studying the thermal lag of regular DSCs, GEFTA proposed touse 14 different heating rates, ranging from 1 to 500 K min−1 andfive different samples masses between 0.1 mg and 5 mg [12,14–18].Following this procedure, only part of what RHC is capable offwould be covered. Therefore the recommended procedure has beenaltered by taking 12 heating rates, ranging from 10 to 1500 K min−1.The sample mass has been lowered considerably, partly because theRHC pans are much smaller than conventional pans used in DSC,but also to keep temperature gradients within the sample smallat higher heating rates. Six sample masses were used, between0.05 mg and 1 mg of indium.

Fig. 3 shows the effect of the heating rate for the melting ofindium in T1 and T4P mode, respectively. Results for all heatingrates are shown for the highest mass used (1 mg), as well as for the

one used during calibration (0.25 mg). For higher heating rates, Teo

and Tp shift to higher temperatures due to increasing thermal lag.The higher the sample mass, the lower the resulting heat flow ratemaximum (indicating a stronger smearing of the signal). The effect

S. Wouters et al. / Thermochimica Acta 530 (2012) 64– 72 67

-15

35

85

135

185

185175165155T / °C T / °C

T / °C T / °C

T1 H

eat F

low

/ W

.g-1

T10.246 mg

Endoup

-15

35

85

135

185

185175165155

T1 H

eat F

low

/ W

.g-1

T11.005 mg

Endoup

0

15

30

45

60

185175165155

T4P

Hea

t Flo

w /

W.g

-1

T4P0.242 mg

Endoup

0

15

30

45

60

185175165155

T4P

Hea

t Flo

w /

W.g

-1

T4P1.005 mg

Endoup

F sured2

omapc

oihtfempsi

mmidloT

ig. 3. Melting endotherms for indium samples of about 0.250 mg and 1.00 mg mea50, 350, 500, 750, 1000, 1250 and 1500 K min−1.

f thermal lag is also more pronounced for the sample with highestass, resulting in a bigger temperature shift for both Teo and Tp,

s well as a broadening of the entire peak. Baseline shifts are lessronounced for T4P measurements as the instrument response isorrected for in the T4P calibration.

Fig. 4 gives a clearer overview of the effect of the sample massn the thermal lag at different rates. All six masses are depictedn T1 and T4P mode at 250 K min−1 and 1000 K min−1. For equaleating rates, the T1 and T4P graphs are plotted on a common scaleo ease comparison. For each heating rate, the upward shift in Teo

or increasing sample masses is clearly visible. A similar but largerffect is seen for the peak temperature. With increasing mass, theaximum heat flow rate (expressed in W g−1) decreases and the

eak broadens, but the melting enthalpy (the area below the peak)tays nearly the same. At higher heating rates the peak maximums higher.

Fig. 5 shows Teo and Tp versus the heating rate for the threeasses shown in Fig. 3 (data obtained from the experiments in T1ode). On the second ordinate, the melting enthalpy (�h in J g−1)

s shown, along with the average (black line) and the standard

eviation (dashed lines) for all measurements. Using these, out-

ying values are easily detected. All plots were made on commonrdinate scales for the three samples to permit easy comparison.he observed Teo and Tp increase with increasing sample mass

in T1 and T4P-mode. Arrows indicate increasing heating rate: 10, 50, 100, 150, 200,

and heating rate, indicating the increasing thermal lag. The effecton Tp is roughly 2 times higher than on Teo. For small masses, theincrease in thermal lag for both Teo and Tp seems to be quite linear.However, for higher masses and higher rates, it becomes clear thatthe slope is decreasing in both cases. Results and trends seen inT4P data are very similar (not shown). For the first two samplemasses, the range over which Teo and Tp change as a function ofthe heating rate is nearly identical for T1 and T4P mode. However,for a high mass, the effect of thermal lag is much smaller in T4Pmode: at 1500 K min−1, the shift in Teo with respect to zero heatingrate is nearly 3 K lower in T4P mode than in T1 mode. For Tp, thedifference is nearly 5 K. As a lag is still measured, the use of theT4P calibration did not fully compensate the thermal lag effects.The time constant as calculated from the slope of the tangentat zero heating rate equals 0.0045 min or 0.27 s (for the 0.25 mgsample).

