Tobacco Documents Online

l" gl "a Z ii .1 ad 5 1" Z o w- < ..J o lO o < b.- O UJ Z Z t~ ua I," ta I- o n- rt -2" < Z z L.) 15 4 97 Dear Tilford, You sent three reports on smoke pH to Dr Btundy, see your letter of February 28. .Mthough I am not involved in Dr Blundy's area, I have had an interest in smoke pH issues over a number of years. The reports were circulated to me by a colleague. I have some comments. Please pass them on to any interested colleagues. Combining B.T. Thompson's reports of February 19 and February 26 ("Whole smoke pH of B&W brands") gives pH results for a total of 72 B&W brands. On my count, all brand average results were in the range 4.7-5.0, with: 26 brands @ 4.7 31 brands @ 4.8 11 brands @ 4.9 4 brands @ 5.0 One of the reports also records pH for exemplar Burley, Flue-cured and Oriental straight grades of 5.3, 4.9 and 4.6 respectively. This compares with mean results for 10 Burley grades, 15 Flue-cured grades and 8 Oriental grades of 7.6, 5.6 and 5.1 respectively CE.D. Massey, RD.2130, 11.11.88). The narrow range of your 72 brand results and the small difference between your straight grade results will of course be due to the choice of a whole smoke method, rather than the Cambridge filter (particulate phase) method used in RD.2130. The large amount of carbon dioxide (42mg/cig from Kentucky 1R4F) in the vapour/gas phase tends to dominate that phase's pH. A saturated solution of carbon dioxide has a pH of about 3.7. When vapour and particulate phases are combined, as in the whole smoke pH test, the effect is to reduce pH levels and contract differences between samples, relative to particulate phase methods. Lower "tar" cigarettes will of course have less carbon dioxide in their gas phase, but particulate to vapour phase balance will not change enough to give signifiant differences in whole smoke pH between "tar" levels. Although a Cambridge filter can have alkalinity problems, this does not affect smoke particulate pH results at the relatively high smoke yields obtained from the single grade cigarettes of RD.2130. (The JD. Backhurst reference in one of your reports notes this alkalinity issue but states that exposure to air removes it,) We now use electrostatic precipitation of particulate phase rather than Cambridge pads. In BTC, we have usually used particulate phase pH results for two main reasons. The first is that there is very little discrimination between widely different blends when whole smoke is used, as you have shown. Hence whole smoke measurements are of little practical use. The second is that the concept of"pH" as applied to smoke is a difficult one because the term refers to a property of a solution, which smoke is not. Thus any smoke pH result must be qualified by a definition of how the smoke was brought into solution. The resulting "number" is a property of smoke and called "smoke pH" for convenience 566560242