The effects of heating rate and sample mass on the thermal lag(on the Teo for the melting of indium) are compiled in Fig. 6 fora range of sample masses and heating rates (T1 and T4P mode).The dashed and solid lines resulting from a fit of all data should be

regarded as guides-to-the-eye. The mathematical model used forthe fit is still under development, and will not be discussed here.The importance of developing a reliable model is that it permits forany combination of sample mass (or total heat capacity) and rate

68 S. Wouters et al. / Thermochimica Acta 530 (2012) 64– 72

-10

0

10

20

30

40

50

60

70

170165160155T / °C T / °C

T / °C T / °C

T1 H

eat F

low

/ W

.g-1

T1250 K.min-1

Endoup

-10

0

10

20

30

40

50

60

70

80

185175165155

T1 H

eat F

low

/ W

.g-1

T11000 K.min-1

Endoup

-10

0

10

20

30

40

50

60

70

170165160155

T4P

Hea

t Flo

w /

W.g

-1

T4P250 K.min-1

Endoup

-10

0

10

20

30

40

50

60

70

80

185175165155

T4P

Hea

t Flo

w /

W.g

-1

T4P1000 K.min-1

Endoup

Fig. 4. Melting endotherms for indium measured at 250 K min−1 and 1000 K min−1 in T1 and T4P-mode. Arrows indicate increasing sample mass: 0.049, 0.135, 0.246, 0.447,0.653, and 1.005 mg.

0.049 mg

155

160

165

170

175

150010005000

T eo

and

Tp /

°C

1.005 mg

1500100050025

26

27

28

29

h / J

.g-1

0.246 mg

1500010005000Heating rate / K.min-1

Fig. 5. Extrapolated onset temperature Teo (�), peak temperature Tp (©) and melting enthalpy �h (+) for indium samples measured in T1-mode plotted against heating rate.For �h, the average and standard deviation are indicated by solid and dashed horizontal lines, respectively. From left to right the sample mass increases: 0.049, 0.246, and1.005 mg.

S. Wouters et al. / Thermochimica Acta 530 (2012) 64– 72 69

153

156

159

162

165

150010005000

Heating rate / K.min-1

T eo /

°C -

T1 m

ode

156

159

162

165

168

T eo /

°C -

T4P

mod

e

Fam

td

mmadTi

3

ccaioshlrocwc

bGtTiefsipsocscz

ta

-60

-40

-20

0

20

40

60

170160150140130Temperature / °C

T4P

Hea

t flo

w ra

te /

W.g

-1

SmB N

SmB N

N I

N IEndo

up

-60

-40

-20

0

20

40

60

180120600-60Temperature / °C

T4P

Hea

t flo

w ra

te /

W.g

-1 C SmB

C SmB

SmB N

SmB N

N I

N I

Endoup

Fig. 7. Thermal transitions in BCH52 as measured by RHC in heating and cooling:T4P heat flow rate versus temperature. (Top) full temperature range showing the

ig. 6. Influence of heating rate and sample mass on the extrapolated onset temper-ture of the melting of indium samples in T1 and T4P mode. In both series sampleass decreases from top to bottom: 1.005, 0.653, 0.447, 0.246, 0.135, 0.049 mg.

o apply the temperature corrections needed to the experimentalata.

For the T1 mode measurements shown in Fig. 5, the observedelting enthalpies �h stay within ±1 J g−1 for all heating rates andasses, with standard errors of about 0.5 J g−1 or less than 2% for

mass of 0.25 mg. This indicates the heat flow rate measurementoes not significantly depend on heating rate and sample mass. In4P mode, a slightly decreasing trend for the melting enthalpy withncreasing heating rate was observed above 750 K min−1.

.3. Symmetry check

As the RHC is intended to use at both high heating and highooling rates, the symmetry of the thermal lag in heating andooling was investigated. In case the RHC behaves symmetrical,

thermal lag correction model constructed using data from heat-ng experiments could also be applied to cooling experiments. Inrder to verify this, a symmetrical transition, one displaying nouperheating and supercooling effects upon heating and cooling,as to be used. For first order transitions, such as the crystal-

ization of indium, this is not the case, as supercooling effectsesulting from crystal nucleation can be expected. As for secondrder transitions, supercooling effects are absent, the use of liquidrystal secondary calibration standards showing no supercoolingas recommended by Höhne et al. [14–17] for FSC symmetry

hecks.In this paper, three substances displaying liquid crystalline

ehaviour were used, M24, BCH52 and HP53, as proposed byEFTA [18,20]. Samples of approximately 0.25 mg were subjected

o six heat–cool cycles with scan rates of 50–750 K min−1, in both1 and T4P mode. Using scan rates above 750 K min−1 leads toncompletely resolved transitions, which negatively influences thestimation of the extrapolated onset temperature. Only resultsor BCH52 or 4 4′-ethyl-4-(4-propyl-cyclohexyl)-biphenyl will behown in this paper. The recommended transition of BCH52 for cal-bration on cooling of conventional DSCs is the nematic to isotropichase transition at 164.8 ± 0.4 ◦C. As the RHC is working at highercan rates than previously used for this material, the suitability ofther transitions in BCH52 was also checked. The heat flow rateurves for the heating and cooling of BCH52 at different rates arehown in Fig. 7. For the heating traces, the sample was alwaysooled at 250 K min−1. The temperature was calibrated using the

ero heating rate calibration discussed above.

The first order solid (crystalline) to smectic B (C–SmB) transi-ion observed around 0 ◦C cannot be used due to a supercooling ofpproximately 40 K for the SmB–C transition on cooling. Apart from

crystalline (C) to Smectic B (SmB), SmB to Nematic (N), and N to isotropic (I) tran-sitions. (Bottom) enlargement of the SmB–N and N–I transitions. Arrows indicateincreasing rates in heating (full) and cooling (dashed): 50, 150, 250, 350, 500, and750 K min−1.

the recommended (though small) nematic to isotropic (N–I) transi-tion around 164 ◦C; the SmB–N transition near 146 ◦C also seems fitfor the analysis of the thermal lag symmetry.

For the evaluation of the thermal lag symmetry using the SmB–Ntransition, the extrapolated onset temperature Teo and peak tem-perature Tp are plotted versus the scan rate in Fig. 8 for both T1and T4P-mode, along with linear regressions made separately onthe data for heating and cooling. In case of perfect symmetry, bothtrend lines should fall on the same line [18]. This is a more strictevaluation as compared to evaluating the deviation of the data froma single linear regression encompassing both heating and cooling(as e.g. in [20]). For Teo, the maximum absolute deviation betweenthe regressions for cooling and heating remains within 0.3 K and1.0 K for T1 and T4P, respectively. For Tp larger deviations are noted,highlighting that Tp should not be used for thermal lag symmetryanalysis. Indeed, for Tp the additional lag with respect to the onsetof the transition depends on the magnitude of the heat flow rateat the peak. For Teo in T1-mode, the zero heating rate extrapola-tion (the intersection of the regressions with the ordinate) is nearlyidentical for heating (145.87 ◦C) and cooling (145.74 ◦C). Besides,also the slopes are very similar: 0.0054 min (0.32 s) for heating and0.0056 min (0.34 s) for cooling. These slopes correspond to the timeconstant in the presence of the 0.25 mg sample. T4P results of Teo

also display a symmetrical behaviour: the differences in temper-ature intersection and slope are slightly bigger than in T1 mode:0.16 K (within the 0.2 K experimental error on the temperaturecalibration), and 0.0011 min (or 0.06 s), respectively.

70 S. Wouters et al. / Thermochimica Acta 530 (2012) 64– 72

y = 0.014x + 145.16

y = 0.0054x + 145.87y = 0.0056x + 145.74

y = 0.011x + 146.98

135

140

145

150

155

10005000-500-1000

Scan rate / K.min-1

T eo,

T p /

°C -

T1 m

ode

SmB NT1

y = 0.0098x + 145.9

y = 0.0034x + 146.73

y = 0.0045x + 146.56

y = 0.0082x + 147.68

135

140

145

150

155

10005000-500-1000

Scan rate / K.min-1

T eo,

T p /

°C -

T4P

mod

e

SmB NT4P

Fmr

e2utsfn

dnalTtThip

tm±i

y = 0.012x + 163.67

y = 0.0058x + 163.56y = 0.0050x + 163.99

y = 0.010x + 164.29

154

158

162

166

170

10005000-500-1000

Scan rate / K.min-1

T eo,

T p /

°C -

T1 m

ode

N IT1

y = 0.0081x + 164.38

y = 0.0038x + 164.34

y = 0.0038x + 164.67

y = 0.0069x + 164.93

154

158

162

166

170

10005000-500-1000

Scan rate / K.min-1

T eo,

T p /

°C -

T4P

mod

e

N IT4P

ig. 8. Thermal lag symmetry evaluation using BCH52 for T1 (top) and T4P (bottom)ode showing Teo (�) and Tp (©) for the smectic B to nematic transition versus scan

ate.