/. tl "r D ii .J i.I 5 Z .,I o tJ t) o O 0~ oA Z z )- n U o Now smoke is a two phase system, panicles and gas/vapour. Smoke is not one phase when it reaches the smoker and therefore we have tended to view "whole smoke pH" as a step too far in what is already a distortion of the concept ofpH. Gas/vapour contributes little to sensory properties, as judged by smoking a cigarette through a Cambridge filter. Most of the positive tastelflavour properties come from particulates and therefore we have focused on differences in particulate pH when exploring sensory properties and theories. Looking at a range of single grade results from RD.2130, combined with blend data on the same grades in RD.2108 (WDE. Irwin, 22.2.88), 91'/, of the variation in condensate pH could be explained by linear regression on blend nicotine and blend total nitrogen (Kjeldahl). Condensate pH can be expected to increase with the concentration of nicotine in smoke, given its chemical properties, see a~so P.C. Bevan, P.35, 8.3.94. Condensate Kjeldahl nitrogen (N-H bonds) can also be expected to be basic and to increase smoke pH. Condensate concentrations of nicotine and Kjeldahl total nitrogen are closely related to tobacco values, as will be seen by plotting results in RD.2130 and RD.2108. Of course, other factors, such as different acid levels, may be found to change pH when blend nicotine and total nitrogen are held constant, but they are not "needed" to explain the pH range of the wide range of tobaccos in the above reports. A further caveat is that relationships of the type described above should not be expected from a set of samples which do not have such a wide range of chemical properties, say a set of brands rather than straight grades. In this respect, root technology will increase total nitrogen, but compared to typical USB blend levels of around 3% total nitrogen (as measured in BTC, see Report P.8 l, 12.3.96 for example), root's contribution is small and little effect on pH might be expected. Unfortunately, I have not been able to find examples with a proper non root control to test this comment, though examples must exist. However, Marlboro, Camel, Lucky Strike and Richland from up to 5 markets were tested for condensate pH (electrostatic) and other properties in RD.2205 (W,D.E. Irwin, 26. l 1.91). The tables in that report have an error. The brand listed as "Lucky Strike" from USA should read "Richland". Other Lucky Strikes were indeed Lucky Strikes. I quote my comment on page 29: ...a "There is no evidence that condensate pH results are higher for those brands which .< • ~ appear to have added DAP. Perhaps surprisingly, condensate pH is fairly strongly z correlated with condensate nicotine concentration (NTR). This positive correlation depends to some extent on the relatively high condensate pH and NTR results for the 77". 7_ Brazilian brands." O [..) A point related to this topic is that smoke ammonia is likely to be formed from many blend substances that register Kjeldahl nitrogen., not just from blend ammonia. This seems to be supported by the two figures on the last page. The symbols mean. 555 and 555L D and DL L and LL M and ML C and CL State Express 555 and 555 Lights U-K export Dunhill and Dunhill Lights UK Lucky Strike and L Strike Lights US Export Marlboro and Marlboro Lights US Camel and Camel Lights [ JS 566560243

z,I I: 3 J l,I 5 I: Z o w o 1I o O ul Z Z U o Ilg These results have not been reported yet, they refer to early '96 samples. The first figure, smoke ammonia concentration v. blend total nitrogen is much more convincing than the second, v. blend ammonia. Only the "L" point in the first figure spoils the pattern. I know this is only a correlation, but it is more than reasonable to expect a causal basis - pyrolysing compounds with NH groups is Likely to include ammonia as a prime product, though I cannot quote a reference. I do recall that protein, whose nitrogen can be half of the total nitrogen, is a substantial ammonia producer based on work here where protein was reduced with enzymes. In summary, there may be a belief in some quarters that root materials contribute the major percentage of smoke ammonia. I think that is not correct. For these same results I calculate that for Marlboro, Camel and Lucky Strike, smoke ammonias are 0.17, 0.18 and 0.15% respectively of total blend nitrogen smoked to the butt mark. For Marlboro Lights, Camel Lights and Lucky Strike Lights the respective values are 0.14, 0.13 and 0.14%. The two Camels had blend phosphate levels compatible with no added phosphate. I conclude that the efficiency of ammonia formation from the total nitrogen of a blend that probably has no root materials is about the same as from a comparable blend/design with root. If your colleagues are interested in exploring any of the relationships in RD.2108 and RD.2130 it might help ifI give a guide to the tables of results. RD.2130, page 6, Table 2, gives condensate pH results in thetast column. The previous column gives total nitrogen concentration in the PMWNF. Nicofine/PMWNF is a further two columns to the left. RD.2108, page 17, Table 3, gives tobacco total nitrogen and nicotine values. Best regards, Derek Irwin CONFIDENTIAL 566560244

o = :£ 3 iJ o J &l 3 Z E ,.) o L) (J o t,., o (/1 Z ID t~ ua U t- 13 ISmoke ammonia per unit NFDPM v. Blend total nitrogen I eML 3 +I LL Z 25-- 2 E 1.51- 1 2 eC • CL eM •LL eL • 555L I ~ t,1 I - 2.2 2.4 2.6 2.8 Blend total nitrogen (%) ISmoke ammonia per unit NFDPM v. Blend~ 3 2.5 O rl 2 E 1.5 /+ •ML •C •CL eM • LL eL e ~.D • 555 • 555L 0.1 ! O.2 I~ ammonia (%) t 0.3 3.2 0.4 CONFIDENTIAL 566560245