In both T1 and T4P modes, the difference in the zero heating ratextrapolation for heating and cooling is much larger for Tp (nearly

K for the intercept and 0.16 s for the slope). Again, the evaluationnderlines the peak temperatures should not be used to evaluatehe symmetry of the system [20]. Also for other recommended tran-itions, such as the nematic to isotropic transition in M24, resultsor Teo show a clear symmetry of the transition, while Tp results doot show the same extent of symmetry.

For the nematic to isotropic transition in BCH52, the absoluteeviations from the linear regression are slightly bigger (Fig. 9), butevertheless the behaviour of this transition is symmetrical within

1 K deviation. The differences in the zero heating rate extrapo-ation for heating and cooling are −0.43 K and −0.34 K for T1 and4P, respectively, the intercept being higher for cooling. For Teo,he difference in slopes is small for T1, and even smaller for T4P.he higher zero heating rate extrapolation for cooling compared toeating and the curvature observed for Teo in the T1 data in heat-

ng probably find their origin in the imperfect resolution of the N–Ieak.

Based on the results for all three examined liquid crystals, three

ransitions can be recommended for usage in the thermal lag sym-

etry evaluation in the range of heating rates studied here (up to750 K min−1). For symmetry checks of the RHC, the nematic to

sotropic transition for BCH52; the smectic B to nematic transition,

Fig. 9. Thermal lag symmetry evaluation using BCH52 for T1 (top) and T4P (bottom)mode showing Teo (�) and Tp (©) for the nematic to isotropic transition versus scanrate.

and the smectic B to smectic A transition in HP53 (data not shown)prove to display sufficient symmetry, thus also proving the ther-mal lag symmetry of the RHC. The non-linear behaviour observedat lower rates for the smectic B to smectic A transition in HP53 [20]was not observed at the higher rates used for RHC. The smecticA to nematic transition in HP53 shows a non-linear behaviour forthe rates used in RHC and should not be used in these conditions,though it may be suitable for the evaluation of conventional DSCsat slower rates. The smectic A to nematic phase transition for M24,recommended in literature, cannot be used for RHC as it proved tobe too small for a reliable detection (data not shown). The nematicto isotropic transition in M24 also shows signs of asymmetry.

As symmetrical behaviour for certain transitions was noted, itcan be said that the RHC behaves in a symmetrical manner in heat-ing and cooling, the deviations from symmetry being small enoughto be within the experimental error on the temperature calibration.Therefore, the corrections for the thermal lag estimated from heat-ing experiments are also fit for correcting data obtained in cooling.

3.4. Verification of the heat flow rate calibration

The results of the verification of the heat calibration using theenthalpy of melting of indium samples, shown in Fig. 5 and already

S. Wouters et al. / Thermochimic

0.5

1.0

1.5

2.0

200150100500-50-100-150Temperature / °C

Spec

ific

heat

cap

acity

/ J.

g-1.K

-1

250 K/min

100 K/min

500 K/min

Literature

T4P0.742 mg

0.5

1.0

1.5

2.0

200150100500-50-100-150Temperature / °C

Spec

ific

heat

cap

acity

/ J.

g-1.K

-1

RHC

Literature

T4P

Fig. 10. Specific heat capacity as a function of temperature for polystyrene sam-ples in T4P mode, empty-crucible lines were subtracted. (Top) 0.742 mg PS sample,measured at heating rates of 100, 250, and 500 K min−1 and literature data [21]. (Bot-tom) Average and standard deviation calculated using PS samples of 0.254, 0.266,a[

bdsflrapab2Ssrcctosopshmi

ture range (−125 ◦C to 200 ◦C). Results for T1 and T4P calibration

nd 0.742 mg measured at 100, 250, and 500 K min−1, compared with literature data21].

riefly discussed in Section 3.2, indicate that the heat calibrationoes not show a significant mass or rate dependence over the rangetudied. In this section, the temperature dependence of the heatow rate calibration is further verified using a polystyrene standardeference material for heat capacity calibration from NIST. Resultsre given for two polystyrene samples of 0.25 mg and one sam-le of 0.75 mg, all submitted to a temperature program suitable for

heat capacity determination: equilibration at −150 ◦C followedy an isothermal segment, subsequent heating to 250 ◦C (at 100,50, or 500 K min−1), followed again by an isothermal segment.uch an iso-scan-iso program is typical for heat capacity mea-urements, though usually it is made over a limited temperatureange to reduce errors resulting from changes in the baseline. Theorresponding empty-crucible lines were recorded with two emptyrucibles using identical temperature programs. After subtractinghe empty-crucible lines, zeroing the heat flow rate at the endf the isothermal sections, and dividing by the heating rate, thepecific heat capacities calculated from the RHC experiments arebtained. As shown in Fig. 10, these specific heat capacities com-are very favourably with literature values [21]. The first graphhows the heat capacity data obtained for a 0.742 mg PS sample at

eating rates of 100, 250, and 500 K min−1. For the three measure-ents, the correspondence with literature values is very good. It

s worth stressing that the heat flow rate was calibrated using the

a Acta 530 (2012) 64– 72 71

melting enthalpy of indium only, the heat capacity measurementwas not separately calibrated using sapphire. For each individ-ual measurement, the average absolute deviation on the specificheat capacity in the temperature range from −50 ◦C to 200 ◦C isless than 4% (somewhat larger below −50 ◦C). Very little effect ofheating rate or sample mass is noted when comparing the differ-ent sample masses, heating rates, and heating cycles (not shown).The lower graph in Fig. 10 shows compares the literature value tothe average values and their standard deviation as calculated fromthree samples (masses: 0.254 mg, 0.266 mg, 0.742 mg) measured atthree heating rates (100 K min−1, 250 K min−1, and 500 K min−1).At all temperatures checked from −125 ◦C to 200 ◦C, the exper-imental value corresponds with the literature value within theexperimental error. The average absolute deviation over the fulltemperature range between the experimental specific heat capac-ity and the literature values is 2%. Heat capacity measurementson polystyrene in T1 mode give similar results and conclusionsregarding heat flow rate calibration. These results indicate that theheat flow rate calibration made with indium at its melting pointcan be used safely over the full temperature range. Besides, theresults show that the RHC can be used to measure heat capacitieswithin an error comparable with conventional DSCs at rates thatare about ten times faster and using sample masses about ten timessmaller.

4. Conclusions

The more comprehensive calibration procedure for FSCs,applied here to the RHC, is needed when performing experimentsin conditions where thermal lag becomes important, as it allowsfor a more quantitative interpretation when using different samplemasses and scan rates. Though the zero heating rate temperaturecalibration using three or four melting point standards involvesa considerable number of experiments, the high scan rates makesure it involves a limited amount of instrument time. Do note thatcare must be taken when using the polymorphic benzophenoneand the volatile adamantane. The use of more stable sub-stances, with transitions in the same temperature range, would bepreferable.

Performing indium melting measurements over a broad rangeof heating rates using a range of sample masses permits the eval-uation of the rate and mass dependence of the thermal lag, andenables one to apply post-measurement temperature correctionsto experimental data recorded with the zero heating rate calibra-tion. If a model can be designed that fits the extrapolated onsettemperature data as a function of sample mass (or heat capacity)and heating rate, this would permit estimations of the thermal lagexpected in other experimental conditions. The thermal lag reflectssymmetrical behaviour upon cooling, as verified using the higherorder phase transitions occurring in liquid crystals, so the samemodel could be used for thermal lag corrections in both heatingand cooling modes.

The evaluation of the RHC proved positive: the instrument canbe operated at high heating rates and low sample masses whilemaintaining the performance expected from a conventional DSC.The thermal lag is limited to 6 K at 1500 K min−1 for a 0.25 mgsample, and is symmetrical upon heating and cooling. The heatflow rate measured is accurate, precise, and is (largely) indepen-dent of heating rate, as indicated by a series of measurements ofthe melting enthalpy of indium and as supported by heat capacitymeasurements for polystyrene samples over a large tempera-

modes are similar, however, the T4P mode is preferred as theeffects of thermal lag are smaller and a higher resolution can beattained.

7 chimic

A

Raa“

R

[

[

[

[

[

[

[

[

[

[[

2 S. Wouters et al. / Thermo

cknowledgements

This study was supported by a grant from the Fund for Scientificesearch-Flanders, Belgium (F.W.O. Vlaanderen). TA Instruments iscknowledged for making a Rapid Heating Cooling DSC prototypevailable at the Vrije Universiteit Brussel (VUB) for beta-testing inProject RHC”.

eferences

[1] M.F.J. Pijpers, V.B.F. Mathot, B. Goderis, R.L. Scherrenberg, E.W. van der Vegte,High-speed calorimetry for the study of the kinetics of (De)vitrification, crys-tallization, and melting of macromolecules, Macromolecules 35 (9) (2002)3601–3613.

[2] V.B.F. Mathot, G. Vanden Poel, T.F.J. Pijpers, Improving and Speeding up theCharacterization of Substances, Materials, and Product: Benefits and Potentialsof High-Speed DSC, Am. Lab. 38 (2006) 21–25.

[3] G. Vanden Poel, V.B.F. Mathot, High-speed/high performance differential scan-ning calorimetry (HPer DSC): Temperature calibration in the heating andcooling mode and minimization of thermal lag, Thermochim. Acta 446 (2006)41–54.

[4] S.A. Adamovsky, A.A. Minakov, C. Schick, Scanning microcalorimetry at highcooling rate, Thermochim. Acta 403 (1) (2003) 55–63.

[5] A.A. Minakov, D.A. Mordvintsev, C. Schick, Melting and reorganization ofpoly(ethylene terephthalate) on fast heating (1000 K/s), Polymer 45 (2004)3755–3763.

[6] A.A. Minakov, A. Wurm, C. Schick, Superheating in linear polymers studied byultrafast nanocalorimetry, Eur. Phys. J. E 23 (2007) 43–53.

[7] E. Zhuravlev, C. Schick, Fast scanning power compensated differential scanningnano-calorimeter: 1. The device, Thermochim. Acta 505 (1–2) (2010) 1–13.

[8] E. Zhuravlev, C. Schick, Fast scanning power compensated differential scan-ning nano-calorimeter: 2. Heat capacity analysis, Thermochim. Acta 505 (1–2)(2010) 14–21.

[

a Acta 530 (2012) 64– 72

[9] R.L. Danley, P.A. Caulfield, S.R. Aubuchon, A rapid-scanning differential scan-ning calorimeter, Am. Lab. 40 (2008) 9–11.

10] V.B.F. Mathot, M. Pyda, T. Pijpers, G. Vanden Poel, E. van de Kerkhoff, S.van Herwaarden, F. van Herwaarden, A. Leenaers, The Flash DSC1, a powercompensation twin-type, chip-based fast scanning calorimeter (FSC): first find-ings on polymers, Thermochim. Acta (2011), doi:10.1016/j.tca.2011.02.031.

11] S. Gaisford, A.B.M. Buanz, N. Jethwa, Characterisation of paracetamol form IIIwith rapid-heating DSC, J. Pharm. Biomed. Anal. 53 (3) (2010) 366–370.

12] DIN Specification 91127, June 2011, Recommendation for temperature cali-bration of fast scanning calorimeters (FSCs) for sample mass and scan rate,by the Research and Standardization Group of NaPolyNet, an FP7 program,www.napolynet.eu (2011).

13] R.L. Danley, New heat flux DSC measurement technique, Thermochim. Acta 395(1–2) (2003) 201–208.

14] G.W.H. Höhne, H.K. Cammenga, W. Eysel, E. Gmelin, W. Hemminger, The tem-perature calibration of scanning calorimeters, Thermochim. Acta 160 (1990)1–12.

15] H.K. Cammenga, W. Eysel, E. Gmelin, W. Hemminger, G.W.H. Höhne, S.M. Sarge,The temperature calibration of scanning calorimeters. Part 2. Calibration sub-stances, Thermochim. Acta 219 (1993) 333–342.

16] S.M. Sarge, E. Gmelin, G.W.H. Höhne, H.K. Cammenga, W. Hemminger, W. Eysel,The caloric calibration of scanning calorimeters, Thermochim. Acta 247 (1994)129–168.

17] S.M. Sarge, W. Hemminger, E. Gmelin, G.W.H. Höhne, H.K. Cammenga, W. Eysel,Metrologically based procedures for the temperature, heat and heat flow ratecalibration of DSC, J. Therm. Anal. 49 (1997) 1125–1134.

18] S.M. Sarge, G.W.H. Höhne, H.K. Cammenga, W. Eysel, E. Gmelin, Temperature,heat and heat flow rate calibration of scanning calorimeters in the coolingmode, Thermochim. Acta 361 (2000) 1–20.

19] LGC, Indium – Reference material LGC2601, Certificate of measurement (2009).20] S. Neuenfeld, C. Schick, Verifying the symmetry of differential scanning

calorimeters concerning heating and cooling using liquid crystal secondarytemperature standards, Thermochim. Acta 446 (2006) 55–65.

21] M. Pyda, The Advanced Thermal Analysis System (ATHAS), The Departmentof Chemistry at University of Technology in Rzeszów (Albigowa), Poland,http://athas.prz.rzeszow.pl/.